UCC2897A: ucc2897a clamp mosfet heats and pcb burns

Hi,

You need to measure the waveforms of MOSFETs to make sure their operation correct. One often seen root cause is the dead time is not enough.

But you have to measure the Vgs and Vds waveforms to decide if each MOSFET is in switching mode. Also, together with your transformer, and MOSFETs specs. 

Dear Sir,

 i have attached the waveforms, it would be helpful, please have a look 

After 5V,4.5A load current waveforms changes

10K Ohm resistor is connected to RTDEL - programmed delay is 126ns

at CS pin, 112 milli Ohm resistor is connected, i attached scheamatic also

psu_frtu_pmp22509.pdf

6303.waveforms.rar

Hi,

I suggest you increase Rdel from 10k to 30k and re-test and check if any improvement.

Can you zip your waveforms instead of rar? I cannot open rar file.

waveforms_8Dec20.zip

Hello,

Attached waveforms in zip file,

will check RTDEL = 30K

Regards

Lakh

Can you change C11 to be 33nF, and change C19 to be 100nF?

The converter looks unstable based on your waveforms.

hi,

upto 4.5A load current it works, at 4.8A load current waveform gets unstable,

 i looks like overcurrent limit on current sense pin is triggered

do you suspect tl431 compensation also,

i will soon check and share result

regards

lakh

ok, so let me know how it runs after your make changes.

hi,

i checked with 33nF clamp capacitor  and 100nF capacitor on c19 on tl431,

getting same waveforms

Can you further increase C19 to 220nF, 330nF, etc, up to may be 1uF? I suspect your converter loop not stable? Also, what is your PWM duty cycle, and what the waveform looks like at light load and medium load?

hi,

1)  i tried with 220nF, 1uF,500nF,330nF, got the same waveforms on drain of main mosfet,

attached waveforms at 0A, 0.5A, 2.5A, 4A, 4.5A, 4.6A load

2) upto 4.5A load current, primary main mosfet drain waveforms looking ok,

 when i increase load current from 4.5A to 4.8A, then the waveforms gets changed, 

at this time when i see waveform on current sense pin, there is sharp rise in primary current and voltage on cs pin gets to more than 0.53V,

it looks overcurrent is trigged and devices terminates switching in between and waits for next switching cycle , need your view on same DSO.zip

3) observed  duty cycle at no load current  is  near to 50%, and at 4.5A load current observed duty cycle 60-62%( approx. 2.4uS time ON period out of total 4uS period)

RON = 86.6K

ROFF = 46.4K

4) I have doubt that transformer gets towards saturation and as result primary inductance reduces and primary current increase which leads to 

overcurrent being sensed by CS pin, which gets apporx. 0.53v and terminates switching,

at this time my clamp circuit  and main mosfet heats the most and temperature rises very quickly in few minutes and keeps on rising to more than 125 degC

if i increase the ambient temperature to 52 degC, same distorted waveforms i start to get at lower load current, at 4A

regards

lakhwinder

Hi,

If you suspect your transformer possible saturation when current reaches 4.8A, you may need resolve this first before debugging others.

hi,

with present RTDEL = 10k 

and RTDEL = 33k, 

for transformer it is observation, i am not clear whether it is transformer itself getting saturation or it is not being reset so it gets saturation

Regards

lakh

how may i ensure that it is transformer is getting reset properly, do you see from waveforms, is it proper?

Can you provide CS current waveform at about 4A?

I can only find 4.6A and 4.8A CS waveforms which seem already show some saturation phenomenon. I need to see CS waveform at about 4A to compare at 4.6A to 4.8A and describe to you.

yes,

 below waveform at 4A current

and below is at 4.5A,

regards

lakh

If you compare CS waveforms, you can see with 4.8A, the CS waveform at close to end of each cycle is titling up while at 4.5A, there is no tilting up. The tilting up of the current is usually a sign of transformer saturation. So it looks at 4.8A, your transformer is in saturation for one other cycle, which may explain why your see narrower pulse width every one another cycle.

It looks RDEL = 30k causing too much dead time. I think after you correct saturation problem, you may try change RDEL smaller and try 20k, or back to 10k, if the temperature is ok.  

hi,

how to correct saturation problem?

i used same transformer from coilcraft that is used in pmp22509 reference design

do i need to change some circuit values so it does not saturate

or need a new transformer with higher rating 

regards

lakh

i also had doubt that whether transformer is being reset properly or not, 

You may increase your current sensing resistor value so to reduce the peak current to avoid saturation. Together you may need to increase your maximum duty cycle a bit. 

I do not know why the transformer ok with the reference design, and why not with yours. Maybe the transformer has tolerance issue.

It looks the issue only starting at > 4.5A, and if there is saturation the reset may not work properly in each cycle. 

Ok, I soon check this

Increased the current sensor resistor to 200mE and checked, at 3.5A load current, duty cycle reaches 62%,

increasing load current more than 3.5A, output voltage starts to reduce to lower voltage, duty cycle seems not increasing further

present values for duty cycle

with RON= 86.6k

ROFF = 46.4k

RTDEL = 10k

SLUC662 calculator shows duty cycle near to 75%,

 what we observed from waveform is maximum duty cycle of 62%

how to further increase duty cycle,

which next size transformer should i use

i ordered PA3856.005NL transformer from pulse electronics to check with the design

hi hong,

i) should there be any delay between the main primary mosfet turn OFF and clamp mosfet turn ON,

 waveforms shows that voltage on drain of primary main mosfet starts to rise when it is switched off

and clamp mosfet was on during this rise of voltage ,after clamp capacitor is charged to peak voltage it starts to decay and reaches near to input voltage

then next switching cycle begins

ii) how can i ensure that transformer is completely reset before next switching cycle

regards

lakh

i) ok, then i check soon with  pulse electronics  PA3856.005NL

 it is rated 9VDC to 57VDC input, 5V 7A, 

will let you know

ii) is there any calculation which may be utilized to see if enough demagetization current has gone through transformer during the clamp mosfet ON time and 

main mosfet OFF time , would it be ok to say that lower clamp voltage(higher clamp capacitor) will require more time to demagnetize core and high clamp voltage(lower clamp capacitor) will require less time comparatively