• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Can we talk about series/pass regulators?

I've been doing some reading on series tube regulators lately.

References:

http://www.audioxpress.com/assets/upload/files/bicknell2890.pdf

Tube Based Voltage Regulators - Part 1

http://www.diyaudio.com/forums/tubes-valves/90020-perfect-tube-regulator.html#post1059442

Also Radio Amateur's Handbook and other odds and ends I have on my bookshelf.

I'm interesting in trying something around a 12AX7 and 6AQ5/6BQ5 or similar. I'd like to use it to power a small SET (6V6 or 6BQ5 output, probably) with one regulator per channel. I also have a couple 6CK4s and 6Y6GAs that look like they'd be useful as pass tubes.

What I'm having trouble finding is information on the maximum permissible current through the series tube. Is it as simple as dividing the maximum dissipation by the voltage drop across the series tube? For something like a 6BQ5 with a 150V drop, that gives 80mA, but this seems very high and I don't see any designs operating at or advertising this much current with this tube or similar. Or should I use the maximum permissible cathode current (65mA in the case of EL84 according to some data sheets)? The latter makes more sense, but it's still very high if compared to many of the designs I've seen.

It seems that lower currents allow for better regulation. Is this a function of the series tube Rp as it relates to the voltage drop divided by the current passed (in that we want the Rp to be much lower than this figure, like when calculating for load impedance in an output)? In that context I think I see why there would be practical current limits to provide useful regulation. That might also mean that a higher gain error amplifier would allow regulation for higher currents (within the limitations of the max cathode current). Not sure if thinking about this like a load calculation is correct though.

Can anyone point me to some more reading or shed light on how to spec series regulator tubes?
 
Is it as simple as dividing the maximum dissipation by the voltage drop across the series tube? For something like a 6BQ5 with a 150V drop, that gives 80mA
Basically yes. However, series pass tubes are usually operated in triode mode, in which case it may not be possible to get both low voltage and high current simultaneously.

Or should I use the maximum permissible cathode current (65mA in the case of EL84 according to some data sheets)?
Yes, although a figure of 65mA is ridiculously low for a 6BQ5. It's probably referring to the max idle current in some particular circuit configuration.

It seems that lower currents allow for better regulation.
Loop gain makes for better regulation. Lower currents may imply a higher voltage for the error amp, and therefore more gain, but this depends on the topology.
 
Thank you, Merlin! That helps tremendously. Using the maximum dissipation figure and the voltage drop is what I assumed initially, but I was thrown off after researching designs and seeing that the EL84 as a pass tube is recommended for use below 40mA in most cases.

I think I understand how gain of the error amp works in theory (amplifying the difference between the voltage reference and the resistor divider on the output). And the series tube sees the resulting signal at its grid and varies its resistance to drop more or less volts.

I'm used to looking at load lines, so visualizing what happens with this pass tube is proving difficult.
 
Using the maximum dissipation figure and the voltage drop is what I assumed initially, but I was thrown off after researching designs and seeing that the EL84 as a pass tube is recommended for use below 40mA in most cases.
The figure of 40mA is probably derived from reasons of practicality and staying within the SOA in the face of changes in raw supply voltage (mains variations).
 
Search the forum, lots of tubed based series pass regulators here.

Tubes like the 6336 (both sections in parallel) is capable of quite large currents at a relatively low effective anode voltage. (several hundred mA at > 125V anode to cathode)

6AS7G/6080 and the Russian 6C33 are other candidates.

6W6, 6Y6, 6BQ5 at lower currents with anode to cathode voltages of at least 150V also work well. Feed screen with very clean DC and you get further reductions in ripple and other noise beyond that provided by loop feedback.

Similar deal with KT120/KT88/6550 and EL34, but somewhat higher voltages are required.
 
Thanks, Kevin! I linked to one of yours in the original post. How does the EL84+12AX7 cascode perform with currents in the 50mA range? I think you cited 40mA as the limit somewhere?

I have other tube options (6AS7, 6Y6, 6L6GC, 6CK4) but I'm trying to keep heater current requirements and overall size down. Then again, I'll have to use separate heater supplies anyways I suppose.
 
With conventional power tubes, yes. Some of the high perveance tubes designed for regulator service have pretty stout heater-to-cathode ratings. If you want to go nuts, I've got an extra 6528A laying around that you're welcome to.

Thanks for the tip and offer, SY! Already plenty nuts🙂 I've already got a couple 6528s (and 6080s and 6AS7s). I also have a massive heater supply transformer. I'll save those for end-game immovable object meets unstoppable force level regulation.
 
Thanks, Kevin! I linked to one of yours in the original post. How does the EL84+12AX7 cascode perform with currents in the 50mA range? I think you cited 40mA as the limit somewhere?

I have other tube options (6AS7, 6Y6, 6L6GC, 6CK4) but I'm trying to keep heater current requirements and overall size down. Then again, I'll have to use separate heater supplies anyways I suppose.

I've not run them beyond 40mA or so, I suspect all would be fine provided you bumped up the raw B+ to accommodate.. Note that at a 50mA load with a compliance voltage (V plate to cathode) of 200V that the 6BQ5 would be dissipating 10W which is IMO might be a bit too close to the plate dissipation rating. This is where the concern arises, and internal losses are probably too high for good performance with less than 200V across the tube. Tubes like the 6W6 and 6Y6 come to mind here because of their high perveance and performance at low effective plate voltages..
 
Note that at a 50mA load with a compliance voltage (V plate to cathode) of 200V that the 6BQ5 would be dissipating 10W which is IMO might be a bit too close to the plate dissipation rating. This is where the concern arises, and internal losses are probably too high for good performance with less than 200V across the tube.

Thanks, Kevin. This helps me visualize what is going on. Imagining it on a EL84 triode load line, the 50mA at 150V anode-cathode is much closer to the 0 grid line than the 40mA at 150V. Giving the tube 200V anode-cathode at 50mA leaves more breathing room for the grid to affect the voltage drop and hold the output voltage constant. But that does get close to max dissipation like you said.

I'm imagining the load on the tube as the output voltage divided by the current like in the below.



Am I on the right track?
 
Ok, started putting something together here:



I thought it would be better to put the VR before the series regulator to lessen the load. Also the error amplifier pre reg in order to provide it with lots of available voltage. This is based on one of the designs in the Audio Express article linked in original post.

The thing I'm having trouble with is figuring out the right resistor dividers for the reference. Just having some trouble visualizing what is happening with the error amp. The cathode is biased up a couple hundred volts by the follower, so I think the reg will attempt to achieve the same drop across the resistor divider. Shooting for 150k total to provide a light 2mA load.

The maximum output I could expect with a really light load is around 400V. I'd like to be able to vary from that max down to around 250V if possible. Maybe I need to look for a 75k pot (150V divided by 2mA) and split the 50k and 25k?

Any input from the great minds is much appreciated.

edit: oops, I have my 75k and 25k resistors switched in the schematic. I want 75k on the bottom and 25k on the top.
 
Last edited:
Honestly, the more I look at it, the less confident I become that I understand it.


My thinking is that the cathode follower (right 12AX7) is biased up by the VR tube and the error amp (left 12AX7) is biased up by the drop across the CF load. I may have that wrong though.

See Figure 2 here http://www.audioxpress.com/assets/upload/files/bicknell2890.pdf

I'm basing some of the parts values off of that schematic (the 100k grid stopper was just carried over).
 
Okay, I just reviewed Bicknell's article, and I see what you're getting at. He used the right side 12AX7 as a cathode follower buffer on the reference voltage, and added the cathode bypass apparently for additional zener diode noise suppression. The 100K grid resistance may be intended for yet more noise suppression. Seems like a lot more noise suppression than necessary. Since it's a tested circuit, maybe you should go ahead and build it that way. On a breadboard. Then try the diff-amp version and perhaps find out why Bicknell didn't use it?
 
The one comment I have about the long tailed pair is limited gain, and this is also where I started decades ago.

I would change the 1K resistor from the gas tube back to 100K and add a cap to ground to create an LPF for the reference tube noise.

Something else to note is that you will not be able to get output voltage down anywhere close to the reference voltage. (You can only pull the error amp plate down to the reference voltage plus the tube's saturation voltage.) The pot is fine for adjusting over a relatively limited range though.