DC heaters and grounding

[EFK]

I don't often employ DC heaters, although it certainly can be useful with 6SJ7 tubes. On the occasions that I have implemented it, I usually keep the power tubes on AC and lift the CT, then run the AC through a 25A bridge rectifier w/ a 15000 uf cap across the DC and a .1 uf poly cap to ground off the negative DC side.

I have always read that you *can not* ground both the DC and the AC, and usually lifting the CT on the AC side is the quietest so that's what I stick with.

Attached is a portion of a MArshall DSL schematic. In the lower right corner, you see the 6.3 AC filament winding and it is grounded through two 100 ohm resistors. Then in the upper left corner they run the preamp off DC after using a bridge rectifier, DC filtered and then grounded by spanning the DC w/ two electrolytics in series and grounded at the center.

So since I assume this works, how is it that they can ground both AC and DC? Would one *have* to implement it exactly like this for it to work? i.e., could a CT on the filament winding work instead of two resistors, or is that 100 ohm resistance per side necessary for some reason? Likewise, does the DC need to be "centertapped" through two big electrolytics for it to also go to ground, or could this system work with the common poly cap off the negative side only? Or could two poly caps be used across the DC, assuming ample filtering upstream, and that junction be grounded?

Just curious why this can work.

[Steve Conner]

Would one *have* to implement it exactly like this for it to work? i.e., could a CT on the filament winding work instead of two resistors, or is that 100 ohm resistance per side necessary for some reason? Likewise, does the DC need to be "centertapped" through two big electrolytics for it to also go to ground

Yes. The secret to understanding it is that if the DC output of the rectifier is allowed to be symmetrical about ground, then the AC input can/will/must be too.

[EFK]

Thanks for responding Steve! Ok, so how important are the values utilized? i.e., I would assume you would want the resistors to be pretty closely matched (as you would anytime the artificial CT is used, I guess) - how about the caps off the DC legs? I'm guessing Marshall chose the high value electrolytics to increase the filtering, being as they are using the same values earlier in the chain. Would two non-polarized plastic caps .1 to 1 uf work just as well? Or do the electrolytics being polarized play a role here?

[Steve Conner]

If C1 and C3 are large electrolytics, they can perform both functions: smoothing the DC in the differential mode, and keeping it tied to ground (bypassing it) in the common mode.

I don't see why C2 and C4 are needed: unless the designer decided that he needed more capacitance. (C1 and C3 being in series, their total capacitance is halved from a ripple smoothing perspective, so capacitors placed in the C2/C4 position offer twice the bang per buck)

If you wanted, I guess you could have the bulk capacitance in C2/C4 and replace C1 and C3 with smaller units. The common mode bypass requirement is a lot easier than the ripple smoothing: it's the same as the little capacitor you'll find bypassing an elevated heater line.

[guitarmike2107]

Just latching on here if you dont mind.

So if the 100ohm resistors on the AC side were connected to an elevated DC voltage for hum purposes and not ground i.e. the cathode in a cathode biased amp, then the junction of the two caps after the rectifier would have to be connected to the same elevated DC point?

[guitarmike2107]

Also they haven't dropped any voltage to the heaters after rectification, so those heaters will be running a little hot then, unless I am missunderstaning something else?

[EFK]

Gotcha - so if you used (like I have in the past) a single big 15000 uf cap across the DC for filtering, you could just use little 10uf-50uf caps to ground as your filtering requirements would already most likely be met.

So if the 100ohm resistors on the AC side were connected to an elevated DC voltage for hum purposes and not ground i.e. the cathode in a cathode biased amp, then the junction of the two caps after the rectifier would have to be connected to the same elevated DC point?

That's a very good question! I would be very interested to know the answer to that as well!

[loudthud]

Also they haven't dropped any voltage to the heaters after rectification, so those heaters will be running a little hot then, unless I am missunderstaning something else?

There are always losses when you make DC from AC, especially at low voltages. Look at the attached simulation. For a simulation to be accurate, you need to know what the transformer and rectifier losses are going to be as well as the capacitor ESR. I have just guessed at the numbers. Note that the transformer secondary RMS voltage has dropped to 6.219V because of distortion and the DC at the output is 6.460V. Both within 5% of 6.3V.

Note: Simulation run at 60Hz. The DC output at 50Hz will be lower. To get accurate RMS numbers you must set the length of simulation to an integral numbers of cycles.

[Steve Conner]

So if the 100ohm resistors on the AC side were connected to an elevated DC voltage for hum purposes and not ground i.e. the cathode in a cathode biased amp, then the junction of the two caps after the rectifier would have to be connected to the same elevated DC point?

As far as I can see, yes. And then both the AC and DC heaters would benefit from the elevation.

[GainFreak]

I elevated my heaters after I found out that some tubes will not stand the pressure in CF position. I have an amp with 12VDC elevated heaters and separate AC winding for power tubes. I tied the negative side of the 12VDC to the elevation point and it's OK but I was thinking if it's going to be better to connect all pins #9 together and connect the elevation point there?

[Merlinb]

I was thinking if it's going to be better to connect all pins #9 together and connect the elevation point there?

If it's clean DC then it will not make any difference where you connect the elevation voltage.

If there's some problematic ripple on the DC then I would recommend you use a humdinger and elevate that, rather that bother with pin 9. (If you still want to try the pin-9 idea then only connect one pin-9 to the elevation voltage, don't connect them all together)

[GainFreak]

If it's clean DC then it will not make any difference where you connect the elevation voltage.

All preamp tubes heaters are running on regulated DC (schottky bridge - 4700uF - 7812 - 4700uF) and I guess it's clean because preamp is dead silent.
Thanks for the one pin-9 idea.