I've been reading up a bit on triode shunt regulators and I think I get the gist of how they operate. Here are a couple of good articles that I'm basing my understanding on:
TubeCAD 'Vanilla' Tube Shunt Regulators (you can find a lot more on JB's site, too)
Power Supply Discussion for the T Rex SET (partially a rehash of the above)
In a nutshell, my understanding is that the current through the tube and the load attached sets the drop across the series resistor, bias sets the current through the shunting tube (wrt voltage at anode), and the grid sees AC ripple at the output (usually a cap connection) and so counteracts it using the tube's transconductance (it's an inverting amplifier).
Hopefully I'm on track with the above.
My question then is how do we determine the load on the shunting tube and whether it is appropriate for the tube in question? It seems to me that both the series resistor and the load from the attached circuitry would form the total load on the tube. In that case, is it as simple as the following?
Rseries || (output voltage/load current) = total load
And if that is the combined load (though maybe I'm wrong), should we try to configure the supply so that it is roughly similar to the 2-3x Rp rule of thumb for amplifier stages? Or is a 1:1 ratio between load and Rp more feasible for a shunt regulator?
The tube is amplifying the grid's ripple signal across the series resistor, so I think that series resistor value should dominate the total load if we want the best ripple cancellation (implying that higher current loads are going to worsen the ripple on the output). Adding a (non-inverting) gain stage to amplify the ripple signal into the shunt tube's grid helps with ripple cancellation if the load current is (relatively) high.
Am I on the right track with the above thinking?
The need for voltage headroom and/or relatively small load currents to maximize the load, and/or additional gain stages to increase ripple cancellation when the load current is high, makes the shunt regulator pretty inefficient (not that I've ever heard them accused of being otherwise).
TubeCAD 'Vanilla' Tube Shunt Regulators (you can find a lot more on JB's site, too)
Power Supply Discussion for the T Rex SET (partially a rehash of the above)
In a nutshell, my understanding is that the current through the tube and the load attached sets the drop across the series resistor, bias sets the current through the shunting tube (wrt voltage at anode), and the grid sees AC ripple at the output (usually a cap connection) and so counteracts it using the tube's transconductance (it's an inverting amplifier).
Hopefully I'm on track with the above.
My question then is how do we determine the load on the shunting tube and whether it is appropriate for the tube in question? It seems to me that both the series resistor and the load from the attached circuitry would form the total load on the tube. In that case, is it as simple as the following?
Rseries || (output voltage/load current) = total load
And if that is the combined load (though maybe I'm wrong), should we try to configure the supply so that it is roughly similar to the 2-3x Rp rule of thumb for amplifier stages? Or is a 1:1 ratio between load and Rp more feasible for a shunt regulator?
The tube is amplifying the grid's ripple signal across the series resistor, so I think that series resistor value should dominate the total load if we want the best ripple cancellation (implying that higher current loads are going to worsen the ripple on the output). Adding a (non-inverting) gain stage to amplify the ripple signal into the shunt tube's grid helps with ripple cancellation if the load current is (relatively) high.
Am I on the right track with the above thinking?
The need for voltage headroom and/or relatively small load currents to maximize the load, and/or additional gain stages to increase ripple cancellation when the load current is high, makes the shunt regulator pretty inefficient (not that I've ever heard them accused of being otherwise).
In escence, are like a VR tube, that drains the excess of current in a parallel circuit counteracting with the load demands (as load sinks more current, less flows by the VR or SR), but amplified and with voltages that can be selected externally, not fixed like discharge (glow) tubes.
Que tal Osvaldo, gracias por tus comentarios.
I have used VR tubes as shunt regulators in the past. I only kept the load current very light because I didn't really understand how they should be loaded though. Do you know any rules for how much current a circuit connected to VRs as shunt regulators should use?
I suppose if the load isn't Class A and might change in current demands you have to be careful not to overload the VR's maximum current or cut them off, but I'm not sure how to think about it when the load is Class A and more or less constant.
I have used VR tubes as shunt regulators in the past. I only kept the load current very light because I didn't really understand how they should be loaded though. Do you know any rules for how much current a circuit connected to VRs as shunt regulators should use?
I suppose if the load isn't Class A and might change in current demands you have to be careful not to overload the VR's maximum current or cut them off, but I'm not sure how to think about it when the load is Class A and more or less constant.
OK, gracias.
I only used neon glow tubes for making series regulators of various kinds, but I use between 100 and 1000µA in them. I never made or saw a shunt regulator with tubes, in SS the world famous LM431 is the most common device.
VR's till I know, are capable of regulating about 10 to 30mA at a voltage fixed by the distance between cathode, anode and gas/vapor mix inside the tube.
Search for the data sheets, VR75, 105 and 150, that was the most common voltages, or the miniature 0A2, 0A3, etc.
Salu2.
I only used neon glow tubes for making series regulators of various kinds, but I use between 100 and 1000µA in them. I never made or saw a shunt regulator with tubes, in SS the world famous LM431 is the most common device.
VR's till I know, are capable of regulating about 10 to 30mA at a voltage fixed by the distance between cathode, anode and gas/vapor mix inside the tube.
Search for the data sheets, VR75, 105 and 150, that was the most common voltages, or the miniature 0A2, 0A3, etc.
Salu2.
Jaja, "Salu2" es algo nuevo para mi.
I will research the LM431 and read some VR sheets again. Maybe if the load current does not vary too much, it doesn't not make too much difference if the output impedance of the shunt is very low. The shunt is just like an AC ground for the signal.
I will research the LM431 and read some VR sheets again. Maybe if the load current does not vary too much, it doesn't not make too much difference if the output impedance of the shunt is very low. The shunt is just like an AC ground for the signal.
Te gusta "Salu2 para to2", ja ja ja.
As I mentioned, shunt regulators I never build, but series yes. If you want, take a look in particular design inspired in the SS version, something like the well known "see-Saw" amplifier. Using a good high triode, it works pretty fine. The drawback is that you need a negative power supply, but the advantage is a high plate voltage for the amplifier, which then comes back as a very good regulator.
http://www.diyaudio.com/forums/analogue-source/292628-high-audio-quality-am-tuner.html, post #107.
Respectfully.
As I mentioned, shunt regulators I never build, but series yes. If you want, take a look in particular design inspired in the SS version, something like the well known "see-Saw" amplifier. Using a good high triode, it works pretty fine. The drawback is that you need a negative power supply, but the advantage is a high plate voltage for the amplifier, which then comes back as a very good regulator.
http://www.diyaudio.com/forums/analogue-source/292628-high-audio-quality-am-tuner.html, post #107.
Respectfully.
the plate dissipation rating of the tube you want to use is your guide,
say you want to use the 6bq5, it has a max plate rating of 12 watts...
so budget your current so that the tube does not dissipate more than 6 watts, this is a good starting point....
that tube will be handling the full load current
when the load is not connected, so please consider this also...
this is how i will do it if i were to make one...
A shunt is just a class A amplifier that happens to be configured to cancel noise. It'll still need the same conditions as any other class A stage to be most effective.
AJT and kyle have it right. And I'd like to mention..
Like +any+ shunt regulator, the zero-load condition is the most strenuous for the reg. device, and it doesn't matter if the latter is a gas tube or zener. At zero load, all of the available regulated supply current dissipates power (makes heat) in the reg. device. As the load begins to draw current, the proportion of current is divided between the reg. device and the load. At maximum output, the reg. device is at minimum power, and runs the coolest.
Like +any+ shunt regulator, the zero-load condition is the most strenuous for the reg. device, and it doesn't matter if the latter is a gas tube or zener. At zero load, all of the available regulated supply current dissipates power (makes heat) in the reg. device. As the load begins to draw current, the proportion of current is divided between the reg. device and the load. At maximum output, the reg. device is at minimum power, and runs the coolest.
Hi Kyle, Ok that makes sense to me. I was picturing it as basically a Class A amplifier too. So ideally we want the load to be larger than the Rp.
And thanks all for the comments on dissipation and unloaded condition. That too makes a lot of sense with regards to most of the configurations I have seen.
Hi AJT,
Yes, I mean the load > Rp rule ALSO applies to shunt regs (just like everything else). That was originally what I guessed, just nice to have it confirmed by others.
Yes, I mean the load > Rp rule ALSO applies to shunt regs (just like everything else). That was originally what I guessed, just nice to have it confirmed by others.
Back in the early days of color TV, they had very high mu (2000!) triodes designed as shunt regulators of the really high voltage fed to the picture tube. See http://scottbecker.net/tube/sheets/107/6/6BK4C.pdf which has an example circuit. Scaled down it might be a usable circuit.
Last edited:
tube parameters that i pay close attention to:
1. maximum cathode current ratings...
2. maximum plate (or plate plus screen) dissipation ratings....
for optimum tube life, these must be respected imho....
Another excellent article about shunt regulators for audio circuits titled Shunt Regulator Concepts by John Camille (Chimera Labs) can be found in SOUND PRACTICES Vol.2 Issue 6 (starting at p.43) Highly recommended if you can find a copy.
I couldn't find it either, but would love to read it. Is there a place to order back issue PDFs?
Any thoughts on something like this?
I'm thinking a light load (like a phono preamp). Uses the non-inverting output of a differential amp as the ripple amplifier and feeds a EL84 (decent transconductance for a wee little 9 pin).
Parts values not specified, of course. Just trying to make sure I understand things. I think the Vref on the EL84 could also be a resistor if DC regulation isn't needed? Might need a grid resistor on the non-inverting half of the 12AX7.
edit: added missing grid resistor
edit2: wait a sec, this configuration doesn't quite make sense. The input for the dif amp doesn't need the DC ref above ground, I think. Revised below.
I'm thinking a light load (like a phono preamp). Uses the non-inverting output of a differential amp as the ripple amplifier and feeds a EL84 (decent transconductance for a wee little 9 pin).
Parts values not specified, of course. Just trying to make sure I understand things. I think the Vref on the EL84 could also be a resistor if DC regulation isn't needed? Might need a grid resistor on the non-inverting half of the 12AX7.
edit: added missing grid resistor
edit2: wait a sec, this configuration doesn't quite make sense. The input for the dif amp doesn't need the DC ref above ground, I think. Revised below.
Last edited:
- Status
- Not open for further replies.