PLC810PG intermittent High AC Current problem

13 posts / 0 new

On our second run of 25 boards (first 25 fine), 20% will blow PFC Fet.  I believe this is caused by the PFC chock saturating.  Most boards will operate fine to full power; but some will not.  Moving the PFC chock from a good board to a bad one does not change correct the current tic?  Can you help me understand what the cause is of this intermittent tic is?   Our board’s primary circuit and board layout is nearly the same as your RD189 evaluation board.  One discrepancy I found was that the two diodes across PFC Sense resistor are DL4007 with no I sq t or high surge rating.

I have attached some scope print outs to demonstrate this problem.

AttachmentSize
HighCurrent.pdf625.65 KB

1) What do you mean by "tic"?
2) Re: figure (2), the signal on the .06 ohm sense resistor is supposed to look like the choke current, but it does not.  

3) What is the saturation current of your choke design (what are the core material, core area, turns, and inductance?) 

 I refer to the intermittent  high current peaks that make an audible sound and swing my AC amp meter to the right as a tic.

I have not measured the saturation current, but the supplier calculated 7.5 Amps.  The core is Magnetics P/N 00K5317E060  core is Sendust, Ac(mm2)=152, 55T 5x#24, 576uH with band.

Note,  I get slightly better results with the chock from the RD189 refrence board.

I greatly appreciate your help in understanding this problem.

 

Your waveforms show that the voltage on the 0.06 Ω sense resistor and on the ISP pin do not follow the choke current faithfully.  This could be your problem.  Where did you probe the sense resistor voltage?

I could not find the datasheet of your core, so I could not check the flux density.

I am going to redo this sense resistor and ISP very carefully.  In the mean time I am attaching the data sheet on our core.

Thank you.

Did you follow the PCB layout guidelines regarding the connection from the PFC current sense resistor to the ISP and GNDP pins?

I probed the .06 sense resistor short lead across resistor, the ISP pin was done with short leads pin 2 & 3.  This does have the shields about 2” apart?  Do these measurements have to be taken separately?  When I look at the chock current probe and just the resistor they seem to track perfect?

 

I believe we followed the PCB rules except the Sense resister leads are not side by side.  I have attached our PCB layout for reference.  I tried running some wire traces that ran side by side that did change the behavior, but no correction.

 

I believe you are correct about distortion between Current sense and ISP.  I just want to prove it before I spin another board?

 

Thank you

When you probed the sense resistor signal, was the probe ground connected to the end of the sense R that is connected to the MOSFET?  (as opposed to the end going to the ISP pin)

Yes, the probe ground was connected to the MOSFET Source. 

1) When you measured the sense resistor voltage, did you have a 2nd probe's ground clip connected elsewhere on the board?

 

2) Pls. attach your schematic. 

1) Yes, both sheilds were short leads.  Pin 2 for ISP, and B- at the sense resistor.

2) Shematic attached

Depending on the situation, sometimes you can't see 2 small signals at the same time on the 'scope if the ground point of each probe is at 2 different locations, even if they are the same node on the schematic.  As a check, leave just one channel connected, so only one ground clip is connected to the circuit at one time.

Regarding your layout.  C39 (resonant cap), has high dv/dt on both sides.  The C39 node connected to the LLC MOSFETs has several SMD parts lying underneath and near it (e.g. C27, R54).  Low voltage, high impedance nodes and circuits are very susceptible to noise from nearby nodes with high dv/dt.  One possible solution is to swap the transformer primary pins and also the locations of R59 and C39.  R59 is quiet - it doesn't have high dv/dt.