DC filaments (ECC803) on a dedicated winding

[R.G.]

Great. Thank you for that. So then i would add... what? 2x 100 ohm resistor to ground after the last filaments? Do I need to worry of any type of precision on those resistors, or even the value? The windings are center tapped, so I assume I would just tape them off? AND finally, if I get any custom would transformers in the future, shall I just eleminate the center taps in the specs? In as far as the filament supply anyway.

If you're going for the "artificial center tap", yes, that works. Two 100R resistors in series, across the whole 6.3V winding, with the center connection between the two resistors tied to ground. The resistor value ought to be low, but not so low as to generate a lot of heat. 100R seems to be a good place to start. You don't need much power rating. The resistors dissipate 6.3V squared divided by the resistance, so for 100's, that's about 100mW each. I would use something like 1W just because I'm paranoid.

The physical location of the resistors inside the amp doesn't matter too much. Near the PT, at the end of the filament string, in the middle all work OK. You'd probably want them near each other so temperature differences didn't make them vary, but that is a really, really minor consideration given how little resistors drift with temperature. Wire wound is probably better, but something like metal oxide (MOX) resistors work fine too. As to precision, 5% works pretty well, but 1% resistors are cheap these days, so why not use them.
Yes, just cap off the existing center taps so you don't have them accidentally making contact. Home Depot wire nuts work well for this.
If you're getting custom wound transformers, you're in a whole different league from me. I've never used a custom wound transformer that I didn't personally wind. I just make available trannies work like I want unless I just can't. If you're getting custom wound trannies, do your homework really well on this amp and get your preferred set of conditions down clearly before you order. If you're thinking about custom wound transformers, be really sure before ordering that you've eliminated the other sources of hum in your amp, and then experiment with winding CT versus artificial centertap, Filaments are by far not the only source of hum in an amplifier.

Well, there was a bit of both, but a low hum was the biggest concern. I havent heard the amp for some months now since I tore it down, but I seem to remember it not being 120Hz. If that means anything.

It's something to check when you put it back together. Put a scope on it and see whether it's 60 or 120. They take different fixes.

The first few "prototypes" were sort of layed out similar to a SLO, including the non-twisted heater supply (attached image sn_III below) - with my fourth prototype, I have reverted to twisting, like I always have done before... I was trying something new and this was where I meant I could do a major improvement in the lead dress area (see attached image sn_IV). And admittedly, they could even be twisted better... (solid core 18 guage)

It's always good to be clear about which rabbit you're chasing. Twisted pairs are good in terms of not radiating either E-field or M-field. The E-fields tend to cancel, and the loop area between the two wires is made as small as possible AND tends to cancel from point to point along the string. Not using twisted pair for something like high current heater wiring is something to do only once you have done your experimentation and found that in your particular case, twisting is not needed. On a volume production amp, It may be OK to not twist the heater wires because experiments have found that it doesn't add a lot of hum and it does save assembly labor. I get the sense that assembly labor is not a big consideration to you, so why not use twisted pair. In any case, run the heaters as close to the chassis surface as possible, as that tends to soak up more emitted E-field and M-field.

[Gtr0]

If you're going for the "artificial center tap", yes, that works. Two 100R resistors in series, across the whole 6.3V winding, with the center connection between the two resistors tied to ground. The resistor value ought to be low, but not so low as to generate a lot of heat. 100R seems to be a good place to start. You don't need much power rating. The resistors dissipate 6.3V squared divided by the resistance, so for 100's, that's about 100mW each. I would use something like 1W just because I'm paranoid.

The physical location of the resistors inside the amp doesn't matter too much. Near the PT, at the end of the filament string, in the middle all work OK. You'd probably want them near each other so temperature differences didn't make them vary, but that is a really, really minor consideration given how little resistors drift with temperature. Wire wound is probably better, but something like metal oxide (MOX) resistors work fine too. As to precision, 5% works pretty well, but 1% resistors are cheap these days, so why not use them.
Yes, just cap off the existing center taps so you don't have them accidentally making contact. Home Depot wire nuts work well for this.
If you're getting custom wound transformers, you're in a whole different league from me. I've never used a custom wound transformer that I didn't personally wind. I just make available trannies work like I want unless I just can't. If you're getting custom wound trannies, do your homework really well on this amp and get your preferred set of conditions down clearly before you order. If you're thinking about custom wound transformers, be really sure before ordering that you've eliminated the other sources of hum in your amp, and then experiment with winding CT versus artificial centertap, Filaments are by far not the only source of hum in an amplifier.


It's something to check when you put it back together. Put a scope on it and see whether it's 60 or 120. They take different fixes.


It's always good to be clear about which rabbit you're chasing. Twisted pairs are good in terms of not radiating either E-field or M-field. The E-fields tend to cancel, and the loop area between the two wires is made as small as possible AND tends to cancel from point to point along the string. Not using twisted pair for something like high current heater wiring is something to do only once you have done your experimentation and found that in your particular case, twisting is not needed. On a volume production amp, It may be OK to not twist the heater wires because experiments have found that it doesn't add a lot of hum and it does save assembly labor. I get the sense that assembly labor is not a big consideration to you, so why not use twisted pair. In any case, run the heaters as close to the chassis surface as possible, as that tends to soak up more emitted E-field and M-field.

Well put and I agree. Thanks. I spent a good number of years building Marshall type amps and prefer that method of layout etc. I did try the opposite ends with the tranny thing for the balance and well, to try it out. One should never base these types of decisions on looks are better carrying balance. Well, I tried. If nothing else you learn the best way to do it. ;-)

I would use something like 1W just because I'm paranoid.

Me too, so I tend to use (usually) minimum 1 watt resistors across the whole thing. Now using 5 watt WW in the power filtering (used to use 2 watt MO) as well.

I just make available trannies work like I want unless I just can't. If you're getting custom wound trannies, do your homework really well on this amp and get your preferred set of conditions down clearly before you order. If you're thinking about custom wound transformers, be really sure before ordering that you've eliminated the other sources of hum in your amp, and then experiment with winding CT versus artificial centertap, Filaments are by far not the only source of hum in an amplifier.

Initially I wanted a SLO type power with taps for tube rectifiers... so I ordered that, and it turned out nice, but way too much juice on B+ so I ordered a second set with lower voltage on the secondaries (was having trouble with screens on EL34's) After lots of research I am trying those button based russian 5881's this time and with a slightly lowered secondary (*I am shooting for under 500 this time, first tranny was putting out 515 to screens after filtering/choke)... though I very much like EL34's... just what I have used most of my 30 years of playing.

In terms of researching trannys... I can't say other than the resulting specs (more so now than a few years ago) I am looking for. Of course it is a topic that piques my interest, it's in a long list of things to study on.

[Gtr0]

Oh yes. And one other thing now.... I am going back to basics... and start over on a single channel version first... get that down pat before I go attempting to add multiple channels. I fully understand everything your saying about production vs homebrew amps.... If I can get that single channel high gain amp quiet enough (which is really all I need for my purposes anyway, otherwise I have just about every Marshall) then I can jump into multi channel. I started this way, but I was never quite satisfied with the noise level, BUT I always went with that non-twisted heater layout... so this time I will change that up.

Thanks again.

[trobbins]

Nowadays there is also the option of using a small cheap dc/dc converter for powering preamp heaters from an unregulated dc supply.

The convert could be a step up type, as per from raw 7Vdc with substantial ripple to 12.6Vdc regulated. Or it could be a step down to 6.3Vdc regulated. The cheap ebay modules are pretty nice and simple to deploy. They are non-isolated, so easiest to use a separate 6Vac winding to get a raw DCV to then go through the dc/dc, so that the regulated heater can be ground referenced in some way (eg. neg 6.3V to ground, as there is no AC voltage to worry about leaking through the cathode heater interface resistance, and hence unlikely to be any benefit from using a humdinger pot or elevated DC to suppress residual hum). They typically come with quite a high output current rating, which is good as it allows the module to handle high initial heater current without too many hiccups (where the output current exceeds the regulation controllers limit and it folds-back or re-starts) - hiccuping is not really an issue, as this would just occur during heater warm-up. One advantage of using such a module is that the raw DC input may be able to have substantial ripple voltage (which means you don't have to use huge levels of filter capacitance, and hence the heater winding doesn't have to supply humungous current pulses which can radiate noise out to the other power transformer windings). Another advantage is that the module typically comes with an output voltage pot, making it very easy to dial in a non-standard heater voltage (eg. if you like the idea of running at -10%).

[trobbins]

Just to add a bit on why humdinger pot can sometimes be a benefit - the input grid is a high impedance circuit and can easily pick up hum through capacitance coupling from each heater wire. Some valves have special metal shielding to lower that coupling capacitance within the tube itself - eg. if you look at the bottom section of an EF86. Outside the valve, it is up to the constructor to not unknowingly lay the grid nearer to one heater wire than the other - that is aided by keeping the grid wiring at as great a distance from heater wiring/pins as practical, but also by twisting the heater wires. Even so, there may be an imbalance between capacitance levels, and so the humdinger pot can custom tune a null in the hum level measured on the output signal at idle (eg. if the hum originated at the input stage and was amplified all the way through to the speaker output).

One advantage I get from quickly hooking up my soundcard/probe to an amp output is that spectrum analyser software can easily observe those hum frequency component levels as the humdinger pot is 'tuned' - just an alternative to putting your ear next to a sensitive speaker with the volume max'ed.

[R.G.]

There is one more step in the humdinger evolution. I'd have to go looking for where I saw this, but it was clever.

There may be some level of hum you just can't get rid of by reasonable means. The humdinger addresses this by attempting to cancel it out. Good as far as it goes. But an adjustable cancelling voltage can't get them all, largely because it's a single frequency and a single phase. Hum may be at various phases depending on how it gets into the signal path, and at harmonics of the power line as well as the base frequency.

The circuit was presented as a "filter" but what it really did was to generate power line frequency, and the second, third, and fourth harmonic as a signal, allow the phase and volume of each of these to be adjusted, and mix them into a cancellation signal you could send into a spare input. A few seconds of pot twiddling and hum gets dramatically lower.

This will even cancel a lot of situational hum. Some locations radiate hum in ways that amps can't reject well. But you can tune them out in most cases.

[Leo_Gnardo]

I'd have to go looking for where I saw this, but it was clever.

I vaguely recall seeing this in some magazine back in the 70's. Might have been Popular Electronics. Other candidates, maybe Guitar Player or Recording Engineer/Producer.

[Gtr0]

So is a humdinger set and forget? Or should access be made??

[Chuck H]

So is a humdinger set and forget? Or should access be made??

Well in the case of Fender amps it was sort of in between. It wasn't a front panel control and it didn't even have a knob. It was on the back and had a recessed flathead slot. So you didn't have to remove the chassis for access, but it was designed so you had to use a tool to adjust it rather than just twiddle it at will. The idea, I'm sure, being that a user might be changing some tubes on occasions without the need to remove the chassis from the cabinet. CTS is making pots similar to the originals for this application. I believe they also make one in an appropriate value for bias adjustment. Some guys like to use those locking collet chuck pots for either of these circuits, but they're more expensive.

[tubeswell]

Don’t know if this has been suggested already (as I haven’t read all the replies above), but you could add a pair of anti-parallel 6A diodes in series with one side of the heater winding before you rectify it, which could eat up 0.7vac before you rectify the 6.3Vac. (5.6vac x 1.4142) - 1.4 = 6.5VDC, which is close enough as a possible easy work around.