Hi, Gang,
The standard diodes that protects outputs from an inductive kick / reverse voltage are always mounted at the outputs. Is there any reason why they can't be added to the driver board?
Thanks!
Frank
The standard diodes that protects outputs from an inductive kick / reverse voltage are always mounted at the outputs. Is there any reason why they can't be added to the driver board?
Thanks!
Frank
Because they won't protect as well with more wire between them and the output transistors. Might still work. Connect/disconnect the speakers while playing at full volume to test it.
"Connect/disconnect the speakers while playing at full volume to test it."
You've gotta be kidding.
Must be a low inductance connection, so not on another board.
You've gotta be kidding.
Must be a low inductance connection, so not on another board.
Yes, I'm kidding. Don't take bad advice. Protection diodes should be close to the device being protected.
This is true. Once a neighbor was trying to diagnose whether it was a bad spark plug or a bad point in the distributor of his old station wagon. I jokingly suggested to stick finger into plug wire boot to see if it was getting power. Turns out that evidently the distributor was working.....zzZAP!.🙄
If the inductance is in the load (speaker) then disconnecting it under drive won’t test the diodes. Abruptly removing the drive from the output transistors, with the inductive load still connected, is what generates the overvoltage. This can happen with square wave drive at clipping and if the current limit circuit is activated or VAS gives into limit when driving a “real” loudspeaker. That abruptly removes the drive from the output stage, and the output terminal reaches for the sky to keep the current flowing in the voice coil till the stored energy is dissipated.
Why disconnecting the load is bad for tube amps is because the offending inductance is in the transformer which is still there if you disconnect the speaker.
Inductances in the power supply leads of a transistor amp can cause spiking when the speaker is abruptly disconnected. The bypass capacitors absorb this energy and keep spiking to a minimum if they are properly implemented. Diodes need to be close and so do the capacitors.
I believe the protection diodes are for current limiting case where the output goes suddenly high impedance, hence the discharging current of the load can provoke overvoltage, exceeding the outputs voltage ratings.
Anyway - back to my question of putting the diodes on the driver board - I'd have less than 2" of stranded 16 AWG and 1/2" of PCB trace connecting the outputs to the diodes - is the added inductance that significant? I get that a lot of inductance would 'slow down' the protection, I just don't have a feel for how sensitive this is. Has anyone ever done something like this?
Last edited:
That amount of added inductance is not that big a deal at audio frequencies. The frequency the diodes will switch on and off at is the audio frequency you are driving the amp at. Most of the signals large enough to activate them will be below a kHz in normal use, up to 20 kHz if you do square wave testing at full power. That’s not really fast. Normal recovery diodes are usually used for these, and that itself slows the switching down to microsecond or two. Where those kind of connections won’t work is at class D carrier frequencies - you’ll develop spikes that blow out mosfet gates. You need tens of nanosecond switching times for that, and two inches of wire is too much.
Ah - so if I am going to do this, I might as well use the fastest diodes I can. These will be SMT so something like this ultra fast, low capacitance diode from Digi-Key (1655-1961-1-ND) might be a good choice. Thanks, wg_ski.
Last edited:
No, super fast diodes aren’t really required there. As I said, a 1 us recovery time is plenty fast at audio frequency. 1N540x series are commonly used and they work fine.
And if you’re worried that lower speed somehow affects the sound, remember that when the diodes are active, it is because a gross nonlinearity already exists to cause its action. Speeding it up isn’t going to make the distortion go away.
I appreciate the added detail. Full agreement that when protection kicks in, sound quality has degraded a while ago and we are just preventing damage to the outputs.
I havent designed or made someone else's design where these diodes were added.
Never blown up an amp so can only assume there is no real need for them ?
The amp output is low z anyway.
Never blown up an amp so can only assume there is no real need for them ?
The amp output is low z anyway.
The diodes are for when the amp is not low-Z: clipping, slewing, current limiting. All of these effectively break the feedback, causing the output Z to go up by several orders of magnitude.
Output transistors will go into reverse avalanche if the output voltage attempts to go beyond the rail and diodes are not used. This happens rather infrequently - only during certain types of limiting. The output transistors may very well survive the event - if you do not exceed their reverse-biased second breakdown capability (E s/b on the data sheet - not to be confused with I s/b). It’s spec’ed as a total energy for a single event This is often specified for switching transistors, but NEVER for amplifier types so how much stored inductive energy they can take is anyone’s guess. Put the diodes there and you simply don’t have to worry about it anymore. The transistors won’t avalanche at all at one diode drop’s worth of reverse emitter-collector bias.