My first voltage multiplier PSU. 😎 (...for tube mics)

[rock soderstrom]

Just the action of elevating the heater may sufficiently suppress ac hum if the heater was directly ac powered, but if that path was followed then OP would need to confirm that layout and other factors don't cause measurable hum because of the ac heater powering.

No, I have no intentions of heating my microphones with AC. I don't know any microphones that are heated with AC but I don't know all tube microphones either.

 

[rock soderstrom]

Aim would be to full-wave rectify for both the heater and the B+ supplies, rather than half-wave for each. For B+ supply, the winding can work in to a 'positive' capacitor input half-wave, and a 'negative' capacitor input half-wave - the winding then sees full wave rectification, and one leg of the winding sits at B+/2.

I don't have a transformer with center tap, so I wanted to use two bridge rectifiers at the beginning.
Can you please outline your proposal, I think I don't quite follow you, but sounds interesting.

 

[trobbins]

Link to a quick example I had of a not-so-common powering configuration - so not directly aligned with what you could use, but indicates some options. https://www.dalmura.com.au/static/6EB8 self split with LFO.pdf

A 30Vac winding is used to generate a B+ of 175V, a screen supply, along with a negative bias supply. The B+ uses two capacitor input doublers, but in a full-wave configuration such that one end of the 30V winding is at mid-B+ dc. In your situation, a 17Vac winding may be able to achieve 100Vdc pro-rata.

Heaters are ac powered but the heaters are DC elevated, and the input stage heater has a lower ac voltage variation relative to gnd - this was a practical outcome as the 30Vac winding had a 17.5Vac tap.

I can appreciate your dismissal of using ac for heater powering, but just need to indicate that imho with due care for wiring layout, and perhaps even additional socket shielding, and elevated ac (to suppress hum caused by the Rhk resistance within the tube), and a bypassed cathode bias resistor, the level of residual hum in your amplified output signal could be adequately down in the weeds (if you had the measurement capability to discern the hum level). Unless you have your power transformer and rectifier/filter circuitry in a separate enclosure, there is always a need for careful layout and screening of the PT and rectifier circuitry (especially when rectifying for a dc heater supply) and always a need to be able to measure noise down in the weeds to confirm nothing is eeking into your audio signal path.

 

[abbey road d enfer]

- B+ minimum 60V up to 120V, if more than 120 are possible, nice!

In theory, almost any voltage can be obtained from a multiplier, it takes putting teh adequate number of stages. Just remember that capacitors end up being in series, so if you have six stages, the actuial capacitance will be 1/6th of the nominal capacitor value.

Ergo, according to Abbey and Jacob I would use a bridge rectifier for the heater with my existing transformer, protected by a fuse. Then a CRC filter and a textbook LM317 PSU followed by another CRC filter.

You don't really need a CRC filter at the input. Your money is best placed on the output filter.

 

[rock soderstrom]

..just build it and see what happens irl, that tells you more than anything..

+1

Heavy testing on my 80x100mm TPSU_X prototype.

 

[wlinart]

Looks like an interesting way for powering tube mics, following with much interest here

 

[trobbins]

A suggestion for pcb layout is to connect from one filter stage to the next through the decoupling cap pad - for example C2 neg pad should connect to C1 neg pad directly, rather than to the diode bridge. Same goes for the B+ diode and cap connections.

One other way to try and minimise rectification transients is to add extra solder pads on the bottom of the pcb to fit an smt cap across certain large e-caps, like C11, to make an easier path for transients to loop around (rather than go via the larger e-cap).

 

[gyraf]

..is that heatsink an off-the-shelf-component, that you can be sure to be able to source..?

/Jakob E.

 

[rock soderstrom]

..is that heatsink an off-the-shelf-component, that you can be sure to be able to source..?

/Jakob E.

Yes, this is a standard part, I've been using it for years. There are others (bigger and smaller) with the same footprint, so you can adjust the heatsink to the expected power loss.

The LM317 regulator is also extra positioned on the edge of the PCB, which makes case mounting easy.

In the end, it's a non-commercial project, so if these heatsinks would no longer be available, I'll adapt the design to the ones available then.

When I'm done testing, I'll release the design freely for selfetching in its own project thread.

Thanks for input and help!

 

[rock soderstrom]

Okay, here are some intermediate results.

The circuit works pretty good, the multiplier does what I wanted it to do. The heating can be adjusted well, the heat management also works well.

In the attached sketch you can see the current situation. I did not draw the heater, pretty standard.



Case 1 is without additional load, only the LED and the series resistor load the B+ rail. This is about the same as the situation with a microphone as load. (5654, 5840 or 6s6b-v). The plan works, I see no problem.

Since I am still waiting for the mic capsule, I also simulated case 2 for a small two tube mic preamp (RL=5.6K, 16.7mA + 3.4mA LED current) with a total load of 20mA.

The higher current leads to a 20V loss at the RFilter and results in a B+ of 93.5V. That's pretty decent and could be significantly improved with a choke (50H) with low DC resistance instead of the 1K resistor.

I am satisfied with the results.

 

[rock soderstrom]

I am satisfied with the results.

Well, the reality is different then. The use in a real project (tube mic SELA T12 style) quickly showed the limits of my design.

Originally I had planned to filter B+ with CRCRCR after the multiplier. (see first post)


Since I wanted to realize the power supply on a half euroboard, the space was naturally quite limited.


So I thought a CRC filter would be sufficient, if necessary with a choke instead of the resistor. I already have plenty of capacitance in the multiplier, right? Then everything fits on the 80x100mm board. That was the new plan, sounded plausible to me. (as shown in my first PCB design)

But it was not. The result was an insufficient filtering of B+, which was expressed by a nice and loud hum.

What was going on? First, I tapped B+ at the x3 tap, since I only needed 60V, so way less capacitance . Second, Abbey had already mentioned it correctly, the capacitors in the multiplier are in series with all the resulting consequences. Thirdly, therefore, in real terms, more or less only the CRC filter remains, since the SELA T12 design does not have its own filtering.

This is definitely too little.

I added two more RC filters as a test according to the original plan and the hum was gone.

Tears of joy, microphone sounds very good and is quiet enough for my application. I think the tube mostly defines the noise floor now, but that's another story though....

I learned a lot with this project that will go into my new PCB design. Information will follow.

PS: a 60V shock from such a multiplier is amazingly painful! Thanks for the reminder.

 

[rock soderstrom]

What happens to my first two capacitors? As you can see in the photos, they blow their cheeks and start to bring their insides to the outside.

From my perspective, there is no overvoltage situation. Am I doing something wrong or did I just get a bad batch? Maybe something happens during discharging after switching off?


Blue caps are 1000uf/50V

 

[abbey road d enfer]

There's something wrong in your schemo. It is grounded only at the. output. Is the xfmr grounded?
Now it's no wonder those caps bulge, they pass about 1A continuously. There is one diode missing somewhere.

 

[rock soderstrom]

There's something wrong in your schemo. It is grounded only at the. output. Is the xfmr grounded?
Now it's no wonder those caps bulge, they pass about 1A continuously. There is one diode missing somewhere.

No xfmr is not grounded. It is a textbook LM317 PSU plus a floating multiplier HV part as drawn.
This layout shows better whats going on. C10 and C11 are the blue caps. B+ and F+ are grounded together at the output.

 

[abbey road d enfer]

It's very hard to figure out what happens. Do you have a clean schemo, showing the interconnection with the heater part?
Your handwritten schemo cannot work.
Layout shows D2 goes to C1, C2, C6, C7 ,and C8, whch your handwritten schemo doesn't show.

 

[rock soderstrom]

It's very hard to figure out what happens. Do you have a clean schemo, showing the interconnection with the heater part?

No clean schemo, sorry. The only interconection is the jumper (red arrow) betwen the two grounds.

Layout shows D2 goes to C1, C2, C6, C7 ,and C8, whch your handwritten schemo doesn't show.

Okay, let me check this.

Edit: Yes, true. This is the the GND bus for the heater PSU. Whats wrong with it?

 

[Doug_]

Might need to use low ESR types in these two locations at least. A lot of current will be flowing in and out of these while the doubler chain is charging. Perhaps also limit this inrush current with a low ohm resistor.

 

[rock soderstrom]

Might need to use low ESR types in these two locations at least. A lot of current will be flowing in and out of these while the doubler chain is charging. Perhaps also limit this inrush current with a low ohm resistor.

Interesting! What speaks for this thesis is the following. I have operated parallel to the multiplier PSU board with x6 factor another with exactly the same capacitors and the same principle construction in the last 3 months. This one did not show the problem. The only difference is the reduced x3 factor, so less inrush current! (...and the heater is unregulated passive CRCRC)


But maybe Abbey is right and I have somehow done something fundamentally wrong? As the thread title says, this is my first attempt with multiplier PSU circuits!

 

[volker]

This is what a voltage cascade usually looks like. Not a diode missing, but the first cap on the lower leg is redundant. And like Abbey said, the transformer should be grounded on one side not floating, since this is basically a half wave rectifier.

 

[rock soderstrom]

Thank you Volker for your contribution. I integrated the "redudant" capacitor on Jakob's advice in order to implement a DC heater with bridge rectifier at the same time.

Perhaps feed the tripler through capacitors, to float ground reference and make fullwave rectification for heater possible

..just build it and see what happens irl, that tells you more than anything..

/Jakob E.

And like Abbey said, the transformer should be grounded on one side not floating, since this is basically a half wave rectifier.

Wouldn't this lead to a direct short circuit if I were to connect one leg of the transformer secondary winding to ground?