If you use a single CCS for both cathodes then you are very definitely limited to Class A. One tube can swing up to 2 x idle while the other swings to 0 and vv., That is as good a definition of Class A as any other.
As to not working very well, tell that to the Vaccuum State DPA300B owners, an amp which took out best amp in show, 3 years in a row. It has Push Pull 300B with a common CCS in the filament/cathodes.
Your point about separate CCS's (which is what I did in the BH design) then you are correct, bypass caps are required (to pass the audio signal as you say), and the quality of those caps definitely affects the sound quality.
Credit Yvesm with the damned clever balanced shunt feedback scheme of the BH's, the CCS bias blocks were my own "folly".
The "philosophy" was perfect idle balance with no adjustment required ever.
The compromised overload recovery was a compromise I was willing to accept after listening to it. Ditto (so it seems) for the MANY guys who have built their own Baby Huey with the bypassed CCS in each output tube cathode.
My prototype was originally a Morgan Jones Bevois Valley, the rebuilt amp to my BH design was a significant step up in performance (the Bevois Valley had individual 270R cathode bias resitors on each EL84 with 470uF bypasses)
Between that original BH design and the more complicated fixed bias variant there are well over 50 BH's happily providing good music to satisfied DIY builders around the world.
Cheers,
Ian
Not really. The CCS prohibits AB operation (except for short bursts - without the caps it can't even do short bursts) as it gives a sliding bias: the more the music, the cooler the bias. Why anyone would want to deliberately design a complex circuit which works worse than a simple circuit is beyond me; and then attempt to rescue the situation by disabling the complex circuit by adding a zener!gingertube said:
If you have to patch up a circuit to get it to work, then maybe it is the wrong circuit?
Should have thought that was pretty obvious and has been mentioned earlier, Equal and consistent Bias that doesn't vary with tube aging--All tubes age differently--no matter what anyone says!
The added Zener does Not disable the CCS circuit at, or below certain output levels.
It also Sounds Good in comparison to decoupled cathode-resistors which they replaced.
There is nothing special about a particular current, so no need to maintain it accurately. Choosing a cool current with a bias circuit which ensures it gets cooler during music is a good way to ensure that the bias is always wrong. It may sound different from the simpler and better cathode resistor bias, but whether it "Sounds Good" (sic) would be a matter of taste rather than fact.
Anyway, you described a predictable problem with this circuit and how you patched it up with a zener then asked for our opinions. You now have our opinions.
Anyway, you described a predictable problem with this circuit and how you patched it up with a zener then asked for our opinions. You now have our opinions.
The goal is to set and keep anode iddling currents equal to make the output transformer happy without having to meter and trim them while tubes age.
OTOH, the zener, even alone, is a good substitution for negative grid supply. Just shifting up the cathodes does not look different than shifting down the grids.
(Yes, it wastes some B+ but saves a rectifier and a shunt regulator).
May opinion (already expressed) is that a CCS bypassed with a cap and a ( judiciously choosen ) zener is the best choice:
- stable idle current,
- low bias shift at hi drive,
- so, fast recovery after over load.
No matter the class (A, AB, B, C, even A2 or AB2 ... ) you're designing for.
Wait for a guru realizes that and it will become an incontournable 😀
Yves.
OTOH, the zener, even alone, is a good substitution for negative grid supply. Just shifting up the cathodes does not look different than shifting down the grids.
(Yes, it wastes some B+ but saves a rectifier and a shunt regulator).
May opinion (already expressed) is that a CCS bypassed with a cap and a ( judiciously choosen ) zener is the best choice:
- stable idle current,
- low bias shift at hi drive,
- so, fast recovery after over load.
No matter the class (A, AB, B, C, even A2 or AB2 ... ) you're designing for.
Wait for a guru realizes that and it will become an incontournable 😀
Yves.
OPTs can cope with some DC imbalance.
A cathode zener is a reasonable substitute for a fixed grid supply provided that the zener always conducts so it can maintain a fairly constant voltage. In this design it is intended that the zener only conducts on program peaks. Thus we have a constantly varying bias, with a fairly hard limit. This seems to be a recipe for subsonic IM. To suggest that this might even be a suitable bias circuit for Class C I will take to be an attempt at humour.
So we have a constant (current) bias for low signals (where it doesn't matter too much what the bias is) and a constantly varying bias (apart from a hardish clip) for big signals. This is supposed to be 'guru-level' design?
A cathode zener is a reasonable substitute for a fixed grid supply provided that the zener always conducts so it can maintain a fairly constant voltage. In this design it is intended that the zener only conducts on program peaks. Thus we have a constantly varying bias, with a fairly hard limit. This seems to be a recipe for subsonic IM. To suggest that this might even be a suitable bias circuit for Class C I will take to be an attempt at humour.
So we have a constant (current) bias for low signals (where it doesn't matter too much what the bias is) and a constantly varying bias (apart from a hardish clip) for big signals. This is supposed to be 'guru-level' design?
Wow…
I read through this whole thread, and I was left cheering for DF96's practicality. The zener in parallel with the CCS disables the CCS. Not only do you not get constant current for each valve (which was the design goal!), but you've disengaged the circuit which is also supposed to protect the output valves from over-current since your B+ supply is supposed to be pretty high in voltage.
I'm also with DF96 regarding his interpretation of the benefits of a simple bypassed cathode resistor, and I'll add this: given it is a fixed resistance, using your trusty multimeter, measuring the voltage across it (carefully) during operation is quantitative metrology for the per-tube operating point current draw. Simple. Reliable. Cheap. Good. Rarely can one use all 4 of those words in the same sentence.
The bypassed cathode resistor approach also tracks tube aging reasonably well. The Baby Huey output tubes can be set for 30 ma by choosing the right nominal resistor empirically. After that, as they age and μ falls off, the drop in current flow will increase the grid bias towards the positive which significantly compensates for the aging. Not technically as 'perfect' as a CCS will, but as DF96 points out, good enough in practice.
FINALLY … if you are really interested in "restoring power", then why not change the BIAS constant current point by adjusting the resistors within them, to a higher current-level? At some point you simply have to admit: either the original premise (saving the output valves from excessive current and anode power dissipation and cleverly balancing current flow, even with tube aging) isn't your real goal, OR that it is, but needs compromise based on adjusting the premise, the operating point, to something other than the design goal.
Just saying.
GoatGuy
I read through this whole thread, and I was left cheering for DF96's practicality. The zener in parallel with the CCS disables the CCS. Not only do you not get constant current for each valve (which was the design goal!), but you've disengaged the circuit which is also supposed to protect the output valves from over-current since your B+ supply is supposed to be pretty high in voltage.
I'm also with DF96 regarding his interpretation of the benefits of a simple bypassed cathode resistor, and I'll add this: given it is a fixed resistance, using your trusty multimeter, measuring the voltage across it (carefully) during operation is quantitative metrology for the per-tube operating point current draw. Simple. Reliable. Cheap. Good. Rarely can one use all 4 of those words in the same sentence.
The bypassed cathode resistor approach also tracks tube aging reasonably well. The Baby Huey output tubes can be set for 30 ma by choosing the right nominal resistor empirically. After that, as they age and μ falls off, the drop in current flow will increase the grid bias towards the positive which significantly compensates for the aging. Not technically as 'perfect' as a CCS will, but as DF96 points out, good enough in practice.
FINALLY … if you are really interested in "restoring power", then why not change the BIAS constant current point by adjusting the resistors within them, to a higher current-level? At some point you simply have to admit: either the original premise (saving the output valves from excessive current and anode power dissipation and cleverly balancing current flow, even with tube aging) isn't your real goal, OR that it is, but needs compromise based on adjusting the premise, the operating point, to something other than the design goal.
Just saying.
GoatGuy
This IS an attempt at humour ! 😀
Looking at the flag that appears on your mails, may I expect to have the pleasure to meet you near Copenhagen in November ?
Yves.
Going through this thread years later I can see how we are all amateurs in some aspects of life and how the search for new understanding is part of DIY. I commend Yvesm for his patience since his willingness to conduct empirical experiments got him right in the end.
The main advantage of cathode bias in power stages is its stabilizing effect both statically and dynamically due to parameter variations. This protects the tubes from overload. The CCS accomplishes this much more exactly for the static/idle condition. However, for dynamic conditions, the cathode bias, and even more so the CCS, may ”overcompensate” under certain conditions by driving up the bias voltage. The CCS/cathode resistor with a zener clamp is just a simple way of achieving adaptive bias for a tube with modest bias voltage and cathode current, by restricting the range of automatic adjustment. Much more complicated schemes have been implemented to acheive the same effect for fixed bias, e.g. Prima Luna. Whether is you appreciate this practically or sonically is another thing, but here is the kicker: this idea was patented (together with diod bias) already in 1964 by Smith & Lang. It expired in 1981.
US3129388A - Cathode bias clamp
- Google Patents
Thanks to all - it was educational.
Cheers,
Kranis
The main advantage of cathode bias in power stages is its stabilizing effect both statically and dynamically due to parameter variations. This protects the tubes from overload. The CCS accomplishes this much more exactly for the static/idle condition. However, for dynamic conditions, the cathode bias, and even more so the CCS, may ”overcompensate” under certain conditions by driving up the bias voltage. The CCS/cathode resistor with a zener clamp is just a simple way of achieving adaptive bias for a tube with modest bias voltage and cathode current, by restricting the range of automatic adjustment. Much more complicated schemes have been implemented to acheive the same effect for fixed bias, e.g. Prima Luna. Whether is you appreciate this practically or sonically is another thing, but here is the kicker: this idea was patented (together with diod bias) already in 1964 by Smith & Lang. It expired in 1981.
US3129388A - Cathode bias clamp
- Google Patents
Thanks to all - it was educational.
Cheers,
Kranis
If you want to have cathode bias but don't want the crossover distortion to appear at large outputs (class B) then use a 5W zener + resistor in series then a capacitor across both. The zener takes about 2/3 of the static bias with the resistor 1/3. Its somewhere between cathode bias and fixed bias. The worst thing to do is a CCS - your better off with a resistor.
Another variation is to put the Zener at the "bottom" of the cathode stack, and it can be common to all of the output valves (preferably), with individual cathode resistors and bypass caps, and use this voltage as a pot-variable G1 bias source. Adjust relative contributions of R and Z to taste.
This is handy when restoring antiques from the Golden Age, designed for 115 volt mains in America and now facing 125 volt mains, an era when 375 VDC was common for EL84s, the available valves were better and cheaper, and Power Wars competition was in full swing.
YOS,
Chris
This is handy when restoring antiques from the Golden Age, designed for 115 volt mains in America and now facing 125 volt mains, an era when 375 VDC was common for EL84s, the available valves were better and cheaper, and Power Wars competition was in full swing.
YOS,
Chris
This was an interesting discussion, but I am not 100% clear what the conclusion was!
It feels like it identified conflicting requirements for traditional biassing solutions (fixed/cathode), and more encompassing CCS variants, and the traditional solutions are more fit for 'real world' cases.
Additionally, zeners can have their place to clamp the cathode bias, to limit the possibility of the grid going positive (and the potential recovery issues of that) or to add an element of fixed bias in a cathode bias solution.
If I understood correctly, there is not really a case for making it complicated for output tubes, beyond regulated supplies for fixed bias?
It feels like it identified conflicting requirements for traditional biassing solutions (fixed/cathode), and more encompassing CCS variants, and the traditional solutions are more fit for 'real world' cases.
Additionally, zeners can have their place to clamp the cathode bias, to limit the possibility of the grid going positive (and the potential recovery issues of that) or to add an element of fixed bias in a cathode bias solution.
If I understood correctly, there is not really a case for making it complicated for output tubes, beyond regulated supplies for fixed bias?
