Super Reverb - Bias Filtering

[410_Fanatic]

Hello!  I am currently working to eliminate a faint 60 cycle hum in my Super Reverb and had some questions about limits when it comes to bias filtering.

I was able to trace the hum to the bias supply by tacking a cap from the bias pot wiper to ground, which eliminated the hum (after reading multiple posts here ).  I had previously rebuilt the bias supply with a new 470 ohm resistor, diode and bias cap, but a slight hum remained.

I currently have 100uf/100v on the bias supply, but in reading other threads it sounds like adding additional capacitance might not be a great idea, considering the recovery time for the bias supply should it ever be taxed in any way.  I also read about concerns regarding start up time, but the current setup seems to supply negative voltage much quicker than the time needed for the rectifier to warm up and for the power tubes to start conducting, so I am less worried about this aspect.

All of this his raised a couple questions for me:

• Can I just add an additional cap from wiper to ground or would this be too much capacitance?  In my test an additional 25uf definitely does the job, but that would be 125uf total capacitance
• Am I better off with two 50uf caps, perhaps two stages of filtering is better than one and that is why it was used in some of the later Fender bias circuits?
• The stock schematic calls for only 25uf, the amp had 68uf when I got it, and I have bumped it to 100uf.  Does still having 60 cycle hum indicate any issues further upstream?  Bad PT?  bad rectifier tube, bad silicone diode?  I ask because the rectifer tube (5ar4) and silicon diode (1N4007) are brand new.

Thanks in advance for any insight.

Best,
Mike

[sluckey]

A two stage filter C-R-C is much better than a single cap. I like 47µF directly on the diode and another 47µF on the pot wiper.

[HotBluePlates]

... read about concerns regarding start up time, but the current setup seems to supply negative voltage much quicker than the time needed for the rectifier to warm up and for the power tubes to start conducting, so I am less worried about this aspect. ...

It seems you've already gotten my input on the raw bias voltage itself, and the charging time required to develop it.  So no need to reiterate here.

... it sounds like adding additional capacitance might not be a great idea, considering the recovery time for the bias supply should it ever be taxed in any way.  ...

Overall, I recommend you not worry about this because it probably isn't hurt by the bias supply capacitance.

     "Bias Recovery" and large capacitance has little to do with the capacitance in the bias supply itself.

     We care about "blocking distortion" or "grid blocking" when the coupling cap is negatively-charged by grid current when the 6L6 is overdriven.

     The R x C involved here is the "C" of the coupling caps and the "R" of the 220kΩ bias feed resistors between the 6L6 grids & the bias supply's output.

     Faster recovery comes by making R smaller, or C smaller, or both.

     But making R smaller reduces the gain of the phase inverter stage.

     And making R and/or C smaller reduces low frequency response, and will increase phase shift inside the feedback loop (probably not an issue in a Super Reverb unless there is a defective output transformer).


Hopefully you now see that bias recovery from overdrive is mostly a function of components that aren't in the bias supply itself.

[410_Fanatic]

Overall, I recommend you not worry about this because it probably isn't hurt by the bias supply capacitance.

     "Bias Recovery" and large capacitance has little to do with the capacitance in the bias supply itself.

     We care about "blocking distortion" or "grid blocking" when the coupling cap is negatively-charged by grid current when the 6L6 is overdriven.

Thank you for the clarification, I clearly misunderstood the point you were making about blocking distortion in a previous post.  I read Merlin's chapter on filtering and got stuck on the "reservoir" analogy and how more capacitance means it takes more to keep that reservoir full, and then fixated on a scenario where not enough negative voltage in the reservoir might lead.

I won't worry too much about this then, because I will not be playing around the with values in the phase inverter.

A two stage filter C-R-C is much better than a single cap. I like 47µF directly on the diode and another 47µF on the pot wiper.

Thank you for sharing this.  After your comment I went and swapped the 100uf for a 50uf and then added an additional 50uf on the wiper.  Worked much better than the single 100uf.

Are 85 degree caps okay here?  It seems the standard is 105, but Mouser seems to have a better selection of 50uf 100v caps with 85 degree ratings.  For testing I am using two of the MOD caps from CE, the leads on them are tiny and don't instill confidence so I am looking to replace them with a better cap if I can find them.

And sorry to add more questions - Is the reason this works better because instead of having a single low pass filter with the 470 ohm resistor and 100uf cap, we now have a second order low pass filter with 470 ohm > 50uf >whatever resistance is between terminal 1 and the wiper on the bias pot > 50uf, and that steepens the frequency rolloff?  Is the 470 ohm resistor before the diode part of the equation here?

[pdf64]

It’s nice to be able to use 105C ecaps when they’re available, because our much lower operating temperatures result in cap life being extended.
But they’re a recent thing. Amps have been running 65C or 85C caps quite happily for decades.

[HotBluePlates]

... Is the reason this works better because instead of having a single low pass filter with the 470 ohm resistor and 100uf cap, we now have a second order low pass filter with 470 ohm > 50uf ... that steepens the frequency rolloff?  Is the 470 ohm resistor before the diode part of the equation here?

The 470Ω before the bias filter (and bias rectifier diode) is part of the equation, because it is part of the charging path for the cap.

The capacitor has reactance, that is like resistance but varies with frequency.  The resistor's resistance and the capacitor's reactance form a voltage divider to reduce the AC part of the voltage present.

Increasing the capacitor's value lowers its reactance, and so reduces the AC part of the voltage even more.  Additionally, increasing the capacitor's value increases its charge-capacity, making it better able to sustain its voltage (and avoid ripple) in the face of a load that draws current.

[410_Fanatic]

It’s nice to be able to use 105C ecaps when they’re available, because our much lower operating temperatures result in cap life being extended.
But they’re a recent thing. Amps have been running 65C or 85C caps quite happily for decades.

Thank you for pointing this out.  I see now that some of the Sprague Atoms are 65C and 85C now, I guess I hadn't been paying much attention and thought 105 was the standard.  I'll use 105C if I can find them, otherwise I'll go with 85C.

The 470Ω before the bias filter (and bias rectifier diode) is part of the equation, because it is part of the charging path for the cap.

The capacitor has reactance, that is like resistance but varies with frequency.  The resistor's resistance and the capacitor's reactance form a voltage divider to reduce the AC part of the voltage present.

Increasing the capacitor's value lowers its reactance, and so reduces the AC part of the voltage even more.  Additionally, increasing the capacitor's value increases its charge-capacity, making it better able to sustain its voltage (and avoid ripple) in the face of a load that draws current.

Okay, I think I understand better, the voltage divider description plus the description of reactance makes more sense to me.  Thanks for bearing with me, I can read a schematic and broadly see what the various parts are, but have never really gotten my hands dirty enough to understand what is really happening at a fundamental level.  I clearly have more reading to do  .

I'll stick to the two 50uf caps since I can't hear the hum from across the room anymore.  The amp in question is a 64 Super that has had a rough life, chances are high that I'll be asking more questions about it in the future.  I appreciate everyone's time and generosity with their hard earned knowledge.

[tdvt]

Following this with interest as while this all makes perfect sense, I had not considered the bias supply as a potential source of hum.

Looking back through older Fender schematics of tweeds & brownface amps, the two-cap bias filter was often used on various models (but sometimes not) 

Seems like an easy, not too expensive detail to mitigate a possible source of hum

[sluckey]

I had not considered the bias supply as a potential source of hum.

Normally it's not a source of hum just due to the common mode rejection characterics of the push/pull output stage. But if the tubes are not well matched the common mode rejection does not work perfectly and the AC ripple on the bias supply will not be perfectly cancelled. Same goes for the AC ripple present on the plates. Ever look at the ripple on the B+ feeding the plates and wonder why the amp does not hum terribly? common mode rejection

However, if the bias supply is sufficiently filtered to start with there is very little ripple to deal with on the grids. Adding a second filter cap to the bias supply costs less than $1. Makes sense to me to just do it. Some of the reissue amps are using two filter caps in the bias circuit.

[410_Fanatic]

Regarding common mode rejection - my tubes are only showing a 2 mA difference with my bias probe, so in my case it didn't seem to take much of a mismatch for that hum to appear.  Wondering in hindsight if it might also be this particular set of tubes (almost new JJ 6L6, set to ~55% plate dissipation).

Either way, I'm happy a $1.50 cap solved the issue and makes it so I don't have to randomly swap tubes in order to find a better balanced pair.  I probably would have torn my hair out looking for the source had it not been for this forum.