I have a home-brew VSE RTP3 (schematic here), which I built over ten years ago. I have been getting some noise from one channel, so I thought I'd test the obvious circuit points. The anodes of the line input cathode stage were a bit off the design 150V, so after checking that the current sinks on the cathodes were working correctly, I measured the filament voltages on the five EH 6922s in the line stage after ten minutes or so of power-up. These turned out to be all over the shop, varying between 4.5V and 6.5V. Looking at 6922 data sheets, I see that the filament voltage is 6.3V, plus or minus either 0.3 or 0.6V, depending on the brand, at a nominal 300mA, so at least some of them are emitting well under spec in my circuit.
Now the filaments are supplied, as Allen Wright preferred, by current sources (an LM317 with 4.1R between adjust and output pins). I checked that all the resistors have their nominal value, and the LM317 reference voltages were within a percent or so of 1.25V. So the filaments are all passing close to 300mA, but the deviation from the nominal voltage was surprising.
Has anyone else used current supplies for the filaments of small-signal valves? Have you found significant variation between brands, or any long-term changes in the filament resistance? I'm reluctant to tune the resistances on the LM317s for each 6922, since this makes valve replacement awkward. I'm tempted to substitute the current sources with voltage regulators, but constant-current uses only half the number of parts as constant-voltage.
Any thoughts?
Thanks,
Alex
Now the filaments are supplied, as Allen Wright preferred, by current sources (an LM317 with 4.1R between adjust and output pins). I checked that all the resistors have their nominal value, and the LM317 reference voltages were within a percent or so of 1.25V. So the filaments are all passing close to 300mA, but the deviation from the nominal voltage was surprising.
Has anyone else used current supplies for the filaments of small-signal valves? Have you found significant variation between brands, or any long-term changes in the filament resistance? I'm reluctant to tune the resistances on the LM317s for each 6922, since this makes valve replacement awkward. I'm tempted to substitute the current sources with voltage regulators, but constant-current uses only half the number of parts as constant-voltage.
Any thoughts?
Thanks,
Alex
There was a thread here some months ago where the conclusion was drawn that heating with constant current is inherently less accurate than heating with constant voltage. The reason is the positive temperature coefficient of the filament's resistance: when the filament is slightly hotter than intended, its resistance increases, which reduces the power dissipation under voltage drive but increases it under current drive.
Whether the effect was big enough to worry about remained unclear, but reading your post, it seems like it can be pretty bad. For accurate heating and inrush current limiting, you could use a voltage regulator with a current limiter.
Whether the effect was big enough to worry about remained unclear, but reading your post, it seems like it can be pretty bad. For accurate heating and inrush current limiting, you could use a voltage regulator with a current limiter.
Premature to swap parts. Do you see the noise on the heaters? Constant noise? Tube swaps? Pin cleaning? Jack/cable checks? Can you put normal 6.3v AC back in for testing? Got a scope? Speaker surrounds still intact?
Last edited:
Vacuum tubes are designed and engineered to work off constant-voltage filament power supplies. Remember — as in your original post — when you say “nominal 300 ma”, this means the spec is more or less 300 ma at 6.3 VAC±10% voltage. Because that's the way the valve is designed. The actual amperage could be anything from 225 ma to 375 ma (in my measurements). But the key is NOT the milliamp reading, but the actual running glow temperature of the filament.
Run your filaments significantly below their designed glow temperature spec, and you run significant risks of having the whole valve transconductance curves being far out of spec. Then their amplification is likewise skewed. Accumulating that nonlinearity on each stage of amplification.
So, as I've experienced multiple times, just give the filaments a solid, solid 6.3 VAC or if you prefer, a well regulated 6.3 V DC. Its what they're built for, and the actual current flow makes not a difference in the world.
Just saying,
GoatGuy
Run your filaments significantly below their designed glow temperature spec, and you run significant risks of having the whole valve transconductance curves being far out of spec. Then their amplification is likewise skewed. Accumulating that nonlinearity on each stage of amplification.
So, as I've experienced multiple times, just give the filaments a solid, solid 6.3 VAC or if you prefer, a well regulated 6.3 V DC. Its what they're built for, and the actual current flow makes not a difference in the world.
Just saying,
GoatGuy
Hey, about half of them were designed to work off constant voltage, but the other half TV and AC/DC radio were designed to work off constant current.
Just saying
Hey, about half of them were designed to work off constant voltage, but the other half TV and AC/DC radio were designed to work off constant current.
Just saying
Yes, … you are right, of course.
The desire to get rid of the filament transformer winding, and just string as many tubes as necessary to use 117 VAC line voltage directly was high. Because of competitiveness, every last savings in building a television went to the bottom line. By the grace of the gods, we are at least past that era. No one is using 50L6 and 35CW4 tubes any more. And in particular, this very site will censor our comments if they even hint at suggesting not having a transformer providing either filament voltage, or plate voltage. Too dangerous.
I should have made myself clear.
Just saying,
GoatGuy
Yes, I agree that I wouldn't have expected this much variation.
The 6922s have been in action for a few years, though I would expect their resistance to have increased with ageing, if anything, which would give a larger voltage for a given current, rather than the undervoltage I measured.
Alex
The 6922s have been in action for a few years, though I would expect their resistance to have increased with ageing, if anything, which would give a larger voltage for a given current, rather than the undervoltage I measured.
Alex
The only tube in this four parts row with a nominal heater amperage of 300 mA is the E88CC. The other three ones are 375 mA.
Besides them, there's the PCC88, designed for 300 mA series heating (which per datasheet is also an option for the E88CC, btw.). I've built a tube preamp around twelve PCC88's with their heaters arranged in three strings of four, each one fed from the yet mentioned LM317/4.1 ohms CCS. Works like a charm, but needs some time to heat up, of course.
Best regards!
Besides them, there's the PCC88, designed for 300 mA series heating (which per datasheet is also an option for the E88CC, btw.). I've built a tube preamp around twelve PCC88's with their heaters arranged in three strings of four, each one fed from the yet mentioned LM317/4.1 ohms CCS. Works like a charm, but needs some time to heat up, of course.
Best regards!
Yes, most datasheets show ECC88 and 6DJ8 (equivalents) as 365mA heater, and E88CC and 6922 ('special quality' equivalents - and similar in most other respects to ECC88/6DJ8) as 300mA. The problem is that nowadays sellers do not distinguish between these slightly different valves, not to mention the Russian lookalikes, so what is sold as a 6922 could have either heater current from 6.3V.
OK, so I have experimented with valve substitution.
First of all, I replaced four of the EH 6922s with NOS Sylvania 6922s, and the voltages changed to 6.35V within 0.2V.
Second, I replaced the old EH 6922s elsewhere with brand new EH valves - and no improvement. Still a scattering of low values (5.0V plus or minus 0.4V).
So the conclusion is that the current supplies are not compatible with the actual 6922s I put in (at least the EH branded ones). That's a surprise, since I remembered the filaments reading a little low when I built the preamp, but not necessarily that low.
I did check the unregulated DC supplies for the filaments, and everything looks fine - about 11V, with not much ripple.
So my next plan is to start changing the LM317 current regulators to voltage regulators. I can't help wondering if I've been missing the full potential of this preamp...
Alex
First of all, I replaced four of the EH 6922s with NOS Sylvania 6922s, and the voltages changed to 6.35V within 0.2V.
Second, I replaced the old EH 6922s elsewhere with brand new EH valves - and no improvement. Still a scattering of low values (5.0V plus or minus 0.4V).
So the conclusion is that the current supplies are not compatible with the actual 6922s I put in (at least the EH branded ones). That's a surprise, since I remembered the filaments reading a little low when I built the preamp, but not necessarily that low.
I did check the unregulated DC supplies for the filaments, and everything looks fine - about 11V, with not much ripple.
So my next plan is to start changing the LM317 current regulators to voltage regulators. I can't help wondering if I've been missing the full potential of this preamp...
Alex
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.