diode recommendations

[classd4sure]

Hi,

I just tested a few different diodes in my circuit simulation with results I can't understand.

Diodes tested: 1N4003, and ES2D

They were for the body diode bypass of the mosfets in a dual N channel half bridge, as well as for the half bridges bootstrap diode.

They should be fast, or very fast...correct? With the 1N4003 being a standard diode..I wanted to try replacing it with later, and expected improvement.

So, with all three diodes being 1N4003's, THD=1.1% and a DC component of -0.2 (yeah yeah.....it's a work in progress..)

I then replaced only the body diode bypass ones with the ES2D's and got THD=1.4% and DC component of -0.4 ......strange? They should be alot faster.

So I changed those back to the prior ones, and then tried the ES2D as the bootstrap diode.

THD from that run = 0.92%, DC component -0.299.

I'm basically ignoring the DC component for now..I think the input stage still needs work, but these changes are affecting it.

How is it the ES2D is better as a bootstrap, (also pumps it up a few extra volts), so I get a significant improvement with it there, but as a bypass diode, it's worse????

Can anyone recommend a different model I might want to try for these applications as well?

Last question, how reliable are the THD results with pspice, and what's the lower limit it can be trusted for, if any?

First few runs of the circuit had THD of 480+%....I got it down pretty quick but the last few I expect to take awhile.

This is for a straight-forward UCD circuit by the way, no fancy deviations from patent at all, so for this reason I'm working towards far better results, as I know it's capable.
Thanks alot

Chris

 

[ekaerin]

Chris,
The MOS_FETs you are using, don't they have internal body diodes
by nature. Maybe the 4003s make things worse cause they recover
slower then the internal body diodes of the FETs causing spikes when
the lower FETs turn on.
If this is the problem, wait till it gets hot and it takes even longer for the
minority carriers to sweep away electrons, ie longer recovery times.

Dunno, exactly what the circuit looks like it is just an idea.

Also, lets say you have 10 Ohms to charge the gates, try put another
10 Ohms in series with a diode to discharge the gates. Then you can
experiment with different values to have controlled turn on that is
within the dv/dt limits of your body diodes.

Regards / Mattias

 

[peranders]

Chris, The MOS_FETs you are using, don't they have internal body diodes
by nature.

Is that really possible? I thought you can't get rid of it. Can you explain how such mosfet is built?

Hej and välkommen, BTW.

Chris can you show us what you are doing? 1N4003 is a typical rectifier diode for 50/60 Hz applications. You would probably need some sort of fast recovery diode.

 

[Circlotron]

For a given voltage and current rating, a fast diode generally has a greater forward voltage drop than a slow one. Therefore, the fast diode may not be bypassing the body diode very well because the body diode may have a lower drop than it. This is assuming you only have a single diode across the mosfet.

If this is so, try a second fast diode in series with the mosfet drain and the original fast diode now across these two. The second diode can actually be a low voltage schottky no matter what the switched voltage. In fact if the voltage is not too high, the first diode can beneficially be a schottky too.

 

[peranders]

As a contradiction to circlotron, body diodes usually have higher losses that corrensponding separate diode if you take the same current rating.

As a summery: The problem is complex.

 

[ekaerin]

Per Anders,
I have never seen a MOS-FET device without body diode. It comes for
free by the internal structure.

And yes, it is complex. The easiest thing is just to try another device.

Circlotrons suggestion also works but involves more components that
needs to be selected carefully.

/ Mattias

btw, remember storpotäten......

 

[classd4sure]

Hmmmm

Irf630's, (not the best for the application I'm sure), 9amps diode forward voltage =0.85 to 1.2 volts, while at 9 amps the ES2D should have about 1.3 volts....aaaahhhhhhhhhhhh (seeing the light).

Data sheet doesnt' say what the 1n4003 would be at 9 amps, and I dont' care to test it since it isnt' rated for that much current...it's simply the wrong part for the job, but I'd think it a safe assumption that it would be lower than 1.3 volts.

So that explains alot. How much current would it have to handle before the body diode kicks in....is another question..

Thanks alot, I've learnt something today

Chris

 

[peranders]

1N4003, do we talking about a small 1 A diode?

http://www.onsemi.com/site/products/summary/0,4450,1N4003,00.html

 

[ekaerin]

.....To predict when the body diode kicks in is a hard task.
Even if the parallel ext. diode is faster with lower drop, stray indactance
in the layout will make the race tuff with the body diode during switching.
If you must use a body diode try Circlotrons suggestion:

Put a shottky (low Vf drop=low loss) in the source of the upper FET
and in the drain of the lower FET. Both connected to midpoint.
Bypass from midpoint to rails with a fast low Vf drop diode of your choice.

Now, you force the reverse current through the external diode and
you can forget about the body diode.
There is nothing like a free lunch however. More conduction losses
(Id x Rdson + Vf_shottky) plus extra parts=cost and space.

/ Mattias

 

[ekaerin]

S__t

"f you must use a body diode try Circlotr"

I mean external diode...........

/ Mattias