Fast, faster, fastest...diodes- UF4007 could be a helpful improvement!

[Pete the Builder]

G'day all,

I've noticed in nearly all circuits the ubiquitous 1N4148 and 1N400x diodes, especially as clamping or voltage-doubling diodes.

What you may not realise is that the "super fast" version of the 1N400x family, the UF400x (and specifically,the UF4007) has a reverse recovery time that's over 30 times faster - just 75nS (typical), compared with a "perfect storm" best possible case of 2uS for the older diode types! Amazing, isn't it?

I've been using UF4007s in everything I design for most of the last 10 years, except of course where specific parameters demand a different diode type!

It might be interesting to hear if anyone tries this replacement in your front ends and voltage multipliers, and what sort of results you get. I'm particularly interested to hear if they make a difference to the front ends of the various detector designs. It would be most edifying...

The 1N4148 diodes can also be replaced, in most cases, by the UF4007. The only places I've experienced problems with this type of replacement has been in exceptionally low-current, HF applications, where it appears the UFs don't get enough current to turn fully on, and generate quite a bit of noise. But it's worth trying.

For those of us with "golden ears", any decent audio amps you might have that use the 1N400x series in linear power supplies, might benefit from replacing the 1Ns with UFs. Because the reverse recovery current flows for a much shorter period, some high-frequency distortion may be eliminated in some designs. It's just a tip, and again, if anyone can be bothered, it would be interesting to hear from you if you notice anything at all. (Please note, if you use oxygen-free copper cables, you won't be able to hear any difference, so it's not worth the hassle of replacing the diodes! ). But for anyone with normal amps, it could be fun to try!

I hope this information is helpful in some small way. Please let everyone know if you find it helps.

The oldies are the goodies, right?

Cheers,
PtB

[Midas]

G'day all,

I've noticed in nearly all circuits the ubiquitous 1N4148 and 1N400x diodes, especially as clamping or voltage-doubling diodes.

What you may not realise is that the "super fast" version of the 1N400x family, the UF400x (and specifically,the UF4007) has a reverse recovery time that's over 30 times faster - just 75nS (typical), compared with a "perfect storm" best possible case of 2uS for the older diode types! Amazing, isn't it?

I've been using UF4007s in everything I design for most of the last 10 years, except of course where specific parameters demand a different diode type!

It might be interesting to hear if anyone tries this replacement in your front ends and voltage multipliers, and what sort of results you get. I'm particularly interested to hear if they make a difference to the front ends of the various detector designs. It would be most edifying...

The 1N4148 diodes can also be replaced, in most cases, by the UF4007. The only places I've experienced problems with this type of replacement has been in exceptionally low-current, HF applications, where it appears the UFs don't get enough current to turn fully on, and generate quite a bit of noise. But it's worth trying.

For those of us with "golden ears", any decent audio amps you might have that use the 1N400x series in linear power supplies, might benefit from replacing the 1Ns with UFs. Because the reverse recovery current flows for a much shorter period, some high-frequency distortion may be eliminated in some designs. It's just a tip, and again, if anyone can be bothered, it would be interesting to hear from you if you notice anything at all. (Please note, if you use oxygen-free copper cables, you won't be able to hear any difference, so it's not worth the hassle of replacing the diodes! ). But for anyone with normal amps, it could be fun to try!

I hope this information is helpful in some small way. Please let everyone know if you find it helps.

The oldies are the goodies, right?

Cheers,
PtB

Nothing wrong with the 1n4148. I believe its actually faster than the uf4007. Hard to say for certain since the Uf4007 doesn't specify its switching speed but its certainly recovers a lot faster (only 4ns).

Midas

[Davor]

True, there are only a few pin diodes that may stand in as a 4148 with faster response.

[Pete the Builder]

Nothing wrong with the 1n4148. I believe its actually faster than the uf4007. Hard to say for certain since the Uf4007 doesn't specify its switching speed but its certainly recovers a lot faster (only 4ns).

Midas

Yeah, I was working from a very old (2001) 1N4148 datasheet from Fairchild that didn't actually specify the recovery time! Sorry about overlooking that... I was really surprised at the speed when I found it! Especially since so many of my 1N4148 replacements improved the circuits quite substantially. Of course, I was mainly using them for low-current, high frequency (500kHz+) doublers and rectifiers for HV supplies and so on!

I was quite surprised at how narrowly defined the RRT specs are for all these parts - the spec is only for a particular forward current, with typically 100R load, and tight reverse current. That's always the catch, isn't it?

Your feedback is interesting. It took me about 10 minutes to find a 1N400x datasheet that specified the RRT! Then I checked 3 or 4 different sheets that did specify that parameter, to compare (Fairchild, ON, and DIODE) and they all specified it as 2uS. On the other hand, I can't find a UF400x sheet that doesn't specify the switching speed - which you'd kind of expect, since they're supposed to be super-fast replacements for the 1N series parts. I guess that's the difficult things with datasheets, you have to find the one you want!

FWIW, there are quite a few comparisons of the 1N and UF series on the intertubes. Quite surprising really, when you're used to thinking of diodes as, well, diodes! I suppose it's a carry-over from tech training, where we always idealised the parts!

In any case, I hope all this info in one place helps some others looking for the same thing!

Cheers,
Pete

[ODM]

There's no such thing as a free lunch, though, except for newer and more refined processes for parts. Even still some nonidealities remain.

It's worth a comparison to check the forward voltage vs. current for those fast recovery diodes. Schottky will cure that issue, but will have leakage and lower voltage limits in turn.

The relatively "new" silicon carbide semiconductor parts are showing a bit of promise already.

[Carl-NC]

I've tested a lot of diodes in the front-end clamp, including the UF series and Schottky, and I still use the lowly 1N4148.

[moodz]

I've tested a lot of diodes in the front-end clamp, including the UF series and Schottky, and I still use the lowly 1N4148.

Have to agree there .... fancy schottky diodes actually made the frontend response worse.
IN4148 are a dependable benchmark ... optimising components should be done after you get your design working.