Where would FRED diodes be useful in our applications?

[Mark Hammer]

FRED (Fast Recovery Epitaxial Diodes) are widely available and inexpensive.  I'm a bit agnostic when it comes to diode differences, apart from simple forward voltage, but open to persuasion.  Do FREDs present any special advantage to us as builders/modifiers, or is their "speed" something that really only pertains to frequencies WELL above the range of known instruments?

Just wondering  I had bought a half dozen some years back, and was wondering if I should pull them out of the diodes drawer and put them to work somewhere.

[antonis]

is their "speed" something that really only pertains to frequencies WELL above the range of known instruments?

IMHO, they beat Scottkies only for higher blocking voltage (>200V)..

https://www.ixys.com/Documents/AppNotes/IXAN0044.pdf

[R.G.]

FREDs will probably not offer much audible change. The extra speed is likely not to matter for audio clipping. Their forward drop is not exceptional, and the difference in stored charge and turn-off sharpness probably won't cause audio artifacts. That said, I have not actually tried them as clippers, so this is a guess based on theory.

Where FREDs shine is as power supply rectifiers. Ordinary diodes, even non-FRED fast ones can slam off; that is, the turn-off is very abrupt. This sudden turn off makes any L-C resonances in the wires/traces leading to them ring as squarks of RF. In an audio amp, a solid, high power blips of RF get rectified and detected as twice-line-frequency buzz. It makes for a "hum" that can't be easily eradicated. FREDs fix this in one.

[Rob Strand]

There's nothing stopping you trying some.   I have measured the VI curves in the past I can't remember if they were hard or soft clippers but, as expected, the VI curve is in the general area of other diodes.

Actually, if you look at the plot here the UF4004 and 1N4937 look like they are harder clipping than 1N4004 and 1N4148 but not as hard as diode-wired 2N3904's and 2N7000's
https://www.diystompboxes.com/smfforum/index.php?topic=121964.msg1148986#msg1148986

So you could say they may have something to offer.

[PRR]

....I have measured the VI curves in the past...

Just ran over more data. At first it didn't look normalized but the right side plot runs a best-fit curve.
https://www.diyaudio.com/community/threads/some-common-diode-forward-characteristics.385939/post-7014545

[Rob Strand]

Just ran over more data. At first it didn't look normalized but the right side plot runs a best-fit curve.
https://www.diyaudio.com/community/threads/some-common-diode-forward-characteristics.385939/post-7014545

The fast diode there looks quite different to the common cheap ones.  (Many look like 1N4004's)

If you take two points and compute, based on the diode exponential behaviour,

n  = (VBE2 - VBE1) / (26 * ln (ID2/ID1) ) ; where VBE is in mV and ln() is the natural log

You should end-up with values of n around 1 to 2 for a normal diode, mostly being around 1.8 or so.
For germanium is can easily be 2 or above.
A large n value is a softer characteristic.

For that fast diode I get quite large n values, which indicates it's a funky kind of diode.

It's always a little tricky getting low current data from manufacturer's datasheet but checkout these fairly common Fast Recovery diodes,

https://www.diodes.com/assets/Datasheets/products_inactive_data/ds26001.pdf

Roughly
10mA   690mV
20mA   720mV

n = (720 - 690) / (26 * ln(20/10)) = 1.66

[amptramp]

If you have built any kind of power convertor into your design, FRED diodes are usually required to maintain efficiency.  Some convertors would run extremely hot and have poor efficiency with normal diodes but work well with FRED diodes.