Gossen T4 K80 R1 - Repair

[dl6lr]

We don't have a dedicated repair category, do we?

Well, I got a Gossen Konstanter T4 K80 R1 a while ago. Power supply, 0..80V, 0..1A with CV and CC modes. Someone modified the original item which has two separate rheostats for coarse and fine voltage and current to 10-turns-pots.
Unit worked apparently well after I cleaned and tested it.

20220816_191823.jpg

Now I used the device to load a battery: LiIon 10S5P 36V/17.5Ah. I spent a diode as a safety against back fire and a small resistor of 3R6. Settled it to 41V and 1A and let it cook. All went well, until after some hours it began to smell. The current was on "11" and the transformer very well warm, so I shut it down.

Strange phenomena: Voltage regulation was well working, so no shorted pass element. No response to the current pot. After letting it cool down, I tested again, and surprise: All is fine and working. So I tested again with load and another strange issue manifested: After a while the current regulation stops working, as before. In that situation, the output voltage rises. If I turn CC down a little bit, is starts to switch between CC and CV mode, with CV being increased a little bit. The meter needle is bumping around, as does the current.

20220816_191848.jpg

20220816_191749.jpg

So this must be some kind of thermal issue. I have to take it apart again and try to locate the issue.

Stay tuned...

[Zenith]

We don't have a dedicated repair category, do we?

It mostly works with tags.

Do you have the service manual for it? That's always a good starting point. Just a schematic is good, but it's far better to have the manual with theory of operation, parts list and set up instructions.

I've never seen much Gossen equipment in the UK. The name mainly comes to mind in connection with light meters, from the days when photographers used them.

[mnementh]

Have you tried with a pure resistive load just to see if it makes a difference in the behavior?

mnem

[dl6lr]

We don't have a dedicated repair category, do we?

It mostly works with tags.

Do you have the service manual for it? That's always a good starting point. Just a schematic is good, but it's far better to have the manual with theory of operation, parts list and set up instructions.

I've never seen much Gossen equipment in the UK. The name mainly comes to mind in connection with light meters, from the days when photographers used them.

Unfortunately I have a bad copy of the schematic only, and it's from a very similar device, but only voltage and current capabilities vary. No SM. I had a glance at it and the CC feature is a little bit unfamiliar to me. But I think I can trace it down, but have to find some time. It's a little bit nasty to operate the device without enclosure, as the big heat sink is at about +100V.
Note to self: Beware not to lick it while the PS is on

[dl6lr]

Small update with the schematics. It is from a similar device, but the board numbers are identical, so the boards described by the schematics are the same as in my device. Unfortunately no time to start with the repair yet.

The ZIP file contains a PDF. All my efforts to stitch it to one page reduces the readability further. It seems to be nearly impossible to extract the images from the PDF as a pixel-by-pixel copy to any image manipulation program. I tested GIMP as well as PDF XChange Pro to extract the images, but they all will render it to any given resolution with blurring the original pixels. Even a 1200dpi rendering gives more artefacts while creating tremendous large images. If anyone knows a better approach...

I am thinking of redrawing the schematics.

Regards, Bernd

[tggzzz]

Many programs that embed images in pdf files also reduce the resolution of the image, which is a disaster for schematics.

I've found that linux utilities make it relatively easy to reduce the bits/pixel and convert between images and pdf files:

• scan a document into one jpg per page (300dpi colour)
• convert each jpg to limited grayscale tiffs (different for text or for images)
• convert those tiffs to one pdf file for each page
• concatenate those pdfs into a single pdf.

I presume there are equivalent workflows on Windows.

That gives a single page schematic occupying ~70kB (e.g. x41.pdf), and a single page photo maybe 450kB (e.g. x2.pdf).

I'm not quite sure what your original document or image is, but why not try posting that here and we'll see what we can do. The nearer to the source scan the better, of course.

[dl6lr]

Well, I already attached the "original" scan that I have...

But your example displays the issue quite right: If I zoom in and make a screen capture, the X41.PDF looks like this:

X41_extract.PNG

If I now import the PDF to GIMP and let it render it with 600dpi, I get this:

X41_GIMP.PNG

Now reducing to bw, I get this:

X41_GIMP_BW.PNG

You see, it is not identical to the original. In my case with the noisy scan it is much worse. That is why I asked for a solution to extract the images from the PDF pixel-by-pixel.

[tggzzz]

Perfect? No.

Good enough for most use? Yes.

W.r.t. what's in the pdfs you supplied: GIGO

[dl6lr]

Well, I haven't found the time to even start with the repair, but I took photos and combined them with the available schematic.

Gossen_Platine.jpg

Gossen_Platinenunterseite_bestueckung.jpg

It has been repaired in the past and it has been improved, surely to make it immune against some EMC. The inductor is not present in the schematics and it has at least two more capacitors around the transistor in the regulation path.

Some components have different values, probably due to being a model with different voltage range. Will redraw the schematics...

[dl6lr]

It has been quite some time since my last post. I was working on the repair. But, I wasn't able to fix it, as I couldn't reproduce the issue again.
What I have done:

I deliberatly compared the schematics with my boards and found some differences. I updated my schematics with some measurements and component values.

I arranged the open PSU with an isolation transformer and made me comfortable with multimeter, paper, pen and drinks.
Assuming this is a thermal issue, I loaded the PSU with 10R at 1A (10V out) and let it cook. Nothing bad happened, I let it cook for about 4h, consuming my beverages. The pass transistors stayed warm and balanced (not perfectly, but I will not change it at the moment). The IR camery did not show any hot spots despite those that are expected (resistors in the PSU section in front of zener shunt regulation). The pass elements get a good 100V input voltage, so there is plenty of room even at 80V output. All other transistors stayed fairly cool.
The circuit uses four more voltages, one from a winding with abt. 10V AC to feed the base of the pass elements, this voltage sits on top of the positive output voltage. Two windings with 76V AC provide abt. 100V DC which are then shunt regulated to -6.5V and +12V/+7.5V for the error amplifiers. These voltages sit upon the positive output terminal too.

So I was a little disappointed and used the original battery charging configuration the next day. Fortunately, I have two more eBike batteries to charge.
So I settled it to 42.5V, connected the battery with a diode and set it to 1A. This was the configuration that was failing prior, so I let it charge the battery for 6h. No problems at all, current regulation was fine. So I put everything back into the enclosure.

The next day I charged the battery for another 6h at 1A with no problems, ommitting the isolation transformer and having it cook inside its enclosure. Nothing, nada, zero issues. I am clueless. Self healing.

[mnementh]

So... no flames...?

How disappointing.

mnem

[Zenith]

Transient equipment defects are notoriously frustrating. You investigate but don't find a fault. The equipment] starts to work, which is great, but you don't trust it because you don't know why it started to work or stopped working.

Such faults can be caused by;

bad solder joints, which look fine,

resistors or other components, which have become sensitive to temperature or vibration,

electrolytic capacitors, which usually dry out with decreased capacitance and increased ESR but can behave strangely,

dirty switch contacts.

Perhaps it's not surprising that the idea of Gremlins came about.

[mnementh]

It has been quite some time since my last post. I was working on the repair. But, I wasn't able to fix it, as I couldn't reproduce the issue again.
What I have done:

I deliberatly compared the schematics with my boards and found some differences. I updated my schematics with some measurements and component values.

I arranged the open PSU with an isolation transformer and made me comfortable with multimeter, paper, pen and drinks.
Assuming this is a thermal issue, I loaded the PSU with 10R at 1A (10V out) and let it cook. Nothing bad happened, I let it cook for about 4h, consuming my beverages. The pass transistors stayed warm and balanced (not perfectly, but I will not change it at the moment). The IR camery did not show any hot spots despite those that are expected (resistors in the PSU section in front of zener shunt regulation). The pass elements get a good 100V input voltage, so there is plenty of room even at 80V output. All other transistors stayed fairly cool.
The circuit uses four more voltages, one from a winding with abt. 10V AC to feed the base of the pass elements, this voltage sits on top of the positive output voltage. Two windings with 76V AC provide abt. 100V DC which are then shunt regulated to -6.5V and +12V/+7.5V for the error amplifiers. These voltages sit upon the positive output terminal too.

So I was a little disappointed and used the original battery charging configuration the next day. Fortunately, I have two more eBike batteries to charge.
So I settled it to 42.5V, connected the battery with a diode and set it to 1A. This was the configuration that was failing prior, so I let it charge the battery for 6h. No problems at all, current regulation was fine. So I put everything back into the enclosure.

The next day I charged the battery for another 6h at 1A with no problems, ommitting the isolation transformer and having it cook inside its enclosure. Nothing, nada, zero issues. I am clueless. Self healing.

Chafed wire that you moved while puttering around in there?

mnem

at the wall.JPG

[vk6zgo]

Chafed wire that you moved while puttering around in there?

mnem
at the wall.JPG

Marconi were always fond of including a chafed wire somewhere, or from time to time, a lost 2BA nut to act as a nice fault-causing "easter egg" down the track---why shouldn't Gossen join the fun?

[dl6lr]

Transient equipment defects are notoriously frustrating. You investigate but don't find a fault. The equipment] starts to work, which is great, but you don't trust it because you don't know why it started to work or stopped working.

Yes, and guess what: The fault is back. Of course I have used the PSU several times in between without issues. So it has to go back to the bench for more investigations.

[mnementh]

Murphy's a bitch... and she has puppies.

mnem

"number 9... number 9... number 9... number 9... number 9..." ~the Beatles

[dl6lr]

Looks like I found the problem. The issue got worse the last days and I could reproduce it easily. After measuring around I suspected two foil capacitors with 100nF in the loop filtering. They measure OK, but one has slight leakage and measures around several Megohms. I replaced them with new ceramics to find the regulation now oscillating at several kHz
After trying different caps I got to a stage, where the loop was stable, but starts to oscillate for short periods with variations of the current. I now have a combination of several caps where the loop stays stable.

Will clean up the part of the schematic and show it shortly.

[Zenith]

I'm glad you found the problem and fixed it, especially since you only had a manual for a similar model to use and the problem was intermittent.

You feel so silly when you eventually track down the source of a problem like that and embarrassed about the time it took to find it. There's no justification for that. You've got it working and learned something along the way.

I had a recent problem with a 6V supply using a three terminal regulator to replace the original 6V supply in an instrument. I'd put a 0.01µF ceramic capacitor from output to ground, but sometimes it went into oscillation at several MHz. I tried adding other capacitors, but the problem persisted. In the end I made up a new add on board with a ceramic and electrolytic instead of just a ceramic, and the problem never reappeared. Bad regulator? Something else? Who cares? It works and life's too short for a root cause analysis of such a problem.

[tggzzz]

I'm glad you found the problem and fixed it, especially since you only had a manual for a similar model to use and the problem was intermittent.

You feel so silly when you eventually track down the source of a problem like that and embarrassed about the time it took to find it. There's no justification for that. You've got it working and learned something along the way.

I had a recent problem with a 6V supply using a three terminal regulator to replace the original 6V supply in an instrument. I'd put a 0.01µF ceramic capacitor from output to ground, but sometimes it went into oscillation at several MHz. I tried adding other capacitors, but the problem persisted. In the end I made up a new add on board with a ceramic and electrolytic instead of just a ceramic, and the problem never reappeared. Bad regulator? Something else? Who cares? It works and life's too short for a root cause analysis of such a problem.

LDOs can be tricky like that; RTFM is in order.

From TAoE p617

F. LDO Stability
It’s worth repeating that low-dropout regulators can be
quite fussy about bypassing (see the comments in §9.3.7)
and that there are large differences among different types.
For example, the LDO Selection Guide from Texas Instru-
ments includes a C out column, whose entries range from
“No Cap” to “100 μ F tantalum.” The symptoms of insta-
bility may manifest themselves as incorrect, or even zero,
output voltage. The latter symptom flummoxed one of our
students, who replaced an LP2950 (fixed 5 V LDO) sev-
eral times before the real culprit was identified: he used
a 0.1 μF ceramic bypass capacitor, which is less than the
specified 1 μ F minimum, and also whose equivalent se-
ries resistance (ESR; see §1x.3) is too low, a hazard dis-
cussed in the Application Hints section of the regulator’s
datasheet.²⁵ A more serious symptom of oscillation is out-
put overvoltage: we had a circuit with an LM2940 LDO
(+5 V, 1 A) that was bypassed in error with 0.22 μ F (in-
stead of 22 μ F); its internal oscillation caused the measured
dc output to go to 7.5 V!

...

²⁵
In these words: “Ceramic capacitors whose value is greater than
1000 pF should not be connected directly from the LP2951 output to
ground. Ceramic capacitors typically have ESR values in the range of 5
to 10 mΩ, a value below the lower limit for stable operation (see curve
Output Capacitor ESR Range). The reason for the lower ESR limit is
that the loop compensation of the part relies on the ESR of the output
capacitor to provide the zero that gives the added phase lead. The ESR
of ceramic capacitors is so low that this phase lead does not occur, sig-
nificantly reducing phase margin. A ceramic output capacitor can be
used if a series resistance is added (recommended value of resistance
about 0.1 Ω to 2 Ω).”

[Zenith]

Worth remembering, but it was a standard 7806, not LDO, and the unregulated supply was at I recall 16V, with maybe a volt of ripple. The 7812, also standard, not LDO, showed no hint of a problem in either case.

In the early days 78 series three terminal regulators were notoriously prone to go into oscillation. Having used many of them over the years, it's the first time I've had a problem like that with one.

[dl6lr]

I promised to update the schematics. Here is a zipped PDF with the schematics I captured. I suppose there are still minor errors and I didn't validate every components value.

GossenT4K80R1.zip

The circuit is a little tricky to measure, as there are two error amplifiers with supplies floating on the positive output (with slightly different voltages as of the current measurement shunt), one for current and one for driving the pass elements.

The issue was in the current loop filtering, consisting of R404,C401/R412,R403,C402/R405.

Loopfilter.PNG

The crapacitors seem to have erratic isolation, although I was not able to measure it. They are indicated as 47nF caps, but installed were 100nF foils. With new caps the loop started oscillating at 5kHz, I increased the caps to 100nF and the oscillating frequency was now 8.5kHz. I tried different caps and the best stability so far was found to be 100nF foil with 47nF ceramics in parrallel.

I will later do a stabilty test with pulsed loads from my electronic load to check for any strange effects and the time constants of regulation.

[mnementh]

I vaguely remember reading aboot similar issues with the frontend on umm... Was it the DS1054Zed?

I think it were noise in the attenuator section traced out by a eevBlog member to oscillating SMD 7805s (or a related family of linear regulators) with MMLCCs placed across the rails rather than electrolytic + small ceramic as recommended in the application notes. I think the fellow substituted SMD tants and the problem was neatly resolved.

As I recall it, the problem was that the extremely low ESR of MMLCCs cause the short-circuit error amplifier to oscillate, so pretty much endemic to the basic design.

This was a problem that made it through Ifni knows how many steps of design and several production revisions without ever being caught by the manufacturer.

mnem
*pinnnnnnnnnnnnnnnnnnnnnnnnnnnnng!*