Diagnosing a rectifier circuit.

[axero]

I'm trying to diagnose the rectifier circuit of a faulty switching type power supply unit but the equipment that I have at my disposal is rather limited so I would appreciate your input.

Circuit description
The rectifier circuit (DB107) is connected directly to the mains input (~240V) through a transformer which acts as a "dual" inductance (I've never seen transformers being used this way!), i.e. no transformation is taking place. On the rectified side of the circuit is about 9.4uF of capacitance and a 1 mH inductance. The impedance after these circuits is rather high, i.e. there is no short circuit here. The fuse circuit behind the rectifier is ok, i.e. not blown.

Measurements and diagnosis
When passively measuring the cut-off voltages of the diodes in the rectifier they seem ok. In other words they behave as normal diodes and don't appear to be shorted or broken. But when I turn the circuit on, I can only measure about 3V on the rectified side of the circuit.

Question
Is the rectifier circuit ok and the cheap multimeter isn't measuring properly because of the fluctuating nature of the output voltage? Or do I have a faulty rectifier circuit despite seemingly "proper" cut-off voltages (565 mV in one direction)?

[Jon86]

Can you give more detail on the circuit itself?
I doubt your multimeter is going to be able to measure it accurately if you've got any kind of AC component in it.

[axero]

Here are some schematics on a part of the device. I cannot see how you come to the conclusion that there would be any AC component on the rectified side of the circuit. Fluctuating, yes but not alternating.

Edit: A mistake was corrected in the diagram and the attached image is now updated with this correction.

[mariush]

Before the bridge rectifier there's a simple line filter (common mode choke). Then fuse, then bridge rectifier. Bridge rectifier converts ac to dc with peaks of up to 350v if you are in EU.

You have about 10uF of capacitance after the bridge rectifier. Depending on the load on the power supply, you don't have perfect dc voltage ... remember the approximation  Capacitance =  current / 2x ac frequency x Vripple 

so for let's say 10mA you'd have 0.00001F = 0.01 a / 100 Vripple  so ripple voltage will be about 0.01a / 0.001 = 10 volts.

Viper22 is an off-line SMPS primary switcher which has a mosfet integrated into it. You have here an application note which shows a board with the chip and explains how it works : http://www.st.com/web/en/resource/technical/document/application_note/CD00004467.pdf

You can google the datasheet and see the pinout and test this way if the viper chip isn't burnt out.

You should also test the feedback signal (through the optocoupler) - that tells the viper chip how to adjust its pwm to regulate the output voltage. Maybe it doesn't get any feedback due to transformer or optocoupler faulty.

[SeanB]

If the integrated converter chip has not blown itself to bits initially assume it is good. Change that 4u7 50V capacitor, along with the capacitors on the output side and look again, you probably will either get an output and it will be close to right. If you only get 3V on the capacitors on the primary side ( the 4u7 400V units) then change those, there should be around 100VDC ( 115VAC area) or around 250VDC plus (220VAC area) across each one. While checking the primary side check the 100k resistor is about the right value and the 2n2 capacitor is not faulty. Your diagram is slightly wrong, the chip connects to the cathode of the diode on the right, not the supply side.

If you have changed the capacitors and still no output change the chip and optocoupler, and the TL431 on the secondary side, then it will likely work. If not then the transformer is dead.

[axero]

It is the ripple voltage that I meant when I talked about the "fluctuating nature" of the voltage. I expect the 240 V(rms) voltage to have a peak of 340V. Since the PSU is 11W I expect the current after the rectifier to be no higher than 50 mA so I expect the ripple voltage to be 50V then (if the caps are ok) and the output voltage to be fluctuating between 340V and 290V at a frequency of 50 Hz.

So what I was wondering here is whether my cheap multimeter will measure a high voltage or get "confused" by the ripple due to it's low sample rate (rate of measurement updates per time-period on the display) and not yield a proper readout. I'd expect the multimeter to take a mean value of the measured voltage over the sampling period (time between measurement updates) which means that if that were the case, I would get a proper readout from the multimeter.

Thanks for the info about "common mode choke" filters, never heard of them before! It makes me curious as to how such a filter would react if the two input signals were out-of-phase. As for mains input, I expect it to have only one phase and neutral but for the general case ...?

I feel that I'm straying a bit out of topic here but I really appreciate your inputs! I find the circuit that I'm analyzing very interesting, so perhaps I'll start a separate thread about it. It feels like if I swap out the rectifier, Viper22, optocoupler and all the (electrolytic) caps it will fix the PSU and the replacement would not even be expensive at all. But I'll swap the components out one by one (starting with the caps as suggested) just to figure out what went wrong.

If it turns out to be the rectifier circuit, would you consider it to be wise to replace it with a "stronger" rectifier circuit to prevent such thing from happening again?

Oh, and I have corrected the attached circuit diagram in my prior post. I felt that something was wrong with the diode on the primary side of the transformer right after I made the post. It just didn't make sense to have a rectified input voltage like that. I hope it's correct now.

[megajocke]

Cheap multimeters typically use a dual-slope A/D converter so measure the mean value during the sampling period which is typically 100 ms (5 cycles at 50 Hz, 6 cycles at 60 Hz). Even if the smoothing capacitor is completely dried up you should expect to measure at least 200 V DC on the output of the recitifer. So it sounds like it (or something before it) is broken.

What the AC range of cheap multimeters actually measure can vary widely however. Some just half-wave rectify the input, measure DC and scale the display to show the correct value for a sine-wave. Those give funny results if measuring DC on an AC range, such as showing 20 V when measuring a perfectly fine 9 V battery. Reversing the probes will then show 0 V.

More useful, not mentioning true RMS types, are the ones that AC-couple, rectify, scale and show the mean value. They would at least get you in the right ballpark for the ripple amplitude on a DC supply.

[vk6zgo]

You are over-analysing this:-

Your cheap meter should be able to see around 340vdc,even if it has a fair bit of ripple on it.
When you are seeing 3v, or whatever on the bridge DC output,switch your DMM to ACV & check the Mains voltage applied to the rectifier.
If there is a high resistance connection to the fuse,this will show up as a reduced AC input to the rectifier.

Another way to do this is to remove the PSU from the Mains Supply,& check the resistance between the two Mains input points & the corresponding rectifier pins.---It should be a low value,& both legs should be similar.

If all is well there,then remove the 1mH inductor & its parallel 4K7 resistor---this will isolate everything to the right of these components on your schematic.
Recheck for 340vdc.------If there is nothing there,it looks like your bridge rectifier is faulty,or the 4.7uF cap is effectively short circuited.

When you are testing with your "cheap" meter,remember these are fairly serious voltages you are looking at,so check the ratings of the meter,& in any case,try to make any probe connections prior to turning the PSU on,rest the meter on the table rather than hold it in your hand,etc.

[axero]

To answer my question ...
I swapped a few components out without success but to conclude the thread from which I intended to find an answer to my question; yes I get a voltage readout from the multimeter after the rectified output. I think that I greatly underestimate the capabilities of my multimeter although it is not one of those luxurious Fluke 87s and it is not a True RMS either (which doesn't matter here since I'm measuring DC voltage anyway)...

What I think is faulty
I conclude that it must be the (dual primary) transformer as shown in the diagram that appears to have shorted or something.

What happened
I started hearing a sizzling noise followed by a muffled pop as soon as I plugged the thing in. I discovered that the fuse on the PCB now was blown. I tried to locate where the sizzling came from but it was hard. So I singed a piece of plastic on the transformer with my soldering iron and concluded that the smell is the same as the smell produced during test operation.

My conclusion from what happened
I think what has happened is that the insulation of the transformer gradually has given up and slowly fried either the VIPER22 or the rectifier (or both). The board is slightly scorched around the legs of the VIPER22. Now when I put the stronger rectifier (rated at 2.5A instead of 1A) it gave rise to a current that blew the fuse.

Finding replacements
I don't have a clue how to find a proper replacement transformer. I guess if the transformer was working, I could try to input some (not so high) AC voltage to the secondary coil and then measure the AC voltage of the primaries (here a True RMS meter would be handy). Perhaps I should try different voltages too and make a series of measurements to increase the precision of the estimation of the coils. But I wish to thank mariush for linking to the Viper22A paper, because the implementation on the PCB seems to be almost identical to the reference circuit design described in the paper. In that paper we have a specification for the transformer.

Here's what I'll do to fix it...
But even better yet, thanks to this paper we now know that the circuits are a part of a 12V power supply rated at 11W. So why bother with unobtainable components when we can replace the whole thing with a replacement supply?

So here's what I did; I took a 12V power supply and fed the voltage directly into where the secondary outputs go in the PCB and lo and behold, it's working!

I'll post more about this circuit in a separate thread because I have further plans and I want to reverse engineer the whole damn thing into pieces..

Edit: A Meanwell 15W (RS15-12) is in place feeding directly into the circuit (past the broken transformer) and it works fine now. I expected it to also use a VIPer22 circuit but the heart of it turns out to be a TNY279. Ironically it turns out to cost about half as much as the replacement components I bought for the troubleshooting procedure.