Mean Well RD-125-1224 problem

[neoone]

Hello!

For introduction - I needed some 12V power supply capable of providing 5-10A for checking a project I'm working on (DC-DC converter, it's unimportant for that topic). I was tired of using old PC power supply and i remembered that I had Mean Well switching power supply with 12V and 24V output somewhere in the closet. I have used very similar models with 5V output and for hundreds of them I remember more or less 2 that had some malfunction. Any way I've decided to put this MW RD-125-1224 (12V/7A, 24V/5A) plus MW RS-150-5 (5V/26A) in one big case and also add some current and power measurement in the future creating a power supply for high currents.

So everything was set and nicely wired, power on... and WTF? there's something wrong with RD-125-1224. Somehow I didn't think about checking it first but it was brand new as I remember used only few times. So the problem is that the supply is switching on and off in about 1.2s period and is making "squeaking" sound. I attach some timings. The supply controller is NCP1203 which I'm not very familiar with. It seams that the condition is identical to figure 15 (page 9) - short circuit but I'm pretty sure there is no short circuit at the output. Resistance measured at 24V output is 980 ohms and 450 ohms at 12V output probably due to 2 power resistor that I think are meant to be minimum load. Any ideas what should I check? Is it something else than the IC ? It seams that for a short time (~300ms) output voltage is stable and then some protection is initiated or something.
gate of the transistor:

Opto coupler (12V I think):

24V output:

gate + VCC (IC)

[DrGeoff]

Is there an overvoltage detector on the output rail?
It may be set too low and crowbaring the supply.

[Alex]

Hi neoone.

Looking at the waveforms you posted and the datasheet of the controller:

The time the output is at the correct voltage (24V from the waveform) is actually the 'Leading Edge Blanking' time as described in the datasheet. This is a fancy phrase for 'delay' and the datasheet specifies 230ns for this and 90-160ns for the actual overcurrent signal to switch off the transistor. So this is the delay you are seeing.

Also, on the last waveform, the low supply for the controller (Vcc) has a change of slope at the undervoltage lockout point, about 7.5V.

Therefore I am confident the controller thinks there is a short-circuit condition.

Can you please probe the voltage on Pin 3 (Current Sense) of the controller and post it here, it should not be exceeding 1V peak.

Alex

[neoone]

this is datasheet for that supply and on page 4 you can see O.V.P. Don't know yet how it's build and why on earth could it be set to low.

About the delay Alex, I've noticed it some time after maybe not in details but the last waveform I took after 30minut from the first three and didn't change the text. Thanks for the detailed explanation though. As you said it seams that controller "thinks" there's short circuit. I'll try to post that pin 3 waveform soon although I think I've already tried to probe it and the voltage was very low.

added:
Ok, here are waveforms with VCC and Pin 3 voltage. It shoots up to 900mV. I'm not 100% sure but I think that the shunt resistor is 3.3 ohm.




added:
hmm it's hard to catch but there seams to be some shots exceeding 1V

But one rare peak like that wouldn't case that problem, could it ?

[Alex]

The overvoltage protection is triggered by output channel 1, 12V in this case. The datasheet of the PSU specifies an overvoltage limit oin the range of CH1: 13.8 ~ 16.2V, so unless there is a faulty component(s), there is no way the output overvoltage protection could be set wrong.

Have a look at pin 3, and it would be very helpful if you can also probe Pin 2 on the same screen or as a separate measurement with the same scope settings.

Alex

[Alex]

Ah, thats it then. That pulse (how long is it?) triggers a shutdown cycle. The shutdown threshold voltage can be lower than 1V if the feedback voltage is low. You can post a capture of the FB pin (pin 2) if possible.

Is your power supply jumper set correctly (230/115VAC) ?
Are you sure there are no shorts on the outputs ?

[Alex]

I am not sure the captured waveform is representative of the true signal, possibly a scope limitation.

Is the output voltage of Ch1 (12V) within specifications?
Is your power supply jumper set correctly (230/115VAC) ?

[neoone]

I am not sure the captured waveform is representative of the true signal, possibly a scope limitation.

yes, I though so too.

Power jumper is set on 230V as it should. Few hours ago I've checked 12v and 24V regulation by built in pot and it worked fine (12V output was able to be set somewhere between 11.3 and 12.8 or slightly less).

added:
Pin 2 (CH1) and 3 (CH2)

[Psi]

you could try to catch the signal with only one channel enabled, from memory that would boost the sampling to 1G instead of 500M per channel.
It might give a better view of the pulse.

[Alex]

Thanks for all the waveforms.

Here is your last capture rotated 180deg and mirrored. Channel 1 should now reflect the output voltage.



Clearly, the steady state peaks match and the controller issues gate pulses that can be seen by the overload (current sense) signal on Ch2.

But then the the feedback signal is lost (goes to max which means that the clamping to ground from the optocoupler is lost), and it starts to slowly recover. This makes me think that there is a fault with the overvoltage protection. The falling Vcc is triggering the undervoltage lockout circuit in the controller causing these weird waveforms.

In fact, if you look at the waveform, it looks like a square wave (output working/notworking) with an RC ramp superimposed. The RC ramp could be caused by the output overvoltage protection.

[neoone]

hmm still that doesn't explain what's wrong. Tomorrow (it's now 2.42 AM in Poland) maybe I'll try to draw schematic of OVP on both sides of opto coupler

[neoone]

this is what I managed to draw from PCB, hope it's accurate. I didn't analyze it yet. Components with "NC" are not on the board and "??" means that I don't know or am not sure what is the value.

[Alex]

Thumbs up for your patience and persistence.

The circuit around the TL431 looks like a driver for the LED in U2. A fraction of the output (R64,65 SVR1) is sent to the reference pin of the TL431 which then drives the LED to whatever is required for the reference voltage to be 2.5V. This is a feedback loop for the output voltage with many of those passives for loop compensation.  But yes, it will act as an OVP too as the supply to the LED is the 12V rail.

i would first look at the the thyristor circuit around SCR1 and U3. Connect a multimeter on the 12V rail (Ch1) and desolder say R62. If you see the rail rising too high then switch off immediately.

If there is still a problem, desolder U3 completely and repeat as above. This should completely disconnect the OVP crowbar circuit.

[neoone]

heh, we seem to think similar in many cases I try not to write what I'm thinking so other people have "fresh minds". The circuit with TL431 is more or less standard for those kind of SMPS I believe (I've used similar one in the past). The second part with thyristor was a bit surprise for me though. It looks like it's protecting from the output voltage going above 15V (if the zener is in fact 15V).
Any way I'll try to take your advice and start to play in this part. There are some 0R connections on the board which I didn't draw so some cases it may be possible to desolder one of them. Some of the resistors and capacitors are SMT so I'll try to use them and THT only if really necessary.

[Alex]

Whatever you do, be careful, safety first. This is mains/high energy territory.

[neoone]

"relax" it's not like I never did it before But of course you're right - safety first!

Any way - it seams problem is elsewhere. I played a bit and finally desoldered U3 completely and nothing has changed - voltage at 12V output looks exactly as before. Something wrong around SHR1 ?

added:
I soldered U3 back in place and desoldered U2 just to check what happens. This is how it looks.

It looks that it works as it should - cutoff after reaching 15.2V. And it doesn't make any noice - I'll check if there is any signal at the gate of Q1 (probably it's fully open so there's no sound from trafo).

added:
Ok there are only 3 pulses on the gate before cutoff so nothing to check really.

[Alex]

Possibly. What is that Trafo block? The flyback transformer or a linear PS for housekeeping? In the first waveforms you posted, Vcc seems to follow the datasheet under short circuit conditions (tertiary winding on flyback transformer). But Vcc in the schematic is regulated by ZD2. What is the value of ZD2?

I was also thinking of the waveform on the FB pin. Good to see the OVP crowbar is working at 15V, you took a nice chance there :-)

Using the decaying voltage reference that this creates you can solder just the LED side of U2 now and check SHR1 by comparing the waveform at its reference input from the R divider and its output connected to the cathode of the LED of the optocoupler. They should be very similar with a cutoff at its output cutting-off at 2.5V input.

Whats connected to pin 1 of the trafo block?

[neoone]

TRAFO is a transformer with a lot of other pins which I didn't draw. It's hard to say what is the value of ZD2 - it seems to say 1N4 and something like 700 or 750.

In the meantime I took a different approach to checking OVP block. I soldered back U2 except cathode of the LED and instead connecting it to TL431 I took Zener diode 7V5 to ground. Here is the effect at the output:


So basing on that I think it indirectly shows that TL431 block should be fine be cause effect is the same with simple zener and with TL431.

Whats connected to pin 1 of the trafo block?

what do you mean ?

Now I'm thinking that maybe there is something wrong with current sense...

[Alex]

So basing on that I think it indirectly shows that TL431 block should be fine be cause effect is the same with simple zener and with TL431.

Yes, it seems like the voltage feedback loop is working.

Ok, I see what Trafo is. I presume 'pin1' of trafo is coming from a winding on the transformer. I am not sure why that has been designed like this. If the designers wanted to create a negative voltage on pin 3 of the controller then R7 is much much bigger than the current sense resistors, so it wont have any effect. If it was intended to capture a positive only induced voltage then the current sense resistors provide the same information. Strange.

You can try partially disabling the current sense by slowly pulling pin 3 to ground with a 10k pot and see if this brings the entire circuit into correct operation. Before doing that I would check what is the maximum current for Q1 (part number?) and then measuring that current (As a voltage and then Ohms law) across the current sense resistors with the scope and stay well below that.

You can also try to power the controller (Vcc) by an external 13V power supply. The datasheet specifies a minimum 12.2V for turn-on, but from the Vcc waveforms I dont see this reached, possibly the current protection kicks in before C10 charges up and the regulation circuit begins to function properly (a DC output voltage at Vcc).

As a sanity check, check the HV supply, it should be about 325V.

If all this fails then most likely the controller has surrendered its spirit.

[neoone]

Ok, I see what Trafo is. I presume 'pin1' of trafo is coming from a winding on the transformer. I am not sure why that has been designed like this. If the designers wanted to create a negative voltage on pin 3 of the controller then R7 is much much bigger than the current sense resistors, so it wont have any effect. If it was intended to capture a positive only induced voltage then the current sense resistors provide the same information. Strange.

yes, that was a bit of a mystery to me also.

I will check some things and get back to you.

added:
HV is ok - 324.1V. I made a small mistake on scheme - HV is actually between ZD3 and D4 not on Pin 8. I don't know part number of Q1 because it is in a silicon holster of some kind which covers it entirely. I could consider cutting it...

[DrGeoff]

SCR1 is the overvoltage crowbar. Monitor this part of the circuit and see if it is firing.

[Alex]

Yeah, no problem, thats how I perceived it.

Before you bother altering the current loop, try replacing the controller. It is a simple circuit and (almost) everything else seems to be working fine (including the crowbar OVP DrGeoff that we checked earlier), so I am beginning to think this behaviour is caused by the controller not having the datasheet specs.

You can 'borrow' a controller from the other PSUs, I believe they would have used the same one.

[neoone]

yes they have the same controller and I've been thinking about that since yesterday I've gor few other MW PSs so maybe I will take it after all.

I will post a waveform taken after putting external power to VCC in a minute.


As you can see output voltage is now fine. CH2 shows burst on Q1 gate. There are 2 shots in each burst with their nominal frequency of 60kHz (this is 60kHz version of NCP1203). The problem is that it still makes a lot of noise - probably due to 1.56kH burst frequency.

I have also checked pot on pin 3 to ground (before putting external VCC) but I didn't push it to GND. Nothing has changed on the output but the burst frequency was changing.

[Alex]

I suggest you try changing the controller first before doing anything else.

Also, it is late in Romania, tireness and electronics don't mix well, tireness and mains must not mix!

[neoone]

yes it's getting late. I'm not sure where this last waveform leave us. I will try to replace controller tomorrow - hopefully I won't damage other PS 

off topic: So You're from Romania? Then I believe it's even 1 hour later there than now in Poland