Has anyone had good experience with mixed kinds of cathode bias? I'm thinking here of combining a resistor and a LED or SIC diode or similar. in my case it would be for an input stage. The goal would be to eliminate the cathode bypass capacitor by enabling the use of a smaller resistor like 100 ohms or even less if possible. An example is shown with a SIC diode. Ale discusses SIC diode bias on his Bartola Valves website.
I have tried pure SIC diode bias a few times and in the end preferred a resistor, but combining the two might give good results. I'll set up a comparison when I have time.
One question - does it matter if the SIC diode is above or below the resistor or is there no difference?
I have tried pure SIC diode bias a few times and in the end preferred a resistor, but combining the two might give good results. I'll set up a comparison when I have time.
One question - does it matter if the SIC diode is above or below the resistor or is there no difference?
Attachments
I dont like any sand or batteries at the cathode.I use heater bias with Rods regulators in my phono, line and driverstage and filament bias at the power tube. No cathode C anymore
> Has anyone had good experience with mixed kinds of cathode bias?
hello Andy - I don't think it's a good idea. If you want a small amount of unbypassed cathode resistance, Filament Bias does it very well, and is certainly better than using diodes IME.
You can also mix grid bias with a small unbypassed cathode resistor: this works very well for DHTs that have big bias voltages (10Y, 300B, PX25...). The small unbypassed cathode resistor gives just a little degenerative feedback, which might be just what you need.
hello Andy - I don't think it's a good idea. If you want a small amount of unbypassed cathode resistance, Filament Bias does it very well, and is certainly better than using diodes IME.
You can also mix grid bias with a small unbypassed cathode resistor: this works very well for DHTs that have big bias voltages (10Y, 300B, PX25...). The small unbypassed cathode resistor gives just a little degenerative feedback, which might be just what you need.
The input stage i was considering here is indirectly heated, not a DHT for which I'd use filament bias. I think that's what Rod was assuming. The ECC82 simulation is spot on - I presently use a 220R cathode resistor on an indirectly heated driver tube with a big amorphous plate choke. But for other cheaper builds I might use a resistor load or a smaller plate choke. However, 220R is a typical value and I was looking for ways of reducing this.
I agree with Cagomat - I've always preferred a resistor to any kind of sand in the cathode. I'm just wondering if a single diode would affect the sound if used with a resistor. I'll rig something up and see if I can hear a difference. I'll start with the SIC diode and then try an LED.
The SIC diodes I have are C3D02060F which produce about 0.86v each.
I agree with Cagomat - I've always preferred a resistor to any kind of sand in the cathode. I'm just wondering if a single diode would affect the sound if used with a resistor. I'll rig something up and see if I can hear a difference. I'll start with the SIC diode and then try an LED.
The SIC diodes I have are C3D02060F which produce about 0.86v each.
Dunno… I have had a remarkable amount of good luck with button-cell battery bias. The darn things last at least 10 years these days, are cheaper than dirt, and provide a totally stable voltage the whole while.
However, in this regard, they really are about the same as the yellow to yellow-green cathode bias light emitting diodes. LEDs have a very sharp VF forward conduction point which over an order-of-magnitude IA (really IK) conditions varies by as little as ±3% peak-on-peak
If you note from ArtoSalo's sim, the degenerate local-negative-feedback of a naked cathode resistor has decided advantages in squashing some of a triode's unfettered g³⁄₂ distortion. It is musically competent.
I had the opportunity to build and rather exhaustively compare several first stages (effectively preamp stages) for an over-the-top system, which were not just working with recorded material, but live instrument pickup sources. In the end, it was pretty much neck-to-neck: 12AX7 + cathode resistor bias, or, J302 JFET also with source-resistor bias (equivalent to cathode resistor). I mean we did a LOT of listening. It was that epiphany that caused me to lose my conviction that insofar as musicality is concerned, that tube front ends are better. JFETs are also sweet, clean, and remarkably durable; still, if the goal is vacuum amplification, RK … naked … up front … is excellent.
The LED and battery bias (and lil' negative DC supplies) all resulted in a harsher kind of distortion. That can be addressed with GNFB, but then what's the point? More stage gain? Not really! The difference between say 8× stage and 10× stage, while not insignificant, turns out to be pretty darn insignificant when the upstream stages' gains are suitably revised to accommodate the situation.
Too long of a reply.
Oh well.
⋅-⋅-⋅ Just saying, ⋅-⋅-⋅
⋅-=≡ GoatGuy ✓ ≡=-⋅
However, in this regard, they really are about the same as the yellow to yellow-green cathode bias light emitting diodes. LEDs have a very sharp VF forward conduction point which over an order-of-magnitude IA (really IK) conditions varies by as little as ±3% peak-on-peak
If you note from ArtoSalo's sim, the degenerate local-negative-feedback of a naked cathode resistor has decided advantages in squashing some of a triode's unfettered g³⁄₂ distortion. It is musically competent.
I had the opportunity to build and rather exhaustively compare several first stages (effectively preamp stages) for an over-the-top system, which were not just working with recorded material, but live instrument pickup sources. In the end, it was pretty much neck-to-neck: 12AX7 + cathode resistor bias, or, J302 JFET also with source-resistor bias (equivalent to cathode resistor). I mean we did a LOT of listening. It was that epiphany that caused me to lose my conviction that insofar as musicality is concerned, that tube front ends are better. JFETs are also sweet, clean, and remarkably durable; still, if the goal is vacuum amplification, RK … naked … up front … is excellent.
The LED and battery bias (and lil' negative DC supplies) all resulted in a harsher kind of distortion. That can be addressed with GNFB, but then what's the point? More stage gain? Not really! The difference between say 8× stage and 10× stage, while not insignificant, turns out to be pretty darn insignificant when the upstream stages' gains are suitably revised to accommodate the situation.
Too long of a reply.
Oh well.
⋅-⋅-⋅ Just saying, ⋅-⋅-⋅
⋅-=≡ GoatGuy ✓ ≡=-⋅
I doubt a diode will be better than a cap.
Some amps rely on cathode resistor to setup feedback and gain. You may not want to change out that resistor.
If you are designing from the scratch, put the resistor in the bottom, if you do need one. Thus, you have the option to hook up global feedback to the junction of the resistor and the diode.
Some amps rely on cathode resistor to setup feedback and gain. You may not want to change out that resistor.
If you are designing from the scratch, put the resistor in the bottom, if you do need one. Thus, you have the option to hook up global feedback to the junction of the resistor and the diode.
Last edited:
A diode and a capacitor are almost but not quite identical. Here's why.
If the triode (or any ode) were absolutely linear, AND the source signal was absolutely 'mean symmetric' around zero, then …
… then many things would hold.
The need for any NFB would be eliminated, as the gain stage would be exactly linear, adding no transfer function distortion to the signal, asymmetrically about the zero mean axis. And so on.
Under those circumstances, the mean voltage through RK would also be exactly the same as the quiescent (no signal) voltage. Thus the 'job' of the bias bypass capacitor would be to hold the cathode at a relatively raised voltage, independent of the instantaneous signal across whatever acceptable high-pass band the engineering team designed into the thing.
However, both real-world recorded material and the V³⁄₂ transfer identity of a triode (and remarkably close in a different way for pentodes inside their 'flat curve' regions) results in a mean RK voltage which varies quite a bit with signal intensity and instrument recording compression.
This means that there is a substantial time-varying mean voltage modulation which the CK bypass capacitor picks up and holds. Thus changing the actual bias of the valve in ways neither predictable a priori or desired post priori for that.
Hence why things like lil' yellow or yellow-green (or blue or UV) LED diodes are such a remarkable solution. No capacitor means they 'return to nominal' basically instantaneously. (Just like naked resistors!)
Subjective comparison of various Stage 1 biasing strategies is what resulted in the take-away I listed in the prior comment. Basically, naked resistors are excellent because of the small local NFB they introduce linearlizing the transfer of the valve; At the other end, when 'real' fixed bias is declared desirable, then button cells and appropriately selected LEDs are also excellent solutions.
⋅-⋅-⋅ Just saying, ⋅-⋅-⋅
⋅-=≡ GoatGuy ✓ ≡=-⋅
If the triode (or any ode) were absolutely linear, AND the source signal was absolutely 'mean symmetric' around zero, then …
… then many things would hold.
The need for any NFB would be eliminated, as the gain stage would be exactly linear, adding no transfer function distortion to the signal, asymmetrically about the zero mean axis. And so on.
Under those circumstances, the mean voltage through RK would also be exactly the same as the quiescent (no signal) voltage. Thus the 'job' of the bias bypass capacitor would be to hold the cathode at a relatively raised voltage, independent of the instantaneous signal across whatever acceptable high-pass band the engineering team designed into the thing.
However, both real-world recorded material and the V³⁄₂ transfer identity of a triode (and remarkably close in a different way for pentodes inside their 'flat curve' regions) results in a mean RK voltage which varies quite a bit with signal intensity and instrument recording compression.
This means that there is a substantial time-varying mean voltage modulation which the CK bypass capacitor picks up and holds. Thus changing the actual bias of the valve in ways neither predictable a priori or desired post priori for that.
Hence why things like lil' yellow or yellow-green (or blue or UV) LED diodes are such a remarkable solution. No capacitor means they 'return to nominal' basically instantaneously. (Just like naked resistors!)
Subjective comparison of various Stage 1 biasing strategies is what resulted in the take-away I listed in the prior comment. Basically, naked resistors are excellent because of the small local NFB they introduce linearlizing the transfer of the valve; At the other end, when 'real' fixed bias is declared desirable, then button cells and appropriately selected LEDs are also excellent solutions.
⋅-⋅-⋅ Just saying, ⋅-⋅-⋅
⋅-=≡ GoatGuy ✓ ≡=-⋅
Hello all, i tested with 6E6P-E plain resistor, led and resistor, led and resistor and cap. Winner is plain resistor.
Well, I listened to my input stage in cathode bias with an unbypassed 220R resistor first. Then swapped in a SIC diode with 110R resistor. All this with a 136H NP amorphous plate choke and FT-2 teflon coupling caps.
Unbypassed 220R resistor was smooth and natural.
110R resistor plus SIC diode was a little less natural, not quite as smooth. Plus point was it had a little more definition in the sound. This at the expense of sounding a bit more artificial and slightly sharper.
I'm on the fence here - the extra definition and focus was potentially attractive but it came at the expense of losing the more natural sound of just the resistor. I'll listen some more. Right now I'm 50-50. So I'd say it's worth a try and see how your own ears react to it.
I haven't tried a LED. For 1.7v bias that looks like one red LED, but no resistor. I'm not used to LEDs so any info welcome.
Unbypassed 220R resistor was smooth and natural.
110R resistor plus SIC diode was a little less natural, not quite as smooth. Plus point was it had a little more definition in the sound. This at the expense of sounding a bit more artificial and slightly sharper.
I'm on the fence here - the extra definition and focus was potentially attractive but it came at the expense of losing the more natural sound of just the resistor. I'll listen some more. Right now I'm 50-50. So I'd say it's worth a try and see how your own ears react to it.
I haven't tried a LED. For 1.7v bias that looks like one red LED, but no resistor. I'm not used to LEDs so any info welcome.
> 110R resistor plus SIC diode was a little less natural, not quite as smooth. Plus point was it had a little more definition in the sound. This at the expense of sounding a bit more artificial and slightly sharper.
I'd say you are describing the sound of added harmonics (so, distortion). The harmonics make the features of the music 'stand out' more. But this is not usually satisfactory, because when you play complex music with a 'hot' string sound, or loud choral music, it all gets confused and unpleasant, IME. Try digging out some 20th Century violin concertos, or string quartets, and see if they sound too screechy.
I'd say you are describing the sound of added harmonics (so, distortion). The harmonics make the features of the music 'stand out' more. But this is not usually satisfactory, because when you play complex music with a 'hot' string sound, or loud choral music, it all gets confused and unpleasant, IME. Try digging out some 20th Century violin concertos, or string quartets, and see if they sound too screechy.
I expected the op to answer since the circuit is his idea.
But I can't fully realize what impedances you talk about. Can you give some example?
That sounds pretty spot on. My standard audio demo is opera excerpts from Gotterdammerung and The Cunning Little Vixen, Steely Dan Negative Girl, and Bill Evans Quiet Light from I Will Say Goodbye. I then go on to Spanish piano music from Alicia de Larrocha beautifully recorded on Decca and my ultimate test, Brahms Symphony 2 by Columbia/Walter. The string sound on this recording is only just bearable on the best equipment I've thrown at it, and with any added distortion it breaks up and sounds horrible. So yes, I do have strings, multiple and single voices and good piano recordings. I suspect I'll eventually tire of the SIC diode sound but I'm giving it a bit more time. The single SIC diode moderated by a resistor is definitely better than multiple SIC diodes, which I can't take at all.
.
Last edited: