Do you have experience with SBR® (SUPER BARRIER RECTIFIERS) ?

[Miyuki]

Hi folks,
I want to ask if you have experience with SUPER BARRIER RECTIFIERS from Diodes inc.

They have 300V high current device what should behave like a schottky 
That SBR60A300CT look very interesting device compared to common fast diode, especially as I want go to 100-200kHz range at rather high current and common 200V schottky have too low voltage rating

But their datasheets are not very detailed
Do you used them in some project ?

[graybeard]

It does not behave like a Schottky.  Schottky diodes do not have storage time, these have up to 50ns of storage time.  They have models  you can try and run simulations with to get an idea of how they might work.  I would build a test circuit with some samples and try them if the simulations look promising.

Look at the spec sheet https://www.diodes.com/assets/Datasheets/SBR60A300CT.pdf

[T3sl4co1l]

Shrug, I used a SBR20A200CT last year, didn't seem to get hot at all.

AFAIK, it should be much like a synchronous MOSFET rectifier: at nominal currents, the charge injection is small and recovery is negligible; at high currents (such that Vf > 0.6V say), injection occurs, r_s drops and t_rr rises, perhaps considerably so if this is based on MOSFET behavior.

Which in turn means they're great for current mode converters, but not so great for snubbing.  And presumably they're more than capable, but merely more expensive than P-N types, for simple rectification, fault protection, wired-OR, etc. (With a YMMV, as prices vary of course.)

I think I'll put one in my recovery jig and test it.

Tim

[T3sl4co1l]

Yeah, regular schottky always have a guard ring that handles avalanche (ESD and such), and will inevitably get forward-biased under surge conditions.  Part of the reason I suppose why schottky tend not to have great surge ratings, the active area is relatively small.

Also why SiC schottky have such a bizarre Vf(If) curve -- the high resistance, and higher Vf of the schottky and PN elements, exaggerates the response.

Tim

[T3sl4co1l]

Testing --

Oh, wouldn't you know it, the SBR20A200CT doesn't appreciate up to 800V and um, I think the transistor in here will do upwards of 30A short circuit?  Probably survived a few hits, but with the ~100Hz repeat rate, that's still not very long.
Which explains why I wasn't seeing 100V across the diode, once I turned down the supply to a more reasonable 100V.
Fortunately there are two diodes in one package, eh?

Forward bias 5A for 200ns, at 25°C:


10A


20A


That's -5A/div, so the trough is forward bias.

0. Ringing

This seems to be ringing more than usual, not sure why.  There are three "rings" of note:
N/A. The undershoot just after the trigger is due to dI/dt in the current sense resistor, that's normal.
a. There's also an odd kink on its way there, it's not a smooth drop.  Gets worse at higher current.
b. When the transistor turns on, there's another overshoot, which I don't usually see.
c. When recovery is done, there's a huge ping (sharp recovery).  This is fairly normal, but I do have an R+C in the circuit which usually deals with it...

Wonder if it has anything to do with capacitance?  Which, they don't actually publish...yikes...  Perhaps dV/dt activation?  5kV/us isn't very crazy, but they do have a limit of only 10kV/us.  Also wonder now if I should've been watching the voltage too, maybe this thing has weird dynamic voltage drops?

Capacitance measures 1.1nF with the DMM, so, around a few volts.  Probably a tiny fraction of that towards rated voltage (100-200pF?).  Anyway,


1. Forward Recovery

Doesn't appear to have any forward recovery, at least of any significance compared to stray inductance in the jig (about 30nH?).  High-speed diodes vary, but this is very typical of MOSFET body diodes for whatever reason.


2. Reverse Recovery

- Very little dependence on forward-bias time.  Looks about the same for t_on from 40ns up (the lower limit of my jig).
- Likewise, no effect on cutoff sharpness.  This is NOT a soft-recovery diode!  Probably, plan on using ferrite beads with it.

t_on invariance is interesting.  Injected charge takes time to diffuse into the junction, and for pulses in this regime (< 100ns, for most power diodes), the charge can have a nonuniform distribution.  When reverse bias hits, the depletion zone rips open and all those charges get sucked out, where they hit the electrodes nonuniformly, determining circuit voltage and current.  You can get a recovery pulse that is lumpy, and usually sharper at the end.  This is the phenomenon behind the Grekhov or drift step recovery diode.  At least one guy has [ab]used a common TV (damper) diode to deliver a 1kV, ~1ns pulse into 50Ω, using this effect.  Some diodes are snappier than others; depends on the doping profile.

A lot of power diodes are intentionally quite poor at drift step recovery: a soft-recovery characteristic leads to more recovery loss, but the reduced EMI can easily be worth that much.  (The snubber losses required to use a sharp-recovery diode, while meeting the same emissions level, might indeed be higher!)

- Recovery time does increase with current.
- The increase doesn't seem to have a threshold, at least in this current regime.  Mind that my jig gets slower at higher currents -- dI/dt is not constant, so this isn't great data.

If (A)	dI/dt (A/us)	t_rr (ns)	I_rr (A)
5	~1000	25	11
10	500-1000	40	12
20	600	70	12

Tim

[Miyuki]

T3sl4co1l Thanks so much

you have 5A/div so Irr should be just half that value, about 11A

It is interesting waveforms
So Irr remains fairly constant and rather high compared to little data that was in datasheet
And that recovery slope is over 5000 A/us if I see it right, that can cause a problem

It looks like only benefit from common ultra fast diode is much lover Vf, switching performance is not so superior for 200V and 300V devices

That explain why there is so little information in their materials

[T3sl4co1l]

Oops, off by one, thanks.  Corrected.

I wouldn't put too much stock in the apparent peak dI/dt, because as I hinted, there's ESL in the circuit.  Look at the overall trend instead.

Tim

[Miyuki]

Do you have some common fast diode like DPG20C200PB or similar to see how it behave on same test rig