Filament Power, for larger builds

[systemtruck]

Hello,
I’m taking steps towards a build that will power up around 20 or so tubes in one unit.
It’s all 12.6V heaters.
I have a toroidal “tube transformer” that offers 4A on its 12.6 VCT secondary and I am considering using 3.6A of that 4A.
The HT secondary of same transformer offers 550mA, but I’ll only ever use around 150-200mA of that.

Would be it smart to get a separate 12.6V supply? Maybe something like a Hammond that provides 10A?

I’m sure I can “get function” with using the onboard 4A as mentioned, but will that constant near capacity current draw from one secondary be a real slog on the other secondary that’s providing the HT?

And will it also reduce the life of the transformer?

I might want to run things as cool as possible. So maybe a 10A filament supply is the better choice.

 

[Rusan]

If it's designed and rated for 4A, then a 3.6A load is not only perfectly fine, but actually desirable. Secondaries are designed to produce their advertised voltage at full, manufacturer-rated load. AAMOF, in some cases a severely underloaded filament winding can produce an excessively high voltage that's out of spec for the tube. This could be possible with a 10A transformer. The only time you'll run into problems is when you inadvertently exceed the published ratings by a certain margin. One such possible scenario would be failing to account for the actual RMS current draw on the high voltage secondary, when huge filter cap values are used right after the rectifiers, in a Class AB amp that has higher current draw at full power than you might think.

Ignoring its current rating, a transformer's heat rise is also very dependent on what the core material is. A given transformer stacked with M19 laminations will run as hot when moderately loaded as an otherwise identical trafo using M6 lams that's presented with a heavy load, due to M19's greater hysteresis losses.

All other things being equal, your existing transformer will be fine, especially since toroidals generally tend to run cooler anyway. And, because you're significantly underloading the high voltage secondary, that'll be even more magnetic and thermal headroom to spare.

 

[systemtruck]

If it's designed and rated for 4A, then a 3.6A load is not only perfectly fine, but desirable. Filament windings are designed to produce their advertised voltage at full rated load. In some cases, an severely underloaded winding can produce an excessively high voltage that's out of spec for the filament.

The fact you're underloading the high voltage secondary just gives it that much more magnetic and thermal headroom.

Well that’s pretty cool!
I suppose the opposite is also true.. that if I go over the limit a bit, it’s really not a major issue especially if I’m not using all the HT?
In one alternate scenario, I’ll draw 4.2A of the spec’d 4A. In a third scenario, I’ll draw 4.8A. And that will be the max scenario.
I just worry about affecting performance of the high voltage in some manner.

 

[emrr]

Likewise, if the B+ winding can deliver 2-3x what you will use, B+ will be high and need big dropping resistances or regulation.

 

[ruffrecords]

For 12V heaters in mixers I use a SMPS these days. I spent a lot of time trying different types. You need one that is rated at twice the nominal heater current draw in order for it to be able to cope with the inrush current of cold heaters. In addition you need one that has a hiccup mode of current limiting as this will ensure enough current is available to successively warm the heaters so their resistance continues to drop. Bottom line is I have now standardised on MeanWell SMPS which are readily available everywhere and have a very good reputation. Your 3.6A at 12 V is 43.5 watts so you could use the MeanWell LRS-100-12 model. DC heaters will make your mixer a lot quieter.

Cheers

Ian

 

[gyraf]

If you're running so many of the same type tubes, you'd probably be better off by running them partially in series, like a string of four from a 50V supply. This to reduce the high currents that would otherwise occur

/Jakob E.

 

[systemtruck]

View attachment 141550If you're running so many of the same type tubes, you'd probably be better off by running them partially in series, like a string of four from a 50V supply. This to reduce the high currents that would otherwise occur

/Jakob E.

How does the current math work for that? If each tube in standard setup draws 0.15A on 12.6V, when a string of 4 from 50V powers them so that each gets 12.5V, what current is required? Something smaller than 0.6A I presume? And what if a 0.3A is thrown into a string if 0.15A tubes?

For 12V heaters in mixers I use a SMPS these days. I spent a lot of time trying different types. You need one that is rated at twice the nominal heater current draw in order for it to be able to cope with the inrush current of cold heaters. In addition you need one that has a hiccup mode of current limiting as this will ensure enough current is available to successively warm the heaters so their resistance continues to drop. Bottom line is I have now standardised on MeanWell SMPS which are readily available everywhere and have a very good reputation. Your 3.6A at 12 V is 43.5 watts so you could use the MeanWell LRS-100-12 model. DC heaters will make your mixer a lot quieter.

Cheers

Ian

So if this supply is used, it’s just a straight parallel connection directly from this DC output? The convenience of it being regulated is nice, so that there’s no concern about voltage running high from being under loaded as just discussed.

I have been meaning to look into linear DC supplies too. I worry about switching noise. I’ve used a lot of meanwell supplies for different things, but never in the context of tubes or in the context of so much gain. But I suppose it’s easier to tame than a 60Hz frequency that cannot be “filtered” so to speak.

 

[systemtruck]

Cruising through some data sheets about heaters, this one took me by surprise even though I’ve looked at the rest of it several dozens of times already..
https://frank.pocnet.net/sheets/127/1/12AY7.pdf

“AC or DC
For 12.6V operation apply heater voltage between pins #4 and #5. For 6.3V operation apply heater voltage between pin #9 and pins #4 and #5 connected together.
When operating from an AC heater supply, do not use 12.6V connection if low hum capabilities are to be realized.”

Wow, what?? I had always heard that 12.6V offered quieter performance than 6.3V.
In any case, this statement certainly makes DC sound more appealing.
There are rumors of DC powered tubes not sounding the same as AC though. Nothing empirical I’ve seen.
Also, some datasheets have different graphs for 6.3V performance and 12.6V performance. I don’t have enough understanding to assess what that’s about though.

 

[emrr]

Humbucking, or similar. You almost never see 6.3/12.6 types run on 12.6VAC in pro audio gear.

 

[gyraf]

..five 50V/150mA strings (of four tubes) = 50V/750mA for your 20 tubes total..

perhaps get a 48V/1A standard SMPS supply, trim its voltage up a bit, and call it a day..

 

[systemtruck]

In terms of physical layout… It’s an 8 channel mixer, with 2 busses.
Each preamp channel has 2 tubes. Each master channel has 3 tubes. So 22 tubes total. Everything is 12.6V.
I’m considering making each channel as a module that can be hinged upwards one by one for service / mods. Maybe I could get a 25V SMPS, and run 10 25V lines from it so it’s star patterned from the back of the unit? If I snake this DC all around across the channels, that prohibits this build approach.
I could also make it two large halves of the “faceplate” and be able to lift one side or the other. Regardless, hinging would be helpful going forward, so I want to take some time to carefully choose wire and ties that lets this happen without risk over time.
Or maybe that’s just too nuts and this build needs to never be hinged.

 

[rock soderstrom]

Everything is 12.6V.

I would use a 12V SMPS and give each tube its own dedicated filament voltage.

 

[soapfoot]

How does the ready-made SMPS compare in terms of reliability?

 

[emrr]

How does the ready-made SMPS compare in terms of reliability?

Perhaps more easily replaced, and more likely to need to be replaced.

 

[soapfoot]

Perhaps more easily replaced, and more likely to need to be replaced.

Certainly a higher component count!

My inexpert concern would be availability of a direct drop-in replacement ten years hence should the need arise

Not that we couldn’t make (mechanical or other) modifications to substitute a different unit—the same is also sometimes required when changing transformers

But a large percentage of half-century-old gear still has a working original transformer, and I suppose only time will tell whether off-the-shelf SMPS units hold up as well.

My hunch would be to suspect they’re well-engineered with a service life target in mind (but I somehow doubt that target is “75 years!”)

 

[systemtruck]

Since they’re so cheap in the ballpark of 15 USD, maybe buy two or even three if it really is perfect for the function? Not sure what would burn out sitting still after 10 years or so unused. I guess the risk would be capacitors drying up.

 

[Cqwet Dbdfte]

A series connection of tubes is not so good for trouble shooting, if a filament breaks.
A separate heater power supply is preferred over mixing in transfomers as some leakage between HV and others is common.
DC heat avoids AC magnetic fields that may couple into sensitive circuits.

 

[systemtruck]

So it seems that, although less efficient than is possible, @ruffrecords ‘s MeanWell LRS-100-12 supply at 8.5A DC 12V could be connected to all tubes in typical 12V fashion, and this is a simple and solid and affordable solution.
In my situation i would choose to split the output from supply into 10 different pathways, each one feeding 2 or 3 tubes, and it would be a star pattern overall.
Do DC heater wires get arranged the same as AC? Twisted tightly and given perpendicular and/or distant entry approaches VS the audio lines to said tubes?

 

[ccaudle]

How does the current math work for that?

Series devices only have one path for current to flow, so all devices in series have the same current. There will be slight voltage variations between the individual devices depending on how closely the resistance is matched.

If each tube in standard setup draws 0.15A on 12.6V, when a string of 4 from 50V powers them so that each gets 12.5V, what current is required?

The entire string will have approximately 0.15A flowing through, and each individual heater will have approximately 12.5V, with the individual voltages varying just slightly depending on the exact resistance. The exact current flowing through the string will be the power supply voltage divided by the total series resistance (which will vary just a little from the nominal value due to device tolerance).
Probably the common case would be to use a 48V power supply, and then each heater will have approximately 12V, which is just 5% below nominal, that should be within device spec (e.g. typical 12AX7 datasheet specifies heater voltage is 12.6V +/- 10%).

As pointed out above some people prefer parallel connection (solves the Christmas lights problem: one goes bad the whole string turns off) which will be the same power, just higher current, so you will need slightly thicker wires than with series connection.

Do DC heater wires get arranged the same as AC? Twisted tightly and given perpendicular and/or distant entry approaches

I think with pure DC it may not matter, but keep in mind that with switching power supplies you will always have some amount of high frequency noise superimposed on the DC, so it would be good practice.

 

[systemtruck]

Series devices only have one path for current to flow, so all devices in series have the same current. There will be slight voltage variations between the individual devices depending on how closely the resistance is matched.



The entire string will have approximately 0.15A flowing through, and each individual heater will have approximately 12.5V, with the individual voltages varying just slightly depending on the exact resistance. The exact current flowing through the string will be the power supply voltage divided by the total series resistance (which will vary just a little from the nominal value due to device tolerance).
Probably the common case would be to use a 48V power supply, and then each heater will have approximately 12V, which is just 5% below nominal, that should be within device spec (e.g. typical 12AX7 datasheet specifies heater voltage is 12.6V +/- 10%).

As pointed out above some people prefer parallel connection (solves the Christmas lights problem: one goes bad the whole string turns off) which will be the same power, just higher current, so you will need slightly thicker wires than with series connection.



I think with pure DC it may not matter, but keep in mind that with switching power supplies you will always have some amount of high frequency noise superimposed on the DC, so it would be good practice.

Naturally that leads me to ask..
When using a DC system, do people tend to add some passive filtering across the output to help reduce what high frequency noise may be there?