Mettler Toledo AT200 Analytical Balance: PSU repair

[nixxon]

A couple of weeks ago, I got a 15-20 year old Mettler Toledo AT200 Analytical Balance (0.0001 g resolution). It seemed to work well for a couple of weeks. While everything was fine, I did notice that the separate PSU got a little warm while connected to the power (IR thermometer read an idling temperature of 36-37 degrees C on top of the PSU metal cabinet). Then a few days ago, the balance suddenly went dead as the automatic sliding doors were closing.

The power plug (primary side of the toroidal transformer inside the PSU) shows an open circuit, and therefore I guess the transformer must be considered dead. There are 4 secondary output wire pairs from the transformer that seem ok. The PSU also has 4 different outputs in one custom and 10-pin plug. (Sec.1: 12.8VDC@55mA, Sec.2: 6.5VDC@350mA, Sec.3: 10.8VDC@65mA and Sec.4: 16.7VDC@55mA).

As I looked closer at the PSU and the PCBA inside, it looked to me that it should be possible to replace the broken toroidal transformer power with something else. As all 4 transformer outputs go into different bridge rectifiers, I figured I could even use some DC source at appropriate voltage level(s) without being concerned about polarity. Each of the 4 voltages are regulated by a separate LM317T (SOT-220).

I connected my (0-15VDC) desktop power supply to the 6.5VDC circuit (the circuit with a LM317T connected to a 240R and a 2k2 resistor). It worked perfectly and gave a steady 6.53VDC output when supplied with an input of at least ~9.7VDC. All the 4 circuits was tested the same way, and they all seemed to work perfectly.

The problem occured when I connected and powered all the 4 inputs at once. Then multiple outputs were low, reading 2-3 volts. I figured out that I could power 3 outputs at the same time, but when I add power to the LM317T labeled "IC2" (the 6.5V one that can handle the most current), the PSU fails. As the IC3 and IC4 circuits (the lower part of the schematic) does not connect to IC1 or IC2 circuits, the IC3 and IC4 traces and components can not be causing the problem.

I have tried my best to draw a schematic of the circuit, using EasyEDA. Please see the enclosed file. I have mostly copied the physical component layout of the PCBA for easy recognition. When "CONNECTOR1 - AC INPUTS" pins 3 and 4 are powered, the components connected to pin 1, "GL1 - BRIDGE" and PTC thermistor "R1 - MOV" gets way too hot. (Desktop power supply set to max 350 mA to avoid damage).

I cannot find any faulty components. I guess I could start to desolder and replace components, but then there is a risk that something may get broken. Will a trained professional, or other enthusiast, like many of the eevblog forum readers, be able to lead my attention to the component(s) that are most likely to be faulty? I would for sure appreciate any tips.

Best regards,
nixxon

[nixxon]

Some more photos

[wraper]

I connected my (15VDC) desktop power supply to the 6.5VDC circuit (the circuit with a LM317T connected to a 240R and a 2k2 resistor). It worked perfectly and gave a steady 6.53VDC output when supplied with an input of at least ~9.7VDC. All the 4 circuits was tested the same way, and they all seemed to work perfectly.

The problem occured when I connected and powered all the 4 inputs at once. Then multiple outputs were low, like 2-3 volts. I figured out that I could power 3 outputs at the same time, but when I added power to the LM317T labeled "IC2" (the 6.5V one that can handle the most current), the PSU fails.

Inputs must be insulated (as with original windings). You cannot just connect them in parallel to the same power supply.

[nixxon]

I connected my (15VDC) desktop power supply to the 6.5VDC circuit (the circuit with a LM317T connected to a 240R and a 2k2 resistor). It worked perfectly and gave a steady 6.53VDC output when supplied with an input of at least ~9.7VDC. All the 4 circuits was tested the same way, and they all seemed to work perfectly.

The problem occured when I connected and powered all the 4 inputs at once. Then multiple outputs were low, like 2-3 volts. I figured out that I could power 3 outputs at the same time, but when I added power to the LM317T labeled "IC2" (the 6.5V one that can handle the most current), the PSU fails.

Inputs must be insulated (as with original windings). You cannot just connect them in parallel to the same power supply.

I was thinking there is a fault on the PSU PCBA, because the PSU was getting pretty warm at no external load before failure (35 - 40 degrees C) and because the primary coil failed into open circuit, possibly due to a tripped thermal fuse deep inside the windings. And, not to forget, there is a soot "shadow" on the wall inside the PSU casing around the IC1 and IC2.

The first time I noticed the problem with hot components "GL1 BRIDGE" and "R1 MOV" while troubleshooting, I actually had all the four AC inputs connected to a 15VAC, 3VA wall wart. I guess the hot "R1 MOV" <--EDIT: Label changed to "R1 PTC" in later schematic thermistor saved a few components at the time.

Now the question is if I can apply DC to the AC INPUTS, or if I have to apply AC. If DC is OK as input, I am thinking of connecting a 9-volt battery to input pins 1 and 2, and 18-19 VDC (Two 9 volt batteries in series)to input pins 3 and 4. I wonder what will happen with this setup. I am very curious what I am going to measure across capacitors C6 or C2. It should be zero volts, as there is no apparent electrical path between the + and - in either circuit. Where should the common ground be drawn to form complete circuit(s)? Normally it should be drawn somewhere along the negative output traces (connected to output pins 2 and 4).

I hope I manage to figure out how this PSU is supposed to work.

[joeqsmith]

Nice balance!  I have the old AJ100.  These things can come in handy.  Good luck with your repair.

[wraper]

Where should the common ground be drawn to form complete circuit(s)?

Rails should be connected together inside the balance itself. I have 2 of such PSUs but too lazy to open them frankly. IIRC one of them has transformer overheating and needs fixing.

[Kleinstein]

The PSU circuit looks like just 4 LM317 based regulator circuits, all separate. There are two obvious mistakes in the schematics: In and Out of the LM317 are mixed up and the MOVs are likely PTC type fuses or less likely inrush current limiting NTCs. 

One can test the circuits with a DC voltage, at least for a slightly lower current. The current rating for the rectifiers might be slightly (e.g. 25%) lower this way. As it is 4 essentially identical circuit parts testing should be relatively easy.

Likely they are connected in the balance though. So powering them from one source will not work.

It is well possible that only the transformer is broken.

[PA0PBZ]

Your schematic makes no sense. the + output of GL1 goes to the input of IC1 (that's fine), but the - output of GL1 goes to C2 and C6, this must be wrong.
Also, the - output of GL2 is connected to the output of IC3, did you swap the in- and outputs?
The - output of the 4 diode bridge is als connected to C2 and C6 - wrong.
And I probably missed a few others...

[nixxon]

Thank you for all your suggestions.  I will read them carefully tomorrow as it is a little late at night right now. I did some more troubleshooting today and found two additional tracks from C6 and C10 to either end of the bridge rectifier close to IC2.

Also the traces leading to CONNECTOR1 DC OUTPUTS, pins 2 and 4 are shorted somewhere. I just cant find where they connect.

Changes are reflected on enclosed files.

Again - thanks a lot

[nixxon]

Where should the common ground be drawn to form complete circuit(s)?

Rails should be connected together inside the balance itself. ...

I measured the resistance between every pin inside the female plug that at a later time will mate with the 2 x 5 pin  "CONNECTOR 2 - DC OUTPUTS" on the PSU PCB. (the other end inserted into the balance) The pairs 1-2, 3-4, 5-6 and 7-8 all had resistance values between 150 and 800kOhms. If nothing is broken inside the balance circuit, it seems to me that the negative pins (2, 4, 5 and 8 ) on the output connector do not interconnect inside the balance to form a common ground.

Now I will try to find out why and where the traces leading to DC outputs 2 and 4 one the PSU PCB have 0.2 Ohm resistance. I was thinking about feeding some current between the two pins (2 & 4) and use a thermal imager to see where the traces heat up. I hope no componens release that special smoke...

[nixxon]

I just found the dead short inside/under "CONNECTOR2 DC INPUTS" pin2 and pin 4?!? 

I connected my "lab" power supply to pin 2 and pin 4 on the DUT DC output. I slowly increased current until I could see a temperature increase somewhere using a thermal camera. The only part/area that heated up was the area around pin 2 and 4. I slowly increased the current up to 2-3 amps (the voltage between pin2 and 4 never exceeded 0.17 V, measured at the pins). It was still only the area around the pins that got warmer, in addition to the current test leads that was connected to the pins. This told me that there was a short between pins 2 and 4.

To verify that the pins were shorted, I moved the current test lead from pin 4 to the negative side of C1 (this is the closest point with a trace to pin 4 that is easy to clip on to). As I increased the test current again, the trace leading to C1 immediately got warmer. After that I repeated this test with the current test leads connected to pin 4 and a point on the trace connected to pin 2. As I increased the test current yet again, the trace leading to pin 2 immediately got warmer. Then I knew for sure that the short was indeed between pins 2 and 4.

I will have to desolder the 10 pin connector and figure out what is going on under there...

[nixxon]

Yesterday I desoldered the 10 pin connector, and discovered that in fact there is a trace on the PCB between negative output pins 2 & 4. This means that the 6.5 V output and the 12.8 V output share a common ground.

I also double checked that the traces between components as drawn in Schematic v. 1.1 were correct. There were no error traces in the schematic v. 1.1. Enclosed is the schematic v. 1.2 where the verified traces are shown with thick lines.

I applied 10.0 VDC to input pins 3 and 4 (for the 6.5V output)and measured a few voltages between components. The measured voltages are shown on the enclosed Schematic.

The 6.5V output circuit and the 12.8V output circuit are interconnected in a very special way. When the 6.5V output circuit is powered by a 10.0VDC input to input pins 3 & 4, the output pins 1 & 2 correctly read 6.53V. At the same time output pins 3 & 4 read 7.14V. I don't understand why the circuits are designed to interfere with each other like this. If anyone would share their thoughts on how this works it would be very interesting.

Finally I powered all the 4 inputs one by one (also with opposite input polarity) by applying up to 15.4 VDC (Bench supply max voltage). All tests were good with correct output voltages. (The input voltage was too low to verify the 16.7 V output, of course). The supply displayed 0.01 A current draw during all the tests.

[nixxon]

On this picture one can see where the positive (pink) and negative (blue) traces go directly from input pins 3 and 4.

It seems I will have to triple check that there really is a trace between the negative connector of bridge rectifier "GL1" and capacitors "C2" and "C6".  <-- Yes, the trace is actually there.

EDIT: I have successfully connected 16VAC from a wall wart to input pins 1 and 2. My desktop power feeds 10 VDC into pins 3 and 4.
Output pins 1 and 2 again read 6.53VDC.
Output pins 3 and 4 read 12.74VDC. Possibly a little shy of 12.8V, due to too low an input voltage. (Voltage between LM317 (IC1) "ADJ" and "Vout" doesn't quite reach the specified 1.25V either, only 1.247)

C1 is adding the input voltages and read  30.5V.
The voltage across C2 (and C6) hasn't changed with dual inputs and still reads reads ~9.5V

I think the time has come to find some old wall warts with suited voltage outputs, connect it all and hope the balance will work again. That is unless there is something wrong with the electronics inside the balance that fried the toroidal primary coil in the first place...

[nixxon]

Your schematic makes no sense. the + output of GL1 goes to the input of IC1 (that's fine), but the - output of GL1 goes to C2 and C6, this must be wrong.
Also, the - output of GL2 is connected to the output of IC3, did you swap the in- and outputs?
The - output of the 4 diode bridge is als connected to C2 and C6 - wrong.
And I probably missed a few others...

Thank you for your points of view. After a little extra checking & measuring, it seems the schematic was just a little wrong (missing the two connections to the positive and negative outputs of the IC2 bridge rectifier (located at the schematic edge coordinates A-5). <-- After re-reading my own post, I just realized that the strikethrough part was known at the time you wrote your reply

As the critical part of the PSU circuit now works, I am about to measure the voltages again while input connectors 1-2 and 3-4 are connected to isolated power supplies. Can anyone explain why the Mettler PSU is designed as it is? Why are the IC1 and IC2 LM317-circuits mixed together as they are? Wouldn't it be better to keep them separate? Maybe it is the best way to create a common ground that is required between the two outputs inside the balance circuits?

Now I understand why C1 is rated 40 Volts, as the two input voltages add up across this capacitor. C4 is actually rated 67 Volts, so I expect to see even more voltages adding up as I look into the "South" side of the schematic (outputs 3 and 4)

[nixxon]

Where should the common ground be drawn to form complete circuit(s)?

Rails should be connected together inside the balance itself. I have 2 of such PSUs but too lazy to open them frankly. IIRC one of them has transformer overheating and needs fixing.

Are your PSU's made for Mettler Toledo balances with that strange 10-pin power plug? If you decide to troubleshoot them, please post your findings. Do you have the balances that go with the PSU's as well?

[nixxon]

I have done some voltage measurements of the IC3 voltage regulator (10.8 VDC output). It seems this voltage regulator is operating in some special kind of reversed mode. Or maybe this is a somewhat regular way of doing it? I have no idea why it is designed like this. Any thoughts? I also found out I have have mixed up the the R10 and R9 resistor values on the schematic (R10 is 120 ohms and R9 is 910 ohms)

I also found that output pin 4 (that is connected to pin 2 on the PCB) is connected to the chassis of the balance.

I tried powering the balance yesterday, and it obviously wasn't happy. The automatic doors were slowly opening and closing in a stuttering way. It seems that the voltage was too low to operate the door motor properly. The Display was also glowing a little faint. Some components in the PSU also got warm. See enclosed thermal image.

I managed to Zero the balance and weigh an item I previously have weighed to 23.1000 grams. The balance showed ~23.0998 grams, witout any warm up time or automatic 3 point linearity calibration.

[nixxon]

Here is an attempt to tidy up the PCB-based IC3 circuit posted above. Is it obvious to you guys that this circuit works?
(The voltage readings are the somewhat messy red letters and lines.)

It looks to me as a very special way of implementing the LM317 voltage regulaton.

EDIT: New image with verified voltages, connections and labels for the LM317 "IC3".
The GL2 Bridge positive pin is directly connected to the output connector pin 7 (positive).
The GL2 Bridge negative pin is directly connected to IC3 "OUT"
The IC3 "IN" is directly connected to output connector pin 8 (negative)

[Kleinstein]

The circuit looks very much like the standard circuit, just the polarity reversed - so either an LM337 or reversed lines at the meter.

[nixxon]

The circuit looks very much like the standard circuit, just the polarity reversed - so either an LM337 or reversed lines at the meter.

I have adjusted the IC3 schematic, and now it seems to be right. The strange connections are:
The GL2 Bridge positive pin is directly connected to the output connector pin 7 (positive).
The GL2 Bridge negative pin is directly connected to IC3 "OUT"
The IC3 "IN" is directly connected to output connector pin 8 (negative)

I may have to install SPICE again to prove and understand how it works.

[Kleinstein]

Really looks like LM337 in this case - for the LM317 the polarity would be wrong. The LM337 also has a "different" pinout, with the input at the center pin. It is well possible the circuit is mixed with some lm317 and some LM337.

[nixxon]

Really looks like LM337 in this case - for the LM317 the polarity would be wrong. The LM337 also has a "different" pinout, with the input at the center pin. It is well possible the circuit is mixed with some lm317 and some LM337.

You are absolutely right! 

IC3 is indeed a LM337T.

All the other IC's (1,2 and 4) are LM317T's and I didn't notice that IC3 was a different beast. 

EDIT: Finally I managed to draw a circuit that makes sense (and voltages adding up, except that "10.8V" at C15)

Thanks!

[nixxon]

Is there a good reason for using a LM337 as IC3? Why not use the LM317 as IC3 to regulate the 10.8V output, like all the other outputs? All of these IC's are floating, aren't they?

[nixxon]

This is the one, non-resettable    thermal cutoff that rendered the toroidal transformer dead  (or saved Our house and lives) 
(I would like it if these Components were resettable)

Maybe I should order a new thermal fuse and rewind the 4 secondary coils with a total of 800 windings...

[nixxon]

Finally, after almost one full year of procrastination, I took the time and started repairing and rewinding the 4 secondary windings of the toroidal transformer. A couple of afternoons later the toroidal transformer was temporarily hooked up to the rest of the transformer PCBA and ready for a test.

The AT200 balance actually works  now

For some reason, a couple of components ("IC2" LM317 and "R1" PTC) on the PCBA are still getting quite hot...