Whilst tongue in cheek the power factor corrected current fed bridge was a semi serious suggestion for an offline convertor. You get PFC along with multiple isolated supplies with good tracking based on what is nominally a simple circuit.
Look Ma, no output inductors. It is similar to flyback. My dig was about the enforced low bandwidth of the control loop. You end up having to use larger than normal, for an SMPS, output capacitors to tame low frequency ripple. That of itself need not be a biggie.
Switches on Marketing Mode. Of course, as we all know, valves sound better when fed with lumps of iron and that is largely down to the 100/120Hz nature of the feed which is what the suggested solution approximates.
Look Ma, no output inductors. It is similar to flyback. My dig was about the enforced low bandwidth of the control loop. You end up having to use larger than normal, for an SMPS, output capacitors to tame low frequency ripple. That of itself need not be a biggie.
Switches on Marketing Mode. Of course, as we all know, valves sound better when fed with lumps of iron and that is largely down to the 100/120Hz nature of the feed which is what the suggested solution approximates.
"elicit".
"illicit" is illegal.
illicit - Google Search
"She admitted to having been in possession of illicit drugs."
elicit - Google Search
"a corrupt heart elicits in an hour all that is bad in us"
This is where progress came to a stop in its phisical form. YouTube
The noise is coming from the crappy chinese powersupply back there near the speaker, its powering the filaments. The noise is not caused by the switching supply that creates the HV. The noise is also caused by the unideally LONG cable to the audio source. The fans wont be a requirement if I redesign the board a bit. Its outputting 55W of power at the voltage measured (round 130mA).
The switching transformer arrived for the mains version, but I hit so many roadblocks that I simply am about to give up this project now for sure. Cant really figure out how to isolate the output from the mains since all the controll circuitry will be on the mains side. Opto isolation is the way but cant figure out how opto isolator functions and cant get my head arround it. Originally I planned to modify a TL431 opto refference from a chinese switching supply but noticed it wont be that simple becase some component values just dont seem right for monitoring 500V output.
The noise is coming from the crappy chinese powersupply back there near the speaker, its powering the filaments. The noise is not caused by the switching supply that creates the HV. The noise is also caused by the unideally LONG cable to the audio source. The fans wont be a requirement if I redesign the board a bit. Its outputting 55W of power at the voltage measured (round 130mA).
The switching transformer arrived for the mains version, but I hit so many roadblocks that I simply am about to give up this project now for sure. Cant really figure out how to isolate the output from the mains since all the controll circuitry will be on the mains side. Opto isolation is the way but cant figure out how opto isolator functions and cant get my head arround it. Originally I planned to modify a TL431 opto refference from a chinese switching supply but noticed it wont be that simple becase some component values just dont seem right for monitoring 500V output.
I can't know what difficulties you are experiencing because you are not giving much away in terms of what your supply actually looks like...
If you want to use a TL431/OPTO then I could guess your component selection problem for a 500V output arises because with your standard transformer you only have a 500V output on the secondary side.
Ideally given you are dealing with Valves you would want a transformer that gives you windings for Anode and Heater. That makes things much simpler in terms of control.
You would regulate the Anode whilst powering the TL431/OPTO from the Heater.
Of course if your commercial off the shelf, COTS, transformer does not have a Heater winding then you get an insurmountable problem. You either have to find a COTS transformer that does or wind your own.
If you want to use a TL431/OPTO then I could guess your component selection problem for a 500V output arises because with your standard transformer you only have a 500V output on the secondary side.
Ideally given you are dealing with Valves you would want a transformer that gives you windings for Anode and Heater. That makes things much simpler in terms of control.
You would regulate the Anode whilst powering the TL431/OPTO from the Heater.
Of course if your commercial off the shelf, COTS, transformer does not have a Heater winding then you get an insurmountable problem. You either have to find a COTS transformer that does or wind your own.
The SMPS has ONLY ONE OUTPUT, and that is the HV. I dont have access to any custom wound transformers so currently the filament is powered by another PSU and for a good while it will be that way. Even if I one day decide to make a combined PSu that will also power heaters I may just copy a chinese design that I would modify to get an output from 5 to 20V at least 40W, but I wont power the entire thing by one tranformer because it will drag the voltage on the HV and I have to choose to either regulate the HV or the filament. And I sure want to have them both stable. In the future I may end up having a chance to experiment with SMPS tranformers that have more windings but currently its not an issue Im having and I do want both rail regulated.
> I wont power the entire thing by one tranformer because it will drag the voltage on the HV and I have to choose to either regulate the HV or the filament.
PC power supplies deliver +12, +5, +3.3(?), and sometimes negative supplies. Only *one* of them is strictly regulated. The others tend to track very well because a SMPS has tight coupling and low-low winding resistance.
Intel specs require +/-5% on 12V 5V and 3.3V under "all" conditions of line, loading, and environment. PC power supply makers do this by monitoring a single output. 5% is not a big deal in tubes, which vary 20%, and can take 10% spread of heater voltage. The 5% spec on PC supplies includes NO load, and I bet if you limit a 6.3V output to, say 2A-6A (avoiding the low and high ends of rectifier drop), you can do better than 5%.
PC power supplies deliver +12, +5, +3.3(?), and sometimes negative supplies. Only *one* of them is strictly regulated. The others tend to track very well because a SMPS has tight coupling and low-low winding resistance.
Intel specs require +/-5% on 12V 5V and 3.3V under "all" conditions of line, loading, and environment. PC power supply makers do this by monitoring a single output. 5% is not a big deal in tubes, which vary 20%, and can take 10% spread of heater voltage. The 5% spec on PC supplies includes NO load, and I bet if you limit a 6.3V output to, say 2A-6A (avoiding the low and high ends of rectifier drop), you can do better than 5%.
Yes, youre right BUT I have not yet made a powersupply of this let alone mains powered and I dont think you read the entire post properly because I stated
> I dont have access to any custom wound transformers
so it is a no go. Also I have difficulty with getting the mains isoolated from the output. So that is the biggest problem currently im having.
And yes I considered handwinding but no I wont do that.
because its nowhere near complete and neither clear to understand (I have not moved things arround so that the schematic can be read clearly at the first glance). I can however send the schematic if you wish, but there is really no magic there because its simply the PWM controller that can be any chip and the bare flyback topology.
Its been a while since I posted here. For those who abslutely hated my idea of creating a powersupply of its own kind and instead tell me to buy a ebay chinese power supply this is great news.
I had 3 tries at this project all inevitably failed and I burned far too much money just to find Im not going make a HV power supply of the said specs.
First LT3751, didnt even get a single pulse out of the chip. Second the all custom power supply (from the video). It burned a dozen mosfets and didnt reagulate almost at all. Third was the TL494, regulates very tight, copes great with resistive static loads, but when loading it with a AB amplifier, dynamic loads will make the controller spazz out go 90% duty cycle and trip OCP before it fries the fet.
All efforts have been proven pointless and just a waist of time and money. I dont have a single idea, where or what can I possibly do now to create such a power supply that would also work instead of blowing up.
Well, the project has inevitably met its end with a big failure sticker on its archive.
I am sorry to hear that you did not have good luck with you SMPS design, however I wouldn't be too hard on yourself. Designing a high voltage high power SMPS is not an easy task and takes a lot of experience. Have you read LT app note https://www.analog.com/media/en/technical-documentation/application-notes/AN118fb.pdf
Not sure what current levels the app note is talking about or what your end goal was. Is there a reason you want to use SMPS instead designing a high voltage regulator for a standard frequency transformer?
Not sure what current levels the app note is talking about or what your end goal was. Is there a reason you want to use SMPS instead designing a high voltage regulator for a standard frequency transformer?
output: 500V 70W
Input: 12-24V
The reason I want to create a useable switching powersupply for power tube amplifier, is the price. No matter how I look at it its less than 50 dollars. Power transformers cost a fortune and Im a student so I cant afford anything really. Now to be honest by this time I spent enough money at the thing I could just buy a transformer but you cant just tune a transformer to your needs with a pot....at least not efficiently enough for me to consider such a linear regulator.
Im givving the thing one last try with the good old trusty UC3842. If that works then I know I need a controller of that kind (mainly looking for LT chips that can do what I need them to do.)
Input: 12-24V
The reason I want to create a useable switching powersupply for power tube amplifier, is the price. No matter how I look at it its less than 50 dollars. Power transformers cost a fortune and Im a student so I cant afford anything really. Now to be honest by this time I spent enough money at the thing I could just buy a transformer but you cant just tune a transformer to your needs with a pot....at least not efficiently enough for me to consider such a linear regulator.
Im givving the thing one last try with the good old trusty UC3842. If that works then I know I need a controller of that kind (mainly looking for LT chips that can do what I need them to do.)
If you are looking for cheap cheap, you could use two lower voltage transformer and back feed one of them with the other. Not exactly the most efficient or elegant method, but lower ripple than a voltage multipliers.
I am sure you are aware but when dealing with high voltage switching one usually needs to add snubber diodes and Baker clamp circuits to the drive circuit.
Why do you want to use such a low input voltage? Are you looking to power this from a battery?
I haven't used the UC3842 but I read it does not have built-in slope compensation.
Wish you the best luck and I'd be happy to help if I can.
I am sure you are aware but when dealing with high voltage switching one usually needs to add snubber diodes and Baker clamp circuits to the drive circuit.
Why do you want to use such a low input voltage? Are you looking to power this from a battery?
I haven't used the UC3842 but I read it does not have built-in slope compensation.
Wish you the best luck and I'd be happy to help if I can.
yes, yes.. I am aware. Feeding transformers back to back? It will weight a metric crap ton. Snubbers are not an issue here, in fact none of my powersupplies so far used a proper primary side snubber network yet.
I run everything on 12-24V on my bench. Easy to manage. Such low voltage will bring the efficiency to basically less than average. Still 75-80% should be alright.
I did plan a full blown AC powered supply. Unfortunately the switching transformer I have does not have any AUX windings and only a single output winding. The voltage feedback will also be a not fun topic here because I need to power the secondary side FB circuit somehow...so far I came up with shunt regulating the supply for that so I dont fry the TL431 with 500V but at least 5mA is required for such a thing and thats 2.5W of power just into heat all the time at 500V.
Lets assume I need a power supply for a modified Mullard 5-20 Amplifier (I built them and by modified I mean the Rk capacitors have been increased from 47uF to 470uF low impedance capacitors (and other small not mattering things that would not affect the powersupply.).
420V at roughly 50W constantly and then the dynamic load of the tubes. It would need heavy frequency compensation at the feedback to not spazz out the controller into shorting things.
The UC3841-3 doesnt have slope compensation youre right, besides I have no idea how to use it properly anyways so it would be just messing up things. Also its way better than the TL494 without current limiting. I realised voltage mode Flyback is pretty wonky. So I will have to look into the UC3843 circuit. I built many forward low voltage to 50V converters at hig power (50W more). Never Flyback. So there is a good chance I will go the mains route this time for the UC3843 method, its simple to implement and has straight forward feedback options. Ill see. Perhaps Ill post a schematic here of the "semi working" powersupply. So you would be able to get what Im talking about.
Thank you for your support. Well see where this goes.
I run everything on 12-24V on my bench. Easy to manage. Such low voltage will bring the efficiency to basically less than average. Still 75-80% should be alright.
I did plan a full blown AC powered supply. Unfortunately the switching transformer I have does not have any AUX windings and only a single output winding. The voltage feedback will also be a not fun topic here because I need to power the secondary side FB circuit somehow...so far I came up with shunt regulating the supply for that so I dont fry the TL431 with 500V but at least 5mA is required for such a thing and thats 2.5W of power just into heat all the time at 500V.
Lets assume I need a power supply for a modified Mullard 5-20 Amplifier (I built them and by modified I mean the Rk capacitors have been increased from 47uF to 470uF low impedance capacitors (and other small not mattering things that would not affect the powersupply.).
420V at roughly 50W constantly and then the dynamic load of the tubes. It would need heavy frequency compensation at the feedback to not spazz out the controller into shorting things.
The UC3841-3 doesnt have slope compensation youre right, besides I have no idea how to use it properly anyways so it would be just messing up things. Also its way better than the TL494 without current limiting. I realised voltage mode Flyback is pretty wonky. So I will have to look into the UC3843 circuit. I built many forward low voltage to 50V converters at hig power (50W more). Never Flyback. So there is a good chance I will go the mains route this time for the UC3843 method, its simple to implement and has straight forward feedback options. Ill see. Perhaps Ill post a schematic here of the "semi working" powersupply. So you would be able to get what Im talking about.
Thank you for your support. Well see where this goes.
I can comment that switchmode control often requires pedantic care in pcb layout, and part selection, and local wiring, and careful management of any pulsed current path and any feedback signal and grounding schemes, and even how to probe to confirm the cleanness of signals. Forums can be a useful way to get free critique on any and all of the above issues, if you have an openness to consider what others suggest, although that typically requires a level of total disclosure if questions are asked.
Last edited:
By all means take this the wrong way but until you lose your fear of winding your own magnetics and otherwise become more robust in your thinking you are really going nowhere.
Mouser Electronics Slovakia - Electronic Components Distributor
ETD39 | Mouser Electronics, Inc. Slovakia
IR2110 | Mouser Electronics, Inc. Slovakia
Mouser Electronics Slovakia - Electronic Components Distributor
ETD39 | Mouser Electronics, Inc. Slovakia
IR2110 | Mouser Electronics, Inc. Slovakia
I am aware that I need to take a lot of things into consideration. I did design a couple switching power supplies all from 12-24V input so I learned some things. Especially the grounding techniques. I design lots of PCBs so I think I should be fine.
I thought it worth making that post - even though this is a diy forum, I would professionally be concerned about a smps designer that gave a project status report with
A nice aspect of smps design is that it covers such a wide range of technical aspects - of which any one can trip up a prototype until adequately appreciated and worked through. PCB design is just one aspect where noise level can increase with an output variable and then reach some limit on a feedback or local supply rail such that the controller gets confused. Feedback management is another area that often takes a lot of effort and simulation to appreciate that any output disturbance or change can be adequately handled.
A nice aspect of smps design is that it covers such a wide range of technical aspects - of which any one can trip up a prototype until adequately appreciated and worked through. PCB design is just one aspect where noise level can increase with an output variable and then reach some limit on a feedback or local supply rail such that the controller gets confused. Feedback management is another area that often takes a lot of effort and simulation to appreciate that any output disturbance or change can be adequately handled.