Low filament voltage and noise problems

[HarryB]

Anybody have experience with low filament voltages creating noise problems in an amp? I'm working on a Peavey Classic 30. The filaments on these amps are wired in series. There are 2 EH 12AX7's and 1 Sovtek 12AX7LPS in the amp. I'm getting 9.37 volts (DC) on the Sovtek and 11.30/11.40 on the EH's (pins 4&5). The noise occurs over the signal going in.

Isn't 11.4 volts still to low for a 12 volt tube to operate properly?

I replaced the Sovtek with a newer EH in an attempt to match the tubes better. The new tube has 10.15 volts on the filament while the old ones have around 11 volts each. The lower voltage follows the tube as I swap it to different sockets. Not a very stable setup.

Thanks for your feedback.

[antik]

It looks to me like there may be 2 problems. First the tubes do not all have the same filament resistance. Second the -36volt supply is obviously not 36volts (11.3+11.4+10.1=33 volts). Causes: low line voltage, inaccurate volt meter, something else is defective and is loading the supply, or a defective component in power supply.

I'd be interested in how much AC on the -36V DC supply. Can check this by changing meter to AC volts and measuring the heater voltage.

[HarryB]

Thanks for taking a look at this. I 100% agree regarding the filament resistance. It confuses me why Peavey would use a series arrangement if there was such variability between manufacturers. A customer would need to use three tubes from the same lot to be certain to have equal voltage to each filament. Everything that I've read says it is important to stay real close to the 12.6 volt spec. I was hoping someone could fill me in on how low filament voltages might create noise, or if it even would.

The tube voltages do add up to the voltage present at R66/R63. I've got -32.3 volts, not the 36 volts listed on the schematic.

The AC supplied to the diode bridge measures 29.4 VAC.

The low voltage starts at the wall.
The AC to the transformer is 117.9 VAC, approx 2% less than 120 VAC.
The B+ at the power tube plate is 322 V, about 3% less than the 332 V listed.
The DC out of the bridge rectifier measures -34.9 V, approx 3% less than the -36 volts on the schematic.
It doesn't look anything is drawing excessive current. The other currents are very small compared to what the tubes are drawing (approx 160 mA).

My meter's check out with one another. One is a Wavetek and the other a Tektronix.

[antik]

The data sheets for the 2 tubes you have show 12 volts as the minimum heater voltage.


At 120 that -36 volt supply should actually be 37.8 for each tube to get 12.6 volts.

I agree it is a bad design. I looks to me like this design would significantly shorten the life of any Mullard/Amperex 12ax7s if someone put 3 of those in there.

Edit: (although the resistor limits current so it might be ok)

[antik]

Something is not right. If the load is only say 0.2A with 29.4V RMS at the bridge input, with a 2200uF filter cap, at 60Hz, the voltage at the output of the bridge should be in the neighborhood of 39.5V DC with less than 1V RMS ripple. Even with a 1.0A load the voltage would be about 38V DC and 3.5V AC

Maybe it's not a 2200uF cap (C45)???

Check the amount of AC on the output of the bridge to see if the cap might be bad or has a bad connection

[HarryB]

Thanks Antik. I'll lift the fuse and put my ammeter in there to see what the AC current draw is and compare it to the DC. Or maybe I'll lift the resistor or cap. I'll have to do that on Friday. It's late. I'm fried!

What formula did you use to figure out the DC and AC components when the cap was placed after the bridge rectifier. I must have skipped class that day :)

I don't think the current limiting effect of the series resistor will protect the tube from damage. The problem as I know it isn't a current limiting issue but that the cathode needs to be heated to a certain temperature for it to emit electrons. If its not heated properly (low current) the cathode coating gets stripped off leaving the tube useless. Previously I was trying to clarify if the stripping (due to low voltage) would cause a noise issue or if the tube would just stop working effectively.

[antik]

Here is a simple formula for calculating ripple voltage. It gives an approximate answer that is conservative (larger). The larger the current, the more conservative the estimate. There are ways to get a more exact answer, but that is usually not required since most of the quantities are estimates anyway.

Vpk = Vrms * sqrt(2) - 2 * Vdiode ; Vpk is peak voltage at the output of the bridge, Vrms is the Ac input to the bridge. Vdiode is the voltage drop across 1 rectifier diode.

For a full wave rectifier:
Vripple = iload / (2*f*C) ; Vripple is the voltage drop as the cap discharges ie ripple voltage peak to peak, iload is load current, f is mains frequency, C is capacitance.

Vdc = Vpk - Vripple/2 ; Approximate value of DC voltage across the capacitor at the output of the rectifier.

[donnyjaguar]

Check the value of R66. I'm guessing its more than it should be. As filaments get hotter the resistance goes down so if you serve up the requisite 36Vdc you might find the variance between the 3 filament voltages a lot less. Its an atypical design, but I've heard good things about this amplifier so it can't be all bad.