EEVblog Electronics Community Forum

Electronics => Repair => Topic started by: Radiohead on March 14, 2017, 07:48:20 pm

Title: Gossen Bench Power supply repair
Post by: Radiohead on March 14, 2017, 07:48:20 pm
Hello Folks,

On a Dutch electronics market I've picked up a Gossen 33K7EU5/2x25R1D, which is a triple power supply with 2x25V/1A and a single 7V/5A output. When I bought the power supply I knew that the display of the second 25V/1A output wasn't working, but when I tested the unit at home, the 7V/5A output only went to 1.5A output current max. The display is based on an ICl7106 and it seems that the ICL7106 is the cause of the display malfunction. The 7V/5A output might not go to it's maximum current output due to a remote sense connection missing on the remote sense terminals on the back of the unit. These are still some guesses, since I haven't found a service manual or schematic on the web to verify this and it seems that the previous owner tried to fix things aswell. I'll try to get a manual or information from the manufacturer, but if anyone has any experience, schematic or manual with this power supply it would be appreciated if you could provide it.
Title: Re: Gossen Bench Power supply repair
Post by: Radiohead on April 14, 2017, 03:29:03 pm
Hello All,

It took me some time, bit I've managed to find the defect in the third power supply output. Since there where no exploded caps, scorch marks or other easy to identify faults I had to reverse engineer the power supply to find the fault. In order to do so, I desolderd some large components and took a photo of the front and backside of the PCB with all the control opamps. With Gimp I've put the two images on top of each other and alligned them. I drew the vias and component pins first and then began drawing the traces between top and bottom. The end result looked like this:


From this colourfull spaggheti I made a schematic in Kicad. Still this didn't quite deliver the insight I had hope to aquire. I couldn't determine the fault. The power supply operated just fine from 0 to 6.8 V, but would not deliver more than 1.5A. After a lot of head scrathing I reverse enginered the darlington output stage and the rectification bridge aswell, but still the fault elluded me. Untill I got curious about the Darlington stage and why there where no emitter or base resistor for the darlington pair. After some searching I found the resistors that prevented the darlington pair from saturating on the main PCB board with the opamps, instead of solderd directly on the TO-3 packages themself. I expected some base resistors on the main power transistors in the darlington pair, but there where non, so I investigated the emitter of the transistor that was driving the other two transistors. They were connected with a wire with some non-shrunk heat-shrink over it to prevent accidental contact with other pins of the darlington configuration. I couldn't see if a resistor was hidding underneath, so I cut the heatshrink open and found a blown (nothing spectaculair though) fuse! The schematic of the darlington pair:


As can be seen, the resistor R22 was able to provide enough current for the main power transistors Q2 & Q3 with the blown fuse to give a false sense of operation, but as soon as any current would be drawn, it would go in constant current mode. So happily I soldered a wire to short-out the fuse and test the power supply. The output would now go from 0 to 7.15 Volt, but as soon as I shorted the output, the current would still be limited to 2 Amps, even when the current limiting was set to 5 Ampere out max!.

So I further reverse engineered the power supply output, and then I found it! A current foldback circuit was present over the current sense resistors! So all that trouble and waisted evenings for nothing but a bloody foldback current limiting protection circuit!  |O |O |O
Title: Re: Gossen Bench Power supply repair
Post by: Radiohead on April 14, 2017, 08:13:49 pm
The only thing left to fix of this bench power supply is the second 25V/1A output's display and that has proven to be quite more difficult than I expected. The display is used to read out the voltage and current of the output, but didn't display anything. The display is built around a ICL7106CPL, an 4049 RC oscillator for -5Volt generation and a CD4030 for selecting the decimal points for the voltage and current mode. Just like the powersupply, I've reverse engineerd the schematics with Gimp:

The previous owner had already tried to repair it, but didn't succeed. The negative supply trace was cut and after I fixed that, I found out why: A negative voltage of -2 Volt was pressent on multiple pins of the ICL7106, which was applied to the LCD display and the CD4030. Also the -5V supply wasn't stable and was around -3 Volt. Therefore I could only conclude that the ICL7106 was broken, so I cut it off and installed a new ICL7106 with a DIP socket, just in case if I need to remove it again. This improved the situation some what, now the display still din't display any value, but the negative voltage was now gone and the -5 Volt was stable.

From application note AN052 from Intersil "Tips for using Single Chip 3.5 Digit A/D Converters" I learned that if the display didn't display anything the oscillator wasn't working. With my scope I checked the OSC pins (pins 38-40) and the oscillator wasn't working at all. I checked the components, but they where all fine. I read somewhere that I could use an functiongenerator to apply a 5V clock signal to pin 40. Since I have no functiongenerator, I've used the -5V generator output pin 12 to create a clock signal on pin 40. The -5V circuit operated on 24kHz, while the OSC components are selected for a 48kHz clock. I've added an additional resistor in parralel to R3 to create a clock of ~46kHz.

Now the display shows the following:

From AN052 it can be read that when the display shows -1666 constantly, the OSC isn't working, eventhough i supply a clock of ~46 kHz. So I'm at a loss what could possibly be wrong with the display.
I've checked the following:

I don't have any other ideas as to why the display isn't working with a new ICL7106, any help or comment is appreciated![/list]
Title: Re: Gossen Bench Power supply repair
Post by: ModemHead on April 15, 2017, 02:10:42 am
I don't think your rigged clock is working since you have constant DC instead of a BP signal. If the clock was running, BP should be a slow (~50Hz or so), 50% square wave, all the time.  A DC bias on the LCD will make segments go dark, but is not a good thing for its health.

Normally the OSC3 pin swings from TEST to V+.  The signal you're injecting presumably does not, it depends on how U3 is powered.

I would recommend focusing on why the oscillator is not running the way it is supposed to. If that doesn't work, nothing works.  The oscillator is quite separate from the integrator section, so I doubt the integration cap C8 has anything to do with it.  If it gets leaky or whatever, it would cause errors in the A/D conversion.

Also, in most applications, a working 7106 (or clone chip) will regulate the COM pin at either 2.8V or 3.0V below V+.  After the clock, this is the second thing I usually check to see that a 7106 is working.
Title: Re: Gossen Bench Power supply repair
Post by: Radiohead on April 15, 2017, 09:37:54 am
BP is used as a digital ground (wrong in the schematics) for the CD4030 and the LCD display. Measured from ground the BP is at 3.6 Volt, but no PWM/clock signal is present.
The common pin is at 2.878 Volt from V= to the common pin. So that's good.
I've soldered back the cap and resistor and removed the connection from the 4049 plus the additional resistor, so it oscillates at 22kHz. The negative supply is now at -3.550V, measured from ground.
All the OSC pins are at ~4.4 Volt, not oscillating a bit, so it isn't strange that it isn't oscillating, there's no logical difference between the three pins.
The R11 resistor of the clock measures 90k in circuit from pin 39 to pin 40. C3 is propbably 100p, but i can't measure it's value in circuit.

If i put a 22k resistor in parrallel with R11 by hand, I see a small oscillation on top of the 4V, but that could be just my own interference created by the bad connection between the resistors. Should I try to desolder and replace C3, since it's value is the only unknown left?
Title: Re: Gossen Bench Power supply repair
Post by: ModemHead on April 15, 2017, 01:31:17 pm
Pin 21 is a digital ground for the 7107 LED version of the chip.  For the LCD 7106, it is the backplane signal which is connected to the common of all the LCD segments. OFF segments are driven with an in-phase copy of this signal, ON segments are driven with a 180 degree out-of-phase copy of this signal. That way an average of 0V is across each segment at all times.

As for the oscillator, with a 9V (15V max) supply across V- to V+, and a good R11 and C3 connected, it should oscillate.  I suppose some sort of major fault on one of the other pins might stop it, but those are the basics. :)  So yes, by all means confirm that C3 is good.

You could pull your wisely socketed replacement chip back out and put it on a breadboard, and test it with these minimum connections to make sure.
Title: Re: Gossen Bench Power supply repair
Post by: SeanB on April 15, 2017, 01:46:13 pm
Check Test is not connected, it should float, otherwise if it is pulled high or low it will stop the oscillator. However I would replace the oscillator parts, and check there is no other connection to them to any other places other than the 3 pins, as the oscillator will run with power connected, and only stop for leakage too great or with test pulled to a logic level.
Title: Re: Gossen Bench Power supply repair
Post by: Radiohead on May 02, 2017, 07:10:35 pm
So I found out why the display wasn't working when i breadboarded the display to test the ICL7106; I had replaced the ICL7106 with an ICL7107, which is the LED version in stead of the LCD version! So I quickly ordered the right ICL and waited for it to arrive. Once the ICL7106 arrived, I tested it first on breadboard. The readings where erratic as I didn't have the right kind of capacitors for the integration capacitor and had to use ceramic capacitors nor did it like the stray capacitance of the breadboard, but it showed that the ICL7106 was working. The ICL7107 on the display PCB was replaced with the '06 and the display came back to life with stable readings. As all my tests were done with a bench power supply, I plugged it back into the Gossen to see if it would still work.

My enthusiasm quickly died when the display once again showed "1--.-" when power was applied. So once again I took the display apart to identify what was wrong. The HEF4049, used to convert 5V to -5V had stopped to oscillate and wouldn't do so under any circumstance. I tried to find the cause of the HEF4049 failure, but all the inputs where well within limits of the ICL7106, so I guess it was a failure to ESD. The HEF4049 would be hard to replace, since it's obsolete and finding a replacement could take longer and would mean I had to order an IC somewhere, so I build a negative voltage converter with an NE555 I had in stock. The NE555 circuit is identical to the HEF4049 from my earlier post, a 24kHz square wave is being fed into C1 to generate a negative voltage.

The display was build into the Gossen power supply once more. The bipolar 555 version was sufficient for the display when the output voltage was read, but when switched to output current, the display required more current than the NE555 could deliver. I searched through my IC stock and found a CMOS variant of the 555, which has higher output currents. With the CMOS 555, the display works properly in both voltage and current display.

Now that the display of channel 2 and the output of channel 3 is working again, I could finally put the gossen back together and add it to my other bench power supplies. Behold my power supply stack:

Title: Re: Gossen Bench Power supply repair
Post by: Radiohead on July 15, 2018, 04:59:51 pm
After receiving a request from someone who is repairing a Gossen PSU for more information, here is some more information: The piece of the circuited board I used for reverse engineering and the schematic I created from the picture.
Title: Re: Gossen Bench Power supply repair
Post by: David Hess on July 15, 2018, 07:25:41 pm
The HEF4049 is just a high voltage CMOS 4049 hex inverting buffers and line driver.  Look for CD4049: (

Based on your schematic, the lower voltage 74HC4049 looks like it would work also.