That king of voltage regulation is par for the course for cap input filters. And it’s probably as high as you would want to run triode-strapped EL34’s anyway. Vg2 rating is 425.
The details with the project are scant to say the least.
It does show using a 320V AC winding.
The two transformers providing B+ are 250VA units.
It does show using a 320V AC winding.
The two transformers providing B+ are 250VA units.
Last edited:
Either the LM317 is shorted out, or the FET is. I tried giving you some pointers to what kind of Mosfet would survive linear operation.
At startup the output capacitor is a dead short, so its only natural that your FET dies with 500V and lotsa current through it. Linear optimized fets are designed to survive this, switching fets just wont.
A maida is overkill for a PP amplifier, a source follower PCB with a big Linear optimized FET is good enough. Any remaining noise is cancelled in the output stage.
But reading further, i can see that you have already figured out a passive supply.
Depends on which datasheet you believe.
Philips originally put a 425V rating for G2, which is what many others copied. Later Philips datasheets list 500V.
Mullard even gives ratings for triode mode: 500V up to 30W and 600V up to 15W dissipation.
JJ is the odd one out as they state 450V, but in triode mode you usually can go a bit higher.
Even if you just use a source follower and zener string for the regulator, you HAVE to include fold back current limiting if you drive a big output capacitor with it. 3 terminal regulators have this feature and people tend to forget that. There are other tricks you can do to make switching FETs survive, but increase the dropout voltage (sometimes a lot). Making a 400+ volt, 300 mA regulator just isn’t trivial. It sort of like high power PA amp design - you have the same considerations. If your linear supply delivers the right voltage under load just call it good if you can stand the unloaded voltage. More available voltage swing will give you more peak power at frequencies where the loudspeaker load exceeds the nominal so it’s not necessarily a bad thing.
How high you can go with modern Chinese EL34’s is anybody’s guess. They may take the full 500 volts in triode, they may run away. Unfortunately I don’t have any recent production EL34’s to relate or experiment with. The only EL34’s I have are some winged C Svetlana’s (some 35 years old now). That used to be called new production, but not anymore.
And if you just want QUIET, regulate and filter the hell out of the front end. I’ve found the driver stage to be sensitive enough that it benefits significantly from a cap multiplier on the zener string feeding a source follower regulator. That’s where you need to reduce noise. At the lower currents that don’t vary too much under load implementing the regulator gets exponentially easier.
Well, the amplifier worked for a short while but hummed slightly.
I tried adjusting the bias but one of the valves wouldn't adjust.
Suddenly that valve glowed bright red and the PCB caught fire.
I'm not going to bother with my experiments with valves any more.
The cathode resistor for the valve in question just blew up.
I tried adjusting the bias but one of the valves wouldn't adjust.
Suddenly that valve glowed bright red and the PCB caught fire.
I'm not going to bother with my experiments with valves any more.
The cathode resistor for the valve in question just blew up.
Bad, valve. No need to shelve the project just yet. Measure continuity in your PP output transformers before making that decision.
Contrary to my last post. I just swapped out the resistor and cleaned up the soot and it seems OK.
The cathode resistors are 10R 3W so some 550mA or more must have passed through it.
The OPTs both measure the same so I'm hoping they are OK.
The cathode resistors are 10R 3W so some 550mA or more must have passed through it.
The OPTs both measure the same so I'm hoping they are OK.
The hum was because the currents were out of balance. Power supply hum only cancels if the currents are balanced. If one of the pair wouldn’t adjust it was probably running too high to begin with, and when it finally red plated and ran away it’s going to take out the cathode resistor. Size them properly and they make good fuses to prevent catastrophic meltdown.
Chinese valves have been known to be bad right out of the box. I’ve had reasonable results with the New Sensor Russian brands. No red plates so far. Your problem just as easily could have been a bad or miswired bias pot. Did you check adjustment range before installing tubes?
Chinese valves have been known to be bad right out of the box. I’ve had reasonable results with the New Sensor Russian brands. No red plates so far. Your problem just as easily could have been a bad or miswired bias pot. Did you check adjustment range before installing tubes?
The bias is also adjusting now. I wonder if the cathode resistor was faulty or just in contact with the ground plane.
Pull every resistor around the output stage/Bias network. And replace them with new ones that are stood off from the board.
R22 has burnt this time so I've no idea how much damage might have been done, particularly to the OPT.
I'm guessing that the EL34 will be toast.
I'm guessing that the EL34 will be toast.
Is there a quick test to see if the EL34s have been spoilt ?
For R22 to fail, the valve must have passed some serious grid current.
For R22 to fail, the valve must have passed some serious grid current.
The DC resistance of the OPT still seems OK but it is very low resistance anyway so difficult to tell.
In my experience, transformers are pretty resilient under short term load. If a winding is going to fail it will probably fail open circuit.
In my experience, transformers are pretty resilient under short term load. If a winding is going to fail it will probably fail open circuit.
Perhaps get set up with a variac and some meters and a test schedule and carefully bring the amp up in stages to confirm sections of circuitry are working before moving forward. Perhaps also add some protection for valve failures.
As I said earlier, the amp had been powered up and was working. It was only when I tried adjusting the bias that things went horribly wrong.
Connect 470k resistors from the wipers of all the bias pots to the negative rail. Seriously, do it. Even if that wasn’t the problem. Eventually it will be, and very well could have been. If the wipers go intermittent you lose bias altogether and you will red plate. I wouldn’t put it past an E-bay kit maker to use as cheap of a pot as they can get their hands on. I only use precision 10 or 20 turn pots with the little metal screwdriver adjust for *any* amplifier bias adjust. And I still put the pull down resistor in there because I don’t like to blow tubes when it’s avoidable. I buy the pots surplus for an average of 50 cents apiece, so you don’t have to spend $10 to get a good pot.