© COPYRIGHT NOTICE All Rights Reserved


(Last updated: 04-Apr-2018 20:28 )

Tektronix 576 Curve Tracer System

 

 

CONTENTS

Part 1) Why using a curve tracer

Part2) Which one to take, 570 or 576?

Part 3) Repairs on my 576

Part 4) Calibration of the Tektronix 576

Part 5) Build a tube interface for the Tektronix 576

Part 6) Loose items, unsorted

Part 7) Testing of two lead devices with the Tektronix 576

Part 8) Testing transistors with the Tektronix 576

Part 9) Testing tubes with the Tektronix 576

 

 

Part 1) Why using a curve tracer

M HAMEG 203-4 oscilloscope has a simple, but very good working curve tracer build in. I found interest ins looking at curves of passive components. It is like these things we think we know it already, but do we really? So I opened my old parts boxes, and tested used coils, old transistors, lamps and diodes, just anything. And the things I have learned sort of pop up again. So I got the taste of this, and I wanted to have a curve tracer which can do more.


Basic curve tracer of Hameg HM203-4. It has no settings, just this one it always takes. This is a 10V Zener diode.

In case you are interested, I can recommend a Hameg Scope for this with a nice little component tester build in. Several Hameg types have that. Just look for this button below the picture tube where it says COMPONENT TESTER. It works really accurate, and you need to know nothing about it's operation. Just connect a diode under test with two banana plugs between ground and where it says "Component Tester". So the red and the black banana plugs you see in the picture. Then press the button and there you are: A two quadrant diode curve. (So forward and reverse in one picture). If you consider a scope for Audio, consider the 203-4. It is a very basic, but usefully model, and the component tester inside comes for free. After playing with this option, I became interested in it, and I thought I wanted to have the Tektronix 570 curve tracer, the first and last one by Tektronix, made for tubes. Still I decided for the Tektronix 576 transistor curve tracer. In the next article, I will explain why. Also show the repair details if the one I bought, and end with some testing on tubes and parts which I have here.

 


The same 10V diode on my calibrated TEK 576

Also I wanted to study a just for my interest, what information can be extracted from components of any kind. Like see the effect of the heater voltage with tubes, and compare two curves charts of a double triode instantaneously. Now of course this can be done with digital curve tracers as well, but that is only half what I intended. A digital curve tracer is very boring to use, because you have to wait all for the curves to be finished point by point.

Another advantage of the analog curve tracers is the fact they work analog. So there is the thickness and luminescence of the trace, showing you the speed of the movement. Particularly when you have components that go into negative resistance (usually parts that ignite, or "fire"), a digital curve tracer can not handle this, because these work by the fixed principle that one X-value, must give one Y-value. This normally so indeed, but just not always, and if that situation occurs, they fail. For a digital curve tracer to display multiple Y-Values at one X-Value, they must have fully analog hardware, and then simply scan X and Y values all of the time. The low cost solution is, take the (X) voltage, and have an A/D converter measure the (Y) current. Fine, but then you can measure only one Y-value, and you can understand, as in the left picture that simplified approach will fail. The Sofia, Amplitrex, u-trace, e-trace, etc, they all work that way.

The best digital curve tracer I have, is the Sofia by Audiomatica Italy. The hardware is software protected against overload, and it switches off reliable with tubes. It should be able to test Field Effect Transistors as well, but I am not going to try it. Resistance of FETs is much lower, and I don't know if that was anticipated in the design.

The AT1000 curve tracer has very robust hardware, more forgiving with mistakes, but the software is has many shortcomings, as it can not work lower than 75 Volts. Resolution below 1mA is in steps of 0,1mA. It is too slow, and curves have no smoothening function. So curves sometimes look like made with a Commodore 64.

There are now some low cost digital tube tracers around, but these are limited in speed, voltage and current. The next lower level would be oscilloscope based testers. Even Tektronix had a module for that for the 7-series scopes, but that is a low power device, hardly suited for tubes.


Part 2) Which one to take, 570 and 576?

There are a few others, but 570 and 576 are the top models.

These differ at following points:

570
576
This junk looks like found in the Titanic, showing how users outbid each other for any price. Sold for 1000 Euro on Ebay Italy, in Jan 2018. It is full of spider droppings, which give those typical white-brown rust pips. Prices up to 4000€ for a good one.
200...400€ for incomplete, 2000...4000€ for 100% tested. A working one "as is" for 500..1000 Euro.
All tube technology
All semiconductor technology
Anode max 500V.
Anode max 1500V.
Can tests small parts, with low voltage, and sensitive current limiting, 1nA per division. Can tests smallest semiconductors. Or test huge power devices, up to 20A, 1500 Volts, 220 Watts.
No such option.
Readout card if the option is included. (not always)
No such option.
Modular construction. So you can replace, or repair the module
Can test tubes, and some FETs. Step function only in Volts.
Can test tubes only with self made adapter and external heater voltage supply. Step function in Volts or mA. Can test with positive or negative voltage.
No safety precautions
Has hazard lamp and safety switch.
No such option.
Can be used with some different interface modules. Such as the standard module, a high power module, or a card interface module.
Quality is good, but not as good as 576.
Reputation for it's superb quality.
No such option.
Pulsed option possible, so you over heat nothing.

Loud fan, which slows down over time until it gets too slow, and internal parts get too hot. The 570 can keep your coffee pot warm.

No fan. Machine gets hardly warm.
Standard round oscilloscope tube, with external graticule.
Square tube, with internal graticule, parallax free. Graticule has an additional cross hair in bottom left corner, where many curve charts begin. This tube is obviously dedicated to curve tracers.
No gold plating.
Some models have almost entirely gold plated components, and PCB's.
Parts are just normal radio parts, and just normal tubes. This is a great advantage for restoring. When the 570 is older, total restoring may be needed.
Some parts are just normal semiconductors, and generic part numbers are not secret. Other parts are often special, but almost anything can be found still, with patience, or for a high price. The technical level of some parts is very high, like laser calibrated resistors, but generally these are problem free.
Reasonable pdf manual available
First class pdf manual available
570 is older.
576 is younger.
Adapters often incomplete
Adapters often all missing, and these cost a lot.
Switch operated. Deck switches are low quality. Any issue can mean trouble. In fact quality is so poor, the Chinese once build a higher quality clone, called GT-2.
Many deck switches are only to operate relays. The relays do the switching. So any switch problems may be solved just by exchange the relay, which have gold plated socket pins.
35kg weight is definitely totally crazy. Postage in a carton box with styropor chips is a joke. Can only be shipped on a palette.

Initially I was interested in a Tektronix 570. However this page is not going to be a tribute to this model. Though they are really nice and very useful, my main objection is the plate voltage can go only up to 500 Volts, and the price is crazy for what you get. When you use a tube at (say) 380 Volts, the curve chart should be up to 600Volt or more. Otherwise it becomes a bad compromise. Better is the Sofia, as it can trace up to 750V. Still a tube like 845 can not be tested under real use conditions at only 750V. You need 1200V or more.

However, objection with the 570 are many. First, the relation between what you pay and what you get is fully wrong. This is caused by people that pay 4000 Euro for them. Which they may be worth or not, decide that for yourself. In Jan 2018 I saw one go on Ebay (Auction nr 232611931723) for 1032 Euro + expensive shipment. It was said to be stored for 20 years, and that seemed like stored in a chicken shack. There was a layer of rust on the whole inside, and from underneath the scope tube socket was white oxide power coming out, and quite a lot actually. Ideally, it was put in the chicken shack in working order by mistake, and all it needs is you to find it in Ebay, and brush away the rust. Not so ideally, it was put there 20 years ago, with such bad defects, the chicken shack was the right place to put it.

You can even conveniently buy such an item, check if the scope tube is at least ok, and if the transformers are still good. If not, sell it on Ebay, and the next one will buy it for a higher price when it is cleaned nicely.

With 570, we have the same situation as with all tube testers, people think they can fix all problems with a soldering iron and a multi meter. Then after they bought it, they find out several owners before them already failed. So you're the next one in line, and good luck with it, because here comes the problem: Either you buy it from somebody who knows MORE about it than you, and he gave up on it. Or, you buy it from somebody who knows LESS about it than you, and you have to clean his mess. I see no other options.

Here is an observation I just made, looking for a Tektronix 5030 scope. I found information and pictures here. Then, I looked on the German Ebay site, and hey.... I see there EXACTLY this oscilloscope for sale, described on W140.com. Same series number! Incredible, but true! Really cheap, the auction number from Jan 2018 is: 182648785851. There are ways to look at old auctions, also with Ebay. Now, on w140.com this scope, with that SAME series number, is described here with the SAME pictures as on Ebay, by somebody who supposedly is an expert with that product. . On Ebay is written he "knows not much about the accuracy". Do you think what I think? And even so a 5030 is what nobody wants to have, but for a 570 they are all haunting.

When the tester was stored dirty, and somebody try to fix the whole thing with contact spray, you will never get happy with it any more. Do not confuse this with the later 576! Though so close related part numbers, the 576 is stuffed with the best and finest parts that were available at the time. There is gold plating simply for everything where they had the choice. Even so, most of the switches in the 576 are just, to switch RELAYS. Some of which have gold plated lead wires. That is a total different way of doing things. The many relays were for the card interface which existed, so most functions of the 576 are basically relay operated and not switch operated. Pull out the interface unit, and you will see there is a whole series of connectors behind it. These allow access to all relays, and other functions. Via those, you can for instance disable the vertical sensitivity as chosen by the knobs, and choose another sensitivity via the relays.

Though errors with tube equipment are usually not so tricky, the 570 has just very many of them. The air ventilator has the habit to slow down when it gets older and there is some risk your 570 was over heated at some point in time. So the 570 is something you should not buy unseen. Bringing it back to instrument condition ( in case you dream about that) means a full restoration. So take it completely apart, replacing every bad item you find inside, and that will be many. Just imagine what to do, after you have de soldered a selector switch, and you find it leaky between all contacts. I mean like some Mega ohms between the contacts. Well you could leave it as is, but no... this is nothing for me. Fixing only what seems broken is no option for me. I tried that out with other tube equipment, but when you repair one thing, soon after the next thing gets broken, and this just doesn't stop. Yet a full restoration of a Tektronix 570 will consume half a year of my time, not to speak of the costs, and many of the objections I wrote down here, do not exist with the 576. ( I fully restored mine in just several days work)

So I had to let go of the 570 idea, I just see too much of the usual risk, somebody sell it for reasons he knows damn well, and I don't want to be his fool. Besides the 576 costs only 1/4 of a 570, it can do 3x higher voltage and current. In fact it's quite amazing to see you can have 1uA per division vertical scale, and even zoom in on that a factor 10, with a delay function as with any oscilloscope, thus getting 100nA per division. Whereas you can also set it to 2 Ampere per division, giving 20 Ampere over the full picture tube. Oh man, the range of this curve tracer is incredible.

Generally the picture tube is a bit flickering, but that is something that belongs to the concept, and they used normal oscilloscope tubes in the 570 and 576 as well. A bit of a pity, because special tubes with very long light up time would have been better. Right now, it triggers to the mains frequency, so the horizontal (voltage) is coming from the mains directly, by using a 1600 V AC transformer, and connect that to the mains via a variac. Even so, the relatively low frequency of the mains is more or less the right one, because a higher frequency would give difficulties with the capacitance of devices. (So called looping problems). So the whole concept makes sense, just not so much the scope tube, but that is how it is.

Quality of the picture tube:

The graticule of the 576 is inside the tube, at the phosphorous layer, so it is parallax free. The tube is flat, square, large sized, and brilliantly sharp. The tube of the 570 is more old fashioned, round and not flat, and I believe the graticule is not parallax free. For the rest, I trust it is as good as any fine Tektronix tube, though myself I find a green scope tube a bit difficult to look at. One way or another, blue is more pleasant.

570 tube. Here you see a picture which is not looping the curves very nice. If it was mine. I would not rest until I have found the cause. Or perhaps never find the reason at all, meaning you have to accept this as is. It's just these kind of issues I try to avoid.

So after all those considerations I choose for the 576. At Ebay, every day you can buy a few, but these are always untested of "expected to be probably good". Many people have parts for sale. Prices of parts are high, but product offerings are many. The 576 has specific issues too, but as far as I know it is only minor things. So even though complexity of the 570 is lower, you can restore a 576 far more easier to instrument condition, because most things inside are still good. They don't make them like that any more. Printed circuit boards are glass fiber, gold plated, with thick solder islands that don't come off when you use a desolder iron. Several boards can be removed easily, because you can pull off all wires with small connectors. Exceptions are the readout card, which can be pulled out from the front after you take off the scope cover. (See next pictures below). Also some boards are underneath a layer of other boards, which is not really easy to service, but I went through that, and it was not so terrible to do.


Part 3) REPAIRS ON MY TEKTRONIX 576

Yes I know, there are people who bought an all working perfect 576 for a low price. My 576 was as expected. I bought from the University of Boston. So probably used a lot when it was new, and then less and less, until it got stored for some years. I received it with shipment damage, due to bad packaging, and they gave it to me for a low price. If somebody has two lower side panels for me... please let mw know. So the ones it stands on.

It was still "working" indeed, but the rest, the more I looked into it, the more needed to be done nevertheless. Luckily there was never worked on the inside guts, except for the readout card, which could pulled out from the front, and some unsuccessful repair attempts were done seemingly by professor Duck and his assistant Goofy. After it didn't work they just plugged it back in, and that was it. Though I am happy, they never tried to "repair" other things. So chances are very high, it was never damaged during it's professional career at the university.

Also, this gives an idea about what is the difference between a "working" one for 800$, and one which is in instrument condition.

What is very helpful is the documentation. There is a first class pdf document. I printed this double sided on 100 gram paper, and used a 26 holes binder. The result is better than a book. There is really everything described in there. All internal part numbers for the semiconductors are listed there nicely with the standard industry replacements.

REPAIRS: Two defective switches

There is the switch to find the zero of the screen. So when you press it, the beam should point to the center of the tube. I needed to wiggle it occasionally and then it worked again. Then I needed to wiggle it more often, and it became a problem. Something similar was with the "invert" switch, which is needed to display curves with negative voltage upside down, so they look normal again. Such curves for instance for NPN transistors are displayed normal, and curves for PNP transistors are displayed inverted. As these are hidden beneath a layer of other boards, and you need to take all knobs off the whole front, and remove all those connector pins from the boards, I had the idea trying to spray some C60 contact oil in the switches. You can access them a little bit by pulling of the knobs. That turned out to be a very bad mistake, and it was the last time I use that contact oil probably. Initially this solved the contact problem, but it made the cause of the problem get worse. So after 3 years, both switches gave up function. The invert switch didn't toggle any more, and the Zero switch sticked out 2mm too far and it had no function any more. So what I tried to avoid, remove all boards, had to be done now anyway, and as I will show later, the contact oil damaged one of the switches.

 


Access to the switches is too poor. Chances some piece parts fall inside and get lost it too high. I decided to take out the whole board.

 


Here you see what boards needed to be removed.

 


Here you can see the problem. A fat green stuff is all over, it sticks like glue .
Also there is an Oxide layer build by the C60 Contact oil.


This was inside. The contacts were sticky, and also there was a dark brown layer of something on them,
which was not very conductive. I think I should better not have used C60 Contact oil.

 

This is the board before repair-

What you can not see, the contact oil has also loosened the bottom of the middle switch, and that made the arm come out 2mm too far. I could clean it, the green stuff resolved in alcohol, the brown stuff would come off with fine abrasive paper. The switch bottom had no great force on it, it could be repaired as you can see on the next picture. I have been looking a little bit for such a replacement board, but it was kind of rare, or too expensive. However the switches as such can be found still from other Tektronix oscilloscope boards, and perhaps from the manufacturer too, but I don't know who that is. They look kind of standard switches to me. Note, to have more room for the repair, those white cubes are relays and you can pull then out of the board just like that.

 

 


This is the board in repaired condition

 

Look at the spray container, the C60 contact oil has created a fat, sticky brown layer with rust. Similar
to the green stuff inside the switches.


No comment :)

 

REPAIRS: Defective lamps in the readout card.

The 576 has a read-out card which is made by lighting into fiver optics with a series of light bulbs. This is one of the best features I have ever seen, and it shows what the Tektronix designers wanted to have, but digital electronics was just not ready for it yet. Though developments were very close actually, and right after these readout cards, first ICs came on the market, so called Character Generators, which could display texts on an oscilloscope tube. However this light bulb solution is amazingly well done, and the card itself is very compact also. Also the cards shows things as they really are. So when you magnify with some factor, the card corrects it's reading. Or the step function (3rd number) can be in mA or in Volts, depending if you test a transistor or a FET. This is done all correct by some arrays to diodes which in the end operate the driver IC's for the light bulbs. You really can't see this is an analog + light bulb solution.

Nice, big indicators, that change their readings with relay clicks, and normal white light. They have build this from fiber optic, and normal light bulbs. There is not one light bulb per dot, but whole groups of fibers are lead to one lamp. That reduces the number of bulbs needed, but there are still some 50 lamps used. So to change from the letter "V" to "A", there are thee light bulbs needed. One will light up all the dots (with separate fiber optic cables) that belong to both letters. The second one lights up the complimentary dots to make a "V", the third one lights up the complimentary dots to make it an "A". The driver IC has a diode matrix inside, and the input is just c "1", or logic"0" and "V" or "A" will burn. All of this becomes very handy when the 576 is programmed EXTERNALLY. Many are not aware of this option! So when you remove the hand-operated unit from the bottom right, you can replace it by a card unit. The card unit will automatically do the complete settings, and you will see that on the dot matrix display. With some technical skills, I think it is not even difficult to operate a TEK576 from a computer that way. All you need is the programmable module, it had all connectors.

The bottom unit is for the steps. For transistors you set to uA base current, and it gives the Beta. For FETS and Triodes, the steps are in Volt, and you get the Gm. Note, a FET and a Triode are essentially the the same for a curve tracer. So even on the very old 570 that was made before FETs existed, you can test FETs.


Readout Logic unit with one cover removed, to replace a defective lamp
The card does all the decoding to the dot matrix. So the inputs are just for: 1, 2.5, 5 etc. When "5" is selected, only the dots to build a digit 5 are back lighted by fibre optics, and the correct lamps are selected by the IC's. Have you ever seen gold plated 16 pin ICs?



Readout unit, lamps with Fiber Optics unit. (click image to enlarge)
There was a lamp defective in this unit. Curiously it pulled 50% more current, but no light. After testing, it appeared to have lost vacuum. So the heater wire gets too cold from the air, doesn't glow any more, and current is too high. I was able to find some original replacement lamps. (ask if you need one).


Somebody replaced a light bulb, but he was not very good at soldering. (This work was done by university of Houston) .
Not pictured here, but this soldering I repaired too, though it was electrically good. The broken lamp was probably suspected here, but it was one of the adjacent lamps that was broken. This must have confused them, and they didn't solder on the broken lamp any more. Well admitted, this was a difficult defect, and the unit works confusing indeed. Here is a typical case of somebody not getting any further, and put it back together for the next person to try. Well that was me.


Look at this hardware quality, all the gold. It is probably very thick too.
With older equipment, when soldering begins to detoriate in general, lots of trouble will come your way. Not so with this beautiful soldering quality from Tektronix. This is the lamps driver IC, when you pull it out, you can apply 5V to the outputs, and see what lamp is broken. So in this picture between the brown wire where it says "AP" and the IC outputs 9...15 you can apply +5V and that will light up one of the lamps. Some lamps give funny patterns, and only the right combination of lamps builds the symbol. So the IC's have some encoding logic inside, and the lamp drivers. So we have an all hardware dot matrix unit here, made with light bulbs. May sound primitive, but on 1968 there was no other option. And really with the white light and brilliant sharp symbols, it is very well done and very useful. I think this was a little bit an art project of the designers also. The readout card was expensive and optional.
I guess it's still expensive today on Ebay.


REPAIRS: Example of a cracked knob. That us because the inside metal part is pressed into there with great force. No good idea for plastics. This seems to happed often, so old knobs are expensive on Ebay. I repaired this knob with JB Weld, which has the same gray color. You can hardly see the repair. So no need to pay 25$ on Ebay.


Sunburn is on many plastic parts, but it came off just by using it shortly. So that proves it was not used for a decade, and was on the daylight. Well, better like that, as in the basement for 30 years. You can see the many scratches on the CRT cover too, but these are actually plastic plates that you can remove conveniently and I machine polished the scratches away.


Some versions have no read-out unit. The read out unit is so wonderful, I would not want to have a 576 version without it.

The Quality of the components inside.

At this point, the 576 differs massively from the 570. The 576 is a new generation of products. The change was made from hand soldered technique, to printed circuit boards, and not just some boards. The PCB's are glass fiber and gold plated in my 576. Augat Brand IC sockets everywhere, and many gold plated IC's even. Though I have seen some from Italy without IC sockets, and without the gold plating. As these sell probably for the same price, make sure you

 

REPAIRS: Cardanic Axis

 


Cardanic Unit broke off. This is from the scope tube brightness. It's crazy but I could just buy original old replacement parts in Tektronix packaging on Ebay. (see below)


Another one broken. This is from the looping compensation capacitor, to the front panel.


1977 NOS replacement parts on Ebay from Greece. Two for 8 Euro. Just what I need :)

REPAIRS: The 5V power supply.

With respect to the Tektronics designers, but the rectifier circuit is strange. The output signal is 5V DC, 100mA to feed 6 TTL IC's. Only 100mA, and they are using a big chunky capacitor of 11.000uF, followed by a series regulator, with a heavy power transistor. Very bulky, and I am sure this could have been done with less parts. Or perhaps it was right after all, read below under ESR meter. Such things do contribute to the massive weight, and I see this filosophy, that "cost, weight and size play no role", throughout the whole machine. But let's not criticize Tektronix for that, because any 50 years old equipment with savings on material cost, too simple circuits, and too small build, are not fun to work on.

In my 576 was a defective rectifier capacitor in the 5V electronics, and the regulator circuit worked very strange. I had a hard time finding out what the heck caused all of this. Whereas in the end this is only a 5V 100mA power supply, feeding 6 TTL IC's with 5V. Well I was lucky, the output voltage was too low. It was 4.2 Volts. When it gets too high it damages the ICs, and they are hard to substitute. Later I learned, other people have the same problem. So when you buy a 576, check the 5V circuit not just to be nicely 5V, and also check for no ripple on the input. As a good 11.000uF capacitor will give no ripple at 100mA. I would say, it is better to check the all rectifier capacitors with an ESR meter (of the kind you buy for 20 $ on Ebay, they work very good). ESR = Equivalent Series Resistance of the capacitor. You can call it Ri if you like. I found it to be in the range of 0,8...1 Ohms Ohms for all power supply capacitors. (apart for the broken one).

The use of an ESR meter

ESR stands for Equivalent Series Resistance. Such a meter cost only 20$ on Ebay, and these get more refined every year. Originally a Danish designer made an Arduino design with this, which worked excellent. He published the code for public use. I made a copy of this long ago. Later, some Chinese said thank you, and took it into production. They improve the graphic interface, but I think not it's functioning. So at the moment you can buy those with a color screen and much better graphic resolution that the first ones. The test routine seems unchanged ever since. All of that for just 20$. Such a meter first ramps up a very small voltage to see what happens. When current begins to flow, it does some other tests, to find out what might be connected, and it continues until any three lead device is identified, and tested. So much better than any 800$ Fluke, which pumps 1mA (at 9Volt open circuit) into any unknown device, that can really damage anything. The ESR tester also uses a low voltage, low current oscillator signal, and it measures current + voltage. When all tests are done, the software can say if it found a diode, a transistor, a FET, and keep PNP and NPS apart. Measure a .resistor, a capacitor, a short, a battery, a zener, or a normal diode, and tell the value. If a capacitor, it will give the value, and the equivalent series resistance. The nice part is, you can leave the capacitor in the circuit, and due to the low signal it doesn't damage any semiconductors. Don't worry about how this is done. The point is: It works. So you just connect it to a capacitor in a circuit. Suppose it is printed on the capacitor: 8000 uF. The ESR meter will detect at the connections for instance 2000uF and 100 Ohms ESR. So definitely this capacitor is bad, though it will have some function still. Or, when it says: 6000uF and 5 Ohms, it is still good. Of course this is not a 100% sure test, but to my experience you are close to 100% in all practical circuits. When ESR is too high, the capacitance itself is too low, that comes together in case of detoriation. When ESR is too high, but capacitance still good, the capacitor has a contact problem inside. For all easy to remove leads, I did solder them off still, and then you can test any capacitor precisely. However, as I said, most of the time you can find the bad capacitors in the circuit, because normally ESR is too high then. ESR should be a few ohms only. There are tables for that, but this approach is relatively new, and I rather compare it with good brand capacitors of the same kind. So you can not compare an bulky 5000uF cap from Sprague, with a Chinese one, and you wonder how they can make them so small and so cheap. . The answer is found in ESR changing already after a few years of use.

A very useful application is test capacitors you would not necessarily test, because they are somewhere on a PCB which works good, and you don't want to damage something. Even new capacitors don't get any better from soldering them in, but un solder 50 years old capacitors, and solder them back in, is not what you should do without need. Then they have been soldered three times instead of only one. WIth an ESR meter, you can quickly test every capacitor and not desolder it. If you never worked with such a meter, spend 20$ on it, and it will change the way you work on broken equipment for ever.

Still, because one appeared bad, I decided to I un solder all capacitors of the power supplies. There are two PCBs, mounted over them, and you need to remove those, by unplugging the connectors. Also take your chance to correct the connectors a little bit, as some of them got a bit loose ever since 1968 when it was made. My ESR meter indicated all capacitors "like new" except for C759 of the 5V supply. It had so high ESR, it was regarded broken. Other 576 users report C759 broken as well, so it is a common fault error. I opened it up so see what's inside, and it was very wet still. So sure not dried up. Still ESR was "infinite" on the tester. So this is really interesting to see, so many of those old electrolytic capacitors were still perfect ever since 1968. Here comes a point, all good electronic designers know: Lifetime of electrolytic capacitors increases drastically when you have a low AC ripple current through them. So with this power supply, perhaps also 1/3 of the capacitance value might have done the job as well, but then ratio of ripple current and capacitance would have been "3 squared" higher, so 9x higher, and lifetime would have been even more drastically reduced than a factor 9. There are tables for that, in any capacitor meant for power supplies, and you can bet those guys at Tektronix knew how to test capacitors with their own curve tracers. So on the one hand, the capacitors seem extremely over dimensioned, but on the other hand, what is sure, almost every one is still good now ever since.

To remove the capacitors, first you need to pull of the plastic cap, by hand. To get the capacitor out, you can do this from the visible side, but you can't get the screws back in for the replacement. To do this, you can carefully take out the small High Voltage box. No need to remove the oscilloscope tube. Hint: There is one nasty screw that you can't get out easily. Use a normal, flat screw driver, and then it works. You can use that under an angle. Like this I could remove the screw, slide out the box, and reach the capacitors from the other end, with my hand inside. I was able to fit new capacitors very easily inside the cap of the old one. Just saw off the cap. It was almost no work, and result was perfect. You can hardly see the repair afterwards! After I did this, the 5V supply was good again, the digital step generator worked, and the 12V circuit worked better also. Don't know exactly how, but the different voltages relate to each other. Like the 12V circuit produces the reference voltage for the 5V circuit.

 


Power supply caps can be reached after removing this small unit.

 

 


Power Supply board. Diode board is underneath. All ok, nothing to say about it.


Rectifier board. C759 is underneath. All diodes give in-circuit 680mV with my ESR tester.
Look at all that gold,
it must have been cheap in 1986.

All in all this whole power supply is a strange thing, and I would not have made it like this. I would have created the +12V with a 5 Volt 2Watt Zener diode as a reference. Then, take 100mA from the Zener diode to feed the 6 TTL ICs with. As simple as that :) But ok, 1968 is a long time ago. Things were different by then, designers liked to present fancy circuits. Those guys were the kings in any electronics company. I am sure, older analog designers know what times I talk about.

Series Resistors. The high Voltage to the device under test has a series resistor, limiting the power into it. While I was testing a power supply choke, one of those series resistors went up in smoke, when I went from 350Volt to the 1500Volt setting. I can't say much about this yet, first I want to replace it, and then repeat the measurement. It was not because of choke saturation, as the Lissajous diagram was a nice circle when it happened.

 

 

This is the power unit, with it's cover off. On the left are the series resistors, with a thermal protection mounted on it. Note, the inside chassis is nicely sprayed in Tektronix blue. This is waist of money and resources if you ask me. But as I wrote before, 1968 was another time.


The replacement resistor is 2x 4700 Ohms in parallel, giving exactly that strange value of 2350 Ohms. There is not much room in there to mount it, but I think I found a good place. As an engineer I really disagree that they mounted those five heater resistors directly above an array of electrolytic capacitors. Well... nothing is perfect.

 

REPAIRS: Design Error in the Step Generator

Yes you read it correct, a design error. In a Tektronix product. That's definitely what it is. The step generator had the nasty habit to skip a step sometimes, and it began to get worse, until in the end only every seconds step was displayed. I found a very nice article in the internet about this, in German language, which made it a lot easier for me.

It begins already here, at IC U3D. This IC generates a needle impulse every time the mains goes through zero. Now U3A, U3B, U3C, U3D is just an quad NAND gate, SN7400. They must have found it pretty interesting at Tektronix to alter the symbols for those, perhaps showing how they used it. However this altering is full of bad errors. That doesn't matter much for the real function of course. Just I want to point out here, what they printed in the circuit for U3D is a NOR gate, but what is used there is an SN7400 NAND Gate. And that is not the same. This is a totally wrong symbol in the drawing for U3D. Also for U3C, now it is suddenly is an AND gate with inverted inputs. So at a "0" on both inputs, you would get a "1" at the output. That is of course what a NAND gate does also, but NOT for all of the other conditions. So frankly, these symbols in the drawing are dead-wrong. But it's only the symbols. The function is good, when you want to call it like that. Look at what they connect to the output of U3C. A capacitor with a discharge e-function on it. This means the input of pin 12, U3D is an analog signal, and that is not allowed with IC's like SN7400. If you ignore that, the output becomes undefined as long as you are in the "forbidden" zone of the input voltage. Undefined means, the logic state is unknown, or the output may oscillate. Whatever the IC does, you can't tell. And that's what it does indeed. However due to this extremely curious circuit, they cut off the "undefined" phase, and what comes out is indeed the required a needle impulse needed for the step counter. However the needle impulse is analog. So the edges are no steep enough for TTL. And now the problems begin.

So look at the inputs of U22. There are two analog needle impulses there, with some very small phase shift, and the result is a small (now digital...) needle on the output of U22A pin 3. The analog signals on pin 1 and 2 however may sometimes cause a second impulse on Pin3, and that causes a failure with the step counter. After replacing U22 (SN7400) with an 74LS132 the problem was gone. This is a quad NAND with Schmitt Trigger inputs. I am not so extremely happy with this, because I do not know the reason why they made this illegal use of the SN7400. Was it ignorance or did they do something extremely special? We won't know. The machine is from 1968 and the designers probably don't work there any more. In case somebody has more information, please let me know.

When I have more time, I want to see how R8 needs to be adjusted in order to get the best needle impulse at U22A pin 3, and check what is the change when going from SN7400 to 74LS132m, as logic levels of 74LS132 are not the same. Because it's LS, and because it's a Schmitte Trigger.

 


Burned resistor in the test fixture. This was not me!


This was the collector "sense resistor". I selected a very precise one.

WARNING: Later I learned how to burn those resistors. There are two in there, one for each position (left-right) of the fixture switch. I received my 576 with one resistor burned, I soon burned the other one myself. This is how it happened: On the text fixture is CBE for transistors, and two additional holes for whatever it is. When attaching a device to those two holes, you can nicely do AC measurements on it. However when you try to test a high current device there, those 22 Ohms resistors go up in smoke. High current testing I do via the Collector-Emitter terminals, and then it works fine.


Part 4) CALIBRATION OF THE TEKTRONIX 576

Of course you need to use the original calibration procedure, and ideally you own the calibration fixture. Here is just a short cut I took, and it seemed to me it was within calibration anyway, apart from very minor adjustments in the 1% range. Also if you find, calibration is totally bad, you need to look for other things first before you mess up the rest. Like one of the internal voltages is wrong, or something with CRT tube. So never try to "calibrate away" a problem, without fixing it at the root. I know people LOVE this method, but I hate it, as it never made me happy. So any problem, find the root cause first, and fix it before calibration.

 

Here is how I went, but again this made only sense as the machine was pretty good condition anyway. If it is full of problems, better fix first these issues one by one, and do not waste time on calibrating an instrument with problems inside.

  1. First: Adjust the -75V as precisely as possible. Next time when there is a reason to take out the power supply board, I want to replace the -75V adjustment pot meter by a 10-turns pot meter. Reason is, the 1-turn pot meter is much too sensitive, and the whole calibration of the 576 depends on this -75V, it says so in the manual. If you just touch the pot meter, it jumps 0.1V. Which is ok, because you can have -0.3 Volt deviation, but as said before the whole calibration depends on it.
  2. Check all other voltages. These were all within tolerance. The CRT High Voltage was 4200 Volt instead of 4kV as it should be. Measured by me with 100:1 oscilloscope probe, connected to a battery operated DMM. This gives an incredible high impedance, probably higher as Tektronix was s using. So the 4200 Volt I I think is ok, and besides there is no tolerance specified in the manual. So this 5% I am going to let it be as is.
  3. I had to to a very small correction of the trace rotation.
  4. The Astigmatism and geometry I set to the "best of experimental" as I could do. Well I guess my Color TV repair experience helped me a little bit. The calibration manual says to adjust it for a nice round spot when fully defocused. That's what it was when I tried that.
  5. The horizontal and vertical amplifier calibration needs a special fixture, which I don't have of course. But to begin with, you can do the centering of each three without the fixture. Compare th 1's, the 2's and the 5's centering without signal, and the spot should be in the center of the tube. When the CRT spot is in the center always there is no need for correction. With my 576 there was only the 5's spot one millimeter off center. After that was readjusted, the spot was exactly in the middle at any range, vertical and horizontal.
  6. The sensitivity of the amplifiers, you can also check with an adjustable DC supply and a precise voltmeter attached to it. This is not in the handbook, but all you need to do is, set the 576 like if you would measure a Zener diode. So put a 20V Zener on and display the curve. Then, remove the Zener, set the sweep voltage to 15V, set the current limiter to 0.1 and attach a 0...25V power supply instead of the Zener diode. The horizontal deflection will now be the same as the DC power supply voltage. Like this you can check any range. If you notice that anything with a 5 is too high, you can adjust one of them, and check it again. Then adjust it such that all "5" ranges as an average give a good reading. Same for the 1's and the 2's. After that, put back on the 20V Zener diode. Of course's it's voltage is not extremely accurate, but whatever it is, it should be the SAME for any sensitivity if you don't change the sweep voltage. So when it's 19.9Volt at 1mA at the 5V per division, it should measure also 19.9V at 10V per division. You can repeat that with some other diodes, like a 10V or a 50V. Don't go below 8V as such diodes have often a bad characteristics. So take 10V and above. If that leads you to the conclusion, the 2's read 4% too high, re-adjust it. So correct it not absolute to the Zener, but 4% to where it was, using a 10V Zener. Repeat the whole thing, and you will come at a point where all voltages make sense at any sensitivity. After this I was done with the horizontal amps. Note: Voltage reference TUBES seem to read 2% too high on the Tek 576, I think that's because they're not made for AC, and the charge and discharge makes the 567 read 2% too high. It's only that moment just before they BEGIN to turn off, they indicate the rated voltage. If you feel uncomfortable attaching a power supply, just try to display the "curve" of a 9V battery, just by treating it as an unknown device. I mean that is a legal test, and you soon will see a 9V line in some way. Try it out! After you understand why this is so, and why this does not empty the battery, you may as well take an external power supply.
  1. After the horizontal amplifier is believed to be absolutely accurate, you can use a precision resistor of 1000 Ohms, and it should give a 45 degree line, at 5V-5mA, or 1V/1mA etc. Be extremely careful to re-adjust the TEK vertical amplifiers by this method, because you can not adjust it better as the reference resistor is. A 1000 Ohms reference resistor is expensive but I use one that I checked it against many multi meters, and I believe it's better than 0.1%. This gave the much wanted 45 degrees line right away. Telling me the vertical amplifiers need no adjustment, and confirming once more my 1000 Ohms calibration resistor.

 

Part 5) Build a tube tester interface for the 576

The 576 was never intended for tube testing. As you can see clearly, at Tektronix they must have thought the days of tubes were over at that time. Well we all thought that, though it was wrong. It is a pity, because for testing triodes all it lacks is a heater voltage. This can all be simply added to the fixture with two banana plugs and an external power supply. Probably the superior versatility of the 576 will allow some testing options, the 570 can not do. I expect to find that out later, and will update this page when I know more.

On the left here you see the Quadrants. These are numbered in electronics counterclockwise, for whatever the reason. Here you see zener diode tested. So the forward voltage is in the first quadrant, with positive voltage and positive current, whereas the Zener part is in the third quadrant, with negative voltage and negative current. These kind of curves are displayed in the "AC mode", so for testing two lead devices.

Three lead devices have a control input, which the 576 can supply either with a voltage or with a current. So you see, this is already something the 570 can not do, but a control current may be interesting when checking positive grid voltage driving a tube. Just to name something. Also the 576 can produce a sweep with a negative voltage. This is interesting perhaps sweeping the grid voltage, and stepping the Anode voltage. Or work with a higher external anode voltage and show al least one curve. So this would show a 2nd. Quadrant curve like this. Though I really have no idea right now if this will be possible, and I want to look at such options all on the end of this project, and not at the beginning.

Things to take care of, and what we need add to the set up when testing a tube with the 576.

  1. Interface. There are three connections on the interface board. C, B and E, for Collector, Base, Emitter. Now FET's are simply connected there as well, and tubes also. Just have have set up the tube tester for the right kind of sweep direction. So referring that to a transistor it means, we need to set the "PNP-NPN" switch to NPN, or (which is the same) to the "+" sign. That will give a positive voltage sweep.
  2. Grounded Cathode. We want to test the tubes normally with a grounded cathode. So choose "grounded emitter". Though of course you can choose grounded gird as well (choose grounded base), but I have no purpose for such a curve The 570 and 576 can do this just re-ranging the leads, but 576 can also do this with the selector switch. Well at first I am only interested in plain, normal triode curves, so the ones you can see all over the internet. These are in the first quadrant, meaning a positive voltage sweep.
  3. Heater Voltage. The 576 has no external heater voltage of course. We need to heat the tube under test with an external AC or DC power supply. When using AC, there is a complication when testing Directly Heater tubes. We need a centered connection on this AC voltage, which becomes the cathode connection for the 576. I tried it without center connection, but that doesn't work at all. Reason is, the anode is sweeped with 50 (or 60) Hz, same as the heater voltage. However with DHT tube, the virtual cathode is always to be thought as the center of the heater. However that virtual cathode is now sweeped with the mains frequency as well, and that messes up the tube curves. When using DC, there is no mess with the curves, but the virtual cathode stays always the center of the heater, and this virtual cathode is not grounded. This creates an offset of half the heater voltage. Also with DC, we need to work with a center tap, so two DC power supplies are needed for DHT. For indirectly heated tubes, this doesn't matter.
  4. Screen Grid voltage. For testing pentodes, we need an adjustable screen grid voltage. This a small current power supply only, but we need it. Perhaps it is a an idea, to derive that DC from Anode voltage, using a resistive divider and a potentiometer, and design this circuit such that the screen voltage is always lower than the Anode voltage. It needs to be tapped at a point inside the 576 before the anode current is measured.

 


This is to copy and print. You can color mark specific set ups on it. Some wise guy sells printed Versions of this on Ebay and sells it as "cards". (it's just a copy from the manual, but he seems to get 7$ for one)


Conclusion: The TEK576 is a highly recommended, extremely useful machine.

Well it took me three weeks, every day 1 hour, but it was fun doing. What I can tell you, having a curve tracer really makes you understand two-lead devices better. As far as transistors are concerned, it amazed me that Germanium transistors have nice curves too. I always thought they were "primitive" or so, with bad curves and lots of distortion on it. However, curves of OC71, AD150, AC126, they were all so nice looking, and hey.... a LOT better than any pentodes I have seen.

The Tektronix 576 is so nice build, it is unbelievable. I would not know what to improve, apart from the weight which is really crazy.

Interesting idea: but I am not going to follow up on this: If you buy the programmable fixture on Ebay, you can access all connectors and wiring to control the TEK576 by computer, using some USB relay cards. May be you don't know, but most settings are relay controlled anyway. So when you choose 20V per division, it's just some relays you operate, also when you use the rotate knob. The knob itself doesn't do the switching, but it switches the relays. WIth the programmable fixture you can access all those relays yourself.

 

Part 6) Loose items, unsorted

Prices of the Tek 576. From what I found, it was it least build for 21 years from 1969 to 1990, and last units were sold in 2000. Perhaps somebody has more information for me. In 1969 it was more expensive than a Chrysler Rambler 6-Cylinder, two door Sedan Coupe. In 1990 it was sold for the price of a Mercedes Benz. It seems as if they tried to stop sales of it, by price increases, but that didn't work.

Catalog prices:

$2125 (1969)
$2800 (1973)
$5200 (1977)
$8775 (1982)
$11455 (1985)
$17050 (1989)
$18440 (1990)

Some specific parts of the circuitry works relatively difficult. It seems to me, the designers had great fun, trying to invent smartest possible circuits. Which is good and bad. What is good about it, such circuits WORK, as they are made by people that know the classical circuits, and then try to outdo them. What is not so good, when there is a defective part, you will spend a lot of time finding out how the "clever" circuit works, which is very difficult when the circuit is malfunctioning. So you have funny signals on a funny circuit, and that can give you quite an head ache. Well, the good thing as I said before, the used parts are generally very good, and nothing much seems broken anyway. The machine is well build from very high quality material, well designed, and made with good consideration somebody needs to be able to service it, and take it apart.

Mechanical parts contains plastic parts that detoriate, or becomes cracked. However real problems from cracking plastic part are small, and it seems replacement parts can be found still. Even the plastic cardanic connections for the pot meters, that can only break off, I found some NOS replacements on Ebay within minutes, and they were even cheap.

Weight is crazy and excessive. They really took no effort to save on weight. So you will see two pot meters in the back, connected with 30cm long metal rods to the front, using cardanic coupling pieces, and the pot meters mounted on thick aluminum panels. Protection caps on the internal high voltage units are from (too) thick metal, etc. Everywhere you look, it's twice the weight needed. So in the end the whole item gets twice as heavy. You can hardly ship it. So beware, you can NOT put this in a card box. This needs a wooden palette to ship. If the post man refuses to hurt his back, and simply drops such a box, I can't blame him.

Lamps are expensive. The "type 349" lamps that illuminate the scope are expensive, they cost 2...8$ one on Ebay.

The red hazard lamp on the right burns the voltages exceed the maximum of 15 Volts. So far so good, but if this lamps breaks, there is a safety precaution switching down the high voltage. They used the safety switch of that, so the one that is underneath the Test fixture. There is a relay which only is "on" when the switch of the safety cap is closed. However when the hazard lamp itself is broken, the tester behaves as if the safety cap is still open, which by itself makes the yellow warning light burn. You can try that out, just set up a test, and while it's running, pull out the hazard lamp. The same moment the tester shuts downs. So yes, really safe, and at 1500 Volts, that is a good idea. On the other hand, I wonder how many 576 were send in for repair, because of this. This behaves like some defect with the high voltage, it refuses to switch on. So you think your 576 is defective, but it's only the hazard light bulb that is broken.

The fuses are at the back. There is also a thermal fuse inside on the high voltage series resistors, which will reset after some time. You can choose from a low, medium, or high mains voltage setting. So save the instrument, best is to take the lowest one.

The high voltage for the horizontal sweep, up to >1500 Volts peak, is generated with a variac, and actually they use the sine wave of the mains voltage to do the horizontal sweep. The nice part is, you can not damage this circuit so easily, it has a thermal break out, which you can re set. By using the variac you can conveniently ramp up from 0 to 1500 Volt, or whatever maximum you take. Also there is current limiting on this, by an selectable series resistor bank. This is strong, safe, effective and will take some amount of abuse and errors.


 

Part 7) Testing of two lead devices with the Tektronix 576

End of Life, GR150DA tube, apparently working.

Here you can see why it may be a good idea to replace the GR150 stabilizer tube of a Funke tube tester. This tube apparently worked normal in a Funke W19 tube tester and it produced the 150 Volt stabilized voltage. This tester tests the tubes with Ug=0V or -2 V, and in this condition specially low Ri tubes react sharply on just a few Volts difference with Anode voltage. So you really want 150Volt and not 148Volt, for the W19 tester to be accurate. GR150 is amazingly accurate, WHEN.... it is a new one. Any used up GR150 apparently work good, but they have much too high internal impedance, and the 150Volts becomes up to 600 Ohms. Which can be in the range of the tubes tested, so that is a bad thing.

Now, that I have a curve tracer, I can see the difference between a used and a NOS tube, clearly. The NOS behaves like ideal. It has no Ri, and not round cut off like OD3 stabilizers show on this curve tracer. So clearly, this GR150 tube is a very special quality. Behavior is more like a Zener diode, and better than any other stabilizer tube I tested. Though problem with a Zener is very large temperature drift. Wheras this stabilizer has no drift at all. I have learned from analyzing the GR150 with the the 576, there is no replacement for the GR150. So replace it by some other stabilizers, and you will never have such a nice, truly vertical load line with Ri=0. Replace it by Zener diodes, and Ri is close to zero, but all Zener diodes have such a terrible temperature drift problem, they can not seriously be used here as precision voltage reference.

Just remember what difficult and complex things they have been doing in the 1960's with early digital volt meters, to have a precise 10V reference in there. Hewlett Packard put the voltage reference IC in a miniature oven. There are voltage references GR... tubes with the size of a cigarette. I have some here. These give 100Volts, and with a 10:1 divider that gives a near perfect voltage reference. I have one the first HP digital volt meters, and it says in the hand book, they need 30 minutes warm up time to be within specified tolerance. Well these GR.... devices need just 60 seconds warm up time. Much better! Well, but I guess I am too late with this suggestion :)

Now, that I have a curve tracer, I can see the difference between a used and a NOS tube, clearly. The NOS behaves like ideal. It has no Ri, and not round cut off like OD3 stabilizers show on this curve tracer. So clearly, the GR150 tube is a very special quality. Behavior is more like a Zener diode, and better than any other stabilizer tube I tested. Though problem with a Zener is very large temperature drift. Wheras this stabilizer has no drift at all. I have learned from analyzing the GR150 with the the 576, there definitely no replacement for such a fantastic tube. So replace it by some other stabilizers, and you have not such a nice, truly vertical load line. Replace it by Zener diodes, the load line is better, but these have (TERRIBLE) temperature drift. I suppose GR150 can only be replaced by a very precise electronic made 150V voltage reference.

The coloring of the light.

On purpose I have changed the color balance of those two pictures, such that it shows the color of the GR150 correctly. I noticed old, used up GR150 show the normal, dark orange color we know also from Neon lamps. NOS GR150 however, seem to have some other gas mixed with the Neon too. They light up with some "pink" tone + the normal dark orange neon color mixed. It is such tubes that show this remarkable absolutely ZERO Ohms load line, as I have never seen before with any gas stabilizer whatsoever. At the end of life, this coloring changes to normal dark orange Neon color, and Ri becomes 600 Ohms. So it seems to me, this added fraction of Gas, gets used up until it is gone. Perhaps this is some small form of radio activity which disappears after long enough use, but I have to say I really don't know what causes this.

Philco NOS Selenium Cell Type 179440-E

Curves of a NOS Selenium cell. This is only the forward voltage, the beginning of the curve is at the bottom left corner. Such a nice technology actually from before the day of semiconductor "chips". They just created a semiconductor from layers of metal oxide, not even hermetically sealed. Just painted the device and that was it. Provided you can accept some reverse leakage, and for high voltage you had to serialize many, they worked fine. Higher current was no problem. Actually very interesting devices, long before Germanium diodes.

Break down voltage is an unsharp limit. It just begins to leak somewhere around -70V and it gets more and more. I don't know if that is very good for the cell, so I stopped that, and no pictures.

I expected this Selenium to be able to get warm, because it has cooler plates. At some 200mA it begins to warm up a little bit, and it smells faintly like new paint. Same as when you warm up an NOS steel tube for the first time.

Above 400mA the current drops within one second, for some 10%. SIt seems that is too much current. ,When I reduce current back to 200mA, the cell gains the 10% within a few seconds again and stays like that.

I was unable to find data about this Selenium cell. If someone has a data sheet, please let me know, it would be interesting to test it versus the original specifications.

So my guess is, this is a 250mA cell, and it can produce something like 70...100Volts DC at a forward voltage of 9 Volts. I don't know how much leakage current is allowed. It seems the forward voltage is 4 Volts, and dynamic resistance is 12.5 Ohms. (so 80mA per Volt)

1N4148, general purpose, small signal Silicon diode

A small signal diode is nothing else but a smaller size chip of the same wafer. So it has less capacitance and less leakage. For that is has a higher voltage drop at higher current, which logically everybody will agree with. Saying so, it should also have a higher voltage drop at medium current, and in fact it will do so at every current. So the meaning of a small signal diode, unlike many think, is not to work better at lower voltage. Because if that is what you want, you better take a low leakage type power diode. The purpose of a small signal diode is only one: It has lower capacitance, so it works better at high frequency, which of course is important for many applications.

It can be seen, the diode begins to wake up above 130mV forward voltage. Above that, dynamic resistance begins to develop. Below that, dynamic resistance seems endless high. So it's impossible to rectify a small signal below 130mV. It seems to need this minimum voltage before something happens inside the chip, which makes it a diode.

 

Unknown, small signal Germanium diode

Shown here at the same settings of the previous diode. Please read the text of the above Silicon diode first. A Germanium diode, curiously, does not seem to need a minimum voltage before it becomes a diode. It is clearly visible it has some dynamic resistance at ZERO Volt. The angle is appr 45°. So Ri=50mV/1uA = 50kOhms. When going to NEGATIVE voltage, like only -25mV, you will see Ri will be 100k, and when going to +25mV Ri will be only 25k. This has a rectifying effect already.

Make note of this really very strange effect: For rectifying very small signals with a Germanium diode, you can work much under the forward voltage. In fact, totally without forward voltage, and apply 25mV AC on it, as in the above example. If you do, a DC voltage will result! To explain this more simple: Suppose you have a plain network of one Diode and a capacitor, and apply 25mV AC on it. At the positive half cycle (see diode curve) 1uA charge current will flow. At the negative half cycle 0.4uA discharge current will flow. The result is, the capacitor will be positively charged to the point where there is no Ri change any more. From the curves again, that is somewhere around 35...45mV. So just normal rectification, only at very high impedance. As long as you put only a capacitive load on it, this makes a working rectifier.

It seems to me, you can even use the Germanium diode at a NEGATIVE voltage. So from -5mV to -50mV there is a significant change in dynamic resistance. That will make the discharge of the capacitor greater than the charge, and a negative rectified voltage will result.

This effect, I have seen only with Germanium diodes. So the "diode" effect is already there at zero volt, and even below, it stops around -60mV.

This explains why crystal receivers work so nicely. As the piezo ear plug is a capacitive load only, it even needs a 680k Ohm resistor in parallel as a bleeder for the capacitor. (So it can discharge too). Whereas the diode has 50kOhms as shown here. So it can drive 680k Ohms easily.

Unknown, 50 years old Silicon power diode.

This is a gold plated, glass sealed silicon power diode. It was inside a very old amplifier, provisionally soldered in there by somebody. I exchanged it just because I wanted four of the same diodes in there. From there it came into my curiosities box, some 20 years ago. What does the TEK576 tell me? The diode is fine and no break down at 400 Volt. I didn't want to go any further. It doesn't seem to like current above 1 Ampere, that makes the forward voltage go up too much, but it doesn't even get really warm from it. So lets call it a 600mA, 400V diode.

So I know now this diode is still good, I exchanged it without need.

Neon Lamp ignition.

This neon lamp tested here, once was the decimal point of an old nixie display.

Here is a zoomed part of the ignition. (you can zoom in with the 576). This way of making it visible is only possible with an analog instrument.

I have no good explanation for what I see here. If somebody knows more, please send me an email.

Note, for this picture, the X-Axis was shifted down to the bottom of the display by setting the tester to "NPN transistor".

At least the observation is, when the negative resistance develops (at A) the resistance itself is too high to reach point B straight away. Also you can see the speed of the movement, because when it moves fast, the line is thinner. Look at position B, here is where ignition begins, and the line moves up diagonally. You can see, at first it moves slow, the line is thick. At that point, the resistance is -125kOhms, estimated, and with that it can not reach the bright spot in the middle directly, the angle of -125kOhms is wrong for that. So some kind of longer path is unavoidable.

This is hard to understand. Well at least for me.

Testing of a very high inductance coil.

So you think you have a good inductance tester? These may have problems with very high inductance coils, with relatively high internal resistance. Such coils are made to have a lot of inductance in the first place, and resistance came second. Like that, they can be made small.

Such a choke is the Lundahl LL1670 grid choke. You can definitely NOT TEST that coil on an induction tester. Well you can, but you get too low result. I had similar complaints from buyers of the LL1670, using expensive electronic testers. Here is why:

The unusual high internal resistance of this part (5k) upsets the impedance bridge I am using. Though this bridge is amazingly accurate, I can not find a clear balance point at the expected value of 540H. As you can see below there is a balance point at 104H and it's the only one it finds.

So I tested the coil for it's time constant, which is the one and only true way. This is a relatively difficult set up, charging the coil with 0.8mA DC, with a square wave generator at 0.5Hz. This frequency is like DC for he coil, and it charges at the positive cycle which lasts 1 second. Then it discharges via the same square wave generator, when it goes to zero Volt. The discharge current is an E-Curve, and at 62% fall time or rise time, the R/L time of the coil is reached. Due to it's high internal resistance it can not discharge fast, so the scope is flickering and unpleasant to work with and to look at. A very nasty measurement that was. But it worked and I found 550 Henry indeed, same as printed on the choke, but with quite some tolerance on the measurement, because it reached the limits of my oscilloscope.

Then, I asked Per Lundhal from Lundahl Transformers, and he told me, a possible way to measure such coils is by putting a very high AC Voltage on it, with a Variac, until the rated current flows. The current and a voltage will give the impedance.

When this impedance is much above the DC resistance, you can neglect the DC resistance, and Inductance results from * L = Z / (2 * Pi * f * L). This resulted in 700 Henry. Per confirmed to me, the result is correct if measured this way, and internal capacitance gives some deviation. He verified the measurement t for me, using just a normal Volt and Ampere meter, and he found 700 Henry also. The advantage if the Tek 567 is only you need no set up with external instruments or Variac. All of that (the Variac too...) is already build in.

With a 90° phase shift part, you get a Lissajous figure, but as you can see, it is not a nice circle here. I have not deeply looked into the reasons for this. At least it is not saturation, because it looks the same at every voltage, also 10 Volts. Perhaps it is distortion of the mains voltage. That remains invisible with resistors, but becomes painfully visible when there is 90° phase shift between voltage and current.

Testing of a very low inductance coil.

This is a coil from a switched power supply, so intended for 50kHz or more. Such a coil will be quickly saturated at 50 Hz.

At small signal, the picture will be a normal oval shape, typical for an inductor. Though I need to go to very low voltage for that, and the coil begins to pick up noise. So such a picture is not a good one. Very confusing, and I will not show it here.

At higher voltage the coil saturates quickly at only 50hz, as can be seen here at appr. 1.8 Volts. So you will see the current go up faster above 1.8Volts. In a way, this indicates the behavior of the B-H curve too. Where the current rises suddenly faster, this is where the flattening of the B-H curve begins. With some math, the curve as shown here could be calculated into the B-H curve.

In the second scope picture, the effect becomes clearer. This is the same test, just done at higher voltage, and you can see the coil saturation completely and only a resistor is left. So it ends with a normal straight line.

Note: The pictures have a thin and a thick line, but that is because of the exposure time of my camera. It would need longer exposure time, but then the picture gets unsharp.

Part 8) Testing transistors with the Tektronix 576

This is it's primary function, and it can do so many things, it surprises me every time how nice and accurate it does so. I feel a little bit like somebody using an oscilloscope for the first time. So you make a sweep with from 0V to a maximum positive voltage for NPN transistors (or whatever other device), or switch it for a negative voltage sweep. In accordance with the four mathematical Quadrants , the negative sweep is done from the right to the left, and the (negative if course....) current from top to bottom, but in case that is too confusing for the user, the picture can be rotated 180° optically. Or, you can do an AC sweep, displaying Quadrants 1+3 at the same time. (A quadrant is this here)

 

 

Siemens AC151 Germanium transistor, with collector and emitter reversed.

This is curious, I just tried if this works at all, and it does! Actually it works also with a triode, when you reverse grid an anode. Just such a tube has very low gain, but distortion is a LOT lower than normal, and also anode dissipation gets limited to grid dissipation here.

Now with a transistor this works a lot better. It amazed me. Probably such a transistor is not a very good one, but at least curve charts here look nice to me and Beta is not so bad. (One division, so 10x). It is unknown what maximum current and voltage it can do like this, and I didn't want to destroy the transistor to find out. So I stopped at where you can see in the curve chart. Note, what is tested here as collector-base junction, is normally is the base-emitter junction. This means the original base-emitter is now connected in reverse direction, and I don't expect the break down voltage of this diode to be very high, and perhaps break down is destructive too. So I stopped at 10V 5mA, but the transistor responded still very predictable and nicely.

 

BC558B Silicon transistor, with collector and emitter reversed.

Obviously also silicon transistors can be used reversed, but the curves look not very nice any more. Silicon transistors have a reverse break down of the the base emitter junction at 7 Volt. So when we use the whole transistor reversed, what is now the collector has a break down voltage of 7 Volt, as you can see from the second picture. I turned the amplitude knob such that break down just begins with the first curve (200uA), and also the zero current line, show this break down. At higher amplitude there would be a vertical line at 7V, as the curve tracer goes into current protection mode.

ASZ16 Germanium PNP Transistor. 60Volts 8Ampere 30 Watt.

ASZ16 is a work horse from the year 1959, for power supplies and high peak current. The curves here, show he curves look reasonable in the low power range. So here at 25 Volts, 300mA. When going to higher voltage, unlinearity becomes absolutely terrible. Fair enough, it's only a transistor for switching. This here is just to show at least the transistor is normally working. In the next pictures, I will show some unconventional use, with that same transistor.

ASZ16 data sheet

ASZ16 Germanium PNP Transistor, used as NPN transistor

This is quite strange, but with the TEK576 it is just a flip of a switch. (see the green arrow at the left). So why not do it. This inverts all voltages, and it means a PNP transistor gets simply used as NPN transistor. To display the sweep still from the left to the right, the whole picture tube gets electrically rotated 180° with the "invert" switch. So the green arrow is the electrical invert, and the yellow arrow is the optical invert. The curves look not nice, and they are only nice at very low power, as shown here. The curves seem to sink away under the horizontal axis. (This is not the TEK576, this is by the transistor itself).

With a supply voltages of 20 Volt, the current rises sharply, and it gets all highly unlinear. This is not shown here, but even then, it still works.

A GERMANIUM TRIODE!

For this take a ASZ16 Germanium PNP Transistor, connect it simple reverse, so as NPN Transistor. The base-emitter is now in the reverse direction, it uses no current.

You think this doesn't work? I also thought this.

It is so confusing, trying semiconductors (and tubes) with swapped connections, or reversed voltage. Instead of the device going up in smoke, like I expected with semiconductor, and specially with so called "primitive" Germanium transistors, it just shows an interesting function. Some of which are strange, and some of which look not bad at all.

Moreover, the devices afterwards always were undamaged.

Well, I have pictured all settings of the TEK576, so it can be replicated when somebody is interested. You know how triode curves of a Germanium transistor look k like? Well here it is, a solid state triode, and surprisingly linear.

Look at the second picture, this is the same configuration, but now at 20x higher voltage. The curves become so nice, you need to realize these are triode curves, in a more or less ideal shape. So they are almost straight lines, absolutely evenly shaped, and the very low Ra makes the lines almost vertical.

From the second curve chart, I find following TRIODE data:

1) Gm70mA/0,2V =350mA/V

2) Gain is 1x. So an input step of 2V gives an output step of 2V also

3) Ra = Gain/Gm = 2.8 Ohms

So gain is disappointing, but output impedance is so amazingly low, this can directly drive a speaker. Meaning there is a lot of power amplification still.

Part 9) Testing tubes with the Tektronix 576

This is a Russian 6SN7. The normal mode is as normal as can be, there is nothing to say about it.

In the second picture, I have reversed Anode and Grid. This sounds unbelievable, but it really works very nice with some tubes, With most tubes there is little or nor useful operation area. With some it works very nice. With 6SN7 seems nice at very low voltage. Interesting is ultra low distortion, a magnitude better than the normal way. Even so, this works here at 6 Volts supply voltage. But it works, and it's an amplifier tube, and I know some people are doing it. (read here)

Part 10) Testing other items with the Tektronix 576


State-1 Non Conductive state


State-2 unlinear resistance Appr 50 Ohms.


State-2 unlinear resistance Appr 50 Ohms.

 

A self made Coherer.

When you are interested in history of electronics, the coherer is a real specialty. This was the first detector of radio waves, even discovered before people had an idea what radio waves really were.

In 1835, the Swedish scientist Peter Samuel Munk, noted how iron filings became conductive after a static discharge in some distance. He did not know what caused it, and nobody had an application for it. He was so close to the discovery of radio transmission, and yet so far away...

In 1890, the French teacher Édouard Branly constructed a sensitive detector of radio waves. The Italian inventor, Marconi build an improved coherer, which could detect radio waves across a few km at first, but later models would overcome 750km at sea. By using ships to repeat the messages, it was possible to transmit messages across the ocean, fully without the use of electron tubes yet. Coded not in ASCII, but in Morse, but for the rest there it had all the elements of email already.

A working coherer is very easy to make yourself. It is nothing but a small container with iron filings, but you need to set it up the right way. It needs a certain shape, adjustment, and the right voltage across it. I used for this a plastic hose, two Zinc plated screws that fit it, and some self made iron filing. It worked right away, and with lots of trial and error, I got it working detecting the small burst of radio waves coming from a piezo lighter. So this brute device, you see on the left, filled with raw iron filings, is in the range of mega Ohms impedance as long as you measure with a voltage below 4 Volts. So you can't use your standard Ohms meter for that, as these devices use often 9 Volts. The set up needs a variable power supply, a small 6 Volts lamp in series as current limiter, and an ampere meter. Of course the TEK 576 offers all of that, but you can as well set that up another way.

State1. Here comes the amazing thing, which will puzzle you straight away. Build this thing as you see here in the left. Press the iron filings stiff together, and then release it 1mm. It is ready to use now. Test it at 3V and no current will flow. Nothing! You can shake the filings, use a magnet, or do whatever you like. I called this state-1 in the picture.

State2. Increase the voltage slowly. Unexpectedly the device becomes suddenly conductive, which an S-Shaped curve. This is an instable condition, but resistance is still positive. There is no way back from here. If you remove the voltage, the effect will stay. If you increase the voltage, impedance will lower further. The device remembers this like a Memristor. I called this state-2 in the pictures. However this is not yet the coherer effect. You can already now lower the impedance with radio bursts from a piezo lighter. This gets of course nicely visible on the TEK576.

State3. There are two ways to get the device in State 3. The first way is just apply a higher voltage for short time. The device with immediately switch to State 3. The second way, is generate a burst of radio waves near by. This will have the same effect.

Note, state 2 will normally not appear after a burst of radio waves. It will rather jump from state-1 in state-3 right away.

Negative resistance. This is what takes the Coherer from state-2 in state-3. So at higher voltage resistance goes down.

Memory effect. Unlike a normal negative resistor (like a neon lamp), there is a permanent memory effect, also when the power is switched off. So, a Resistor that remembers it's current. I don't think I should call it a Memristor, but what is it?

Reset. Once in state-3 you can reset the Coherer by tapping on it. It will jump to state-1

For good functioning you need a few things. So just a little pile of iron filings, and poke two wires in it, will not work. The filings must be in a cylinder, and you need some contact surface. I used big flat screw heads for this. There should be no pressure on the filings, but also it should be not too loose inside. Next is, you need the right bias voltage. It worked best when was close before State-2 begins. So when I see small current begins to flow, I reduce the voltage 0.5V and tap it, to give it a reset. Then is ready to detect. You do need a "good" lighter, so one that radiates a lot of dirt when you use it. From 7 pieces I found, only one worked very good, and four did not work at all.

All in all this requires a lot of trial and error, but I can tell you it was extremely satisfying when I found the Coherer working. I was able to switch a lamp on, just by firing a piezo lighter in 1cm distance to the coherer.


Planned before I die

Curves of a burning Flame.

There was a patent awarded to Lee de Forest for this in 1905.

 

 


© COPYRIGHT NOTICE All Rights Reserved