Hi guys,
I'm collecting info for my future amplifier building (i have decided to go with a KT88 SE amp).
I'm a good electronic engineer (but I have no experience at all with tubes), so my question is:
Is it possible to realise an high frequency PWM power supply for my amplifier? I mean, an SMPS running at, let's say, 50 Khz?
A power supply like this one will be as small as a laptop power supply (something like 2 cigarettes packets), with regulated output and practically no ripple.
Your opinion?
Any problem you can see?
Is there anybody that has already built something like this?
Thanks in advance,
Giovanni
I'm collecting info for my future amplifier building (i have decided to go with a KT88 SE amp).
I'm a good electronic engineer (but I have no experience at all with tubes), so my question is:
Is it possible to realise an high frequency PWM power supply for my amplifier? I mean, an SMPS running at, let's say, 50 Khz?
A power supply like this one will be as small as a laptop power supply (something like 2 cigarettes packets), with regulated output and practically no ripple.
Your opinion?
Any problem you can see?
Is there anybody that has already built something like this?
Thanks in advance,
Giovanni
Yes, it has been done, so it should be possible. You didn't open your question by saying that you're an experienced switch-mode power supply designer, so I assume you aren't. In which case, the answer may be more problematic and making a PSU that can deliver 400V at quite a bit of current without any nasties and without suddenly dying unexpectedly is likely to be rather harder. You will definitely need a good oscilloscope and probably a current probe to assist during development. Oh, and a lot of reading stuff about switch-mode power supply design.
Do report back with results - I'm sure a lot of people would like to exchange their kilos of expensive iron for a nice light power supply.
Do report back with results - I'm sure a lot of people would like to exchange their kilos of expensive iron for a nice light power supply.
It is worth going through the schematics at www.davidberning.com to see some implementations of switching supplies for tube amps. The "support" link will get you to the schematics of a variety of commercial amps from the 1970s until the '90s. He has also adapted this technology to replacing the output transformer.
BTW, I love Bergamo! Branzi is impossible to get here in the US.
BTW, I love Bergamo! Branzi is impossible to get here in the US.
A SMPS for a vacuum tube amplifier is definitely possible. If you are building the amp for your own personal use, go for it. If you are building the amp for eventual sale, you will get a lot of resistance from the "sandophobic" (fear of silicon) crowd.
I have experimented with SMPS for automotive vacuum tube amps with moderate sucess. I have a design working currently that puts out 300 volts at 200 mA. It is a push pull converter using a SG3525 chip from ON Semiconductor. This type of design is well suited for 12 volt operation, but not for 120 or 240 volt use. The transformer came from a 12 volt 15 amp power supply, I hooked it up backwards.
I have experimented with "off line" flyback converters with more smoke than success. I was using a NCP1651 IC from ON semi. The power supply works OK, but blows up violently without warning. Sometimes it will run continuously for days, and then BANG! It is gone. I have not had time to figure it out.
I saw a Chinese 300B SE amp on e-bay that featured a SMPS, nobody bid on it.
I have experimented with SMPS for automotive vacuum tube amps with moderate sucess. I have a design working currently that puts out 300 volts at 200 mA. It is a push pull converter using a SG3525 chip from ON Semiconductor. This type of design is well suited for 12 volt operation, but not for 120 or 240 volt use. The transformer came from a 12 volt 15 amp power supply, I hooked it up backwards.
I have experimented with "off line" flyback converters with more smoke than success. I was using a NCP1651 IC from ON semi. The power supply works OK, but blows up violently without warning. Sometimes it will run continuously for days, and then BANG! It is gone. I have not had time to figure it out.
I saw a Chinese 300B SE amp on e-bay that featured a SMPS, nobody bid on it.
I am.
But I never used tubes before, and I really do not know how they behaviour.
To build an SMPS means work with RF issues, and I don't know how this can influence the functioning of tubes.
I can reduce RF to almost zero, shield the SMPS, regulate its output, but I do not know if this is enough to use it with tubes.
I know nothing about tubes, and so it could be, for me, that SMPS cannot be used.
This was my question.
BTW: I built in the past an SPMS that has an output of 4100V, 350mA (almost 1500W), so I should have the necessary skill to build a 400V, 300mA SPMS.
Thanks,
Giovanni
If you can build the supply, go ahead. Put it in its own shielded box, like the power supply inside a computer. The same filtering scheme that is used in a PC power supply should work OK. I would put CLC filtering on each output, only the C's should be chosen to have a low ESR at the switching frequency. Standard electrolytics are useless. The L should be a ferrite core type like the ones used in typical SMPS's. All outputs (including filaments) should be filtered DC, and you should be OK.
First decide what type of tube amplifier you want to build. Then determine its power requirements. There will be at least two supplies needed. The plate (high voltage) supply, which will be somewhere between 200 and 1000 volts. This is the one that you will tie the feedback loop to. It will need the tightest regulation, and lowest output impedance. You will also need one or more filament supplies. These come from an auxiliary winding on the transformer. The regulation inherited from the main winding is good enough. Some amplifier designs require a negative voltage bias supply. This can come from an auxiliary winding, but typically should be regulated by a linear regulator to insure stbility and low noise.
Many of the modern digital audio amps incorporate SMPS's, and if you think about it a digital amplifier is really a modulated SMPS with a low pass filter on its output.
First decide what type of tube amplifier you want to build. Then determine its power requirements. There will be at least two supplies needed. The plate (high voltage) supply, which will be somewhere between 200 and 1000 volts. This is the one that you will tie the feedback loop to. It will need the tightest regulation, and lowest output impedance. You will also need one or more filament supplies. These come from an auxiliary winding on the transformer. The regulation inherited from the main winding is good enough. Some amplifier designs require a negative voltage bias supply. This can come from an auxiliary winding, but typically should be regulated by a linear regulator to insure stbility and low noise.
Many of the modern digital audio amps incorporate SMPS's, and if you think about it a digital amplifier is really a modulated SMPS with a low pass filter on its output.
So you are an experienced SMPS designer. Excellent! Well, you have lots of possibilities open to you. As tubelab.com has pointed out, you will need a high voltage supply of at least 300V, and this needs to be regulated and with low ripple.
You also need low voltage supplies at high current for the heaters, and I suspect that these should come from an entirely different switcher. They don't need tight regulation because the load is constant, but they do need particular attention to high frequency noise and switching artifacts. The reason for this is that noise on the heaters will be capacitively coupled to the cathodes. However, assuming you can make this supply quiet, you have an option open to you that isn't open to others. You could use a cathode follower output stage. The advantage of this is that it has local feedback to give the low output resistance. It has two disadvantages. First, it needs a driver stage with a very high HT (typically >700V), but that's not a problem for you - you can just stack a pair of 350V supplies. The other problem is the heater to cathode insulation. Because only a very limited voltage is tolerated between the heater and the cathode, they ideally need to be tied together. But that means you'd need to drive the typical 1nF interwinding capacitance of a conventional transformer. But you can make an HF transformer with far lower interwinding capacitance.
You also need low voltage supplies at high current for the heaters, and I suspect that these should come from an entirely different switcher. They don't need tight regulation because the load is constant, but they do need particular attention to high frequency noise and switching artifacts. The reason for this is that noise on the heaters will be capacitively coupled to the cathodes. However, assuming you can make this supply quiet, you have an option open to you that isn't open to others. You could use a cathode follower output stage. The advantage of this is that it has local feedback to give the low output resistance. It has two disadvantages. First, it needs a driver stage with a very high HT (typically >700V), but that's not a problem for you - you can just stack a pair of 350V supplies. The other problem is the heater to cathode insulation. Because only a very limited voltage is tolerated between the heater and the cathode, they ideally need to be tied together. But that means you'd need to drive the typical 1nF interwinding capacitance of a conventional transformer. But you can make an HF transformer with far lower interwinding capacitance.
Here's a mobile one I developed a while back:
http://geek.scorpiorising.ca/contrib/Geek/smps_50W_12V-350V_FWD.png
And the thread on the trials and tribulations:
http://geek.scorpiorising.ca/GeeK_ZonE/index.php?topic=1092.0
I *never* want to go through that again!
http://geek.scorpiorising.ca/contrib/Geek/smps_50W_12V-350V_FWD.png
And the thread on the trials and tribulations:
http://geek.scorpiorising.ca/GeeK_ZonE/index.php?topic=1092.0
I *never* want to go through that again!
Thanks for the help...
EC8010: I've already heard about the "Cathode Follower" configuration (and I really like it!!!), but I'm a complete newbie to tubes, so I do not understand completely how it works (I'm not sure I've understand how tube output impedance works), and I fear a CC is too complicated for me as a first project.
I'm not able to draw a CC schematic, as I can only "copy" (and slightly modify) existing tube schematics, for now.
I will go for a less complicated SE project, perhaps with a sort of a local feedback I have in mind (I will try it before to share).
For the heaters, I have a stock of high quality traditional split bobbin transformers, each one with two 6.3 V 2.2A outputs: I will use one of those for each channel, using one output for the KT88 and the other one for the pre tubes.
They are very small, light and well done (they came from Signal Transformers): I can hide them under the chassis.
My idea is to use a traditional heating method (for the main reason I already have the trafos, and they are for free) and build and HV power supply to remove the big, ugly (sorry, folks, I know I'm an heretical but I do not like big iron masses) power transformer.
I read the thread: they are experimenting the SMPS without any documentation or calculation: the most difficult part of an SMPS is to build the transformer, you MUST make some serious calculation on it before to start, or you'll hardly will have success, unless you're lucky.
If you know how to calculate the trafo, all the rest is quite easy.
Using a trafo from a monitor as it is it is not a good idea....
😀
As for every electronic project, it can be really hard to realise or really easy: it depends on the knowledge you have.
Designing an SMPS (I mean, a little SMPS with only a secondary output, not an industrial one) is not more complicated than build, for example, a good Williamson amp.
But I'm not able to build a Williamson amp!
Thanks again for your help
EC8010: I've already heard about the "Cathode Follower" configuration (and I really like it!!!), but I'm a complete newbie to tubes, so I do not understand completely how it works (I'm not sure I've understand how tube output impedance works), and I fear a CC is too complicated for me as a first project.
I'm not able to draw a CC schematic, as I can only "copy" (and slightly modify) existing tube schematics, for now.
I will go for a less complicated SE project, perhaps with a sort of a local feedback I have in mind (I will try it before to share).
For the heaters, I have a stock of high quality traditional split bobbin transformers, each one with two 6.3 V 2.2A outputs: I will use one of those for each channel, using one output for the KT88 and the other one for the pre tubes.
They are very small, light and well done (they came from Signal Transformers): I can hide them under the chassis.
My idea is to use a traditional heating method (for the main reason I already have the trafos, and they are for free) and build and HV power supply to remove the big, ugly (sorry, folks, I know I'm an heretical but I do not like big iron masses) power transformer.
I read the thread: they are experimenting the SMPS without any documentation or calculation: the most difficult part of an SMPS is to build the transformer, you MUST make some serious calculation on it before to start, or you'll hardly will have success, unless you're lucky.
If you know how to calculate the trafo, all the rest is quite easy.
Using a trafo from a monitor as it is it is not a good idea....
😀
As for every electronic project, it can be really hard to realise or really easy: it depends on the knowledge you have.
Designing an SMPS (I mean, a little SMPS with only a secondary output, not an industrial one) is not more complicated than build, for example, a good Williamson amp.
But I'm not able to build a Williamson amp!
Thanks again for your help
Your heater scheme sounds good. I should warn you that SE amplifiers are more sensitive to power supply problems than push-pull, but as SMPS design is your speciality, trading difficulty in something you don't know about for difficulty in something you do know about makes sense.
croccodillo said:
Smps noise within an amp SE or push-pull is hardly ever an issue....the B+ ripple frequency is so high compared to 100/120hz.(I've come across many equipments with an smps inside and a quick check on the noise meter gives quite acceptable circuit noise performance down to -75/80dB below 1 watt with input s/c)....it's RF charge injection via the chassis and mains leads that the tuner heads and other sensitive equipment picks up the hash. I agree that RF experience and physics is vital to smps design. I'm suprised to hear the transformer as a "problem" bit. The trouble with designing smps proper with safety interlocks and watchdogs is that it takes way longer to get it right than designing a bog standard tube amp .
If I was a beginner with glass I'd rather crack a proven design, get it going for your collegues to admire then you will be the better.
Understanding tube characteristics is easy..(go to Steve Bench's site at Duncan amps , Svetlana alias at tubeshop.com and study various tube curves).
richj
If you have the engineering knowledge to build a multi kilovolt SMPS without making smoke, you can build just about any tube amplifier that you want. You will find that tubes are pretty simple after you have gained some knowledge and experience.
I would recommend the book Valve Amplifiers by Morgan Jones. I don't recommend this book to someone without a good knowledge of electronics, since he covers the entire first year of engineering school in one chapter, but the rest of the book outlines the design of vacuum tube amplifiers quite well. There is a companion book, Building Valve Amplifiers that deals with the mechanical aspects of amplifier construction, if that interests you. The author seems to have a bias toward push pull amps, but the theory is applicable to all tube amps.
If you build an amplifier design from one of the popular schematics, it WILL work. Then you can modify it to your liking. I see no problem with using a SMPS. Using seperate filament transformers, under the chassis, is a good idea, I have done this.
I would recommend the book Valve Amplifiers by Morgan Jones. I don't recommend this book to someone without a good knowledge of electronics, since he covers the entire first year of engineering school in one chapter, but the rest of the book outlines the design of vacuum tube amplifiers quite well. There is a companion book, Building Valve Amplifiers that deals with the mechanical aspects of amplifier construction, if that interests you. The author seems to have a bias toward push pull amps, but the theory is applicable to all tube amps.
If you build an amplifier design from one of the popular schematics, it WILL work. Then you can modify it to your liking. I see no problem with using a SMPS. Using seperate filament transformers, under the chassis, is a good idea, I have done this.
- Status
- Not open for further replies.