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

I've been designing switching power supplies of various sorts for over 40 years.
Then we're almost on par: a bit less than 40.
I do not like center tapped PP staff that came out of space design where opposite switch could open at any given moment. Neither HB (PC PSU) which relies on secondary side inductor for regulation and does not work at low load.
It's fare to say that as shown yours's pretty quiet and primary commutation could be rather soft if Ls between primaries is somewhat on a high side. It's probably the case for low input voltage. It provides some boost at idle which is not relevant if some load is present.
Anyhow I merely tried to steer TS from flyback which has ugly form factor for a current well exceeding one for LLC and being not that easy to furnish. Not knowing experience it's hard to suggest something particular. Well documented solution should always do.
In case like you've presented I would go for something like
https://www.irf.com/technical-info/userguide/ug-0601.pdf Without secondary sync rectifiers.
If you insert resonant capacitor in series with primary, you'll get, with some effort, ZVS at primary regardless of the load and ZCS in rectifiers. Expired by now US7145786B2 Point of load sine amplitude converters and methods.
There's some magnetizing current circulating to obtain ZVS, that's it. It's perhaps multi quasi resonant by definition but who cares.
Cheers!
Alex
 
Last edited:
I actually have used a classic inverting flyback DC-DC design to provide a few watts of negative supply for headphone amp designs powered by a switching adapter. A CMOS gate package was used to implement a constant on-time controller, with ultimate short-circuit protection provided by the adapter feeding the DC-DC converter. A couple of my early compact SE amp designs used a quasi-resonant flyback utilizing an ST L6565 controller. The AC-DC supply is nothing I would foist on the world due to concerns about safety and liability, but it worked well to power my "Shrine" SE UL amplifier with hybrid front end and 1625 outputs.
 
Use an isolation xfmr and LT8312 development board:
 

Attachments

  • LT8312_Demoboard.png
    LT8312_Demoboard.png
    41.9 KB · Views: 134
Thanks for the schematic but looking at post 1, I mis that it is adjustable. Ripple =?, and isolation transformer is adding extra cost.
1) It's adjustable -- the value of the resistive network tagged to the "FB" pin. See the datasheet/
2) Ripple will be a function of the current drawn. To reduce ripple the value of the output inductors has to change. You can post-filter the supply as well with something like the Janus Regulator which is really a ripple eater and quite effective.
3) Isolation transformer for safety. I believe that DIYAUDIO frowns on purely "off-line" switchers.

The chip is tiny, obviously. If you don't want to build it in its entirety, you can get the development board from ADI or DK.
 
A (isolation) transformer as well as optical device for the feedback are mandatory for any mains SMPS.
You basically can't get around making your own transformer.

To keep it simple, I would go for some kind of PFC stage, that keeps the voltage relatively steady for the following stage.
If you want to keep it straightforward, you could have a look at the IR(S)2153
Above 70W, PFC is required.

The IR(S)2153 will work without PFC, but regulation is rather poor.
On the other hand, the regulation of a linear transformer power supply is also not great, and even worse with a tube rectifier.
 
Ad the moment i use this one.

https://www.aliexpress.com/item/100...daa,tpp_buckets:668#2846#8115#2000&pdp_ext_f={"sku_id":"12000015963231639","sceneId":"30050"}&pdp_npi=2@dis!EUR!!45.69!!!!!@211b5db316551372258983283ee191!12000015963231639!rec

View attachment 1072981

But heater and anode supply share ground. It is not possible to lift heatersupply.
There is a 30mV rimple at 62Khz. I like to compare if there is a change in sound when rimple is below 1mV.

I already tried LR8N with fet to bring down rimple, but rimple just got 15mV down(-3db).

Further i have not found a good smps diy design with great specs for hifi gear.
I have just bought this. There is a nice delay (20 secs?) and then the HV comes up slowly. This latter is very good.
less good is the shared ground. That is, they are interconnected.
there are other boards that have a floating filament power. Much better.
I will feed through a standard tube power supply choke to get all residue away.

and many Riaa or pre boards etc have a HV section with a HV smoothing series transistor; that will help if the ground plane is clean . .
several years ago I used an simple SMPS for the filament in a RIAA amp. The residue was visible in the output . I dropped the whole idea. [do I fall in the same trap twice???]
 
Last edited:
  • Like
Reactions: Koifarm
Today made a schematic with through hole parts of a Isolated linear regulated highvoltage SMPS.
The LNK6769E Offline switcher has over/undervoltage and current protection, slow/auto start, primary side regulation and thermal shutdown.
Swiching frequency is 132khz. And frequency jittering reduces EMI.
After the transformer you can preset unregulated voltage (200V,300V,400V) to reduce power dissipation of the linear regulator.

HVSMPS 1a.jpg
 
Today made a schematic with through hole parts of a Isolated linear regulated highvoltage SMPS.
The LNK6769E Offline switcher has over/undervoltage and current protection, slow/auto start, primary side regulation and thermal shutdown.
Swiching frequency is 132khz. And frequency jittering reduces EMI.
After the transformer you can preset unregulated voltage (200V,300V,400V) to reduce power dissipation of the linear regulator.

View attachment 1078967
And how could we increase to 2 amperes? Run 10 of them in parallel? LOL
 
Your right, at the output i forgot an electrolytic capacitor.

C5, L2, C6 were not so expensive or do mean something else?

Tomorrow i get all the parts and i will make a prototype for testing before i design the PCB.

At the moment estmated cost are 45 euro.
 
Last edited: