PMP8740: Opto isolated feedback

Hi Venkatesh,

I am afraid you need to add snubber on secondary side, due to the fact every transformer has leakage inductance.

This leakage inductance is generating lot of loss if not properly snubbed.

You can see it in the voltage of all rectifiers on secondary side.

From the latest waveform you sent, I see the voltage spike reaches 650V, so very likely your diodes are working at breakdown voltage, involving high dissipation.

Please follow the example in PMP7246, available at the following link on TI.com:

https://www.ti.com/tool/PMP7246

On page 1 of this schematic, all sync-FETs Q7....Q10 are in your case the SiC diodes.

Here the snubber network is simply D2, C21, C22 and C23, and the dissipation of leakage inductance energy is accomplished by R1, R2, R3 and R4.

Please select the capacitors C21....C23 so that at each turn off (or clamping spike) the associated delta-V is restricted to less than 3%-5% of Vout, therefore 11.4V-19V (you will need several capacitors, rated at least at 650V; the best are film capacitors). If the capacitance value is too big, you may accept a delta-V up to 10%.

Then the diode D2 should be also a SiC rated at 650V (5A to 8A rating should be OK). After that, please adjust all resistors R1....R4 so that the voltage on cathode of D2 is lower than the maximum rating of all SiC diodes. You will probably need to dissipate 10...20W on these resistors.

I see also that all SiC diodes are rated for 8A, while your output current is 8A....I think this is a bit on the edge, and I would consider to replace these SiC diodes with ones rated at 15A or more. 

Regards,

Roberto

Hi roberto,

I will do that

SiC diode current rating is 20Amps roberto,voltage  is 650V only.

I think that is the issue at full load it may reach 650 and above volt leads to damage

Thank you

Venkatesh 

Hi Roberto,

How are you,

I have added snubber at Diode played R & C values finally reduced the Ripple

Before snubber

After snubber added(RC)

But resistor is heated up at full load

Now problem is at full load,converter operated in DCM mode only

what could be the reason,attached secondary voltage and primary side transformer current

i did changed Tmin resistor value but no use,what could be the cause for it

Thank you 

venkatesh B

Hi Venkatesh,

I still see the voltage on SiC diodes reaching 790V: is that too high, and dangerous? Did you add a series RC snubber or a clamping (RCD) network? From your waveforms it looks like you only used a RC snubber, which is not clamping SiC diode voltage.

Second question: you say the converter works at full load in DCM mode: do you mean actually "burst mode"? I see bursts in the screen shots.

Also, the red line is primary side winding current of the transformer? I am wondering why it's not trapezoidal, but it's almost a square wave: if so, it looks like the output inductor has very high value....can you please share again how many uH you have on output inductor?

Thanks,

Roberto

Hi i used only RC snubber only.

Earlier it reaches 1000V, now it is in 790V,sic diode rating is 1200V ,i thought of its sufficient.

Output inductor is slightly higer 240uH calculated,used inductor is 280uH ,hope higher L is okay 

Yes it is in burst mode,but it should be in CCM right?if i give input is 300V DC load fully then that time it behave in CCM,any voltage lesser than 350V DC input i can able to achieve CCM at full load. But when i give 400V Dc input output is in burst mode at full load,what could be the reason?

Where should i place Diode in case of RCD clamp option,share any calculation if available, for RC snubber i tried trial and error method and found values

Thank you

Venkatesh 

Hi Venkatesh,

Since you use 1.2 kV SiC diodes, therefore 790V peak is sufficient margin.

Also 280uH inductance is fine. 

Burst mode: if you achieve CCM at 300V or lower Vin, it means probably you reached maximum duty cycle and the converter runs in open-loop. If that is the case, you may just have your converter unstable at 400Vin.

In order to check it, can you please reduce the gain of the loop, and make it slower?

RDC clamp, please follow the schematic of PMP8740 or the UCC28950 EVM.

Thanks,

Roberto

Hi roberto,

How to reduce the loop gain?

In PMP8740 output is 32V,62A for us 450V,8A output

share any excel sheet calculation available if any

Thank you

venkatesh B

Hi Venkatesh,

The loop gain can be reduced by reducing the resistor Rf, described in cell 140 of the UCC28950 Excel calc sheet, see link:

https://www.ti.com/tool/download/SLUC222

In your case, if the voltage loop is active, it's R229. Please reduce it from 68.1 kOhm to 10 kOhm, and increase C244 from 2.2 nF to 220 nF.

If, instead, the current loop is active, you may want to reduce R38.

Thanks,

Roberto

Thanks roberto i will do that

Hi roberto,

I have disabled voltage feedback path(open),reduced resistance from 100K to 1K but no use

let me know other parameter to check

Thank you

venkatesh

Hi Venkatesh,

OK, you disabled the voltage loop, therefore your converter is working as constant current source.

May I ask what is your load? Is it a resistor, an electronic load, a battery?

If it's an electronic load, is it set to CC, CV or CR mode?

Thanks,

Roberto

Sorry, I selected wrongly the button "TI Thinks Resolved"

Question: your output electrolytic capacitor is only C120 = 68 uF, right? Is it a bit too low,? Maybe you can increase by adding a second 68 uF in parallels to C120...

Its okay,roberto 

Output side filter C is 470uF roberto

Thanks 

OK Venkatesh,

Anyway your control loop section is different from mine. Can you please increase the value of C22 capacitor? Please increase it from 100 pF to 100 nF. In my schematic I don't have this extra pair pole-zero...

Regards,

Roberto

okay i will do that roberto

Thank you

venkatesh B

Good morning roberto,

changed that capacitor still same response,is there any other parameter to change?

thank you

Good morning Venkatesh,

Sorry that it takes so long to fix it: since I am not very familiar with the way you close the loop around UCC28950 (I understand you took it from TI.com, but still I prefer my circuit), can you please change the section around UCC28950 and optocoupler and make it similar to my schematic, shown in PMP8740 project?

This way we can reduce the number of variables.

Please let me know if that is feasible or you prefer to stay with the circuit you are currently using.

Thanks,

Roberto

Ok roberto i will do change and see is that possible in another board

Thank you

Hi roberto,

I have one doubt is DCM and Burst mode both are same or not

From my understanding Tmin, CS & DCM threshold are responsible for this operation

One more doubt is i am probing feedback OPAMP section,Only enabled Current control

while turn on,pin 7 is becomes 7V,then current starts to build,when it reaches 6Amps,becomes CCM mode,voltage at pin 7 becomes 12V

same repeating for 8A full load,but Pin 7 always 7V even at full load

what could be the reason

Thank you

venkatesh B

Attached the schematic

hi roberto

any update

Ok roberto i will probe and check it 

Thank you man

Hi roberto,

CCM/DCM issue got fixed

I played changing Pole/Zero R & C values 

finally it is fixed

Now i want to optimize snubber(RC) and may add RCD,CUrrently RC snubber is placed 110 W loss,i may require to put RCD if require

Can i post it separately for that or i continue in this roberto

I am very glad and appreciate your support throughout tjis time,i may handshake you if u are nearby,thank you so much

Thank you

Venkatesh

Hi Venkatesh,

Glad to see you are progressing and now the converter is stable :-), and thanks for your kind words.

I believe we can continue to post on this thread, since very likely this will help also other engineers.

In terms of RC snubber, I am not sure I have already sent you this document about snubbing; in case I have not yet done, please access the technical article:

https://www.ti.com/lit/ta/ssztbc7/ssztbc7.pdf

Here a method to select both R and C in the snubber is explained.

For RCD clamping, you should already have documents: if not, please let me know and I will attach a link or document as well.

Regards,

Roberto

Hi roberto,

yes i did RC snubber as per calculation and tested also.there was almost 100W loss happens,snubber resistor is too bulky to handle that much power.

Kinldy share RCD calculation tool if any and do share any lossless passive snubber available 

Thank you

venkatesh B

Hi roberto good morning

Hi Venkatesh,

As basics for snubbing I would have a look to the attached (and pretty old) PDF document:

Snubbing methods slup100.pdf

Active snubbing is even more complex, so I don't have documentation for that, sorry.

Regards,

Roberto

Thank you roberto,i will check it

HI roberto,

I am proceed with 1200V Diode now,everything seems ok

when i give 400V DC from RPS(Regulated power supply) PSFB DCM to CCM converted at 1A and above

but when i give 400V DC from Front end PFC(Interleaved PFC circuit) PSFB still operate at DCM mode even at full load

attached the Transformer waveforms 

what could be the reason for this

Thank you

venkatesh

Hi roberto,

r u there

Hi Venkatesh,

I can see from the waveform you attached that the time distance (period) is practically 10 msec. I believe your mains frequency is 50 Hz, correct? This way, on the output of the PFC Boost you will have 100 Hz ripple, therefore 10 msec period. It means the current waveform you measured is correct, and maybe is a consequence of pretty remarkable 100 Hz ripple of the output of PFC stage. Can you please measure this pea-peak ripple and let me know?

Thanks,

Roberto

Oh yes it is 100Hz only

ok i will measure PK to PK ripple at PFC output side as well as PSFB output side and share you

Thank you