GZ34 arcing

Just switched on the 300B amp and after about 30secs the rectifier started arcing, lasted about one sec followed by one very quick arc and then seemed to be ok.

It’s a new JJ GZ34 and it arced low down near the base rather than in the plates.

Should this happen on a new GZ34? read it can be down to high input capacitance on the supply?

No and yes, it could be down to excessive capacitance. But would need to see cct. to confirm.

60uF is the stated data sheet max for a capacitor input filter.

I’d expect the arching to be around the plates than the base though, so could simply be a faulty tube.

Rectifier arcing is usually down either to them being asked to pass too much current (something downstream’s gone awry and is drawing too much current) or because they’ve got some gas in them somehow (the glass is micro-cracked, or a gas bubble trapped in a spot-weld has burst). As Pete says, the first cap after the rectifier is supposed to be 60uF max, but that’s assuming the rest of the circuit is operating close to max current and voltage. If the rest of the circuit can be guaranteed to be gentle on the rectifier then it is possible to get away with a bit more than 60uF. It’s not good practice though.

Arcing at the bottom of the valve can be due to a bit of debris dropping out of the main structure and bridging across the wires where they feed through the glass base.

1 Like

Think I’ll swap it out with a 5R4WGY potato masher I’ve got spare

Know JJ had some QC issues years ago but thought they were quite good now

You’re aware they’re not equivalent, right? And have 4uF stated as the max capacitance for C-input.

The 5R4Wxx datasheet specifies a very low first capacitor - just 4uF - although experience says you can get away with more. They have a big forward voltage drop though, so you’ll be under-running the 300Bs with a lower HT voltage.

1 Like

Ah in the spec he says “rectifier tube 5U4G or 5R4GW or GZ34”

(I assumed the 5R4GW was a Japanese to english thing)

I went for a JJ GZ34 as it came with no valves but in the photo he was using 5R4GW

1 Like

Without knowing the circuit it’s difficult to be sure.

There are plenty of designs that aren’t very good and don’t observe specs, so I wouldn’t necessarily trust recommendations without being sure. Especially if risking nice vintage tubes. (New tubes are mostly utter crap anyway)

As Graeme says, the voltage drop will be significantly higher with 5R4 meaning a different operating point and less power. It will sound different for sure.

I’m specifically using crappy old shuguang 300B and new JJs as I’ve read his build quality can be a bit crap so didn’t want to risk anything nice or NOS

1 Like

Looks like it is 55uF in mine.

Probably ok, but if I’m designing I’d personally use less to be in the safe side (or use choke input).

I’d probably not not be risking vintage tubes!

think it’s got a choke as well

This was my first thought on seeing it was a new rectifier, albeit not a Russian one at least…

Short-lived and unreliable rectifiers are my #2 (:poop:) reason for giving-up on all-valve gear - sick to fucking death of watching them fritz-out on startup.

Only exceeded by the misery of suicidal SETs… (#1 reason)

Yes, but I’m guessing it’s not choke input.

The issue is the current spikes into the cap. A choke input draws a more constant current whereas charging a capacitor results in large spikes.

1 Like

assumed it is an input choke as there’s only one and it’s inline with the 5V tap?

The difference is whether the filter after the rectifier is CLC or just LC - the third and fourth cases here (CLC is a pi filter though, not a pie one !):

Filter circuits - Inductor Filter, LC filter, CLC or PI filter, Capacitor filter | D&E notes.

The choke will be in the same position in either case.

Looks like an LC?

If any cunt says that won’t work as they aren’t connected to ground they can fuck off :slight_smile:

1 Like

Nope, CLC (plus whatever, second C is off the right of the drawing). Depends if you have enough spare voltage to add resistance in from the transformer
(or prior to the cap) to limit the current that can flow.