IRS27951 / IRS27952

[nigelwright7557]

I reduced Rmin to 2k2 and it fires up every time now and holds 6 amps per rail.
The recommended value of 3k9 forced the 27951 to reset.

[lorylaci]

Originally Posted by nigelwright7557

I reduced Rmin to 2k2 and it fires up every time now and holds 6 amps per rail.
The recommended value of 3k9 forced the 27951 to reset.

You should scope the lower FET gate, and tha half bridge output, to check for ZVS or ZCS switching.

Although there should be no connection between Rmin and the trigger of overcurrent at startup, if the soft-start is set right.

[nigelwright7557]

Originally Posted by lorylaci

Although there should be no connection between Rmin and the trigger of overcurrent at startup, if the soft-start is set right.

I looked through the spread-sheet again and altering my transformer also altered the start components and I had missed that.
I fixed the start up components and the circuit powers up every time even with the spread-sheet specified Rmin value of 3k9.

[nigelwright7557]

|I have built up another pcb and sadly the 27952 cant have been soldered in right and it blew mosfets, zener and 27951.

I only had high Rdson mosfets so put those in and replaced zener and 27952.
The circuit powers up ok now so long as there is no load.
I have ordered some more low Rdson mosfets and will try them when they arrive.

With my next pcb revision I will make longer pads for the 27952 so it is easier to solder by hand.

Before I power up a new pcb I have to buzz out every 27952 pin for connection to the pcb and for shorts between adjacent pins. Its a real nuisance, why cant IRF supply DIP.

[nigelwright7557]

Originally Posted by lorylaci

Although there should be no connection between Rmin and the trigger of overcurrent at startup, if the soft-start is set right.

Backing off Rmin was allowing it to power up because it drew less current as it was further away from resonance.
Fixing the power up components now makes Rmin the correct value.

[nigelwright7557]

Originally Posted by lorylaci

You should scope the lower FET gate, and tha half bridge output, to check for ZVS or ZCS switching.

How do I know if there is ZCS or ZVS ?

[lorylaci]

Originally Posted by nigelwright7557

How do I know if there is ZCS or ZVS ?

Previously I posted a simulation picture where there was inductive operation (possibility of ZVS) but the dead-time was to small to achieve complete ZVS:


You can see in the red signal (lower FET gate) that there is miller plateau caused spike, the on the yellow signal (half bridge) that the transistion is very steep.

If you have capacitive mode (possibility of ZCS), then you will see the same, even if you have big time.

Even if you in capacitive mode that does not mean you can certainly achieve ZCS . (as if you in the inductive mode ZVS is a possibility with good dead-time)

Here is a very good link about ZCS resonant converters: http://schmidt-walter.eit.h-da.de/sn...g/snteng4b.pdf
If you in capacitive mode then the periodic time of your control signal must be longer then the resoant network periodic time. (FYI: IRS27951 is intended for ZVS resonant controller, it cannot be used as ZCS resonant converter controller)

[nigelwright7557]

Thanks for the explanation.

Things have fallen apart again. I blew up a set of mosfets, a 27952 and a zener diode.
I replaced them all but I am having trouble getting the 27952's soldered in again.
I have blown up two so I need to order some more. The blown 27952's power up in very odd modes, one mode has a very long HO with a very short LO. It should be a 50/50 pulse.

I have started work on a mark 2 pcb with much longer pads for the 27952, this means I can put solder on the pads and then run the soldering iron up to the pin taking the solder with it. This should put less stress on the 27952.
I also had trouble getting high power rectifier diodes and they were a silly price. I have replaced them with a 600 volt 10 amp bridge rectifier which is much cheaper.

[lorylaci]

Originally Posted by nigelwright7557

Thanks for the explanation.

You're welcome

Originally Posted by nigelwright7557

I replaced them all but I am having trouble getting the 27952's soldered in again.

Nigel, first practice soldering, I haven't blown up any SOIC case for about 3 years.

Originally Posted by nigelwright7557

much longer pads for the 27952 .. This should put less stress on the 27952.

That is a good start. I alway use bigger pads than reccomended. This ease soledring, also bigger pads can dissipate more (so better thermal resistance.

Originally Posted by nigelwright7557

I also had trouble getting high power rectifier diodes and they were a silly price. I have replaced them with a 600 volt 10 amp bridge rectifier which is much cheaper.

Where did you put those? On the secondary? Get them out ASAP!!!
For the secondaries you have two choices:
- ultrafast rectifiers (<100ns recovery time) for diodes >100V
- Schootky diodes (they have "no" recovery time) for diodes <200V
You CANNOT use normal slow (>1us) rectifiers! They will BLOW UP!
(i can buy MBR10100s for about 100ft (0,3 GBP))

[nigelwright7557]

Originally Posted by lorylaci

You're welcome


Where did you put those? On the secondary? Get them out ASAP!!!
For the secondaries you have two choices:
- ultrafast rectifiers (<100ns recovery time) for diodes >100V
- Schootky diodes (they have "no" recovery time) for diodes <200V
You CANNOT use normal slow (>1us) rectifiers! They will BLOW UP!
(i can buy MBR10100s for about 100ft (0,3 GBP))

I put the bridge on the primary side.
I used ultra fast diodes for the secondary as it is over 100KHz.