Soft recovery diodes

[Rhodesplyr]

I remember about 10-12 years ago when the DIY Hi-Fi community discovered and adopted soft-recovery diodes as the standard for solid state rectified power supplies, and, from what I've read, Schottky diodes are even better. I was wondering if that trend had ever found its way into musical instrument electronics.

The idea was that using soft-recovery diodes would reduce the broad spectrum switching noise of common diodes, making things like snubbers unnecessary. For example, these days, you can buy UF4007 diodes, the "ultrafast" soft-recovery version of the 1N4007.

I've done a couple of diode upgrades in Hi-Fi equipment where it did seem to yield some improvement, but I tried it in one of my solid-state rectified Leslie 122 amps, and it made no difference whatsoever--just to prove that I don't believe in "magic parts" solutions.

The reason I was thinking about this was because the Mesa Boogie amp I was working on recently has some noise in the input stages, and while some of it looked like typical background thermal noise of a high gain stage, the scope trace also showed very regular vertical spikes that looked like diode switching transients. The lower voltage power supplies are on the main audio circuit board, and, being from 1988, it uses standard 1N4006 diodes.

Please don't stereotype me as a believer in audiophile snake-oil. I'm really not. On one hand, I like to make use of advances in component technology if they are worthwhile, but, on the other hand, I try to maintain a healthy skepticism.

[Alex R]

It would be interesting to hear if fitting UFs changed the trace (and the noise). Whatever theory says, there are grey areas in audio electronics wherein differences that might not show up in specs do change the subtle listening experience. On the other hand, people over 45 and all guitarists should stay out of such discussions because their hearing is FUBAR anyway :-).

[Steve Conner]

I've tried various different kinds of diodes in power supplies of both tube and solid-state amps.

The deal, as I see it, is: Provided the power supply is designed correctly, standard recovery diodes won't add any noise into the audio. Therefore, changing them out for fast/soft/Schottky/whatever audiophile grade diodes, won't make any audible difference.

If the designer screwed up, though, it might well make some difference. You would hear a buzz in the audio because the switching spikes repeat 100/120 times a second. But his screw-up would probably make the unit fail EMC before it degraded the audio: most of the energy in recovery spikes is in the medium-wave radio band, not audio. To pass EMC you need to put snubber capacitors across standard recovery diodes.

Even with ideal diodes, you can still get 100/120Hz spikes into the audio path by screwing up the design. The reason is that the diodes conduct in sharp spikes of high current anyway, and the more "audiophile" the equipment (oversized toroidal transformers, oversized filter caps) the sharper and more vicious the spikes are. But again to emphasize, with proper design you can make them inaudible. I'd bet this is what's happening in the Mesa Boogie mentioned above.

I'm sure you'll want to try it anyway, so on to diode choice. Schottky diodes aren't available in really high voltages, except for the new silicon carbide ones, which surely must have some audiophile kudos, if only on account of their price. IMO, Schottkies are only really useful for DC heater supplies, where their low volt-drop reduces heating and so on.

The MBR340 is a decent "jellybean" schottky diode, rated 3A, 40V. I've also used the 1N5819 but I can't remember its ratings. The UF400x and UF540x are convenient fast diodes that can directly replace the 1N400x and 1N540x. If you want fast diodes for a solid-state power amp, there are a multitude of TO-220 packaged things meant for SMPS use. I used the STPR1220F once, but for every SS amp I've built since, I've used a regular bridge rectifier.

One place Schottkies REALLY make a difference is the commutating diodes in Class-G power amps. You can measure a decent reduction in THD with them.

[Rhodesplyr]

If the designer screwed up, though, it might well make some difference. You would hear a buzz in the audio because the switching spikes repeat 100/120 times a second. But his screw-up would probably make the unit fail EMC before it degraded the audio: most of the energy in recovery spikes is in the medium-wave radio band, not audio. To pass EMC you need to put snubber capacitors across standard recovery diodes.

I'm thinking about this question in terms of older equipment, so that raises the question of when EMC rules and standards were formulated, disseminated, and enforced. The vast majority of equipment that's come across my workbench has standard recovery diodes without snubbers. So my question is more in the context of retrofitting lower noise diodes into old designs vs. incorporating them into new designs.

I agree with your guess about what's happening in the Mesa Boogie amp. It has unfiltered AC running across broad swaths of the PCB, and it uses marginally filtered DC to switch Lead/Rhythm modes via LEDs/LDRs. I have a feeling that there's some coupling happening between noisy control voltages and low-level audio, especially where the leads are close together at the LDRs. One clue to this was that some previous tech had attempted to increase the filtering of the DC control voltage supply. In fact, I called Mesa Boogie to check to see how much noise was normal, and the tech I talked to freely admitted that this amp (made in 1988) was a noisy design, adding that low noise really wasn't a concern for them at the time.

One place Schottkies REALLY make a difference is the commutating diodes in Class-G power amps. You can measure a decent reduction in THD with them.

They've also proven useful in replacing germanium diodes in older radio tuner detector circuits, but that's certainly a more obscure application than Class G amplifiers.