1. Home
  2. Electronics Design
  3. Diode Recovery Tester

Diode Recovery Tester

formatting link

formatting link

This more or less replicates the conditions specified in most datasheets, and is very adjustable: pulse duration, forward current, ramp rate, and commutation voltage are all adjustable. (Well, test power is external, so you can use an adjustable supply for that.)

Connect an inductor between the labeled terminals, then connect a diode.

Warning: short circuit current is not limited, beyond what the transistor can do. Testing a diode backwards may explode it. (That said, a 1N4007 will handle 80A for a few microseconds just fine.)

The snubber circuit clamps open-circuit voltage, so the circuit does not need to be stopped while swapping out diodes. (TVS are impressively robust and handle this service easily!)

There are only small differences between the circuit shown, and what I've built. The big one is the transistor, where I used some beefy salvaged IGBT (RJH6075 -- it "doesn't exist", but it's probably 600V, 75A, and 200 or

300nC). It's suitable for most diodes (which are tested in the 100-1000A/us range, for whatever reason), but too slow to see interesting behavior.

Observations:

- I haven't seen any spooky (SRD-like) behavior, on any of the diodes I've tested, in the operating range I was testing at. Dang.

- Zener diodes are generally pretty fast (t_rr < 50ns), and quite sharp (sharpness factor < 0.3?). Might be useful?

- Most diodes do have t_rr varying with t_on, but Irm varies too. This seems an inefficient way to sharpen a pulse. That is, unless Irm gets disproportionately large at high dI/dt, which seems likely, and presumably leads to the region of SRD-like behavior.

- Some junctions have t_rr varying with t_on, even for t_on > 10us. I interpret this as, it takes a few recombination times for charge to "soak" in and reach steady state. (Accordingly, it should take as long for Vf to stabilize, but Vf is ~logarithmic with charge(?), so it would barely be noticeable.)

- I think the slowest diode I have is HIB120 (high voltage, 3A size; can't find any data), t_rr > 5us.

- Horizontal output transistors are unimpressive as diodes (unsurprising given their ~us t_stg, I guess).

- I have some surprisingly fast diodes: MUR880E for instance, lives up to its datasheet. Others I don't remember at the moment, but have salvaged from power supplies and TVs, with similar performance. And schottky of course, which do indeed exhibit nearly (or actually) zero recovery.

Ooh, now I can test if SiC schottky exhibit nasty recovery drool at pulsed currents (i.e., where Vf is high enough to forward bias the guard ring PN junction).

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com
Reply to
Tim Williams
Loading thread data ...

Here's a DSEP29-06A at 10A, 60V, and 1350 A/us:

formatting link

Ch1 is 10x probe, 20V/div. Ch3 is 5A/div.

I'm not sure how much of that is stray inductance in the diode, and in the current shunt resistors. Probably at least the voltage step is diode stray (which I'd eyeball around 20nH; a 14V step at 500 A/us is 28nH, so that's probably most of it).

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website:

formatting link

Reply to
Tim Williams

There is really something to be said for color-coded traces!

One complication here is that the forward conduction time profile can greatly affect sunsequent reverse recovery behavior. Most data sheets assume DC forward conduction before the reverse measurement. The forward timing effect can affect high-frequency switchers, sometimes in a vey bad way.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics
Reply to
John Larkin

My apologies for not using different color electrons in the CRT!

Not that I've seen, in the better switching diodes -- it seems Pt doping and electron irradiation do a superb job taming charge effects.

Compare:

Type: BYW98-200 If = 5A t_on > 1us Vr = 100V dI/dt = 500 A/us Results: t_rr = 28ns I_rm = 7.4A No apparent drool. But, this particular model is so sharp, it's ringing there, and I can't tell.

Type: C3D04060 If = 5A t_on > 1us Vr = 100V dI/dt = 500 A/us Results: t_rr = 12.8ns I_rm = 3.4A Recovery does not vary significantly with If (up to 20A), which is nice. Probably, the guard ring is easily forward biased, but is still pretty fast all its own.

And that's a rather old type, BYW98. MUR880E is just as fast, but softer (some drool: enough to help with ringing). Some other random diodes in here also do very nicely.

In contrast, I've a 5TU47 here (Toshiba 1500V 5A CRT "damper" diode, t_rr =

0.6us) that does, well, exactly that. Check this out,

If = 10A t_on > 1.8us Vr = 100V dI/dt = 100 A/us Results: t_rr = 600ns I_rm ~= 10A Recovery is pretty slow as-is, but it's the tail of drool (current draw persisting after voltage has commutated, which only takes 200ns) that would really kill efficiency, if you used this in a switching application.

For short on-times, its recovery is approximately proportional to stored charge, and the drool is non-existent:

If = 10A Vr = 100V dI/dt = 500 A/us

t_on = 100ns Results: t_rr = 100ns I_rm = 18A Still well within forward recovery during t_on.

t_on = 200ns Results: t_rr = 150ns I_rm = 26A

t_on = 400ns Results: t_rr = 240ns I_rm = 30A

Note t_rr and I_rm continue to rise. The trend is flattening out, until by about t_on = 1.8us, it's more or less saturated.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com
Reply to
Tim Williams

Right. It gets interesting as the forward-bias time is reduced. As you would expect, the stored charge goes down so reverese recovery time goes down. But the waveform of the reverse recovery can change, namely the diode can start to look like a step-recovery diode, with an insanely fast snap-off edge. That can cause problems.

As far as I know, the first exploitation of that effect was in the HP

1430 12-GHZ sampler ca 1966. They forward biased the step-recovery diode for a few ns (instead of previous DC bias) to make the reverse recovery faster. The Grehkov DSRD effect is similar, in power diodes.

Here's a 2KV pulse. A diode is forward biased at 48 volts, and the current ramps up linearly to about 100 amps or so. Then it is reverse biased from a 400 volt supply, through a small inductor. At about -200 amps, it snaps off.

formatting link 

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics
Reply to
John Larkin

Checking my archives, I had done this already -- though not with the same adjustability and capability of the newer circuit:

formatting link
I wonder what dI/dt this had achieved. Couldn't be too much, it was air wired (I have a picture of that floating around somewhere, too).

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website:

formatting link

Reply to
Tim Williams

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.