Bassman style amp blowing rectifier diodes

[BungleSim]

Hello all,

Full disclosure: I work for a company in the music industry and I inherited this amp project. The original engineers that worked on it cannot be reached for insight. This is a Bassman style circuit with 5881s.

Now that that is out of the way, the problem I'm having is that the D4 and D5 rectifiers are burning out. This happens somewhat randomly and is hard to reproduce (I have yet to reproduce it on my own but I have seen at least 5% of the amps built fail in this way), but from what I can tell it happens after closing or opening (?!) the STANDBY switch. I have one recorded account of a user playing the amp with no issues, then switching into standby and leaving the amp alone for at least a few minutes, and then coming back to find fuses F1 and F2 are blown without him touching anything. At least that's what he reports.

I am suspecting that most of the time that the rectifiers blow is after closing the STANDBY switch and the inrush current of the reservoir caps is too much for the diodes to handle, causing them to fail short and then causing F1 (3A SB) to blow. However, this theory brings up the following questions which causes it to fall apart: 1. Why doesn't F2 (1A FB) save D4 and D5? These are 1000V/3A rectifiers. 2. How can the diodes fail with the STANDBY switch open, like with the user above?

Any ideas? Is it possible C24 is causing issues? I'm happy to remove C24 but since I haven't been able to reproduce the fault, it's hard to test whether it is the culprit.

Any and all help is appreciated. Thank you!

The rectifiers are 1N5408 and their datasheet is attached.

PWR_SCH.png

[sluckey]

What voltage is present between the red wires on the PT?

[BungleSim]

What voltage is present between the red wires on the PT?

420VAC before closing STANDBY
409VAC after closing STANDBY (with tubes, 8-ohm dummy load connected, no signal in, all knobs at zero)

[pdf64]

I can't see anything obviously wonky, but would be better to put the HT fuse and standby switch on the AC side of the rectifier, rather than the DC side http://www.valvewizard.co.uk/fuses.html
A trickle charging bypass resistor for the standby switch may be beneficial http://www.valvewizard.co.uk/standby.html
As would a 100k safety resistor on the bias trimmer http://www.valvewizard.co.uk/bias.html
An NTC current inrush limiting thermistor might be considered, eg CL90 https://www.amphenol-sensors.com/en/the ... -cl-series

[SoulFetish]

I'm with pdf on this one. The schematic doesn't show anything out of the ordinary (even though there are improvements that could be made to the design, namely moving the fuse to one of the legs of the HV secondaries on the AC side.).
First, there is no reason that a 1N5408 should be failing in this circuit with any regularity. Second, it seems strange to me is that only one side (D4 & D5) of the bridge seem to be failing. If it were a problem with a faulty component downstream (ie C24), you would think the both phases would be affected due to the randomness of flipping the standby switch.
Can you check the orientation of the diodes and wiring of the AC connection and confirm that they are all connected correctly?

[BungleSim]

Can you check the orientation of the diodes and wiring of the AC connection and confirm that they are all connected correctly?

Everything is connected correctly. These amps go through a lot of testing before going to a user (initial power supply measurements, bias, output signal test on scope, hi-pot, ground bond, burn-in, then audio test), so any "oopsies" like wrong orientation or wiring would've been caught. Also, this amp was played for a while before it failed.

I replaced all the diodes with the same type and I've been trying to get the amp to fail all day with no luck. Just been flipping the power and standby switches on and off in different meaningful combinations.

[pompeiisneaks]

This is also one of the major design flaws of using a standby switch. The inrush current of using one can really harm components. OTOH there's 0 chance not having one will have any negative impact on a well designed amp. This is why I don't install them on my builds...

~Phil

[pdf64]

...I replaced all the diodes with the same type and I've been trying to get the amp to fail all day with no luck. Just been flipping the power and standby switches on and off in different meaningful combinations.

Maybe there was a bad batch of diodes, perhaps sourced from non-traceable vendors?

[BungleSim]

...I replaced all the diodes with the same type and I've been trying to get the amp to fail all day with no luck. Just been flipping the power and standby switches on and off in different meaningful combinations.

Maybe there was a bad batch of diodes, perhaps sourced from non-traceable vendors?

It's entirely possible. I'm requesting the datasheet from the PCB vendor.

[xtian]

Isn't the reservoir cap(s) the next suspect? Like, shorting under surge/high voltage?

[SoulFetish]

(Just spit ballin’ here) perhaps the standbyswitching may be causing high voltage transients in the secondary. The resonance caused by the capacitance of the 1N5408 diode junction in combination with the inductance of the winding might be problematic in this circuit. Can you get a scope on it to capture the diode switching alone and the the physical switching of the standby to see of there are any spikes?
If this is the case, a proper RC combination across the winding might damp the resonance.
But this is where I would call in the bigs like trobbins or RG who have more expertise with such things.

[Phil_S]

Bypass the flippin' switch. (Pun intended.)

[trobbins]

The symptoms seem to suggest switch contact bounce may be a contributor. The switch contact would initially see a half cycle charging current peak that could reach circa 10A level, depending on the transformer effective resistance, and the immediate load current of the hot output stage. The switch sees current pulses of the same polarity (as does the fuse), so any bounce could cause arcing and/or sequential current pulses with repeated high peak levels. If the transformer secondary has substantial leakage inductance then switch contact bounce at a time when charging current pulse peak is quite high could exacerbate a few concerns, like diode reverse recovery, and diode PIV stress, due to abnormal dI/dt levels in the transformer secondary causing winding terminal voltage transients (at a time when off-state diodes have max PIV stress).

Some more detail on the B+1 loading, and transformer winding dc resistances, could help scope the technical side more.

I doubt it is related to diode thermal failure from over-current, as 1N5408 is used, and would more suggest PIV is at risk. If it was PIV related then some options could be to use 2x 1N5408 or UF5408 in series, and add a C-RC snubber across the HV secondary, and replace the standby switch if the amp has been in service for some time (as the contacts may have degraded).

Ciao, Tim

[BungleSim]

Some more detail on the B+1 loading, and transformer winding dc resistances, could help scope the technical side more.

What would you like to know about the B+1 loading? I have the OT transformer spec so I can look up or measure any data you need.

The measured DC resistance of the HT secondary (RED-RED) is 67.2 ohms.

[BungleSim]

If it was PIV related then some options could be to use 2x 1N5408 or UF5408 in series, and add a C-RC snubber across the HV secondary, and replace the standby switch if the amp has been in service for some time (as the contacts may have degraded).

I don't think I can fit double the diodes, but I'm interested in hearing more about this C-RC snubber. What does that look like?