Damn modern rectifier tubes

[diydidi]

So what you guys are saying, is that the standby switch can stay if adding the helper diodes?
Alternatively remove the standby switch?

[nevetslab]

Fender uses BYD33V diodes (1400V) at least in newer Princetons. I think it is advisable to exceed 1000V of a 1N4007 in this function. I use BY269 (1600V) for it.

They keep the entire length of pins. Possibly to reduce mechanical stress and keep them away from the heat source.

After navigating all of the sources on line to drag you where you didn't want to go just to download the Philips BYD33 series soft-recovery avalanche diodes, I finally got the data sheet. These don't appear to be the globular body, so further digging need

BYD33 Soft Recovery Rectifier series.pdf

[nevetslab]

The Marshall designed built AC30TB of the 90s wins my gold award as the worst standby ever, as it combines hot switching with a wiring error, apparently long standing, that sent HT current via the GZ34 rectifier heater (DC output taken from pin 2 rather than pin 8). So the start up current surge stressed the rectifier heater, occasionally leading to the unusual dead heater failure mode.
Of course, taking the DC output from pin 2 only really matters with an indirectly heated type such as a GZ34.
Edit - just seen in the other thread that you’re already on top of this issue

And I continue to find them with that HT wiring off of Pin 2 thru the heater instead of Pin 8. They did finally get it right on the AC30-CC2X and the AC30HW series.

Yesterday, having found Antique Electronic Supply again had J/J GX34/5AR4 rectifier tubes, I was about to buy some, but, as I had just sent a matched Sextet of J/J KT88's with one tube having no similarity in transconducance back to them, I wanted to have the rectifiers hitch a free ride with the replacement KT88's. Those may have disappeared by the time they get to re-testing the tubes and comparing my install data with their Apex matching methods. I hadn't heard of this wide-spread failure of 5AR4/GX34 rectifiers until Randall blew the whistle and everyone else added to it.

I DID forget to remove the Mains fuse and clip in my small 5A magnetic circuit breaker, which cost me a mains fuse and another HT fuse when the rectifier tube blew that came in with the amp.

[Pedro Vecino]

After navigating all of the sources on line to drag you where you didn't want to go just to download the Philips BYD33 series soft-recovery avalanche diodes, I finally got the data sheet. These don't appear to be the globular body, so further digging need

I think that the reference that I have put is only generic. Here they show one with suffixes added:

https://darrenriley.com/store/fender...tr-0057351001/

[trobbins]

If one looks at mains voltage tolerance, and some safety margin from 1kV PIV, then a transformer with secondary greater than 280-300Vac (per side) should perhaps use 2x 1N4007 in series to manage the full PIV (for a leaky valve rectifier anode). Luckily many have found that 1N4007 actually have a much higher PIV than 1kV, with leakage just starting to increase into the breakdown region around 1.5kV - so the 1N4007 has itself lots of margin. And as some have also found, any diode in the 1N4004 to 4007 range has been using the same raw diode for a few decades now - I certainly have a batch of 1N4004 that show the same PIV performance as 1N4007.

[Leo_Gnardo]

I think that the reference that I have put is only generic. Here they show one with suffixes added:

https://darrenriley.com/store/fender...tr-0057351001/

Just had a look. ONLY $4.89 each. AND out-of-stock. Harrumph X2. Darren Riley indeed.

I'll stick with mounting 2x 1N4007's in series for each cathode.

[nevetslab]

Ditto on that ^^^^^^ I'll rush right out and buy a 100 pcs at that price!

[Helmholtz]

Luckily many have found that 1N4007 actually have a much higher PIV than 1kV, with leakage just starting to increase into the breakdown region around 1.5kV - so the 1N4007 has itself lots of margin. And as some have also found, any diode in the 1N4004 to 4007 range has been using the same raw diode for a few decades now - I certainly have a batch of 1N4004 that show the same PIV performance as 1N4007.

I learned not to rely on typical - but not guaranteed - component properties.
In the 80s we had a promising new design where the deadtime (necessary to avoid cross-conduction) of the BJT half-bridge chopper depended on the storage/recovery time of 1N4004s in the drive circuit.
Lots of diodes measured, many lab samples tested.

Soon after sales start we had complaints about failing products.
Analysis showed that some of the diodes were actually switching considerably faster than "typical", resulting in transistor shorts.
Production stopped, design canned.

It is no secret that 1N4004s to 1N4007s grow on the same wafers. Difference is only in selection regarding reverse breakdown voltage.
The 1N4007s will have passed a 1kV test.

Furhermore I learned that so-called "plastic-rectifiers" have a considerably higher failure rate (FIT value) than glass-passivated (GP) or glass bodied types - especially at elevated temperatures.
I typically avoid the cheaper "plastic-rectifier" types.

[Randall]

One thing I should have paid attention to is the PT in this amp is a sub and is putting out 435 - 435vac and a 550v B+. That is an issue for another thread, but perhaps this is why my TAD rectifier is arcing? I got another one, and under the same circumstances it does not arc, and seems to be stable.

[Helmholtz]

Max. allowable input voltage for a "real" GZ34 is 2x550Vrms.

[Leo_Gnardo]

One thing I should have paid attention to is the PT in this amp is a sub and is putting out 435 - 435vac and a 550v B+. That is an issue for another thread, but perhaps this is why my TAD rectifier is arcing? I got another one, and under the same circumstances it does not arc, and seems to be stable.

With your observations & Helmholtz's specification quote, I'd have to say your TAD tube is clearly defective.

[Helmholtz]

Of course the 2 x 550Vrms shouldn't be exceeded in standby.

[Randall]

So would this be a good case protection diodes?

[Helmholtz]

So would this be a good case protection diodes?

Certainly.

But even a 1500V diode might not suffice here.
You could use 2 x1N4007s in series per side.
Merlin recommends to wire a 10nF/1kV ceramic cap across each diode to make sure the reverse voltage is equally shared between the diodes.

[Randall]

I have installed the series pair for each side. Do I understand Merlin recommends a ceramic cap for each diode, as in four diodes and four caps? And to be clear, 10nF as in .01uF, correct?