6000W By IRS2092

[Workhorse]

Originally Posted by Baldin

Ok, did the clamping:
COM to LO to VCC, and VS to HO to VB.
Did it with 4 x 1N4148 as this was what I had.
No improvement what so ever

VS is already clamped to COM through lower bodydiode and external ES1D in parallel.

Don't know what is wrong

IN4148 is too much fragile for handling peak currents atleast use BAV21 or a bigger UF4007.........If you have access to SMD packages then go for it.

[Baldin]

Originally Posted by Workhorse

IN4148 is too much fragile for handling peak currents atleast use BAV21 or a bigger UF4007.........If you have access to SMD packages then go for it.

Hi Workhorse
Yes I know. My thinking was; they are fast, and they might last for a few trials, and it any of them burn, I'll know there is a large current flowing, hense showing that this could indeed be the source of the problems.
Nothing bad has happened .... therefor this is NOT the source of my current problems.

In future designs, these clamping diodes will be in though ... using fast tough SMD types

After further work, I can't help to think it has something to do with the use of LT1711. My daughter board with either MAX913 or MXL1016 works without problems. The current drawn by the LT1711 also seems quite excessive ... being 30 mA DC and AC peaks of same magnitude. This also stresses the small regulators made by single FZT657/FZT757, which becomes very hot.
I'm driving the LT1711 +-5V at both input stage and output. This should be possible according to the data sheet, but I'm begining to thik that this is maybe not a good idea after all (and not really necessary either).

But I like the look of the output of the 1711 ... it's just so square and clean ...
Looking at the output of either MAX913 or MXL1016 it looks nice on the unloaded output, but loaded by the levelshifter circuit it looks a bit strange (see picture below .... second picture is the output of the LT1711).

[Workhorse]

What type of level shifter you are using?

[Workhorse]

Originally Posted by Baldin

After further work, I can't help to think it has something to do with the use of LT1711. My daughter board with either MAX913 or MXL1016 works without problems. The current drawn by the LT1711 also seems quite excessive ... being 30 mA DC and AC peaks of same magnitude. This also stresses the small regulators made by single FZT657/FZT757, which becomes very hot.

After going thru several datasheets of high speed differential/complementary output comparators such as[LT1016, LM360] they seems to be current hungary, nothing to surprise

Try to use SOT-223 package for regulators with good copper plane for sinking heat

[Baldin]

Yes of course, the faster, they are the more current hungry ..... but +-30mA + 30mA spikes is quite a lot.

FZT657/FZT757 are in fact SOT-223 .... but I could give them a bit more copper though

See post #125 for the level shifter:
6000W By IRS2092
It works ok, but could be faster .....

[Workhorse]

Originally Posted by Baldin

Yes of course, the faster, they are the more current hungry ..... but +-30mA + 30mA spikes is quite a lot.

FZT657/FZT757 are in fact SOT-223 .... but I could give them a bit more copper though

See post #125 for the level shifter:
6000W By IRS2092
It works ok, but could be faster .....

Try connecting in Common Base fashion , you are using common emitter type which is plagued with storage time effects.

[Baldin]

Hi Workhorse

Sounds right that a common base type would be faster, but I'm not quite sure how that would look .... could you draw that up?

Thanks in advance, Baldin

[Workhorse]

Originally Posted by Baldin

Hi Workhorse

Sounds right that a common base type would be faster, but I'm not quite sure how that would look .... could you draw that up?

Thanks in advance, Baldin

Omit Q1 & R2 and join R3 directly to output of comparator. If possible then decrease value of R3 to 560ohm and R5 to 680ohm.



BTW why are you using logic gates when your comparator is already giving you 2 out of phase outputs, simply level shift the both complementary outputs of comparator and add Deadtime and feed directly to gate driver chip.

[Baldin]

Hmmm, Workhores that just works so much faster (in sim so far ) .... thanks a lot

Well the logic is there mostly due to history ... started out using a comparator with only one output ... I'll loos it in next round, just going for two level shifters and two following RCD circuits ... less components and more compact design ...

Best regards Baldin

[Baldin]

Ok, tried it out in real life ... just kept R3=2k2 and R5=2k7 for startes.
Both MAX913 output and Level Shifter output looks much better.
Speed is faster also, caused the switching freq to shift from 270 kHz to around 290 kHz.

Will sim a little to find lower values for R3 and R5, which will not burn, to get even better and faster shifting