Reverse recovery

I just grabbed my old TI discretes data book. On the data sheet for the

1N4148 they specify 50mA forward current for the reverse recovery time test.

So what "always" are you talking about?

--
Tim Wescott
Wescott Design Services

Unfortunately the absurdity doesn't seem to extend to signal diodes, a quite delicious absurdity as they wouldn't even withstand the 1A, at least not continuously. It does appear to extend in the other direction, despite arbitrarily powerful diodes.

I am hard pressed to find a *power* diode which differs. Rarely, graphs at rated current are included at least. I did not look at large modules or hockeypucks, which don't seem to be listed on Mouser, at least not "ultrafast" types.

10A 200V, If = 0.5A, no graphs

16A 200V, If = 0.5A, no graphs

15A 200V, If = 1A, with graphs

15A 1200V, graphs. Seems to me, IXYS at least has nice graphs, if only they'd use useful numbers up front.

50V 30A, If = 0.5A, no graphs

600V 120A, If = 0.5A, with graphs

600V 280A, If = 1A, with more rows and graphs

A sufficiently representative sample?

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

I can certainly see your point with the ones that are rated much above

10A -- not to mention 100A!

The real question is -- how much current does it take to make sure that the diode is saturated with minority carriers, thus making the reverse recovery time measurement representative of what will happen at rated current?

--
Tim Wescott
Wescott Design Services

What's even more bizarre, if they do that on the really big diodes (>100A), capacitance will be so huge relative to the 1A reverse current that they'll look even *slower*! Same as calling a schottky "4ns" or something, which is based only on junction capacitance and test current.

Presumably, all junction diodes use approximately the same current density (allowing for differences in voltage and speed), so they'll be fairly well saturated at rated current. Somewhere up the knee, where they start looking resistive (1~2V).

The diodes that provide a graph usually have parameters of If, dI/dt and Tj, and end up in the 200ns range for something like dI/dt-snubbed hard switching (like CCM buck/boost). It would be nice if they'd just use that number to begin with.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

y l

hat

I had same experience when selecting diodes, If budget allows and you need high voltage, you may consider SiC diodes. As far as I know, reverse recovery charge/time doesn't increase with temperature.

Best regards,

Wim PA3DJS

without abc, PM will reach me very likely

Reverse recovery time of what? SRD are spec'd at lower numbers, like

10 or even 3 mA in some cases.

John

Especially fun is a power schottky that has PN guard rings. There may be no reverse recovery at 1 amp, but tons at higher If.

John

I disagree, it is arbitrarily insane..

Why not be more realistic and measure at rated current and at a commonly used (for THAT diode) current?

for a particular diode the relationship between stored charge and forward current or di/dt is a direct one. Reverse recovery time shows an inverse relationship to Ir/If in current-contrled situations.

Jedec published data for early fast rectifiers included a typical test circuit drawing and graphs illustrating typical stored charge data for different junction temperatures, forward current and di/dt. The relationship to temperature may be non-linear; increasing with temperature to a point where it begins to reduce and is process dependant.

While the one amp recovery time spec will often be quoted, larger-rated devices will often include data for at least one other higher current level.

Jelly-bean components will be supplied with jelly-bean paperwork and should only be applied where jelly-beans will function adequately. A good designer will work with what's available.

RL