Opinions on various protection methods

[MORE_Guitar_Solos]

Hey everyone,


I'm kinda re-entering the game again after a hiatus from playing/building of about 10 years, and I'm noticing the costs of spares has gotten pretty ridiculous and this has me considering installing some protection on my black panel Bassman and home-built AC15 and 5E3. The Bassman is a touring veteran without a failure and the AC15 has been a studio/practice space amp since it's "not loud", but will probably be getting some gigs as these days not loud is kind of required in most clubs.


I'm curious about the utility of the following additions for these amps:
1. Backup diodes on tube rectifier (AC15 and 5E3)
2. B+ fuse (all amps)
3. Flyback diodes on power tubes (do these make sense on cathode biased amps?)


Never used any of these in the past and never an on-the-road failure, but shit is getting so expensive I think some protection might be wise. Probably opinions are divided on some of these, but I'd appreciate any wisdom y'all have to share.


Thanks in advance
Dave

[pdf64]

My view that 1&2 are a default, essential good practice, there's no point not doing them.
Marshall's greatest contribution to the valve guitar amp is introducing the concept of the HT (B+) fuse.

3 I don't like, they alter the tone and the typical failure mode is unsafe, ie they short and subject all the expensive stuff to fault current.

Plus there's better methods of protecting the OT, my preference is a load resistor soldered across the OT secondary, value 20-30x the nominal load impedance, wattage rating 1/5th-1/2 of the maximum power output (into the intended load).

The point is to ensure that the secondary never becomes open circuit, as that's when a resulting inductive back emf spike might become damaging.

[Merlin]

3. Flyback diodes on power tubes (do these make sense on cathode biased amps?)

Flyback diodes are definitely recommended for push-pull amps, cathode biased or not (diodes are not needed for SE amps). The most common cause of OT failure is flyback voltage breakdown, and those diodes provide outstanding protection against that. A resistor across the secondary is also recommended as pdf64 said. However, I disagree with him about tone, the diodes do not alter the tone, unless it's the tone an output transformer a few moments before it sparks over!

[MORE_Guitar_Solos]

Thanks for the excellent replies, guys.


Couple of questions:


For the flyback diodes, I understand these go from plate to cathode, and on a fixed bias amp this is essentially plate to ground. On a cathode biased amp, should they connect to the top of the cathode resistor or to ground?
Also, best place for the B+ resistor: Between rectifier and reservoir cap?
Finally, any recommended diode types for these applications? I have some 1000v 1N5408's on hand, assume these would be sufficient?


Thank you!

[MORE_Guitar_Solos]

Plus there's better methods of protecting the OT, my preference is a load resistor soldered across the OT secondary, value 20-30x the nominal load impedance, wattage rating 1/5th-1/2 of the maximum power output (into the intended load).

The point is to ensure that the secondary never becomes open circuit, as that's when a resulting inductive back emf spike might become damaging.

One of my amps has multiple taps each with a jack (4, 8, 16 ohm), rather than a selector switch. Put a load resistor on each?

[pdf64]

Plus there's better methods of protecting the OT, my preference is a load resistor soldered across the OT secondary, value 20-30x the nominal load impedance, wattage rating 1/5th-1/2 of the maximum power output (into the intended load).

The point is to ensure that the secondary never becomes open circuit, as that's when a resulting inductive back emf spike might become damaging.

One of my amps has multiple taps each with a jack (4, 8, 16 ohm), rather than a selector switch. Put a load resistor on each?

To ensure current can always flow, only one is needed.
I suggest 470R soldered directly between the 16 ohm and common wire ends, wattage as appropriate to the amp.

[Merlin]

For the flyback diodes, I understand these go from plate to cathode, and on a fixed bias amp this is essentially plate to ground. On a cathode biased amp, should they connect to the top of the cathode resistor or to ground?
Finally, any recommended diode types for these applications? I have some 1000v 1N5408's on hand, assume these would be sufficient?

Always plate to ground. You want the diodes to have a reverse voltage rating of 3x B+ or more. 1N5408 or 1N4007 are rated for 1000V so they're fine for B+ voltages up to about 330V, but for higher voltages you can use two in series. That's exactly what I have in this prototype amp where my B+ is 470V (the diodes are wired to the ground terminal on my 16ohm winding because that was a convenient tie point for me)

[pdf64]

... the diodes do not alter the tone ...

... You want the diodes to have a reverse voltage rating of 3x B+ or more. 1N5408 or 1N4007 are rated for 1000V so they're fine for B+ voltages up to about 330V, but for higher voltages you can use two in series. That's exactly what I have in this prototype amp where my B+ is 470V (the diodes are wired to the ground terminal on my 16ohm winding because that was a convenient tie point for me)

I tried that arrangement (3x1N4007 string per push pull side) on a couple of my amps.
I got a bit suspicious about the tonal impact, so with my old AC30, I put the ground returns of each string on a pole of a dual pole switch.
Thereby allowing a fairly instant AB comparison of the diode / no diode tone.

With an amp used clean, I couldn't detect any difference. But cranked up (into a real speaker, Fluxtone Model 10), switching in the diodes caused a noticeable smoothing of the tone (I think my hearing tops out at about 10kHz).

It's certainly not a night and day difference, but I found it noticeable, and I'm normally the guy that's mystified by people claiming to hear differences in cap / resistor etc brands.
The change in the amp's output waveform was apparent on a scope.
Unfortunately I didn't record any audio or take scope photos.

So it may be worth your while trying a similar AB test 

Additionally, when I've checked overdriven output valve anodes with a 100:1 probe, the back emf spikes look to be well over 2kV, so I'm not sure 3kV of diode would be sufficient in the case of, eg a dodgy intermittent connection in the OT secondary circuit (though the back up safety resistor might mitigate that).

[printer2]

Smoothing of overdrive tone? That sounds like a win. If too much, bias cathode biased amps to give some cross over distortion. Fixed bias, your on your own.

[MORE_Guitar_Solos]

... the diodes do not alter the tone ...

... You want the diodes to have a reverse voltage rating of 3x B+ or more. 1N5408 or 1N4007 are rated for 1000V so they're fine for B+ voltages up to about 330V, but for higher voltages you can use two in series. That's exactly what I have in this prototype amp where my B+ is 470V (the diodes are wired to the ground terminal on my 16ohm winding because that was a convenient tie point for me)

I tried that arrangement (3x1N4007 string per push pull side) on a couple of my amps.
I got a bit suspicious about the tonal impact, so with my old AC30, I put the ground returns of each string on a pole of a dual pole switch.
Thereby allowing a fairly instant AB comparison of the diode / no diode tone.

With an amp used clean, I couldn't detect any difference. But cranked up (into a real speaker, Fluxtone Model 10), switching in the diodes caused a noticeable smoothing of the tone (I think my hearing tops out at about 10kHz).

That's super interesting, pdf64, and I'm one who DOES NOT want any additional "smoothing" of the tone! Thanks for sharing this!

[stratomaster]

A common protection diode is the R3000.  It has a datasheet junction capacitance of 7pF.  About half that of the 1N4007.  I'd have to imagine the capacitance of 3 in a row adds instead of adding inversely as series caps would.  In that case it, the 45pF of 3 series diodes may indeed be sufficiently high capacitance to make an audible difference under certain conditions.

I'd say stick with the R3000 for lowest capacitance and retaining the flyback protection.

There's also the HV3, but that is at 15pF.  Probably just on the cusp of audible given the popularity of 22pF caps in certain Marshall circuits for stability and smoothing.

[Merlin]

Additionally, when I've checked overdriven output valve anodes with a 100:1 probe, the back emf spikes look to be well over 2kV, so I'm not sure 3kV of diode would be sufficient

Those spikes only exists when the diodes are not fitted. The diodes themselves never see those back EMFs because they're clamped by... the diodes! They only see the normal audio voltage.

I'd have to imagine the capacitance of 3 in a row adds instead of adding inversely as series caps would.

Why would you imagine that? The diodes in series share the voltage, and junction capacitance scales non-linearly with reverse voltage, so the total capacitance is still less than one diode alone. Even a single 1N4007 is less than 6pF in our scenario, which is negligible. But if you like the sound of extra 'grit' provided by voltage spikes that are trying to kill your OT, then leave them out I guess   

[stratomaster]

Why would you imagine that? The diodes in series share the voltage, and junction capacitance scales non-linearly with reverse voltage, so the total capacitance is still less than one diode alone. Even a single 1N4007 is less than 6pF in our scenario, which is negligible. But if you like the sound of extra 'grit' provided by voltage spikes that are trying to kill your OT, then leave them out I guess 

I'm not advising leaving them out, just using a single diode of sufficient voltage to perform the task.  In making the assumption of additive capacitance with added diodes (akin to adding cable length, at least in my mind) I was attempting to reconcile the observed effects on the sound vs the datasheet.  The only phenomena I can think of would be the snubbing effect of the diode capacitance.  If that capacitance is responsible for the change in tone, then it makes sense to try and minimize that capacitance when fitting the flyback diodes. BUT if I'm wrong and the capacitance is not the reason for the change in tone, then disregard entirely. AND I'd be curious to know what explains the observed change (beyond a placebo effect).

I'd rather the diodes be there than not--effect on extreme high end not withstanding.

[Merlin]

I'd be curious to know what explains the observed change (beyond a placebo effect).

I presume it is the sound of the back EMF spikes either being there, or not being there, no?

[stratomaster]

I presume it is the sound of the back EMF spikes either being there, or not being there, no?

Makes sense. I was unaware the diode(s) were conducting that frequently.  I thought the spikes themselves were relatively rare and exceedingly destructive, and only turned the diode on when exceeding the breakdown voltage--basically protecting the transformer from the worst of it--but possibly dying in the process.  I need to read up more on what they're actually doing then.

If my understanding isn't too far off, then perhaps balancing diode reverse voltage vs protection level vs change in sound would be a worthwhile endeavor for those that want the protection with minimal impact on sound?

[pdf64]

I'd be curious to know what explains the observed change (beyond a placebo effect).

I presume it is the sound of the back EMF spikes either being there, or not being there, no?

Yes, that's my take.
Overdriven into a real speaker load, there's obvious significant back emf spikes on each leading edge, so twice per cycle.
They seemed to make a definite contribution to the tone of an overdriven amp.
The diodes reduce the spiking.
But amps have been coping with the spiking for decades, so it seems to be a non issue. Rather I think it's the massive spikes that result from an open circuit, or intermittently open circuit, secondary that cause insulation damage to the OT primary circuit (output valves / sockets are part of that circuit).

The other failure mode that seems to kill OT primary insulation (but due to hot spot overheating) is an output valve short, when there's no or inadequate HT fusing.

[MORE_Guitar_Solos]

A common protection diode is the R3000. (. . .)
There's also the HV3, but that is at 15pF.

Do these go by other names/codes? My usual supplier is Amplified Parts (Antique Electronic Supply) and I find neither of these there.

[stratomaster]

Not that I know of.

Here's where I get mine.
https://www.mouser.com/ProductDetail/Rectron/R3000-T?qs=m6PtEeoja6gmP37yljRwMQ%3D%3D

https://www.mouser.com/ProductDetail/Diotec-Semiconductor/HV3?qs=OlC7AqGiEDkkpzp89F7IGQ%3D%3D