PMP8861: Can't get the reference design stable.

Hi Teun, 

I have reached out to the designer to see if we can get you some help to fix this issue. Please give us a few days.

Thanks,

Robert 

Hi Robert,

Any news yet?

Hi Teun,

Looks like the converter is not starting.  Can you provide a scope picture of VDD and VOUT on the same plot?

thanks,
Brian

Orange VDD

Blue VOUT

Hi Teun,

Looks like the VDD voltage is dropping out.  Instead of the OUT (gate drive) waveform, can you capture a picture of the output voltage of the converter?

Also please check the connections of L2, D3, and D5.

thanks,
Brian

Sorry this is VOUT then.

The connections for the BIAS voltage seem to be correct. Looks like the minimum voltage of 8.5V could be exceeded. But what causes that?

Hi Teun,

The BIAS voltage should be provided from T1 auxiliary winding through D3, D5 and L2 after the output voltage rises high enough.  In your picture, the output voltage is up and regulating, but the BIAS is not being provided from the T1 auxiliary winding.  Something must be wrong with this part of the circuit.

Can you check the voltage on the anode of D3 during startup?

thanks,

Brian

Hi Tuen,

This measurement looks like the ground connection on the scope probe is not connected. Can you please check to measure the waveform between anode of D3 and PGND?  

This waveform should look like a square-wave, and not go below 0V.

thanks,

Brian

Hi Brian,

That actually is the waveform between D3 and PGND. Could my auxiliary winding be polarised the wrong way? I did not use the original pulse transformer since it was not available.

Hi Teun,

Rethinking my earlier comment... this waveform will go below 0V for a portion of the switching cycle, but it still does not look correct.  It is possible that you have the polarity wrong.  Can you add a second channel to the scope image showing the voltage on the drain of Q7/Q10 on the same plot showing voltage on anode of D3?

When Q7/Q10 are ON, the voltage on anode of D3 should go high.

thanks,
Brian

Hi Teun,

Can you please zoom in on the time scale (2us/div) so that we can see the individual pulses?  The power supply should be switching at 200kHz.

thanks,

Brian

Hi Brian,

This in on the off section on D3's anode.

And this one on the on section.

Hi Teun,

Looks like the resonant frequency of the clamp is not correct. 

What is the primary inductance of your transformer? It should be around 40uH.

Also, the voltage on the aux winding seems to be fluctuating.  Is pin 6 of the transformer tied to PGND?

Do you have a datasheet or drawing for the transformer that you are using?

thanks,
Brian

Hi Brian,

I have been using this transformer:

https://www.coilcraft.com/getmedia/498498a8-0185-4168-8009-38193155936d/poe120pl.pdf

Pin 6 is on PGND.

So then I should try to change the value of C8?

I assume you are using the POE120PL-12L_ version of this Coilcraft transformer series?

The problem is the turn ratio is wrong for the 18-72V input range used on PMP8861.  Primary:secondary should be 1:1.  The Coilcraft part is 2:1.

What is the input range for your application? 

thanks,

Brian

Yes the POE120PL-12L is used, but I only used 1 of the primairy windings. 2-4

18-72 input range would be great, but 36-72V would be ok since these were the first prototypes we did. And just wanted to see them work first.

Hi Teun,

The problem is that the inductance is way too low now.  It would be better to use the POE120PL-24L version of the Coilcraft transformer to get the correct pri-sec turn ratio and close to correct inductance.  However, you will need to provide an external VDD for the controller to allow it to keep running because the AUX turns are too low.  At least this way, you should be able to test your prototypes.

Hi Brian,

I tried using the POE120PL-24L transformers. Now it is stable and working but only at +/-8.4V max. Is it possible to get it up to 12V with this winding ratio?

Hi Teun,

The POE120PL-24L has a 1:1 turns ratio (primary : secondary), same as the original transformer used in the PMP8861 design, so it should be fine.  You say that you have +/-8.4V output?  What do you mean by "+/-"  the output should not change polarity, and should just be steady with a positive voltage.

thanks,
Brian

Hi Brian,

I didn't change anything from the design at first. The POE120PL-24 gave me a stable 8.38V output. After that I changed R18 and R22 for trimpots. I can play a bit with the output voltage but can't get it much higher than 8,5V I can however turn it lower than that.

Hi Teun,

That is strange.  Can you send scope pictures of the switching waveforms on the drain of Q7/Q10, and drain of Q5/Q9?

thanks,

Brian

Yellow Drain Q7 & Q10

Blue Drain Q5 and Q9

Please check the voltage level and waveform on the FB pin of the UCC2897A.  Also please check the voltage waveform on the CS pin of the UCC28987A.

Blue CS 

Orange FB

I have fiddled a bit with the trimpots an got it to about 10V. The input voltage was at 50V but I noticed that turning that up to 60V, the output went up from 10V to 11V. Also it already draws about 5,5W doing nothing.

Hi Teun,

When the output is below the regulation level, the FB pin should pull all the way to 5V, but the FB pin appears to be stuck at around 2.8V. 

Please check the following:

1. Verify R24 is 1k on your board.

2. Make sure the only components between the FB pin and PGND are the optocoupler and filter cap C16.

3. Measure the voltage on pin 2 of the optocoupler (U2). When the output is not in regulation, this pin should be around 0.7V below the output voltage.

thanks,

Brian

Points 1 & 2 check.

I set the trimpots back to the standard values of 10k and 2.61k.

And on pin 2 of the optocoupler I measure 1.7V with the output at 8.75V. This is under no load.

With a 2A load I measure pretty much the same. But output voltage drops to 8.11V.

I tested all this at an input voltage of 50V. But I noticed that under no load the converter is not stable when I increase the input voltage to 60V. And it starts to make an audible noise.

When I have a load however it is stable at 60V input and higher. The converter works for the most part it seems because I can stil draw up to about 8A.

Looks like you might have the wrong part installed for U3.  Make sure you have the TL431 and not a TLV431 (or some other variation of this part.)

Yes that seemed to have been the problem! Thank you. Converter now seems working.

But it is still not stable when the input voltage is turned up to 60V or higher when there is no load. Under load it works fine.

Here is a screenshot of the Q7/Q10 Drain signal under no load at 60V input:

And here is one of the same signal under 1A load.

Looks like the synchronous rectifiers (Q1 and Q5/Q9) are not turning on.  Check the gates of these MOSFETs to see if there is any gate drive voltage present.

I think the voltage is there but something is interfering with it.

Under load it looks like this.

And the converter always seems to have an audible hiss somewhere. It's just much less loud when it's stable. Couldn't hear it before over the noise of the fans of the electronic load.

Are those pictures gate voltage of Q5/Q9, or Q1?  Can you show both gate voltages? 

Those were Q5/Q9.

This is without load. Orange = Q1 Blue = Q5/Q9.

This is with a 2A load.

Strange.  Looks like there is drive on both sets of synchronous rectifiers.  The resonant voltage on the active clamp seems kind of high though.  What is the part number of the capacitor that you used for C8 (clamp capacitor)?

I don't actually know. I have it specified as a 250V, X7R, 47nF, ceramic capacitor. I will have to contact the manufacturer for the exact part number. But I didn't think it would be that critical.

Could the value be the issue? Since the primary inductance of the POE120PL-24L is lower than that of the original transformer?

I had already increased the clamp capacitor value which resulted in a significant improvement. Now that there is also more output capacitance it seems to be quiet.

Thanks.

Yes. Thank you.