The SOFIA Digital Tube tester
by Audiomatica Italy

Compare also with other digital tube testers

This tester was commercially speaking not a success, but I think that was only because Audiomatica was ahead of it's time. Recently I was informed, only 80 pcs were ever made. Their idea was, to sell it to companies, but buyers were mostly private. Mr. Maurizio Jacchia, President Audiomatica told me this via phone, and this customer group they did not target at the time for the Audiomatica. They did have an unfinished Windows version working, but this was not continued, as they obsoleted the whole product suddenly. I had an order for one, and they told me it was the last one they would build. Then, they cancelled my order even, though it was confirmed originally. Well now it is almost 20 years later, and I found a very good used one, and another one with small problems.

I keep on saying, this was a big mistake to obsolete it. But I see also some behavior of them which caused this. When you see today, how many Amplitrex are sold, and believe me, Amplitrex is not half as good, then you can see Audiomatica was 10 years ahead of it's time. So it might have been better for them to bite through the sour apple of migration from DOS to Windows, not make the price so excessive. Then they should have sold it direct, and they basically would have had the sales that Amplitrex has now. Which I estimate a few pcs per week.

The best of the Sofia it it's natural, intuitive and easy way to operate it. This is a fine piece of software design. Combining with DOS speed with 16bits A/D converters, this tester is a pleasure to use, after 20 years of service. It doesn't have this (much!) too loud fan of the AT1000, it boots 5x aster, it plots a complete curve chart 20x faster, and much more resolution. With the AT1000 when you draw a 10 steps curve chart with high resolution, I surf the internet until it is finished, and then it doesn't even have the normal tube test, that is additional, you need to wait once more. Or twice for a dual Triode. With the Sofia you say "start", the Hazard lamp switched on, the RS232 LEDs flashes a few seconds, hazard lamp goes off again, and 15 seconds later, it's all completely done. That is fun to do. So curves and dynamic test results come up in one screen. Using the AT1000 is boring compared to this. I enjoy using the Sofia every time, and there is nothing pleasant about working with the AT1000.

The things this software can do makes you silent for a wile, and then you say... wow! So I say today, those who write tube tester software, and not look at this classic software first, have quite a risk to re-invent the wheel. (and not make it round :) Oh, and yes it can be downloaded for free, and live test reports I can send you you also. So that are curves,. you can zoom in and out. By zoom in, they are re-drawn with higher resolution. Yes, in DOS, but you don't see any of that, it feels the same as Windows.

To say it upfront, this is my favorite tester, but I have my reasons. With this very old Sofia, a test is performed much faster as with the new made AT1000, and with MUCH higher resolution too. Wow, what difference that is! With the AT1000 when I measure a dual triode with -say- 2x 10 curves, it gets so boring I surf the internet while the AT1000 tests. With the Sofia, you have no time for that, it's finished while you look at it, and that is test data and curves all together. And it's done in 16bits instead of 10 bits. Is the scale not as you wanted? With the AT1000 you must start all over. The SOFIA you can just crop or zoom the scale. Horizontal or vertical only, or both. Amazing feature and very very practical.

The screen 4) is full of information and all soft keys work and do something. What you need to do, is boot your computer first. However, since you need to boot only the BIOS, and no need to start Windows, you are ready quite fast, even though the PC is an older DOS one. So then when you enter in DOS the run code for program which is "Sofia21", the screen nr1 pops up almost without delay. Then click your choice, like: Matching of already tested tubes, Quality Control, Edit tube data, etc, or as here in this case: "Measure" (which is tube testing). So then the screen 2) pops up and there you go. Select a tube from the database, and screen 3) tells you what cables to plug. (This plug action is not needed with the Amplitrex, BUT... for that the Sofia can do any connection, whereas Amplitrex is limited, and you fill find several tubes it can't connect, and you need adapters).

When you switch off in DOS, you just press the "off" key on your PC, and you're done immediately. Much Better than Windows!

When you want to test and match a series of -say- 100 tubes, the high speed and natural comfort of the Sofia software, makes it the undisputed king with great distance to modern products. These all seem to have snail speed software, compared to the SOFIA. Moreover, you can save the curves, and let the Sofia do the matching now, or at any later time. So it tells you for instance: Pair#1 is Tube 4+12, Pair#2 is tube 17+49, etc. So it comes up with 50 pairs, and even sorts them by matching quality. The matching is not just done, by sorting the plate current in a table. No sir! It does so by overlaying the curves best possible way, using a mathematical approach, which is very similar to the way the RMS result of a wave shape is calculated. It works indeed with a Root Mean Square method, showing the designers of this software knew their math. Isn't that a good idea, the software designer knew his math. With all respect to the Amplitrex, I don't get the feeling with this product. If I want to do this with the Amplitrex (overlay the curves) I print them, put several sheets on each other and hold them against the light. I can do so with maximum four sheets of paper and with more sheets you can't see what're doing anymore. Doing so in the computer age, I feel a bit silly, and for myself here, it is matching which I use the AT1000 for most for. This 20 years old Sofia does this job in just seconds. Anyway each tester, surprises you with nice features, so even today, in 2013 as I re-edit this text, for me the SOFIA is like a relict from the good old days of user friendly software engineering. It reminds me of the older Hewlett Packard instruments, that surprise you also with user friendly option. Still today, the SOFIA works so nice, with high accuracy, and adapting to the user, instead of YOU having to adapt to the machine. However, if you are prepared to adapt to the machine, and dig into the second layer of functions, so to say the specialties, it still amazes me every time with nice things this software can do. One more thing, the annoying "run time error" of the Amplitrex (meaning the software crashed), the SOFIA doesn't have that kind of a problem. When you do something wrong, you can correct it, and there is never a crash.

The hardware is beautiful made. This was an expensive very high quality case.

About shorts and fuses (skip this when not interested).

First let me start with the Shorts risk with a Sofia, because you always have to keep this in mind when using it. When you forget about the "shorts tests" with any historical tube tester you make a beginners mistake because these testers will be seriously damaged by a shorted tube. With computer controlled testers, you have little influence on this. The Sofia does have an overload test, and a heater circuit test, but that is all. It is not enough. I cannot say it is fully unprotected, and indeed it switched off when I made an error with the cabling (tube connections). Yet, the previous owner warned me, the tester needed a few times a repair, after testing hundreds of SOVTEK KT88, and some shorted ones were amongst those. Since he used the tested for KT88 high quantity testing mainly, he gave up on the Sofia. He had it repaired once more, and then sold it to me in a factory repaired condition. Now I don't know what he has been doing, but from my experience with people, tubes and equipment, I learned that so called "professionals" (and companies) make serious mistakes as well. Well still I regard it a strong warning he gave me, and I will not put any questionable tubes on the Sofia, and not use it as a work horse. Only use it only for tubes that are basically good, but I need the curves from. When I have time, I will put a protection series lamp in the gird circuit, same as Hickok uses it, and I have seen that on other testers too. The other power supply outputs I will protect with normal fuses and put reverse diodes over it. That seems all missing. The most common short is cathode to heater with KT88. Since the cathode is grounded in the Sofia, that will short circuit the heater, so the fuse hopefully helps here. Another common short is grid to cathode, and most likely the a series lamp will cover this very good, since a low current lamp is known to be the fastest fuse there is. From my R&D days I remember an extremely useful method to protect high voltage motor control circuitry. You use a fast power thyristor, which shorts the DC voltage directly after the capacitor of the rectifier circuit, and in the path is a normal fuse. Very crude, but the most useful method I have ever seen, and low cost too. When you fire this thyristor, you switch off the High Voltage to all electronics with a delay of less than a micro second. Then, some 10 milliseconds later the fuse will blow. Interesting, the thyristor survives this. Even if you blow the thyristor, the damage is a short circuit, so it still does it's function, and all you need to do is replace it. For firing the thyristor you can use any detection circuit you define yourself. Such as a current overload detector of the heater supply in this case. Well... these are just ideas. I will first put a fast fuse inside.

I once had a FACTORY NEW Electro Harmonix 6922 tube with a hair-thin wire wrapped around all pins, in a circle! So all pins were short circuited. Don't ask me what that was, but it came out of a box with 100 factory new tubes, from Russia. Also I have still here in my memorabilia cabinet a new Electro Harmonix 6SN7, also from a 100 pcs factory box, with the socket one position rotated, so that tube can never have worked. I do not try to prove here, they do lousy testing at EHX. These were tested with the Russian L3-3, and it survived those, though the 9-pins all together shorted 6922 made some smoke come from the inside. Well it survived that. Actually the Sofia would have spotted this mistake too, because it will not switch on the High Voltage, after the tube passes the heater current test. Heater current is measured, and must reach it's correct value before the rest of test program can start.

Another possibility is, a tube that appears good at 250V will short circuit at 500 Volts. This can happen if the tube is gassy. This I already had, and the tester switched off normally and detected the error.

The rest of the story.

The software

The weakness is, it runs under DOS which is a pain in the ass to set that up, but ok I have it done, and spend the Sofia it's own DOS computer. This is an Hewlett Packard Kayak 800 from the year 2000, with a multi boot option for DOS and W2000. These used to be available on Ebay, but they were never really cheap. It's one of the finest real DOS machines, because it has a factory installed dual boot system. I have not tried it, but I am pretty sure it can run Windows7 because it has 1Gb of RAM. Though the W2000 works still fine, it hooks up in the network still, and it has a USB port even. After the tests are done, I can re-boot in W2000 and copy the results on a USB stick and take it to a modern PC. From there I open the results again with a DOSBOX virtual machine. The regular virtual machines, they all don't work. The good thing is, once you have the software working, it is the best piece of test software I have ever seen. Even under DOS, the speed beats anything I have seen before, and Graphics quality is excellent. If you have an XP machine, the Sofia software runs fine, but only for looking at test result files (the .pte files). I use this option a lot, because once you have a .pte file, you can plot the curves on your screen once again, and move the cursor all around to analyze the curves, and get gain, transconductance and Rp at a random point. Also matching can be done afterwards, or repeated with a few of the tubes removed from the list. Still for actually measuring, you need DOS always.

What is good about the Sofia, and seldom mentioned, is the fabulous matching software. You can save up to 200 curves, and the program calculates which curves fit best to each other, using a mathematical error function. I mean you can do this in Excel for one single operating point, but for complete curves it is like programming a piece of software. The Sofia does this in 10 seconds for 200 curves. That's DOS :). So we have to say it, the software is GREAT and the best part of it, and in REALLY compensates the trouble using DOS. Furthermore, graphic capabilities of modern LCD monitors seems the perfect thing for this software. I don't know how this works, but it looks if the pixels from the DOS Program fit directly 1:1 on the pixels of the LCD monitor. The result is a crystal clear picture. (see at the end of this page, but these are compressed jpgs. The originals are sharper even)

A mystery remains.... At the power supply main board is a large connector, unused. It is of the same kind as the top board. So one connector connects the power supply to the microcomputer board, another (similar) one is empty and unused. Who knows what is was intended for! If I was to say, I would connect a relay board to it, so the cables at thee deck get obsolete, but that is really only just a guess.

It has no tube data base inside, just 10 tubes to give you an idea how to do it yourself. However, that works fast an easy, and after a while you have what you need.

Software Bugs Ж

I found only one.

The heater voltage is connected with one end to ground, and the other end is heated with DC. For indirectly heated tubes that is no problem, but this gives an issue when testing directly heated tubes (DHT). The reason for this is, with DHT, the cathode's electrical reference is by definition the middle of the filament. So you lift up the cathode reference point by by half of the heater voltage. Not much of an issue when the heater is +2.5V and the grid -45V, and low gain tube such as a 2A3. The error is 1.25Volt only and you would not even notice at this low gain of 4x. However, it is a major issue with a tubes like RE134 or the Emission Labs 30A.

Read more here

The hardware

This is made in non-SMD. You will appreciate that when you have a repair. There are very few connectors, which devices are always a long term problem in old electronics. The electronics don't get hot, and it needs to cooler fan. The fact the electronics stay cool, it the ticket for a long life. The printing on the IC's has not been removed, and these are all standard parts, that still can be bought. The circuit boards are clean and tidy, and doing service and repairs is not hard to do. The A/D converters part has many potentiometers, but probably most are just for the offset of the Op amps, and when needed I can dig into this. So far this is not needed, and I won't touch it. The core of the A/D and D/A converters are ICs used for sound cards originally, these work very well, and can still be found on Ebay. So are all other parts. The microprocessor is a first generation 8 bit working at y 6mHz. Given it has only 1400 Byte of code to take care of, this speed is more than enough. I expect the tester itself has a start up routine in the firmware, and after that reports itself via the RS232 ports as ready to go. It takes only commands to set the voltages, and responds with plate current, screen current and heater current. All higher functions are managed by the DOS software, which is not run inside the tester but on the PC.

The PC port is RS232, and here comes a bit of a problem setting up modern PC's for an RS232 port. Though I was able to use a USB to RS232 cable under Windows7, the accuracy under a non-windows DOS version and a real RS232 port is better. For the rest of the DOS story, read the software part.

The case is beautiful quality, larger and heavier as you might expect.

Very useful, and simply forgotten on many other testers is the High Voltage "HAZARD" lamp, indicating the high voltages are switched on. (I passed on this idea to Ronald Dekkers, the maker of the uTrace. He has already included it for newer versions).

A major shortcoming of the SOFIA, is the tube connections have to be made by plugging in the cables. Well some protection comes from this too, since you cannot plug in two cables in one hole, so an error you usually see, but this stays kind of touchy. Best is to make colored overlay cards for the tubes you want to test. That works faster, and is error-proof. An advantage we have here too, since these are all banana connectors, and you can access the tube connections easily for whatever purpose you need it. Like connect external meters to it. Even put in a current link.

The deck panel is removable easy and quickly, and you can add more tube sockets to it.

The specifications

Now looking at the specifications this makes you amaze. This speaks for itself. It can even test 6C33 which is usually a problem with other testers because of the heater power. Also you can make curves of the 845 and 211 up to 700V. Not extremely much for this tube, but you come in a meaningful area of the curves. Even a GM70 with it's 20Volt, 3Amp heater is possible.

Sofia tube tester specifications
Resolution:      16 bit 
Plate voltage range:      0...700V
Plate current rage 0....250mA
Grid screen voltage range:      0...700V 
Filament voltage range:      0...24V
Filament max. supply current:      5A 
Grid voltage range:      +10... -150v 

Technical Article about the Sofia

Click here for full article

Test procedure for technicians, using an oscilloscope and test software.

Compare also with other digital tube testers

Known problems: Very few, but some are SERIOUS. .

It does not need much fantasy to understand what product must have inspired the makers of the Sofia. Here is a picture of the Tektronix 370A