DC filament supply - SMPS or linear ?

I am designing two tube units with a filament current requirement of about 3.5A at 6.3V and I am debating which power supply topology to use.

Initially thought a simple linear supply with a 1084 regulator would do the job, the problem is a power transformer with about 6A winding, which would be too big for the 2U rack chassis I am using. The power transformer also needs winding 250V @ 130mA for the HT and 12V @ 1A for relays and LEDs.

A possible solution to overcome the size limitation would be using an EI transformer for ht and aux supplies and a separate toroidal transformer for the filaments,
However, even with the relatively lower hum radiation from the toroid I am still a bit concerned about noise since the input circuit works at microphone level.

I have started considering SMPS for powering the filaments, this is the first time I use them with tube circuits.
First of all I don't know if having the SMPS inside the same chassis with the audio circuits can create noise problems? This would be my main concern.
I guess there shouldn't be a problem with mains frequency hum but what about the switching noise, can it be picked up by the sensitive high impedance circuits and if that's the case could the issue be solved with some steel shielding rather than expensive mu metal ?

Another issue, I need to stick to 6.3V tubes and the choice of SMPS module is a lot more limited than 12V.
I am wondering if a 7.5V module with a resistor on the output would be the best thing to do despite the extra heat generated.
I have read some of the adjustable SMPS can be modified by altering the resistances in the adjustment section, but I am not sure if all modules allow that and if there are potential side effects to consider.

Something like Meanwell RSP-75-7.5 mounted inside the chassis and an EI transformer for the ht and 12V fitted on the back panel should get the job done and cost less than a large all-in-one power transformer, and also compared to the solution with separate 6.3V toroid.
From what I understand SMPS don't need additional filtering for filament supplies so I would also save money on rectifiers, filter capacitors and regulators.

Of course I would be interested in any alternative solutions you might want to suggest.
 
I only use mean well smps from the EPS serie for filament or heater now. I use the version ending with a -c to get the shield cage and you can even position them close to a IT transformer without problems. They offer little bit of adjustment and I use two resistors in serie to drop a little bit voltage. On some tube like my 808 they do the powering up in two steps because of current limiting security inside, so I use a separate switch to turn on the heater first and then the hv ps. I’ve been using that on tubes from the 30s and 40s without any degradation and it’s amazing if you need to get the lowest noise floor possible in an amplifier.

i tried different things including expensive dedicated heater boards for 808 45 845 2a3 and this is the best solution I could find and it’s really cheap, even if you need one per tube in case of DHT.

Don‘t use any additional filter with those at the output or input. Those smps are really well build.
 
Last edited:
  • Like
Reactions: reamp
This is how I use those smps to give you an idea. There is a mistake in the 120v wiring of my switches on the schematic. The first switch should be feeding the smps and from the exit of that switch it goes to my HV switch so you can’t power up the HV without the heaters and filaments.
 

Attachments

  • 202012252217161000.jpg
    202012252217161000.jpg
    579.9 KB · Views: 147
  • Like
Reactions: reamp
On some tube like my 808 they do the powering up in two steps because of current limiting security inside, so I use a separate switch to turn on the heater first and then the hv ps.
To overcome the extra current demand with cold heaters, would it be enough to use 2x the nominal power ?
I am mainly using small signal tubes and no DHT, would it be safe to use just a single switch to turn on both supplies at the same time ?
I am aware of the supposed benefits by switching on the hv only after the heaters have warmed up, but to be honest with linear supplies and small signal tubes I never did and did not get the impression this shortened tubes life in a noticeable way.

Don‘t use any additional filter with those at the output or input.
I see they have an EMI filter built in. No need to use a filtered IEC socket to avoid noise injected back into mains ?

What about protection against spikes or surges, do they need additional components on the input for example MOV ?
 
They have mov inside and and well protected.

I use 45w smps for 25- 30w heater so yes around 1.5 to 2 time the nominal power.

You can use single switch but they take a second or two to power up. Using Twi switches is really cheap and easy to do. All my amplifiers have the 1 -2 switch powering scheme so I'm use to it.
 
If you are lighting a large number of tubes, like that one gigantic Carver model with, what was it, twenty 6550's on it, then the SWMP for the heaters makes a lot of sense. Just be sure that the supply does not spray HF garbage into the audio stream.
 
I connect my amps directly on 118db eff compression drivers so I have to keep the noise as low as possible. On the speaker terminal 1mv is still too much. The small smps are a big help and simplify the circuit a lot and save space, specially with dht heater pulling 25-30watts.

Note that in my full linear à l'ancienne circuit, I mention 17mVRMS accross the 300B tube heater, and NOT the speaker terminals, where I have zero mV ripple checked with bench-lab TRMS DMM and scope.

That said, my most efficient speakers are also rated at 95dB/W/m, and not 118dB/W/m... Just a mere 23dB below 🙄 ! Then I imagine your embarrassment, yes... 😕

Just be sure that the supply does not spray HF garbage into the audio stream.

Yes : that's what I would check too.

T
 
I'd rather deal with noises in the high khz range behind a bandwidth limited output transformer and then bandwidth limited speakers than 60hz, 120hz ... That are in the audible range.

And I understand you meant at the heater, it's probably a 0.1-0.2mv residual at the speaker terminal depending of tube gain and output transformer ratio if SE. I take than any day over AC on a dht heater.

Probably low enough for most speakers anyway.