I need low-current voltage regulated supply for a tube preamp/driver circuit.
Kicker is I need voltages in the +- 100 to 200 voltage range, and I want to use my existing power transformer that only has 400-0-400 CT output. Maybe I'm dreaming.
Anyway, I already have nice high-voltage current source devices, the IZCP10M45S. Is the a decent, simple way to use current sources for voltage regulation?
Kicker is I need voltages in the +- 100 to 200 voltage range, and I want to use my existing power transformer that only has 400-0-400 CT output. Maybe I'm dreaming.
Anyway, I already have nice high-voltage current source devices, the IZCP10M45S. Is the a decent, simple way to use current sources for voltage regulation?
You can use a CS to send current through a zener diode which regulates voltage. Eventually boosted by a power device.
But it will be horribly inefficient.
Say you start with 400VCT that delivers at least 520VDC, you need 150V, at 50mA. You need to lose 370V @ 50mA = 18.5W into the current regulator.
A new transformer is much better. Or, if your transformer has dual primaries, rig it for 220V and run it from 110V, that gives you about 250VDC.
Jan
But it will be horribly inefficient.
Say you start with 400VCT that delivers at least 520VDC, you need 150V, at 50mA. You need to lose 370V @ 50mA = 18.5W into the current regulator.
A new transformer is much better. Or, if your transformer has dual primaries, rig it for 220V and run it from 110V, that gives you about 250VDC.
Jan
Last edited:
Thanks Jan. Yes, inefficiency I expected. But if I only need to drive a couple of 12AX7s, I don't need much current.
I was hoping there was a way to jig the CS control to make it regulate voltage, but maybe all it winds up being is some sort of shunt regulation.
CS into high voltage zeners is ugly. I'll see what else I can do. Maybe just change the circuit to avoid the need.
Or maybe some sort of Maida arrangement with an LM317 or TL783. Trying to keep it simple, drat.
I was hoping there was a way to jig the CS control to make it regulate voltage, but maybe all it winds up being is some sort of shunt regulation.
CS into high voltage zeners is ugly. I'll see what else I can do. Maybe just change the circuit to avoid the need.
Or maybe some sort of Maida arrangement with an LM317 or TL783. Trying to keep it simple, drat.
You do not need much current. I doubt you really need voltage regulation for audio.
400-0-400VAC is brutal. Are you sure??
Anyway: 400VAC makes 550V DC. Most 12AX7 stages can eat 300V. 550V-300V is 250V drop. "Couple" is 4 triodes and about 4mA. 250V at 4mA is 62k and 1 Watt of resistor. 4 stages of C-R-C-R-C-R-C filtering should be plenty clean. 62k/3 is three 22k 1 Watt resistors. 20Hz is 22k against 0.5uFd. Aluminum is cheap. Buy 50uFd (47u) caps for 0.2Hz corner, so three sections makes -150dB down on ripple (not counting parasitics).
400-0-400VAC is brutal. Are you sure??
Anyway: 400VAC makes 550V DC. Most 12AX7 stages can eat 300V. 550V-300V is 250V drop. "Couple" is 4 triodes and about 4mA. 250V at 4mA is 62k and 1 Watt of resistor. 4 stages of C-R-C-R-C-R-C filtering should be plenty clean. 62k/3 is three 22k 1 Watt resistors. 20Hz is 22k against 0.5uFd. Aluminum is cheap. Buy 50uFd (47u) caps for 0.2Hz corner, so three sections makes -150dB down on ripple (not counting parasitics).
Hah. Yes, your analysis is correct, so far as it goes. Here are several point to explain what's going on.
I'm going back and looking at a DC negative feedback approach from the output tube cathode current to the LTP bias. Had trouble getting that to work before, we'll see.
Maybe I'll use a reversed small low voltage power transformer connected to the Hammond 278X's 5v heater winding, since I'm not using that, but it could still sag with heavy output.
And your point about low current has me thinking, I've never really run the power numbers, maybe I could get away with a resistor/zener approach. Hmm...
Meh, what to do, what to do...
- The high voltage is for the output tubes. I'm working on a class AB2 circuit that will end up with 7027a or bigger tubes in at least quad configuration. I want them to be able to handle bursts of power.
- I don't need the regulation for filtering, I'm not worried about that. I need it for driver voltage stability. I'm working on a direct coupled driver with source-connected mosfets to the output grids. So far my Spice design works, but it's clear the direct coupled driver will need to be stable, because it's the bias for the outputs.
I don't expect the negative current to vary much at all, but the positive will when the OT grids are overdriven. OTOH, the positive is less sensitive, so long as the negative doesn't bounce too much. I think...
- I'm not planning to run the 12AX7s anywhere the 580 volts the output tubes get, or even the 400v their screens will get. I'm looking at +120 and -170 for the voltages to run the 12AX7-based LTP phase splitter, with either an SS current source or a mosfet for the cathode coupling. It's a sweet circuit.
Balance is beautiful with that approach, and I have all driver voltage swing I could ever want.
I'm going back and looking at a DC negative feedback approach from the output tube cathode current to the LTP bias. Had trouble getting that to work before, we'll see.
Maybe I'll use a reversed small low voltage power transformer connected to the Hammond 278X's 5v heater winding, since I'm not using that, but it could still sag with heavy output.
And your point about low current has me thinking, I've never really run the power numbers, maybe I could get away with a resistor/zener approach. Hmm...
Meh, what to do, what to do...
You can use a capacitive supply as a lossless dropper, just like in a mains supply, except it will be isolated.
The cap supply will behave like a crude CCS, you can use it to feed a zener string, followed by additional filtering/regulation as needed.
The cap supply will behave like a crude CCS, you can use it to feed a zener string, followed by additional filtering/regulation as needed.
Detailed nits: I don't know any grid-current rating for 7027. 807, yes. 6L6, yes. (I am aware most 7027s are 6L6 guts in different glass; why not just use 6L6?)
6550 will make more power without grid current than 7027/6L6/807 in grid current.
Regulating tube voltages leads to cascading troubles. Regulate just G1 and the amp is hypersensitive to line voltage. G1 and G2 is a little better. G1 G2 and Plate... the requisite voltage regulator is equivalent to another power amp which never drives the speaker. IMHO you may as well oversize(1) the audio amp and not expect so much of it.
(1) We can never afford to over-size audio power like we would like to. Buy a 200W amp, the user turns it up to 300W. Buy a 600W amp and now they need 1200W.
6550 will make more power without grid current than 7027/6L6/807 in grid current.
Regulating tube voltages leads to cascading troubles. Regulate just G1 and the amp is hypersensitive to line voltage. G1 and G2 is a little better. G1 G2 and Plate... the requisite voltage regulator is equivalent to another power amp which never drives the speaker. IMHO you may as well oversize(1) the audio amp and not expect so much of it.
(1) We can never afford to over-size audio power like we would like to. Buy a 200W amp, the user turns it up to 300W. Buy a 600W amp and now they need 1200W.
Good ideas, thanks for your comments!
Well, I have the 7027s... Main reason. Kind of like the farmer with two chickens. They've been nagging at me. I'd love to play with KT120s or 160s, too. Don't have those. I do have a couple of 807s I may try, but I need the right sockets. This is an experimental amp. If it doesn't work, ok, if it does, cool.
ISTM most tube manufacturers don't bother to spec G1 current because few people want it and they don't want the hassle. Doesn't mean it won't work. I'm just looking for a way to reduce/soften clipping and extend guitar pick pulse headroom, really.
As to regulation, you raise very good points. The intent is just to stabilize the fixed bias delivered by the driver stage. Everything else can hang loose. My spice model works nicely with supply regulation, but I'm starting to thinking I need to take a different approach to stabilizing driver inherent DC stability.
As an aside, I've tried a IXCP1045S CCS and a IRF820 high voltage mosfet for LTP current source, and finally figured out why the mosfets didn't work well. They do NOT like working at the low current level of a pair of 12AX7s, delivered crappy current regulation. The CCS seems to work nicely for current regulation but is harder to configure for DC feedback.
Still exploring other ideas.
Well, I have the 7027s... Main reason. Kind of like the farmer with two chickens. They've been nagging at me. I'd love to play with KT120s or 160s, too. Don't have those. I do have a couple of 807s I may try, but I need the right sockets. This is an experimental amp. If it doesn't work, ok, if it does, cool.
ISTM most tube manufacturers don't bother to spec G1 current because few people want it and they don't want the hassle. Doesn't mean it won't work. I'm just looking for a way to reduce/soften clipping and extend guitar pick pulse headroom, really.
As to regulation, you raise very good points. The intent is just to stabilize the fixed bias delivered by the driver stage. Everything else can hang loose. My spice model works nicely with supply regulation, but I'm starting to thinking I need to take a different approach to stabilizing driver inherent DC stability.
As an aside, I've tried a IXCP1045S CCS and a IRF820 high voltage mosfet for LTP current source, and finally figured out why the mosfets didn't work well. They do NOT like working at the low current level of a pair of 12AX7s, delivered crappy current regulation. The CCS seems to work nicely for current regulation but is harder to configure for DC feedback.
Still exploring other ideas.