Diode reverse recovery

[AlfBaz]

I was messing around with some simulations when I got this weird result.
It took me a while to realise that it's probably due to poor reverse recovery.

This is the result from running 100kHz 4Vpk-pk through a 1N4007 into a 500ohm resistor.
1. Is this waveform realistic or is spice exaggerating?
2. What, if any, common diode parameters (datasheet ones) give a hint as to the performance of reverse recovery?

PS other diode models produce a much, much smaller negative going dip

[c4757p]

That's what I got too. Seems realistic to me.

2. What, if any, common diode parameters (datasheet ones) give a hint as to the performance of reverse recovery?

The reverse recovery time. Most diodes list it. Seems like it is left out of almost every 1N400x datasheet, though. I was able to find one that does list it (General Semiconductor), and it is listed as 30 µs, so it definitely can't handle a 100 kHz square wave. The slow recovery makes them good for mains frequency rectification, though, because it eliminates harsh, high frequency noise when they switch off. The UF400x series has a much faster recovery time, if you are looking for something similar but faster.

[ivan747]

What other diodes? A 1N4148 signal diode would be much faster than this rectifier, of course. If you have an oscilloscope in hand, by all means test it. 100kHz in a sine wave shouldn't be hard to generate and virtually all scopes can do 1MHz (to see how the diode recovers more sharply). It should be a fun experiment. I would do it, but I lost my oscilloscope.

[Marco]

Breadboarding it would also be the only way to get some useful waveforms ... modeling of reverse recovery is generally poor.

[c4757p]

OK, I'm a bit intrigued and bored - had to do it myself. Here are some waveforms. Top to bottom: 1N4148, 1N4007, BAT85S (Schottky), input signal. (Ignore the scales at the bottom. They should all be 1V, but I don't have four probes with the 10x indicator pin, so the scope thinks they're 100mV). Forgot to include the UF4007.

(Pardon the crappy picture. Bench was a mess so I set the circuit up right in front of the scope, leaving no room to set up a camera...)

I'd say the SPICE reverse recoveries were quite accurate, actually. It also did a good job with the relative amplitudes of the signals. (I used the PSpice models in the graph I attached earlier, which you can easily find online and which work fine with LTSpice. They seem to be significantly better than the LTSpice ones and include just about every device out there.)

[BravoV]

Even schottky recovers really fast, still has some losses, now what about those so called "zero" reverse-recovery loss SiC diode like the one made by Cree ?

Zero loss ?  Is this just a marketing jargon ?

[Bored@Work]

As it is stated in one of Linear's application notes about a particular 5A, 40 kHz switching regulator

The LT1070 will eat 1N914 and 1N4001 diodes and not even burp. Diode currents, especially during startup, can exceed 5A. This takes care of the 1N914s. The 1N4001s will last for a little while, until the heat generated by their horribly slow turn-off characteristics causes them to self-destruct.

[SeanB]

Just hype, they are just very fast recovery devices, with low stored charge. The problem with the 1N400x devices is that the reverse recovery differs from manufacturer to manufacturer and from batch to batch quite widely. It is not tested for on production except by sampling to see it is less than some worst case limit. I would not use a 1N400x above 1kHz, if I need the fast recovery use the UF400x devices. Even at 1kHz the losses in a 1N400x device are rising.

[AlfBaz]

OK, I'm a bit intrigued and bored - had to do it myself. Here are some waveforms....

Thanks heaps, I was hoping someone would try it 
All my gear is in storage and at this point in time I have no where to set it up

[AlfBaz]

As for how this came about I was messing around with w2aew's diode switching circuits in pspice and just picked the first diode that came up

[Jay_Diddy_B]

Hi Group,

Diode Recovery is a function of forward current and the rate the current is changed from positive current to negative current, -dI/dt.

In order to publish short times on their datasheet, manufacturers can pick unrealistic conditions. For example testing a 10A diode at 1A. They normally publish their test circuit.

They also use a circuit where the inductor is replaced with a resistor. The set the value so that the reverse current is 2x I fwd. This gives very misleading results.

Several years ago I built a test fixture so that I could compare diodes under the same conditions. This is an LTspice model of the fixture.



V1 is used to set the forward current in the diode under test. The combination of V2 and L1 sets the -dL/dt. A low value resistor R2 allows the current to be observed using a scope.

I  have arranged the forward current to be stepped through a series of values. You can see how the recovery time changes as a function of forward current.




I have attached a zip file containing simulation for the MUR460 and the 1N4007.

Jay_Diddy_B

[Jay_Diddy_B]

Hi Again,
Just for completeness here is the resistor based tester:


And the results:


Jay_Diddy_B