REPORT ABOUT A COMPLETE RESTAURATION OF
A FUNKE W19S TUBE TESTER

Everybody  who is working with more than one tube tester, will probably have noticed that they do not show the same results.   So I decided to restore my tube testers  one by one.    This is the report of my Funke W19 restoration.

About the W19S:

The Funke W19S is along the finest tube testers I know. It has several advantages you will hardly find with other testers.  I will briefly list them here, these are as far as I know,  but I may find some more later.

Test cards. Message from the department of mistakes and errors department.

I read in the internet, that some of the test cards have mistakes. Don't worry, all testers have some errors in the tube data. With W19 I never believed it until I found one myself. I will list the corrected cards here. So please send me those if you have some!

Card 304 Now, this is a disappointment. This is no mistake, it is a piece of sh... and a big one. Obviously Funke never tested this card, or they would have found the mistake, same as I did.

It makes me remember this joke.... a guest in an expensive restaurant calls the waiter to complain he can not eat the soup. The waiter says everybody is eating this soup, so there can be nothing wrong with it. The guest replies: if you try my soup, you will understand I can not eat it. The waiter refuses. The guest insists, and the waiter looks around if nobody sees it, and agrees to try. Then he says.... "there is no spoon". The guest says: I told you so, I can not eat this soup.

Note the RES1664d is a military tube from the 1940. You know what means? I draw two conclusions from that.

  1. The German army never tested their 1664d tubes, because with this card it's impossible. I wonder how that went in reality. I really wonder.
  2. Funke only calculated the data on this card, never verified it with a real tube if their calculations were right. If they only tried, they would have noticed the missing hole, AND the short circuit that comes if you put a pin in a WRONG punched hole that I closed. Interesting it works fine now, with corrected holes. So they calculated it correct, but the conclusion is, other cards that you may use and think are good, may be in fact only calculated and never verified with a real tube as well. The proof for this way of working is found in card. Look here ar the result: Card 304

Anyway still the database they use is very good.

The restoration:

The principle of  restoration. tube electronics  is simple.  You need to decide where to put your focus on.  Is your main concern is originality,  or is it getting a precise function with best reliability?  For me it was the last.

Suppose you have a radio from 1922,   it would be no  good idea to take out the old capacitors, and replace them by modern ones. For the function that is certainly better, but not for a collector.  The collectors's challenge is to repair the old capacitors the original way, or sometimes don't repair it at all.  Leave it all rusty as is, and just look at it, and imagine all the years this equipment has seen.   The original historical tubes,  have  more value than new replacements,  even though the new tubes  perhaps work better.    So originality comes first with collector's instruments.

With an instrument,  that you intend to restore for daily use,   the story is a different one!  We want to USE this piece.  For an old, high quality tube testers there is no new replacement on the market.  So you need to restore an old tester to get a "like new" one.  The focus is going to be on using precise and reliable components.  So here it is like this:

So with this repair list in mind, I bought my W19 on Ebay.   So the only requirement was  that is was "working and complete".    When I bought it, it was supposed to be "very good" an  "like new", and I payed 670 Euro for it, from a surplus company in 2003. You don't get them for this price any more. Of course it was full of problems.  Not the kind you would  notice right away. After receiving it, I checked all functions, and many worked unsatisfactory and some not at all.  The rotary switch had problems, and it seems "Mr Clever" tried to fix it by randomly bending on the contacts. It turned out  it was only "working and complete", and not more than that.  Well at least I knew now, why the selller disposed of the tube tester. Since I knew where to start the repair, this encouraged me to keep it. After a short email with the seller, he admitted he only put in a rectifier, and test it quickly, and that was it. I think he really was only the seller. He offered me 130 Euro back, and I agreed. So the decision was quickly made, I kept it, and it was to be taken apart  COMPLETELY.   

Now the problem with taking it apart is,  that they must have thought at FUNKE this would never be necessary.  So the whole stuff is soldered in there permanently and you can NOT get the electronics out in one piece.  You have to separate the deck part from the case part.  What makes it worse, the case is used to hold everything together. Arrrgggh!  It is really a crazy job, repairing a W19.     

So what to do...  I took about 50 detailed pictures first, and with those in my PC,  I felt confident to solder the whole wire tree off.  Don't think the tree is nicely organized.  Some wire are going underneath here, and some other underneath there...   Well,  I wanted to re-solder every joint anyway.   And guess what.... three solder joints I found were bad. I could just  pull out the wire! They must have had contact still,  but I could just simply pull  the wire out! On the stabilizer tube socket, one contact was so cracked,  it was almost broken off.   This was a construction error.  Hard to describe here,  but they way they did that,  I can only say:  Of course this brakes off after many years.   It's shown in the third picture. <== I would recommend all W19 owners to check this point.

Here is how I received it.   Nice on a first look,  but full of hidden small defects, as we will see later. You can enlarge most pictures by clicking on them.


1) Bearing of rotary switch not original, and 2mm play
2) Rusty nuts, were not tight enough, but too rusty to adjust.
3) Rotary switch locking pin works not correct.
4) Ceramic Isolation of resistors covered with rust layer.
5) Rusty screws.   No problem, but not nice
6) Cracked solder joint on regulator tube.


Click this image, to enlarge!
Enlarge, and look at the transformer screws..   bad... bad... !

I made dozens of pictures like this,  to be sure I have all details for building it up again

Click this image, to enlarge!

Enlarge, and look at the bad solder joint.  This happened form putting in and out the rectifier tube. This may have troubled the technicians at the army! 


The original electrolytic capacitor was still good. Though some white stuff had leaked out,  it was still ok on a capacitor tester,  even with 160V bias.  No electrical leakage.  Anyway,  after 40 years of use,  it was taken out.  I replaced it by a Mundorf foil capacitor. It had the same dimensions. 


After removing the deck  plate,  this stayed inside. 




1) This contact is supposed to be closed in this switch position
2) Bearing has 2mm play
3) what's this V- shaped rusty hole for?

The switch bearing is probably not original.  It had 2mm play, and most contacts were  not closing reliable.   The previous technician must have had lots of trouble with it, because all contacts were bended upside to correct this, but this switch is so complicated,  that doesn't work well.  Thesis was a monkey repair.  I am SURE this is why this FUNKE W19  was sold on Ebay.


So here we go and take the whole mess apart.   Enlarge the image and you will see how badly the fingers have been bended.  I think with the results  nobody was happy.


What a mess!

Here you see the restored switch.  It runs smooth and makes a nice sound.  It feels like a very high quality switch again now.  All contacts were polished.  The not original bearing was adjusted better, and now there is no axis play any more. 


From this picture you can see  that needle is unbalanced.  When horizontal it is at zero.   Now... theoretically you can work with an unbalanced needle, as long as the meter is fully horizontal.  Since the Funke W19 has the meter mounted under an angle of some degrees,  you will get an un linearity of the scale.   I measured it is 8%  in the middle of the scale. At 45mA tube will read 49mA. I think this may be an original inaccuracy, but when doing a precision recalibrating job, this  is now unacceptable and must be fixed.

CLICK THE ABOVE IMAGE FOR FULL DETAIL OF THE METER BALANCING

Here is a warning when you are going  to try this with your own W19:    Only do it when you are really sure you can do it.  I calculated you need to apply the balance weight with an accuracy of 15  microgram.  Each piece will give one scale division.  15 microgram is very very extremly  little.  For this process, first I took away some of the original weight, which was obviously just "solder".  So now the needle had much too little balance weight   The needle fell  completely down  when held as in the picture above, and would be balanced when horizontally .   Then I applied little pieces of resin as balance weight,  with the solder iron, using a good lens to work under.   With this you reach the limit of your own accuracy, and get it as good as in the above picture, best way.  So like 2..3 divisions you can get after some practice.  I put just a fraction too much resin on it.   Then I mounted a fine needle on the solder iron tip.  With this hot needle tip I could remove microscopic fractions of resin.  It pulls out  thin hair wires like that.  After this the needle was perfectly balanced.   Enlarge  the image to see the balance weight.

So now,  after the mechanics are calibrated, we can do the electronics. First I took out the little window in the middle, polished it again, and glued it back in without any option for dirt to get inside.  That was not so well done originally and dirt particles were on the meter scale.   Note the blue resistors I put in on the LEFT. Click here to enlarge picture so you can see it better. It was some improvisation finding the right values from standard resistors, but I found them.  With those the full scale for the voltage measurement was re-calibrated. The Current measurement needed only two very small corrections.  These are the two green resistors I put in.  There is one on the top, just in the middle.  Try to find the second one yourself ; )  It is on the picture.   So far for the full scale calibration. As a reference I used my Agilent digital equipment.  

The meter is based on a 250uA full scale meter, and with an interesting array of resistors  they make it either a volt meter, or ampere meter, each with several scales.  What I would like to mention here,  is how the original calibration was (obviously) done.   They first put in the wire wound resistors,  and had an interesting way to change their values.  For making the value smaller they just pressed on a small part the wound coil, so some windings get short circuited.  Of course when going to far, you could reverse it by pulling the coil slowly.   So in a way these coils were used as adjustable resistors.  When it was done, they coils were encapsulated with some kind of resin.  I don't know exactly what material that is, but it appeared to be in fine condition still, no cracks or anything.  So the resistors that were still ok, probably will stay good  longer than I will live.  That are the series resistors of the voltmeter.   The shunt resistors for the ampere meter were made of few windings thick wire.  They just files off some thin pieces to make their value larger.  To me it looked like damage at first,  but under the microscope you could see it was done intentionally.      So what looked like messy work, when opening this meter, was in fact the original calibration work, and no mess at all. 

The meter on the right was used for the calibration of the Funke W19 meter.  It give 6 digits, and it is officially calibrated by the Agilent calibration service. So that one is  my reference.  The one on the left is uncalibrated, but you it is a pretty expensive one also.  These are available with calibration certificate too, but I don't have one. It shows 0.1%  deviation from the calibrated Agilent meter, as you can see.   So it is fine for me, without calibration.  Both meters are only three years old.

Note, the analog meter you see,  is on 250Volts (Lower scale).   This is the from Fluke power supply, that I used to generate the test voltage. 

Here is the finished W19,  in the sunshine of 2005,  and ready to be re-born.


After cleaning,  the bottom looks like this.  Note that all high impedance connections have a 100 Ohms resistor in series with it, which is not on the circuit diagram.   Any monkey mistake will blow out the resistors, and save the transformer that way.  The resistors will also prevent oscillations with power pentodes,  and protect  against short circuited tubes as well.