Using Bias Tap for Heater Elevation

[stratomaster]

An idea occurred to me that by duplicating the bias circuit in say an AB763 and simply reversing the diode that one could create a variable elevated center tap for use with the filament heaters.  I'd probably have to play with the resistor value upstream of the diode since there's more hanging off of it, but in theory this would enable the dialing in of DC offset that gives the best hum rejection.  A bias balance type of circuit could even take the place of a humdinger.

Seems like added complexity but it has the benefit of being adjustable, available at power on, and agnostic to any standby switching employed. 

My main concern would be say a heater short affecting bias. 

Any legs to this idea?

[stratomaster]

Of course fixed resistors could be employed if no adjustment is needed. So for the cost of a diode and 1 or 2 100v e-caps hanging yet another thing off the HT line can be avoided, and there's no danger of elevating the heaters too high since they're limited by the tap voltage.

[pdf64]

Good idea!

[PRR]

> enable the dialing in of DC offset that gives the best hum rejection

Cost you a dime to drop main B+ through a resistor to a pot and find out if "dialing" has any advantage.

[Williamblake]

Also if you use your B+ from the preamp it already is very nicely filtered.

[stratomaster]

Also if you use your B+ from the preamp it already is very nicely filtered.

But downstream of a standby switch.  What happens to the center tap voltage reference when on standby?

[PRR]

But downstream of a standby switch.  What happens to the center tap voltage reference when on standby?

This is just for hum reduction? Then who cares if it goes-away in standby?

[pdf64]

I was thinking that too, but then thought when standby gets flipped to operate, the h-k voltage will be higher than intended for a few seconds.

[66Strat]

I think it is a non-issue. Heater and cathode (follower) voltage should both rise and fall proportionally with the DC voltage.

EDIT:

Worst case condition. Cathode voltage is not raised as soon as the heater voltage. Heaters are raised 70 or so volts above cathode. H/K rating is 100 volts DC, per GE data-sheet. I don't see this rating being exceeded.

[stratomaster]

I finally thought this through and am considering adding it to a Super Reverb I recently came to own. 

I think I can fit everything on the stock eyelet board.  I'm using this image for reference as it was the clearest photo I could find for marking up. Basically just move the chassis ground to the other side of the existing cap (100μF @ 100v in my case), add a diode for the positive swing, bypass the cap with a resistor, and add a cap in place of the link for the new bias filter.  If I add a second pole to the bias pot wiper then I can decrease the bias cap to 50μF @100v without adding noise.  The MOD cap will fit in that space without heroics. As will the 8μF MOD cap fit on the bias pot.

Is there any benefit to taking the positive voltage after the 470Ω resistor instead of before it?

[Merlin]

If I add a second pole to the bias pot wiper

Are you talking about the existing output tube grid bias pot? You'll never get a positive voltage from there!
I don't really see the purpose of this. You could just replace the lower resistor in a normal elevation divider with a pot...

[stratomaster]

If I add a second pole to the bias pot wiper

Are you talking about the existing output tube grid bias pot? You'll never get a positive voltage from there!
I don't really see the purpose of this. You could just replace the lower resistor in a normal elevation divider with a pot...

I was inarticulate. I've abandoned the idea of an adjustable heater elevation.  Instead I wanted to use the positive swing of the bias tap to generate the DC voltage for heater reference.  Since quite a few vintage Fenders and their offshoots have the 2x100Ω CT attached to the pilot light, the location seemed ideal and it wouldn't take much to shoehorn the elevation circuit into the existing eyelet board.

The second pole I'm referring to is basically just a cap off the bias pot wiper much like you illustrate on your webpage on bias. It has nothing to do with creating a positive DC voltage. That comment was more about physically fitting a second cap on the existing bias board and decreasing its capacitance to reduce size, but adding the cap to the bias pot wiper to make up for any noise induced by shrinking the first bias cap.

My question is: is there any advantage to taking the feed for the positive supply after the 470Ω resistor?  It is a much cleaner layout if I take it direct from the bias tap, but if there's good reason not to--say current limiting, then I'll live with a diagonal diode.

[Merlin]

Oh I see. Yeah I don't see a problem with that.

[HotBluePlates]

Is there any benefit to taking the positive voltage after the 470Ω resistor instead of before it?

Before = More Voltage
After = Less Voltage

[stratomaster]

Is there any benefit to taking the positive voltage after the 470Ω resistor instead of before it?

Before = More Voltage
After = Less Voltage

I understand that, but I was looking for a bit more so I just modelled it in PSUD2.

It looks like the presence of the 470Ω resistor drops ripple across the 47k resistor by about 40% and drops current by about 60%. Voltage drops by ~ 10%.

Small values of both current and ripple to begin with, but notable increase in performance. Worth messing up a nice rectangular layout? Probably not.

I didn't model the effects of the bias supply that would share the same resistor though as I'm unsure how to model this exact arrangement in that program. I'll have to dust off LTspice for that, I guess.

[HotBluePlates]

Is there any benefit to taking the positive voltage after the 470Ω resistor instead of before it?

Before = More Voltage
After = Less Voltage

I understand that, but I was looking for a bit more so I just modelled it in PSUD2.

It looks like the presence of the 470Ω resistor drops ripple across the 47k resistor by about 40% and drops current by about 60%. Voltage drops by ~ 10%.

Small values of both current and ripple to begin with, but notable increase in performance. ...

I didn't model the effects of the bias supply that would share the same resistor though as I'm unsure how to model this exact arrangement in that program. I'll have to dust off LTspice for that, I guess.

Dude... it's a non-linear AC Voltage Divider.

Don't overthink:  do you want "More DC Volts" after the diode, or do you want "Less DC Volts" after the diode?

[stratomaster]

Follow up. It worked as intended. Had to go radial for the bias cap for fit.  70v elevation.

[stratomaster]

Another update.

Adapted this idea to a model with solid state rectification--in this case a 1970 Twin Reverb.  Have not tested at full voltage yet, but on the light bulb limiter I'm getting just shy of 62V elevation taking the voltage after the 470Ω dropper for fit (and out of curiosity). I'm expecting this to work just as well as it did in the Super Reverb above once done.

[kagliostro]

Mr. Merlin, I am honored that you used an old diagram of mine from 2014 as a basis to present your example 😁


Franco