Two transistor forward diode getting very hot

[cupoftea]

Hi,
Doing Two Transistor Forward, off Mains FWB. No PFC. 100-265vac. 130Wout, 24Vout, 120kHz.
Np/Ns = 22/12. Sandwich wound. No Fan.
Secondary diodes are STTH5L06D (TO220 metal tab)

The two sec diodes are both side by side and screwed to a flat plate of anodised aluminium of size 15cm x 10cm by 2mm thick.
(OK, its got a 90 degree bend down the middle but lets not go into that).

At 115VAC and full load all is well, all temperatures are below 65degC.

But go up to 240VAC, and the freewheel diode starts getting well hot. Goes to 100 degC within minutes.
Then blew up a bit later...(I hadn't yet thermocoupled it when it first blew up as I thought the alu plate
would be a good "cooler" for it.)
(All with prototype PCB in open lab ambient (22degc))

So I believe the main cause is reverse recovery, also I hadn't expected the alu plate to be so bad as a heatsink.

Are there different types of "aluminium" with differing thermal properties?
This alu "Plate is actually a right angle bracket used to screw the supply into the unit.
It feels rather hard for aluminium.
Also, I notice quite a temperature gradient along it. Hot near FETs... much cooler at ends.

STTH5L06D

https://www.st.com/resource/en/datasheet/stth5l06.pdf

In scopes attached...........("Two tran fwd1" & "Two tran fwd2")
light blue = Freewheel diode volts
Red = Lower fet vds
yellow = Upper fet vds
Darker blue (noisy) = primary current

Also for "Two Tran fwd3" scope shot...
light blue = Power diode volts
yellow = freewheel diode volts

--- Updated Thursday at 9:06 PM ---

Am actually thinking of parallel SiC diodes, or non-sandwich winding so the leakage-in-sec can staunch the rev rec current....Would you concur?
The voltage reflected to the secondary is so high that this diode is getting "roughed up" by rev rec currents , a bit like like a PFC diode.

Reckon two if these in pllel should do the trick....

https://www.smc-diodes.com/propdf/S3D04065A%20S3D04065F%20S3D04065E%20N2721%20REV.B.pdf

Any cheaper ways?

 

[KlausST]

Did someone invent schematics?

 

[D.A.(Tony)Stewart]

Yes there is a tradeoff between strength and conductivity in Alum alloys. 1100 is pure AL.

ref

 

[cupoftea]

Thanks, please find schem and LTspice as attached.
Sim of course does not show rev rec...and that appears to be what is killing the output freewheel diode.
Choice between rewind non sandwich, or 2 pllel sic fets.
LLC not an option due to wide vin range.
Its going to be 3 phase / 3 diode bridge 100-265vac, with 220uF storage cap.

 

[Easy peasy]

The diode you have chosen is quite a slow part Trr = 95nS worst case and this is only at 1A forward and 50 A/uS falling current at turn off.

In your design the forward current is 6A and the falling di/dt would be in the 100's of amps per uS

These diodes are also slow to turn on, and the turn off Ipk generally gets worse with temp - hence the thermal runaway - also 1V x 6A half the time = 3W + sw losses - each - so good heat sink to air needed . . . i.e. < 30 deg C rise for 15 watts - check out a heatsink catalog for what is needed . . .

But mainly - Get some better diodes - e.g. MBR20100CT for each diode.

also - snubbers are always a good idea - as is slowing the gate turn on of the fets . . .

--- Updated Friday at 1:27 PM ---

or MBR10150 single diode in a TO-220 - remember the snubbers.

 

[aliarifat794]

Increase the thickness and surface area of your aluminium plate. Improving airflow can significantly enhance heat dissipation. Apply thermal paste or a thermal pad between the diodes and the heatsink to improve thermal conductivity.

 

[Easy peasy]

Also - please learn to take sharp clear pictures of your scope - so that V/div & timebase nS/div can be easily seen - use the HOLD function on the scope to freeze the action so you can power down the ckt and take a pic.