Who can/will help me with highvoltage SMPS design for tube preamp.

Actually I chequed the "rimple" meaning when I saw it first because it looked weird, but it's an archaic form that can be used instead of ripple so I just thought you liked playing with words. I bet nobody talks about electricity today in Ampere's or Maxwell's language, but that doesn't mean they were wrong about their stuff.
I wasn’t inferring any thing other than saying you would be unlikely to get high quality/quantity of search results when using a phrase like ‘high performance, low rimple power regulator’ or at least not as good as low ripple. Yeah I saw the archaic form you mention too, but I don’t think it was deliberate. I also don’t believe it says anything about Koifarm’s intelligence, or lack thereof. I’ve seen enough of his posts to know that would be folly; before we even talk about what poor form that would be for me personally to suggest that someone making an obvious typo, or simple mistake (not even really a mistake, just a quirky term to use) in their non-native tongue, was anything more than that.
 
Last edited:
Today i looked at the side of Power Intergrations.
They have an online smps design app called PI expert.

I tried to design HV smps but the limit was 100V.
But connecting 4 windings of 100V get you 400V.

I would connect them with a switch to choose 100,200,300 or 400V output and only one output filter.
Maybe behind the smps a Kubota HV regulator.

Schematic below looks promising to use.
I only have to find true hole parts.


FullSchema.gif


Making the transformer also looks easy.

600535D7-E8A4-469B-A068-9AE9340D4B5F.gif
A0330411-1807-46F4-8C0C-CE35F861F931.gif
 
Last edited:

alexberg

Member
2009-10-01 2:25 pm
The absence of 50/100hz resp 60/100hz products. No 50/60hz transfer true housing to microphonic tubes.
With all do respect, ignorance is the bliss.
SMPS of any kind is a rather low loop gain class D amplifier powered with high ripples DC source. You will get 50(60)/100(120) and a multiples of thereof as well as intermodulation products on conversion frequency. Most ferrites for power electronicsc do have substantial magnetostriction thus you getting mains' modulated vibrations @ conversion frequency and multiples of thereof. Glass envelopes, having high stiffness so to speak will transfer this to the electrode system -> electrons do not have inertia in that case. Thirdly, if not resonant, the converter of yours will provide you with a lot of conducted and induced EMI. To beat those will be the art of its own.
While it's feasible - high sensitivity scopes are the best example - it's not the sensible way to go.
Often than not people report harsh sounding of amps due to intermodulation that comes from unusual sources like detection of EMI, self - oscillation or HF ripples from PSU.
If all of the above would not stop you from proceeding, I would go to ONSEMI/TI and check what they have on LLC.
For those proficient in English as doctor Higgins's the humble servant of yours is not a native speaker.
The last and not the least: even though SM PSU of yours operates, it does not infer that it's as proper as one would think.
 
  • Like
Reactions: 1 users
High voltage SMPS for tube amps and preamps has been done, Manley Labs use it now (a design by Bruno Putzeys altough) in their gear for a few years. And it sounds good enough to be in almost all their pro audio studio gear, and in headphone amps they build. But it seems not to be easy to do it. I'm no expert, but saying it's impossible to do it right is ********.

https://www.manley.com/manley-power
 

alexberg

Member
2009-10-01 2:25 pm
High voltage SMPS for tube amps and preamps has been done,
skipped
I'm no expert, but saying it's impossible to do it right is ********.
Rather self-evident, for EE, hurdles and obstacles to overcome were merely pointed out .
On the other hand the way to go is also suggested, namely LLC which could make it somewhat easier.
I bet reproducing most of good audio gear on avail is not that simple without substantial time, knowledge and money investment.
 

NickKUK

Member
2019-12-28 9:16 pm
Today i looked at the side of Power Intergrations.
They have an online smps design app called PI expert.

I tried to design HV smps but the limit was 100V.
But connecting 4 windings of 100V get you 400V.

I would connect them with a switch to choose 100,200,300 or 400V output and only one output filter.
Maybe behind the smps a Kubota HV regulator.

Schematic below looks promising to use.
I only have to find true hole parts.

Making the transformer also looks easy.

Your supplies aren't regulated and the danger and are only referenced by a single voltage feedback. It's possible to overload a non-regulated supply and the SMPS will not be aware of it. Not sure of the secondary winding isolation if you connect them all in series. I'm assuming it's rated at continuous 400+Vdc.

You'll start finding currents for 24Vdc to 400Vdc boosting become very large for the inductors. So it's probably better to run a mains PFC SMPS -> DC link voltage (say 380Vdc) then run separate DC-DC converters off the DC link. You still have the issue of overload but the separate boost/bucks can take that into account. The PFC then monitors the link voltage.
 

alexberg

Member
2009-10-01 2:25 pm
Are there some remarks about the schematic i posted before?
Let's start with the x-former. There's such a thing as Dowell, you need to partition windings horizontal (per picture).
I would add shields if you do know what to do with them
LLC topology looks nice, but a lot op 400V schematics have 5 to 10V ripple. 😇
I've spent 3 minutes on onsemi site. attached is the results in terms of ripples.
People here are genuinly trying to help which some won't accept.
94.png

Design done for 400V 0.5A. 13mv 50W yeilds 8mV at 20uf film cap as output filter.
It's highly unfortunate that you need to be given a fish instead of a rod, but here it is:
https://www.onsemi.com/design/resources/technical-documentation?rpn=FAN7688
LLC-Resonant-Half-Bridge-Converter-Design-Tool-for-FAN7688 xlsx

https://www.onsemi.com/pub/Collateral/AND90061-D.PDF
 
Last edited:
Alexberg, thank you for the info and links. Luckily i am on vacation and have a lot of time to read.
But there is quiet impressive schematic.

939F995A-29EC-47DB-8FEB-6BC2359C0325.png


As you seen i started with a unregulated SMPS witch also have overload protection according to the manual. And after that i like to use a linear regulator to get ripple below 1mV. Maybe 100uV is possible.

Schematic of linear regulator.

B3CD39BE-B128-4579-820C-0172972E844B.jpeg

This looks a lot easier to achieve.

@NickKUK The way with the DC/DC converters is not my thing. It looks like the kodabmx build. I like to have one nice PCB( max 100 x 100mm)..
 
Last edited:
With all do respect, ignorance is the bliss.
SMPS of any kind is a rather low loop gain class D amplifier powered with high ripples DC source. You will get 50(60)/100(120) and a multiples of thereof as well as intermodulation products on conversion frequency. Most ferrites for power electronicsc do have substantial magnetostriction thus you getting mains' modulated vibrations @ conversion frequency and multiples of thereof. Glass envelopes, having high stiffness so to speak will transfer this to the electrode system -> electrons do not have inertia in that case. Thirdly, if not resonant, the converter of yours will provide you with a lot of conducted and induced EMI. To beat those will be the art of its own.
While it's feasible - high sensitivity scopes are the best example - it's not the sensible way to go.
Often than not people report harsh sounding of amps due to intermodulation that comes from unusual sources like detection of EMI, self - oscillation or HF ripples from PSU.
If all of the above would not stop you from proceeding, I would go to ONSEMI/TI and check what they have on LLC.
For those proficient in English as doctor Higgins's the humble servant of yours is not a native speaker.
The last and not the least: even though SM PSU of yours operates, it does not infer that it's as proper as one would think.
Alexberg, i like to reply on this one. I measure when using the SMPS with the classD amp between 20hz and 20khz no interference products higher then -100db or even lower. About HF i shall make some measuements after my vacation. I have done already a lot of linear vs SMPS comparisons. And i could not hear the difference between linear or SMPS.
 

alexberg

Member
2009-10-01 2:25 pm
The point's it's easier said than done as well as not seeing smps shortcomings
(specifically if you were mere comparing products on the market, not self made ones).
Are the topic still tube preamp or class D power stage?
SMPS itself is the class D power stage and not the one you are used to:
in self-oscillatory ones loop gain is rather "high" to say the least
which is not the case for PSU where only DC performance may be on par.
 
Last edited:

alexberg

Member
2009-10-01 2:25 pm
Schematics shown has two auxiliaries and sync rectification which are unreasonable unless you're after the power density per say, especially @ 400V Uout. Using secondary type controller may require some start-up aux ps but you can always salvage wall-wart from older phones to make things easier.
P.S. I would recommend to look for Nicola Rosano on LLC in particular.
 

NickKUK

Member
2019-12-28 9:16 pm
Today i looked at the side of Power Intergrations.
They have an online smps design app called PI expert.

I tried to design HV smps but the limit was 100V.
But connecting 4 windings of 100V get you 400V.

I would connect them with a switch to choose 100,200,300 or 400V output and only one output filter.
Maybe behind the smps a Kubota HV regulator.

Schematic below looks promising to use.
I only have to find true hole parts.


View attachment 1073724

Making the transformer also looks easy.

View attachment 1073726 View attachment 1073727

Thinking about this further - you could simply stick with 100V on each winding, connect in parallel and then use a voltage multiplier diode network to get to 400V. The load would be spread evenly and you get smoothing built in from the voltage multiplier network.

I still don't like that it's only using a feedback coil and not voltage based feedback on the load..