APC 2200RMXL half output amplitude

Anonymous user · ‎2021-06-29

I get a funky output waveform from my apc2200XL when its on battery. See attached. The amplitude is half of what it should be, and there are spiky shark tails on it.
I have checked the transistors on the big heatsinks and they seem ok (at least not shorted).

Load is a big lamp (maybe 250W) so (nearly) purely resistive. Graph shows another UPS output and the line voltage (but right now I've forgotten which one is which - doesn't really matter).

Batteries are sparkling new and fully charged. When on battery there is also a hissing noise from one (or both - hard to tell) transformers. When attached to servers/computers as load the ups complained about overload and shuts down (makes sense - the servers has auto ranging SMPS which draws twice the current at half the voltage I guess).

Any ideas? Is there somewhere I can get a service schematic for this one so I won't have to guess so much?

Anonymous user · ‎2021-06-29

I get a funky output waveform from my apc2200XL when its on battery. See attached. The amplitude is half of what it should be, and there are spiky shark tails on it.
I have checked the transistors on the big heatsinks and they seem ok (at least not shorted).

Load is a big lamp (maybe 250W) so (nearly) purely resistive. Graph shows another UPS output and the line voltage (but right now I've forgotten which one is which - doesn't really matter).

Batteries are sparkling new and fully charged. When on battery there is also a hissing noise from one (or both - hard to tell) transformers. When attached to servers/computers as load the ups complained about overload and shuts down (makes sense - the servers has auto ranging SMPS which draws twice the current at half the voltage I guess).

Any ideas? Is there somewhere I can get a service schematic for this one so I won't have to guess so much?

See Answer In Context

Anonymous user · ‎2021-06-29

This is more complex than I thought.
I powered it from two lab psu's and tried to start it up. Looks like I get bad gate signals to half of the H bridge.

Anyone got a schematic for it? PCB # is 640-0735G rev 7. Copyrighted 1996 so its only about 14 years old... :-).

Looks like its hard core level shifters built from thru-hole bips and timed with capacitors... none of theese sissy prefabricated H-bridge driver IC's...

voidstar_apc · ‎2021-06-29

Purely conjecture, but the jagged edges on your scope trace may be the UPS hitting a current limit. The trace starts getting choppy around 40V and keeps returning to 40V... so perhaps it's limiting itself at 6.25A? You could test that by using a different watt bulb.

Anonymous user · ‎2021-06-29

Since transistors dont make noise by themselves (except maybe for the final pop), I figured the Nasty Waveform in the picture with the small peaks passing trough a big coil would make the hissing noise.

Where is the inverter located? There are four heatsinks with 5 transistors on each -> 20 transistors. None of them seems to be shorted. Is this the inverter?

I'm not familiar with this ups, but isn't it supposed to re-generate the input waveform on its output (there are two types, the "on line ups"=re-generating and the "back-ups"=switch over with a relay, yes?).

I guess I have some reverse engineering ahead of me. !!!!!.

Anonymous user · ‎2021-06-29

Ah, that explains some things at least 🙂

The battery is connected directly to the transistor heatsinks. The transformers are connected in series and connected to two of the heatsinks.
I would assume that the transistors generate a 48V amplitude signal feeding the transformers that transforms it to 230VAC.

The age is... well... discontinued I think. It has been standing around and collecting dust for a while for no obvious reason. So I bought new batteries for it, but then it turned out to be a bit unreliable. So I decided to try to fix it. It's a spare ups in any case, we don't really need it right now but it simply feels bad to throw it away (especially since I just bought new batteries for it in the assumption that it worked).

Checked the capacitors and they are ok.

Anonymous user · ‎2021-06-29

I'll check it, but the thing I can see right now is that I don't get gate signals for half of the H bridge. Sounds odd to turn off half the bridge on overcurrent as this would actually increase the current (powering autoranging SMPS).

The two I get (A and D) looks fairly ok (a bit slow maybe).
See the attached picture for the working gate signals (Hbridge_A_D).

The other half of the H bridge (B and C) don't get any gate signals. I just see some small noise. This probably explains why I get half the voltage out of the transformer (half the voltage in, half the voltage out, right?).
See the attached picture for one working and two not gate signals (Hbridge_A_B_C).

Seems to be bips + RC 's doing the timing so maybe a bip is broken.

And.. oh, I did ask for a schematic already? Right.

dario_ramos_apc · ‎2021-06-29

Hi

I think you don´t see the B&C signals because you are trying to see, at the same time A, B&C. You are triggered by the A signal, your time base is set to see the 50KHz switching frequency, and the B&C signals appear 10 milliseconds latter to construct the next 50/60Hz half cycle. Try to see only crossed signals: A&D or B&C.

Although the bridge inverter is not working write I think that the two halve cycles are there. If they were not there, a very large amount of DC current was feeding the transformer, it saturates the iron, a large noise was eared and, probably, the UPS stoped with fault led lighted.

I have some experience with APC UPS problems. With the SU2200, when they become old, some problems with pre-drivers appears witch results in a power bridge malfunction and final death. I made some reverse engineering to solve this situation. I give you some tips from my experience:
- problem starts (visually) with R38 and R43, both 1Kohm 1/4W, running hot. Running hot they rise its value. Check it and change them if necessary.
- Change capacitors C34 and C35, both electrolytic 22microfards/16Volt. They do the floating power supply for driving the upper transistors A&B.

Component numbering is from normal SU2200 and not the RM model. I think is the same.
Good luck

Message was edited by: dario_ramos

Anonymous user · ‎2021-06-29

For a while there I was really hopeful, but then I measured A and B on a longer timescale.
One half is simply not there. The picture shows B and A in peak detect mode.
Good idea though. And I'll check the other components as well, thanks.

Yeah too bad APC won't let us take a peek in the schematics. For a product this old it wouldn't hurt. Or at least just the circuit around the H-bridge? Pleeease?

Ah well never mind, this is kindofa fun hobby project anyway. Normally I repair sputter power supplies and they seem to have some things in common at least. I'll do some more digging.

Anonymous user · ‎2021-06-29

Ehm. Embarrassing. The reason I could not see a waveform was that it was, in fact, not turned on. It was in some sort of stand-by mode.
When I removed the load (the transfomer) and turned it on I shortly (a second or so) see a pulse train on both pin 6 and 7 on IC14 (which is an APC IC so I dont know anything about it).
Then the UPS stopped, turned on the "overload" led and started beeping. Probably because it could not see any output.

Sooo I measured A and B on the gates. And look what I found.
B is behaving as it should. So does A BEFORE the gate resistor. But at the gate I see only some noise! As it should not draw any current from the gate I am now focusing on the A leg of the H bridge (well the D leg behaves in the same way actually).
I've found no shorted pins but I'll investigate it a bit further.

dario_ramos_apc · ‎2021-06-29

Forgive me but I don't believe in what you are seeing. Reading your time base, 40 ms/div, your picture is showing some sort of 50Hz signals????

It seems you are approaching to explode something and not to a solution.

Trying organize our minds:
1 - You said that you had an inverter problem! So, I think you are working from batteries or some kind of DC source. Don't put mains voltage at UPS input because, if you do it, you are confusing yourself. Although that UPS is not an on-line one, the inverter is always running. Without mains as a backup inverter, with mains as a battery charger.

2 - In this post you say that the UPS tries to start, you see some pulse train and, after that it stops indicating overload. I assume that you have no load at the output and, as you said in the past, you are feeding the UPS inverter with a DC power supply. If it is the case, because the inverter as an important inrush start-up current, your power supply may be protecting itself, the DC voltage goes down and ... the UPS stops. This is not so unusual. Verify if you have the same behaviour with batteries.

3 - You saw pulse trains at IC14! Yes, it is this IC the origin of the four signals to the power bridge. Pins 6&7 one side and 3&4 the other side. Pins 7&4 the upper signals and pins 6&3 the lower signals. All those signals, on those pins, are GND referenced, and you can see them simultaneously.

4 - You tried to see the driver signals near power transistors.
CAUTION:near the transistors the driver signals are referenced to different circuit points with different potential and you have to work with care to avoid electrical conflicts and to see what you want.
Near the power transistors, the driver signals are referenced to the transistor's sources. So, driving signals for the LOWER power transistors, are GND referenced and the driving signals to the UPPER power transistors are referenced to each side of the output inverter (heat sinks points where output transformer is connected). To avoid any kind of misunderstanding and electrical conflict, it is good practice to use galvanic isolated channels or see one signal at a time, with correct reference, when trying to see the UPPER driving signals.

Good luck

Anonymous user · ‎2021-06-29

Hi Dario

First of all the setup. Thanks for the pins of IC14, this helped. I am using two small lab supplies as voltage sources. Mains is not connected. I have disconnected the two transformers and also the H bridge cap bank to avoid current inrush (as you pointed out, the lab psu's current limits otherwise and the UPS dies). No risk of lightning special effects / explosions though.

Standby mode:
Measuring the pins on the IC shows a "high" frequency signal on the lower transistors (pin_3_6 standby and T_lower_standby). I guess this is to keep the transformer at a fairly known state.No current is provided to the transformers from the "batteries".

Start mode:
ok lets press the start button. A long beep, a pause, the fan starts, some signals appears and then the PSU stops with a constant beep (overload). Since the transformers are not connected I am not surprised that it complains.
The signals I see for a short while are pin3_6_oper and pin4_7_oper. They make sense in that they are exclusive, but as you say they are very slow. I hardly believe it either but this is what I'm seeing.
The upper transistors didnt fit into this message but they are in the same time scale, mutually exclusive but with a higher amplitude (around 60V actually).

The dead time between the transistors don't seem to be very long and might be a problem. But in a way the H bridge behaves as it should when I've treied it now (though it has not been on for that long).
I will add the transformer in series with a 100 ohm resistor to see what happens.

Message was edited by: onbattery

dario_ramos_apc · ‎2021-06-29

Ok! Now it's better!
We don't know where the problem is but, at minimum, we understand the picture.

Let us explain a little more to the APC guys how does his UPSs work:

Standby mode:
The high frequency signal applied to the lower transistors is correct. I think they are present to charge the floating capacitors C34 and C35 that I ask you to inspect. They have to have energy to start the bridge when necessary.

Start mode:
The signals are correct. Now I think to realize why they are so slow. The bridge control, probably, is not a fixed frequency. So, transistors are drive on and, because there is no output voltage nor over-current detection, they stay on until the end of the half cycle of the low frequency.
One thing I don't understand: In the first picture you posted your mains voltage and output inverter had an exactly 20 ms period / 50 Hz. Now, your signals period are closer to 16ms / 60 Hz???

The inverter inrush current:
Initially you have a natural inrush current to charge the large capacitors.
The inrush current that deregulates the DC voltage and makes the UPS to stop is not related with the capacitors. It has it's origin in the transformer. It is the same phenomenon why you trip a magnetic circuit braker when you connect a transformer to the mains.

Now let us see if the drive signals do something to the power stages:
Put a resistor in place of the transformer - between the two heatsinks - and inspect if you see on their terminals a low frequency square wave when you power on the UPS. Exactly in accordance with the low frequency signals you saw.
Use a resistor that could use a current near the maximum that you have (R=48/Imax).
Inspect the good "saturation" of the transistors.

Finally don't forget:
- mechanical components are a source of problems
Why not admit that you have relays with welded contacts?

Good work

Anonymous user · ‎2021-06-29

I checked the fuses and the relays at an early stage, at least I thought of that :-). Can't say if the relays are in a bad shape but at least they are not welded. I tested it a bit brutally by applying a voltage directly to the coil (I know, this may be tough for the driver but I assumed it's open collector and besides the psu was floating relative to UPS gnd) and they switched as they should. Checked with a beeper as well and the NO beep(on) /don't beep(off) as they should. SPDT models shift connections as they should too.
I could have missed one of course.

I have a feeling some thing has changed though. Earlier in my text I had, for instance, no good gate signals but now I have. It may be one of those irritating errors that come and go or a bad solder point or some bad connection. I HAVE been poking around a lot after all. Maybe it'll work now if I plug it in. We will see soon...

I have measured the H bridge output. Looks nice. I loaded it with a 100R power resistor in series with the transformers (see attached setup2). I got the signals as in the picture "Transformer_in" which looks just as I'd expect it (slow time base but you had an explanation to it so I'll stop thinking about that).

What mystifies me a bit is why I can't run the whole UPS in the current setup. I mean, sure I have not connected any batteries but the lab power supplies should be able to deliver about 2.5A. And with the series resistor they don't go into constant current mode. It's the UPS stopping after a second or two - beeping and flashing the lowest battery bargraph led. All cables are connected (except I have no real load or mains power) but should it not start up and generate the 230VAC output from the lab supplies?

I think I'm about to move on to the scary secondary high voltage circuit.
A short at the output (varistor or whatever) could cause problems... but then I'd expect a fuse to blow. And there isn't really much there on the secondary side that could break... Or is it...? 🙂

dario_ramos_apc · ‎2021-06-29

Hi

You are getting crazy with your UPS ...
If you had already checked the relays ... ok!

Let me try to clarify your doubts.You say:

*1 - What mystifies me a bit is why I can't run the whole UPS in the current setup*
Forget! It is impossible that the UPS could run in that situation!
The major evidence is that the transformer's primary resistance is a short circuit compared with your resistance. So, the voltage at transformer's primary is near zero then you can't have output voltage.
I don't know but let me guess. The transformers of your UPS can have a magnetizing current - no load at output - of 4~6 Amp. This current is essentially reactive - don't corresponds to active power - and is supported by the large electrolytic capacitors and the battery.
Because the active power necessary to the transformer is lower than 2.5A*48V=120W, the UPS would run if you get the limitation resistor out! The answer is YES IF THE INITIAL INRUSH CURRENT DOESN'T DEREGULATE THE VOLTAGE.
It was possible to run completely your UPS on bench and with your PSU with little transformers with the same voltages.

*2 - I have a feeling some thing has changed though*
It was not the first time and wouldn't be the last that by magic effects the equipment starts running and we have to wait the problem becomes more clear later.
One of this magic effects is the welding relays becoming apparently good because of the mechanic vibration we stress them when desassembling.
Check if it is true. Assemble the UPS and try it.
Before do it, check the following:
- At the output of the inverter, in parallel with the transformer's secondary, you have large polyester capacitors. I don't know the value but you easily see them because they are large (3 or 4 micro/400V). Take the out and confirm it's value. If they lost capacity the inverter can oscillate the output.

After assembled the UPS and if it doesn't run:
- Inspect again - now with the real thing - the output signals of the bridge. The low frequency is no more present I guess.
- See the differential voltage of each primary transformer and compare them. They have to be more or less equal.
- The secondary of the transformers connects to the board with two connectors that are pin to pin shorted (please confirm) putting the secondaries in parallel. Try to start with one connector at a time.

Try hard and good luck

Message was edited by: dario_ramos

Anonymous user · ‎2021-06-29

Hi again.
I actually measured theoutput and for the few seconds the UPS was "up" I got half the expected voltage out. Have no graphs this time - I forgot the files on the office computer.
I sort of understand why the connectors are in parallel on the secondary side, but why is there four output nodes? I haven't reverse engineered the output circuitry yet and maybe it's easier to simply ask - you seem to know these old beasts by the back of your hand. 🙂

Also I got some disturbing news. This UPS originates from a company we (well, a subsidary) had a contract with long ago. A former manager told me that "oh, there was an UPS we discovered wouldn't run well on battery. It was like that from when it was delivered but we didn't find out until it was too old for warranty repair and so we just shelfed it. It might be that one you are having on the workbench".

If he is right I'm trying to find a fabrication error which is a bit tough... I am still suspecting one of the transformers (at least I have not verified it's NOT one of the transformer). Maybe I can disconnect them and try them from a signal generator to see if they behave the same. If they don't... well... then I'm in trouble.

dario_ramos_apc · ‎2021-06-29

Hi Patrik

Now I think you are testing with the UPS completely assembled, with all components and from battery!
You say that the UPS stays on a few seconds, and you see half amplitude output.
Clarify this please:
After these seconds the UPS beeps a lot and signalizes OVERLOAD.
You have no load at output.

As another guy said, your images on the first post denunciates an inverter over current. Why does it exist we don't know. Until this moment the electronic control didn't gave us any evidence.

I asked you to try run the UPS with only one of the secondary connectors connected at a time. Did you?
The transformers have to be connected with correct phases. If they are not correctly connected these strange phenomenons could appear.

How electric part is organized and why 4 poles at transformer secondary connection:
Imagine you develop an interactive UPS with that topology but 24V battery UPS 1100VA. The transformer you use here have, for instance, this main voltages: 20Volt/230Volt.
After this, imagine you develop another model doubling the power. Because power doubles so does the current if you stay with same voltages. One solution to maintain the primary current is double the battery voltage. So, the new transformer has this voltages: 40Volt/230Volt. The primary current is the same as 1100VA/24V and secondary current doubles.
This evolution is compatible with:
- Put the two primary transformers of 1100VA/24V in series to have the primary transformer of the 2200VA/48V UPS. The same current and double voltage.
- Put the two secondary transformers of 1100VA/24V in parallel to have the secondary transformer of the 2200VA/48V UPS. The same voltage and double current.
This explains the mechanism of serial primary and parallel secondary. Naturally you have to chose correct phases!
The four poles are related with stabilization.
The secondary is formed with two windings (two windings = four poles):
- The main winding 230V witch I was talking about in the upper lines.
- The stabilization winding +/-18% of the main winding: +/- 40 Volt
When you have input voltage connected, the output could be directly the input or the input +/-18%. The plus or minus is done putting the regulating winding in phase or opposite phase with the main winding.
The way the stabilization winding is inserted in electrical circuit is done with a part of the relays on the board.
The relays on the board I don't remember exactly if they are 4 or 5. Their main functions are:
- 1 x DPDT - input back-feed power protection. Connects mains phase and neutral to the UPS.
- 1 x SPST - Input relay. Connects mains phase to electrical stabilization.
- 2 x SPST - stabilization winding connection to the main winding. (+/-18%)
- 1 x SPST - output relay. Connects or disconnects power to the output.

This explain your doubts an could be here your problem.
Use this information to realize if something doesn't match.

Message was edited by: dario_ramos

Message was edited by: dario_ramos

Anonymous user · ‎2021-06-29

Hi Dario

Its late and so I feel a bit stupid right now. I'll simply answer your questions and try to read/understand your answers tomorrow...
First of all the secondary measurements. This was still with the lab psu's, and the UPS turns off after a few seconds (starts to beep).
Hey maybe I can make a small movie and post here with my cell phone. I'll try that tomorrow.
This attached picture may or may not say something to you. It's one probe on the P wire and one on the N. Both probes are referenced to chassis ground.

Anyway...

Right now I have NOT assembled the UPS (and with assembled I more or less mean soldered the nasty big capacitrors back and put batteries in).
I don't have any load on it either. I still have the 100R resistor in series with pri windings.

I'll try to put the batteries in tomorrow (too tired now to be tap dancing with high voltage) but before THAT I'll check the transformers separately. Still haven't done that. But the LCR measured that they had the same inductances for the respective winding pairs, so maybe they are ok after all.

Yes both transformers are in parallel on the secondary and in series on the primary side. On the secondary side the colors are connected to the same color (blue black yellow white).
I disconnected one of the transformers on the secondary side but I don't remember the result now so it can't have been very exciting. Maybe it's better to try it with batteries connected as you say.

Thanks again for your time.

dario_ramos_apc · ‎2021-06-29

Hi Patrik

The images you have here are not understandable because, when you are seeing main signals, the reference to be used is mains NEUTRAL.

PAY ATTENTION
Your oscilloscope chassis and naturally the reference alligator of your voltage probes are connected to protective ground. So, if you connect them - the alligator - directly to neutral, two things could happen:
- the neutral of the ups is effectively the mains neutral and you short-circuit neutral to ground tripping differential circuit breakers. The over-current through the probe is originated by your neutral voltage.
- the neutral of the ups is the mains live and you short-circuit live to ground tripping differential and/or magnetic circuit breakers. The over-current through the probe is originated by your live voltage. This is possible if in your country, as it is in mine, you use a schuko type plug/socket to connect to mains and, in consequence, you could reverse polarity.

Naturally if your oscilloscope is battery operated this discourse doesn't make any sense.

HOW DO I DO
My better solution is to isolate the oscilloscope with a 1:1 transformer. Something like 200 VA is more than sufficient.
If I have no transformer I connect oscilloscope to mains with a special power cord that has a cut ground wire.
The last ingredients are: opened eyes and no distractions. If you forget this ingredients you could use your UPS for that it was not conceived for: electro-stimulation.

Good luck

Anonymous user · ‎2021-06-29

I get a funky output waveform from my apc2200XL when its on battery. See attached. The amplitude is half of what it should be, and there are spiky shark tails on it.
I have checked the transistors on the big heatsinks and they seem ok (at least not shorted).

Load is a big lamp (maybe 250W) so (nearly) purely resistive. Graph shows another UPS output and the line voltage (but right now I've forgotten which one is which - doesn't really matter).

Batteries are sparkling new and fully charged. When on battery there is also a hissing noise from one (or both - hard to tell) transformers. When attached to servers/computers as load the ups complained about overload and shuts down (makes sense - the servers has auto ranging SMPS which draws twice the current at half the voltage I guess).

Any ideas? Is there somewhere I can get a service schematic for this one so I won't have to guess so much?