Hi all,
I am designing a tube amp for the first time and I have a basic question:
Does each tube's HT supply need to be isolated by a resistor and cap from every other tube's supply, or can several tubes be supplied from the same common HT supply?
I would like to supply 8 tube stages from the same 260V supply. Is this okay, or should I isolate each on from each other?
Thanks,
Dave
I am designing a tube amp for the first time and I have a basic question:
Does each tube's HT supply need to be isolated by a resistor and cap from every other tube's supply, or can several tubes be supplied from the same common HT supply?
I would like to supply 8 tube stages from the same 260V supply. Is this okay, or should I isolate each on from each other?
Thanks,
Dave
Ideally you decouple (isolate) each section of an amp - for example the output section is decoupled from the driver which in turn is decoupled from the input.
For a good quick and easy to understand approach, read the power supply section of this site
Hi!
This depends on the actual amp and I would also balance it with the overall budget which is spent.
You mention 8 tube stages which sounds a lot, is this a preamplifier?
In a power amp it would be better to isolate the power stages with LC rather than RC. For drivers and preamps RC is ok.
It also depends on the power supply rejection ratio of each stage. In some cases it is quite ok to supply both channels from the same node. In preamps you can also supply two consecutive stages from the same node. Usually each stage inverst phase. So the interaction is less.
If three stages are supplied from the same node, you can experience motorboating. This is a kind of positive feedback through the common supply.
Best regards
Thomas
This depends on the actual amp and I would also balance it with the overall budget which is spent.
You mention 8 tube stages which sounds a lot, is this a preamplifier?
In a power amp it would be better to isolate the power stages with LC rather than RC. For drivers and preamps RC is ok.
It also depends on the power supply rejection ratio of each stage. In some cases it is quite ok to supply both channels from the same node. In preamps you can also supply two consecutive stages from the same node. Usually each stage inverst phase. So the interaction is less.
If three stages are supplied from the same node, you can experience motorboating. This is a kind of positive feedback through the common supply.
Best regards
Thomas
Wow, those were some fast replies!
aardvarkash10, I have just purchased Merlin's books. I have been intently studying them for the last week. I am basing my PSU design on his examples. These books are awesome. He is a great teacher.
I hope he completes a trio of books by writing a book on the power amp output stage. I still have many basic questions about output transformers, power tubes, screens, etc.
I am going to go for a MOSFET regulated PSU, as Merlin describes in his preamp book. I want to have three Voltage levels regulated by three discrete MOSFET series regulators - 420V for the power tubes, 310V for the screen, and 260V for four 12AX7's.
In Merlin's preamp book, he shows several examples that have more than one HT supply for the preamp section. He mentions some limits on current for each HT section, but says nothing about how many tube stages can be supplied by each HT section.
I guess I could supply the first two 12AX7 input tubes with a separate HT section than the next two 12AX7's.
I don't know it this is a tradoff between a higher HT voltage versus some more noise filtering for the input stage. I have no "shop feel" for this part of the design.
yagoolar,
I will look into the PSUD 2 program. I hope it can run under Wine/Linux.
I have discovered LTSpice, which the author has made sure will run under Wine. In a couple of days I have learned enough to model my PSU section and all the tube stages. I am stuck on the output transformer and input transformer for the moment. I have been doing a lot of research on how to make a decent enough spice model.
I have learned more about tube amp behavior in the last few days using LTSpice than I have in a year. What a great program!
Thanks,
Dave
aardvarkash10, I have just purchased Merlin's books. I have been intently studying them for the last week. I am basing my PSU design on his examples. These books are awesome. He is a great teacher.
I hope he completes a trio of books by writing a book on the power amp output stage. I still have many basic questions about output transformers, power tubes, screens, etc.
I am going to go for a MOSFET regulated PSU, as Merlin describes in his preamp book. I want to have three Voltage levels regulated by three discrete MOSFET series regulators - 420V for the power tubes, 310V for the screen, and 260V for four 12AX7's.
In Merlin's preamp book, he shows several examples that have more than one HT supply for the preamp section. He mentions some limits on current for each HT section, but says nothing about how many tube stages can be supplied by each HT section.
I guess I could supply the first two 12AX7 input tubes with a separate HT section than the next two 12AX7's.
I don't know it this is a tradoff between a higher HT voltage versus some more noise filtering for the input stage. I have no "shop feel" for this part of the design.
yagoolar,
I will look into the PSUD 2 program. I hope it can run under Wine/Linux.
I have discovered LTSpice, which the author has made sure will run under Wine. In a couple of days I have learned enough to model my PSU section and all the tube stages. I am stuck on the output transformer and input transformer for the moment. I have been doing a lot of research on how to make a decent enough spice model.
I have learned more about tube amp behavior in the last few days using LTSpice than I have in a year. What a great program!
Thanks,
Dave
Last edited:
This has the foundations of an interest thread. Care to share what you have learnt or designed as it would improve my knowledge. The design books look good but expensive and I would have to wait for ship from US - you didn't spot an ebook version by any chance?
Hi Dave - the advantage of a world-wide membership is round the clock responses!
Three active regulators is probably overkill. The purpose of the regulator here is two-fold - to set the voltage (obviously) and secondly to remove power supply ripple.
imho, the highest level reg feeding your power tubes will do the majority of the work - everything after it can be very basic since they are invariably operating in class A and so are not cursed with variable current loads.
For simplicity, I would regulate the 310V to the screens - they respond well to tight regulation. Downstream from there at your voltage amps and phase splitter, simple dropping resistors and cap decoupling will do very well.
Of course, ymmv!
Thomas,
Thanks for the advice. Your explanation makes a lot of sense. Nice blog, by the way.
I am designing a power amp that mixes two line level signals together - one balanced and one unbalanced. This has caused the tube count to go up.
I intend to slightly overdrive this amp. In order to avoid motorboating, I think I will put the input stage for both channels on the same HT supply and then feed the mixing tubes and driver tubes from a separate HT stage. Both the screen and power tube anodes will be supplied from separate HT stages.
revoli,
Merlin Blencowe's books are definitely worth the money. He knows how to explain a concept without pulling in a dozen other undefined concepts. One idea builds a foundation for the next idea. He shows the math, without burying you in it. I am also getting a lot out of Morgan Jones' two books, "Valve amplifiers" and "Building Valve Amplifiers". Jones tends to make leaps in the math that I find harder to follow, though. The schematics, figures, and graphs are very well done by both authors.
I have a few of Kevin O'Connors books, also. They are full of good information, but I find myself having to re-read sentences and paragraphs several times to make any sense out of them. He sometimes leaves out labels in his schematics, which makes his explanations hard to follow. I do return to his books often though, as I study this subject.
The best explanation I have found of how a tube works comes from an army training manual, BasicTheoryAndApplicationOfElectronTubes-Army1952.pdf (free to download):
Basic Theory and Application of Electron Tubes : Depts. of the Army and the Air Force : Free Download & Streaming : Internet Archive
aardvarkash10,
Thanks for the advice. I may do an RC network to isolate the input tubes' HT from the mixing tubes' HT, rather than another MOSFET stage. In both cases I will need a cap that can withstand the full 260 volts. If I use a MOSFET, I may be able to use a smaller cap and I may be able to have a smaller voltage drop between stages. I also imagine that a MOSFET stage will be more immune to voltage sag, also.
Is there a minimum voltage drop I should design between HT stages?
mike,
I see in Kevin O'Connors designs a standard 10k-22uF, RC network separating stages. Both tubes in a splitter stage will have the same HT, but two common cathode stages, whether DC or AC coupled, will be supplied from separate HT supplies, even though they may occupy the same tube. I think this jibes with your advice.
I would like to have both channels sharing the same HT supplies for their input, mixing, and driving stages, otherwise I think I will be dropping the voltage down way too much and filling the chassis with caps.
yagoolar,
I will definitely give PSUD2 a try using Wine/Ubuntu. I am going to eventually model my output transformer by following this method:
Audio Transformer Data and Modeling
I need to take measurements from the actual transformer, though. So, I will be ordering a Hammond 1650N to try this out in LTSpice. I was able to measure and model in LTSpice a small 6VAC transformer that I had on hand using this method. I hope it works with the Hammond OT.
I think I know enough to make transformers and capacitors smoke, so I will keep studying and refining my design. Learning LTSpice has been great, as I am now able to smoke hundreds of virtual amps before burning up a real amp. More comments would be greatly appreciated.
Thanks,
Dave
Thanks for the advice. Your explanation makes a lot of sense. Nice blog, by the way.
I am designing a power amp that mixes two line level signals together - one balanced and one unbalanced. This has caused the tube count to go up.
I intend to slightly overdrive this amp. In order to avoid motorboating, I think I will put the input stage for both channels on the same HT supply and then feed the mixing tubes and driver tubes from a separate HT stage. Both the screen and power tube anodes will be supplied from separate HT stages.
revoli,
Merlin Blencowe's books are definitely worth the money. He knows how to explain a concept without pulling in a dozen other undefined concepts. One idea builds a foundation for the next idea. He shows the math, without burying you in it. I am also getting a lot out of Morgan Jones' two books, "Valve amplifiers" and "Building Valve Amplifiers". Jones tends to make leaps in the math that I find harder to follow, though. The schematics, figures, and graphs are very well done by both authors.
I have a few of Kevin O'Connors books, also. They are full of good information, but I find myself having to re-read sentences and paragraphs several times to make any sense out of them. He sometimes leaves out labels in his schematics, which makes his explanations hard to follow. I do return to his books often though, as I study this subject.
The best explanation I have found of how a tube works comes from an army training manual, BasicTheoryAndApplicationOfElectronTubes-Army1952.pdf (free to download):
Basic Theory and Application of Electron Tubes : Depts. of the Army and the Air Force : Free Download & Streaming : Internet Archive
aardvarkash10,
Thanks for the advice. I may do an RC network to isolate the input tubes' HT from the mixing tubes' HT, rather than another MOSFET stage. In both cases I will need a cap that can withstand the full 260 volts. If I use a MOSFET, I may be able to use a smaller cap and I may be able to have a smaller voltage drop between stages. I also imagine that a MOSFET stage will be more immune to voltage sag, also.
Is there a minimum voltage drop I should design between HT stages?
mike,
I see in Kevin O'Connors designs a standard 10k-22uF, RC network separating stages. Both tubes in a splitter stage will have the same HT, but two common cathode stages, whether DC or AC coupled, will be supplied from separate HT supplies, even though they may occupy the same tube. I think this jibes with your advice.
I would like to have both channels sharing the same HT supplies for their input, mixing, and driving stages, otherwise I think I will be dropping the voltage down way too much and filling the chassis with caps.
yagoolar,
I will definitely give PSUD2 a try using Wine/Ubuntu. I am going to eventually model my output transformer by following this method:
Audio Transformer Data and Modeling
I need to take measurements from the actual transformer, though. So, I will be ordering a Hammond 1650N to try this out in LTSpice. I was able to measure and model in LTSpice a small 6VAC transformer that I had on hand using this method. I hope it works with the Hammond OT.
I think I know enough to make transformers and capacitors smoke, so I will keep studying and refining my design. Learning LTSpice has been great, as I am now able to smoke hundreds of virtual amps before burning up a real amp. More comments would be greatly appreciated.
Thanks,
Dave
I am stuck, again.
I modeled my active power supply for my amp in ltspice, setting the resistor network for the mosfet gaits, so that I was getting the voltage output I wanted for each tube stage. I separately modeled all of the tube stages using the idealized voltage source symbol/model.
When I combine my modeled power supply with my modeled tube stages, the voltage levels sag over 150V. Any attempts to change the gate regulating resistor networks have little effect on raising the voltage level back up.
I thought that a regulated voltage source gave you a stiff voltage level that sags very little. What am I doing wrong? Does anybody have any thoughts they could share?
Thanks,
Dave
I modeled my active power supply for my amp in ltspice, setting the resistor network for the mosfet gaits, so that I was getting the voltage output I wanted for each tube stage. I separately modeled all of the tube stages using the idealized voltage source symbol/model.
When I combine my modeled power supply with my modeled tube stages, the voltage levels sag over 150V. Any attempts to change the gate regulating resistor networks have little effect on raising the voltage level back up.
I thought that a regulated voltage source gave you a stiff voltage level that sags very little. What am I doing wrong? Does anybody have any thoughts they could share?
Thanks,
Dave
I figured it out! This one hurt my brain.
Tony,
Attached are 2 screenshots of an LTSpice simulation of the power supply, which includes some dummy loads for the valve stages.
In the first screenshot, I have diodes D11 and D13 placed wrong. D11 and D9 share a common node, which caused a large voltage to be dropped, and D13 was also misplaced.
In the second, screenshot D9, D11, and D13 are properly isolated from each other with biasing resistors.
My first step to solving this was to remove D9, D11, and D13 from the circuit, at which point it behaved correctly. Later, I saw the misplacement of the diodes and resistors. I think D9, D11, and D13 actually are not necessary, since the IRF820 already has internal reverse bias protection diodes. Correct?
I will have to think to get clear in my mind exactly why these errors caused the circuit to behave the way it did. Once I saw it, though, I knew it was a (now) obvious mistake.
Please, let me know what you think.
Thanks,
Dave
Tony,
Attached are 2 screenshots of an LTSpice simulation of the power supply, which includes some dummy loads for the valve stages.
In the first screenshot, I have diodes D11 and D13 placed wrong. D11 and D9 share a common node, which caused a large voltage to be dropped, and D13 was also misplaced.
In the second, screenshot D9, D11, and D13 are properly isolated from each other with biasing resistors.
My first step to solving this was to remove D9, D11, and D13 from the circuit, at which point it behaved correctly. Later, I saw the misplacement of the diodes and resistors. I think D9, D11, and D13 actually are not necessary, since the IRF820 already has internal reverse bias protection diodes. Correct?
I will have to think to get clear in my mind exactly why these errors caused the circuit to behave the way it did. Once I saw it, though, I knew it was a (now) obvious mistake.
Please, let me know what you think.
Thanks,
Dave
Attachments
I am planning to build a power amp for musical instruments and mics that have been brought up to line level with a preamp, stomp box, etc.
I will look into increasing the reservoir cap, as you suggest. I need to make sure that I do not have too long of an inrush current - I don't want to fry the rectifier caps. I am now noticing that there is a marked difference in the size of reservoir caps between guitar amps and hi-fi amps. The hi-fi amps choose much larger caps.
My psu design is based on Merlin Blencowe's book, "Designing Power Supplies for Tube Amplifiers". He does a great job of explaining this circuit and the function of the various components. The 10uF caps tied to the MOSFET gates have their effective capacitance multiplied by the gain of the MOSFET. This allows for smaller caps and an extremely quiet HT supply.
Kevin O'Connor's book, "The Ultimate Tone", also describes a similar psu topology. Both are great books, but, like I said before, I think Blencowe is a better teacher. After reading Blencowe's book, I now feel I can get something from O'Connor's explanation.
Based on aardvarkash10's advice, I am using passive RC decoupling past the driver tubes' HT supply. In "TUT", by O'Connor, he says you drop about 1V for every 1k of resistance in an RC decoupling network. An LTSpice simulation shows this to be a pretty accurate estimate.
Thanks,
Dave
I will look into increasing the reservoir cap, as you suggest. I need to make sure that I do not have too long of an inrush current - I don't want to fry the rectifier caps. I am now noticing that there is a marked difference in the size of reservoir caps between guitar amps and hi-fi amps. The hi-fi amps choose much larger caps.
My psu design is based on Merlin Blencowe's book, "Designing Power Supplies for Tube Amplifiers". He does a great job of explaining this circuit and the function of the various components. The 10uF caps tied to the MOSFET gates have their effective capacitance multiplied by the gain of the MOSFET. This allows for smaller caps and an extremely quiet HT supply.
Kevin O'Connor's book, "The Ultimate Tone", also describes a similar psu topology. Both are great books, but, like I said before, I think Blencowe is a better teacher. After reading Blencowe's book, I now feel I can get something from O'Connor's explanation.
Based on aardvarkash10's advice, I am using passive RC decoupling past the driver tubes' HT supply. In "TUT", by O'Connor, he says you drop about 1V for every 1k of resistance in an RC decoupling network. An LTSpice simulation shows this to be a pretty accurate estimate.
Thanks,
Dave
Be mindful that you need a tube in place to pull down the voltage in each RC decoupled stage - so best to use cap voltage ratings that can cope with no tubes present (or place the nominal discharge resistor at the end of the preamp RC daisy chain, or augment with some HV zeners).
Edit: dang - missed the posts on the second page! Anyway, still valid for those of us that just use RC decoupling from output to input, and suffer lots of drop from one end to the other.
Edit: dang - missed the posts on the second page! Anyway, still valid for those of us that just use RC decoupling from output to input, and suffer lots of drop from one end to the other.
Last edited:
trobbins,
I am using RC filter networks for the last valve stages, so what you are saying applies. I am interested in your advice on using a zener.
So, with the zener, you would jack up the negative terminal of the filter cap? For example, if you had a 260V HT, you could put a 160V zener under the cap, and now use a cap rated at around 100V?
I am using RC filter networks for the last valve stages, so what you are saying applies. I am interested in your advice on using a zener.
So, with the zener, you would jack up the negative terminal of the filter cap? For example, if you had a 260V HT, you could put a 160V zener under the cap, and now use a cap rated at around 100V?
No that wasn't my intent with zener. The intent was that if you had say a 300V electrolytic with a nominal 260V HT (all tubes in) at the end of a few stages of RC droppers/decouplers, but if the tubes were pulled out then the 260V would rise above 300V, then put a 300V zener across the 300V electrolytic. The upstream R's in the RC decouplers would (depending on design) limit the cap voltage and not cook the zener. I just had a few 150V 5W zeners lying around so have used them for that application in a few restorations. However it may be an expensive option if you have to go buy them.
Ciao, Tim
Ciao, Tim
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.