Switch Mode Powersupply for tube preamp, DIY.

[ELS]

I want to modify an SMPS I ripped out of some parts TV iirc, the transformer has many secondaries that are wired in parallel so I could just wire them in series to hopefully get at least around 250V, if not a doubler probably would work too.
But I can't get it working, does it have to have the voltage sense feedback for it to work? the feedback circuitry is intact it's just the voltage sense feedback with the opto that's not, it's an STRF6656 based SMPS. STRF6656 pdf, STRF6656 Description, STRF6656 Datasheet, STRF6656 view ::: ALLDATASHEET :::
Why not a linear supply: because I feel like shielding an SMPS would be easier without having a bunch of mu-metal to shield an EI core trafo.

 

[ELS]

It seems that the switcher is trying to start by looking at the voltage on pin 4, but there isn't any AC on the secondary, maybe the resonance is off, the capacitor across the primary was missing so I tried 47pF, 100pF, 270pF, 1nF, 10nF and none worked.I doubt it would be above 10nF, and it's probably not super picky for it to just work somewhat so the problem is probably somewhere else.

 

[Lampie519]

The problem with most consumer quality smps units is that these work close to audible frequencies. This can cause artifacs. Better to use something that uses frequencies above 200kHz.

 

[ELS]

The problem with most consumer quality smps units is that these work close to audible frequencies. This can cause artifacs. Better to use something that uses frequencies above 200kHz.

Well the IC I have lists a frequency of around 100khz, I plan to use a buffered low pass filter in the power rail anyway so that would eliminate any noise on the supply rail itself, all that would be left is EMI but I feel like that would be easy to shield with aluminum or plain steel.

I tried to build a preamp before with an EI core power transformer and even after putting 3 5mil steel plates between the power transformer and OPT it still was unacceptably noisy, ended up putting the OPT outside and far away.
since then I've learned about dampening the secondary of the power transformer, since I was using slow silicon diodes it probably rang like hell, but I doubt fixing that would've made that big of a difference in the noise.

I'm thinking of ways I could make a hot chassis setup, transformerless. But I can't think of a layout that wouldn't be noisy and wouldn't be too big a death trap at the same time. The next best thing just seemed to be a switcher.

 

[Mr. Moscode]

Make an analogue supply in another chassis and keep it away from signal wires. No switching spikes or rf noise - which you don't want on any filaments next to cathodes. Mo' better easier to go this way since it is very low power and you don't contaminate the immediate area, either. Easy to regulate DC filaments, too.

 

[ELS]

Make an analogue supply in another chassis and keep it away from signal wires. No switching spikes or rf noise - which you don't want on any filaments next to cathodes. Mo' better easier to go this way since it is very low power and you don't contaminate the immediate area, either. Easy to regulate DC filaments, too.

I want to use AC filaments anywhere I can, for increased tube reliability.
Making an external supply is too easy , isn't as practical too, the idea of taking the preamp and being able to plug it in anywhere and plug in headphones for example is very appealing. 2 parts would clutter things up.

Are L core transformers any better than EI core ones for EMI? I have a few that I could rewind to use them. But I feel like it wouldn't be that better. unfortunetly all the C-core ones I have are way too small to handle the current.

 

[Mr. Moscode]

AC on filaments has slightly more of an impact on reliability because it causes filaments to vibrate from the AC magnetic field.
How big is this preamp? How many tubes? What's it for? Drive an amp? Headphones only? Post a schematic please.

 

[Lampie519]

the IC I have lists a frequency of around 100khz

There are other IC's or you build your own RF oscillator. Even my heaters are fed by 250kHz AC in this case (center tapped and for direct heated tubes as well).

 

[ELS]

There are other IC's or you build your own RF oscillator. Even my heaters are fed by 250kHz AC in this case (center tapped and for direct heated tubes as well).

And you don't have weird intermodulation distortion? well I guess it is far enough above the audio frequencies to hear it but still...

AC on filaments has slightly more of an impact on reliability because it causes filaments to vibrate from the AC magnetic field.
How big is this preamp? How many tubes? What's it for? Drive an amp? Headphones only? Post a schematic please.

Well I'm basing that thought off of "Getting The Most Out Of Vacuum Tubes - Robert B. Tomer 1960", He writes that a DC filament will slightly shift from the cold startup surge, but only in one direction as opposed to an AC system, which will strain the heater over time and have a higher chance of it breaking.
If you eliminate or reduce the surge current on a cold start then yeah it shouldn't really matter.
Now thinking about it, idk how I planned to get a line frequency AC heater system in the preamp with an SMPS

Current schematic is as such:


I haven't yet finished the schematic, I plan to tune the values when I have it somewhat put together.
I want to use it going into a mixer input, but it should work fairly well for headphones since it has a very low output impedance, around 7 ohms
Tube compliment is 6AK6, 6BL8, nearly 12AU7 *2, 6AL5, nearly 12AX7, and ECC99.
Heater current should be around 3A @ 6.3V, idk the B+ current but probably around 20mA @ ~320V, so total of around 25VA.

 

[Lampie519]

Dc heating is not responsible for high inrush currents. The voltage applied is. So if you use a soft start circuit it is ok to use DC on your heaters.
On your schematic i don’t understand your white cathode follower as it is not connected to the cathode at all. So it will not perform wel, only if you want a high z output to the transformer but then a white cathode follower is overkill.
Any ac above 250khz i found is not audible in the audio range. But that is my experience.
I am an audiophile and nothing on the market is good enough for me so i have to build all mself from D/A converter to amps and power supplies (especially those). I love vintage recording stuff but that is for recording only not for playback.

 

[ELS]

Dc heating is not responsible for high inrush currents. The voltage applied is. So if you use a soft start circuit it is ok to use DC on your heaters.
On your schematic i don’t understand your white cathode follower as it is not connected to the cathode at all. So it will not perform wel, only if you want a high z output to the transformer but then a white cathode follower is overkill.
Any ac above 250khz i found is not audible in the audio range. But that is my experience.
I am an audiophile and nothing on the market is good enough for me so i have to build all mself from D/A converter to amps and power supplies (especially those). I love vintage recording stuff but that is for recording only not for playback.

It's not the DC that's responsible for that, it's just that with DC the surge current will be only in one direction, while with AC it will be in both.

 

[Mr. Moscode]

The 10K R in the white cathode follower will restrict the current drive to the output and I think it may have problems driving headphones.
The WCF was designed for underwater repeaters running shielded cable to compensate for load capacitance. I used them in my hybrid mosfet amp for a while but took them out in later versions and in upgrades. Removal opened up the sound.
Parallel the 2 triodes if you want a nice clean output. Run them at about 1/2 their power rating.
You can use current limiting in the dc heater ckt which will limit inrush. I still think this is too small of an amp to need a smps, unless you plan to carry it around in your pocket.

 

[ELS]

The 10K R in the white cathode follower will restrict the current drive to the output and I think it may have problems driving headphones.
The WCF was designed for underwater repeaters running shielded cable to compensate for load capacitance. I used them in my hybrid mosfet amp for a while but took them out in later versions and in upgrades. Removal opened up the sound.
Parallel the 2 triodes if you want a nice clean output. Run them at about 1/2 their power rating.
You can use current limiting in the dc heater ckt which will limit inrush. I still think this is too small of an amp to need a smps, unless you plan to carry it around in your pocket.

The 10k plate resistor is mandatory for it to work, otherwise the bottom stage just acts a a current sink.
And it will only limit the maximum voltage swing at a given load impedance, so at higher volume levels it will have less bass response. and 7 ohm output impedance with a 4:1 OPT ratio would be enough to drive headphones, it's not a primary concern anyway.

Well I don't want to have a mile long chassis for a simple preamp, so I need a power supply that I can shield.

I found an article with schematics for a modified ATX SMPS of interest:

Unfortunately I didn't find the "see text" text.
I'm curious as to how the ferite transformer was modified as I haven't succesfully ever split the ferite core apart in SMPS transformers, it seems it's glued together
An indian video said that you have to put it under a 60w tungsten lamp for 15 minutes, so I'll try that!

 

[Mr. Moscode]

25 Watts is a small power transformer. You can get something like what you need from guitar amp parts suppliers.

I know how the WCF works which is why I stopped using them. You don't need frequency compensation for short runs. If the load has hi C then there will be HF rolloff. You might be better off with a totem pole P-P output stage. Just my thoughts.

However, it's your project and I hope you get the results you're looking for.

 

[ELS]

25 Watts is a small power transformer. You can get something like what you need from guitar amp parts suppliers.

I know how the WCF works which is why I stopped using them. You don't need frequency compensation for short runs. If the load has hi C then there will be HF rolloff. You might be better off with a totem pole P-P output stage. Just my thoughts.

However, it's your project and I hope you get the results you're looking for.

Well I'm still learning in preamp design, I mainly build guitar amps. this circuit is of the LA-2A compressor, with a different bulb driver circuit, and a vari-mu compressor of my own design.
If you're interested the controls at the top from left to right are as follows: Treshold, Ratio, Release, Attack.

 

[Lampie519]

Your output is not 7 Ohm as you claim, it is at least 1kOhm as the transformer is attached to the wrong side of the 1kOhm resistor unless you want it to be 1kOhm. But you state that if is 7Ohm so this is a contradiction.

 

[ELS]

Your output is not 7 Ohm as you claim, it is at least 1kOhm as the transformer is attached to the wrong side of the 1kOhm resistor unless you want it to be 1kOhm. But you state that if is 7Ohm so this is a contradiction.

That's not quite how cathode followers work, as well as this is not a purely cathode follower circuit, and the OPT has an impedance ratio of 16:1

 

[Lampie519]

Then explain how this one works as it is not making any sense this way. Do you have some reference of this circuit or calculations that explains this?

I see hat your limiter gain stage has no differential input signal therefore the gain here is very low. Better to ground one input instead.

 

[ELS]

Then explain how this one works as it is not making any sense this way. Do you have some reference of this circuit or calculations that explains this?

I see hat your limiter gain stage has no differential input signal therefore the gain here is very low. Better to ground one input instead.

Well a cathode follower has a lot of negative feedback, that's why the high anode impedance of the tube doesn't ruin the output impedance.
The way the cathode follower is hooked up with the 1k cathode resistor doesn't mean that the output impedance will be 1k + the output impedance of the CF, since there's still negative feedback as the grid leak is connected after the 1k cathode resistor.
And also the bottom stage is also driving the output so output impedance will probably be like Ra || Z(CF).
Ra of an ECC99 is around 2.3k. the Gm is 9.5mA/V, so like 100 ohm output impedance.
There probably is some funky things that go on when the output is loaded since the anode gain will increase and so the lower stage will take over.

I doubt I'll have problems with the 1st compressor stage not having enough gain as the supressor grid of an 6AK6 needs just a few volts to go into cutoff. if it is a problem I can swap it for a higher gain tube.
And I need to connect the second input to the cathode for it to work with a carbon mic input. I want to see how a carbon mic would sound going into a compressor.

 

[ELS]

Ok so I'm not reverse engineering an SMPS and the winding counts are confusing me,
the voltage/turns ratio isn't constant, it increases with the increase in turns. I'm taking the DC voltage after being rectified so maybe that's just not the best way to calculate this, but maybe there's some increase in volts/turns the more turns you put on the core?
the numbers I have are these:
1.25V/T @ 4T
1.25V/T @ 6T
1.31V/T @ 11T
1.35V/T @ 13T
1.43V/T @ 17T
1.43V/T @ 21T

Maybe it's to do with the number of "strands" the windings have, since most of the windings are stacked on top of eachother (in series), so the 4 turn winding is like 5 wound in parallel.