• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Hum pot implementation on 2a3

Hello folks - I was wondering if someone could help me clarify what I was reading in a post regarding hum pot use with a 2a3. I have tried to draw the schematic that a poster was describing but think I have it wrong. See photo below and any help is appreciated!

Apologies!! These quotes were from Palustris (left) and Kmaier (right).
 

Attachments

  • humpot schematic_Page_1.jpg
    humpot schematic_Page_1.jpg
    308.2 KB · Views: 781
Last edited:
1. The schematic on the left has finer Resolution (easier to find the spot of lowest hum), but has less Range of adjustment versus using the 50 Ohm pot by itself. Use that schematic if you have enough range to find the sweet spot.

If you are worried about getting optimum performance out of a 2A3, you might consider using very well filtered DC for the filaments. But that takes new filament transformers, and more parts and space.
DC filaments gets rid of the intermodulation of 2X power mains frequency (100 or 120Hz) that appear on each and every music tone/note. The intermodulation tones are generally fairly small, but can be quite large, especially when the music is being played at or near full amplifier output.

I designed and built 2A3 amps with AC filaments. I liked them, and so did those who heard them.
But if I ever build a 2A3 amp again, I will use DC filaments this time.

2. The schematic on the right does not have any DC path for the 2A3.
It shows the self bias resistor connecting to the Center Tap of the 2A3 filament (that is the DC path).
But the 2A3 does Not have a center tap.
Well, there is a 'Pseudo 2A3' from one manufacturer that does have a center tap.
I would worry that the quite expensive 'Pseudo 2A3' might not sound as good as a real 2A3. And if that manufacturer no longer makes that tube,
you would not have a replacement.
 
Last edited:
1. The schematic on the left has finer Resolution (easier to find the spot of lowest hum), but has less Range of adjustment versus using the 50 Ohm pot by itself. Use that schematic if you have enough range to find the sweet spot.

If you are worried about getting optimum performance out of a 2A3, you might consider using very well filtered DC for the filaments. But that takes new filament transformers, and more parts and space.
DC filaments gets rid of the intermodulation of 2X power mains frequency (100 or 120Hz) that appear on each and every music tone/note. The intermodulation tones are generally fairly small, but can be quite large, especially when the music is being played at or near full amplifier output.

I designed and built 2A3 amps with AC filaments. I liked them, and so did those who heard them.
But if I ever build a 2A3 amp again, I will use DC filaments this time.

2. The schematic on the right does not have any DC path for the 2A3.
It shows the self bias resistor connecting to the Center Tap of the 2A3 filament (that is the DC path).
But the 2A3 does Not have a center tap.
Well, there is a 'Pseudo 2A3' from one manufacturer that does have a center tap.
I would worry that the quite expensive 'Pseudo 2A3' might not sound as good as a real 2A3. And if that manufacturer no longer makes that tube,
you would not have a replacement.

I fail to see how the circuit on the left would have finer resolution... it doesn't. The balance solution there is common, i.e., the DC and AC balance are linked. The padded pot certainly helps with minimizing the hum level, as it effectively decreases the impedance of the filament to ground. The circuit on the right (loosely drawn from an old post of mine) is incorrect.

As you noted, there's no center tap on a 2A3. The actual schematic would have the cathode resistor connected to a center tap on the 2.5V filament transformer to ground. This results in a fixed DC balance to ground and an adjustable AC balance to ground, also using a padded pot. The adjustment range for both circuits would be identical, as you can swing the AC balance from one filament leg to the other. Using padding on both pots, a 100 ohm pot would yield a finer resolution over a 50 ohm pot.
 
kmaier,

You are half right. The range is the same, but the resolution with the wiper at or nearly centered is finer.

1. Start with a 50 Ohm pot that is connected to the ends of the filament. Now, connect the wiper to the top of the self bias network (self bias resistor with the bypass cap in parallel) to ground.
With the pot centered, it is 25 Ohms from the wiper to either end. The adjustment is 100%, end to end as the pot is rotated.

2. Now, connect a 15 Ohm resistor to one end of the filament, and the other end of the 15 Ohm resistor to the wiper of the pot, and to the top of the self bias network.
Connect a second 15 Ohm resistor to the other end of the filament, and the other end of that second 15 Ohm resistor to the pot wiper, and to the top of the self bias network.

Using the 15 Ohm resistors, and with the pot centered, from one end of the filament, we have a 15 Ohm resistor in parallel with 25 Ohms of the pot, that is 9.37 Ohms to the top of the self bias network. And we likewise have another 9.37 Ohms from the other end of the filament to the top of the self bias network.

Do a little analysis of that circuit, and you will find that the range is the same when the pot wiper is turned fully CCW or fully CW . . .
But with the pot wiper near the center, the effect is like a gear reduction, the same 5 degrees turn CCW or CW from center will have less effect than the circuit in 1. above.
For example, try with the wiper at 23 Ohms from one end, and 27 Ohms from the other end.
Do that for 50 Ohm pot only, and no 15 Ohm resistors.
Do that for 50 Ohm pot plus the two 15 Ohm resistors.

The range of the pot is the same, but now it is easier to land on the exact sweet spot, as long as it is anywhere near the center rotation.

Yes?
No?
 
Last edited:
kmaier,

You are half right. The range is the same, but the resolution with the wiper at or nearly centered is finer.

1. Start with a 50 Ohm pot that is connected to the ends of the filament. Now, connect the wiper to the top of the self bias network (self bias resistor with the bypass cap in parallel) to ground.
With the pot centered, it is 25 Ohms from the wiper to either end. The adjustment is 100%, end to end as the pot is rotated.

2. Now, connect a 15 Ohm resistor to one end of the filament, and the other end of the 15 Ohm resistor to the wiper of the pot, and to the top of the self bias network.
Connect a second 15 Ohm resistor to the other end of the filament, and the other end of that second 15 Ohm resistor to the pot wiper, and to the top of the self bias network.

Using the 15 Ohm resistors, and with the pot centered, from one end of the filament, we have a 15 Ohm resistor in parallel with 25 Ohms of the pot, that is 9.37 Ohms to the top of the self bias network. And we likewise have another 9.37 Ohms from the other end of the filament to the top of the self bias network.

Do a little analysis of that circuit, and you will find that the range is the same when the pot wiper is turned fully CCW or fully CW . . .
But with the pot wiper near the center, the effect is like a gear reduction, the same 5 degrees turn CCW or CW from center will have less effect than the circuit in 1. above.
For example, try with the wiper at 23 Ohms from one end, and 27 Ohms from the other end.
Do that for 50 Ohm pot only, and no 15 Ohm resistors.
Do that for 50 Ohm pot plus the two 15 Ohm resistors.

The range of the pot is the same, but now it is easier to land on the exact sweet spot, as long as it is anywhere near the center rotation.

Yes?
No?

We are in agreement on using a set of padding resistors on the balance pot. Also, as part of the analysis, you can take the 9.37 ohms and half that, as the two legs are in parallel to the "cathode". The circuit on the right has it's balance pot configure the same way, albeit with a 100 ohm balance pot and a pair of 12 ohm resistors. The difference is 9.67 ohms instead of 9.37 ohms, which is not going to be significant. The only difference (assuming a correct schematic) would be the fixed DC balance via the filament transformer.

So, you stated that the circuit on the left had finer resolution. Based on the above, it doesn't. In short, the 100 ohm pot would allow a finer resolution based on your own description above.

Yes?
No?
 
On a filament transformer without a center tap, can one be created with 2 resistors and the cathode resistor connected from there to ground? Then connect the cathode bypass cap from the humpot wiper to ground? In this configuration, would the padding resistors be needed on the humpot?
 
On a filament transformer without a center tap, can one be created with 2 resistors and the cathode resistor connected from there to ground? Then connect the cathode bypass cap from the humpot wiper to ground? In this configuration, would the padding resistors be needed on the humpot?

Without a center-tap on the filament transformer, the split balance technique would be different. You could implement a separate DC balance and AC balance, or just use a combined one (AC and DC balance linked) as in the initial image on the left.

Linked below is a schematic that shows a pair of balance controls... AC and DC. However, I would note that the DC balance doesn't add that much... however, one could, as noted, use a pair of resistors to provide a virtual center tap for a fixed DC balance and then implement the separate AC balance.

https://www.diyaudio.com/forums/att...gulated-supply-direct-coupled-amp-mq_45se-jpg
 
Kmaier, I was thinking of making a virtual center tap with two 100Rs for a fixed DC balance and using the humpot for AC balance. Should padding resisters be approximately one half of each leg of humpot(100R pot>22R each padding resisters)?

Emk2, I wish I could read the article...looks very interesting.
 
Kmaier, I was thinking of making a virtual center tap with two 100Rs for a fixed DC balance and using the humpot for AC balance. Should padding resisters be approximately one half of each leg of humpot(100R pot>22R each padding resisters)?

Emk2, I wish I could read the article...looks very interesting.

Using a pair of resistors for a virtual center tap and using it for a fixed DC balance is fine, but also realize that the resistor pair does two things: 1- They become part of the self-bias resistor value. 2- They also present an additional load to the filament current. Granted, the added current will likely be small, but if you use a pair of 100-ohm resistors, they effectively act as a 50-ohm bias resistor and should be accounted for in your overall bias voltage/current.

For the AC balance pot, I've always used a 100-ohm 2-watt linear pot. In general, a pair of 12-ohm resistors for padding is what I use. Also note that this setup presents an additional current load to the filament current and should be accounted for. Also note there are two reasons to pad the balance pot. 1- To allow a finer adjustment for the hum level, i.e., not a knife-point 2- provide a lower effective resistance in series with the bypass capacitor, which allows better nulling of the AC hum level.
 
kmair,

I originally said:
"1. The schematic on the left has finer Resolution (easier to find the spot of lowest hum), but has less Range of adjustment versus using the 50 Ohm pot by itself. Use that schematic if you have enough range to find the sweet spot."

A 50 Ohm pot does not have as fine a resolution in the center, as the same 50 Ohm pot has when you add a pair of 15 Ohm resistors to that pot.

Also as has been stated in many of the postings above, unless the pot is far off center, then the majority of the current passes through the 15 Ohm resistors.

The left schematic has a 50 Ohm pot.

The right schematic has a 100 Ohm pot, yes of course a 100 Ohm pot has more resolution with the 12 Ohm resistors. But as already noted, the schematic wiring either needs to be changed, or list the exact model number of the "2A3" or "45" tube (that has a center tap).
I am sure someone on the forum will try and build to the right side schematic without reading the text below, and then wonder why it does not work with a real 2A3 or 45.
 
Last edited:
I used to have 45, 2A3, and 300B Direct Heated Triodes (DHT) amplifiers, single ended and some push pull.
Some of them used AC filaments, some of them used DC filaments.

I will not build another DHT amplifier unless I use DC power for the filaments.
It solves several problems, and I have written about the problems with AC powered DHT in various threads in this forum.
And many others have written about that too, on this forum too.
AC is OK. DC is better.
 
Last edited:
AC or DC... in general, I would tend to agree (on DC), but not in every case. I also wrote up a lot on this topic many years ago. I think it depends, mostly on the type of DHT being used.

For something like a good spec'd 45 DHT, I've been able to get hum levels below 90dB referenced to 1-watt output, using an AC filament supply. Granted, I don't have too many 45s that meet this criteria, but certainly enough to supply my personal needs for life.

For the 45 DHT, I've found that ALL of them have the identical "M" style filament. Using DC on this tube pushes one end of the internal structure at a higher bias voltage than the other, which can be a 5% change. I currently have numerous NOS pairs of 45s that can yield over 80dB S/N ratio referenced to 1-watt is more than respectable.

On a side note, I've read posts where folks have touted amazing sound from SET amps that are less than a few watts... superb low-level detail, etc., yet state that the hum and noise output from the amplifier is quite a bit over 1 millivolt, which has a S/N ratio less than 60-65dB.

Back to the topic however. I also have some original NOS single plate 2A3 DHTs and some Western Electric 300b DHTs. The filament (cathode) structure for these are very different from a 45. Effectively, they are a center-tapped filament. Meaning there's a left side and a right side. One filament pin goes to both sides where they meet in the center of the structure. The other filament pin goes to both outside ends of the two sides. In this configuration, an AC filament is going to be useless as the S/N level will be very poor indeed. A DC filament supply must be used for proper S/N ratio. Also, due to the center-tapped configuration, you don't have a situation where one side of the structure has a higher bias than the other.

I also have some NOS 808 triodes and 3C24 triodes. While these are typical transmitting triodes, they can be used in audio amps in Class A2. However, the helical filament does not balance well for hum with AC, so these also must be used with a DC filament supply.

So, for me... I have some lovely 45 DHT and 2A3 DHT (dual plate versions) amps that run AC filament supplies with fixed DC balance and adjustable AC balance that achieve over 80dB S/N referenced to 1-watt output. They are near dead silent with higher efficiency speakers as well. If I can avoid the added complexity of the DC filament supply and achieve the required specification, I'm fine with it... otherwise I'll employ a DC supply.
 
kmair,

As you well know, there are at least 2 effects of using AC filaments on DHT tubes.

1. The AC current through the filaments creates time varying magnetic fields.
Those (multiple) filament wires are attracted to the steel plates at a 2 times line frequency rate (at the rate of Alternations). Any imperfection of the filament's placement between the steel plates on each side of the filament wires, will cause the filament to vibrate toward the closest side of the plate.
This modulates the DC current at a rate of 2 X line frequency.

2. When the AC voltage is instantaneously at 0 Volts, the distribution of DC current to the plate is relatively constant along the filament wires. Consider the transconductance at this state.
When the AC voltage is instantaneously at the crest (peak) of the sine wave, the distribution of DC current to the plate is Not relatively constant along the filament wires. Consider the Different transconductance at this state.
The issue is that the transconductance is changing at the rate of alternations of the AC (2 x the line frequency).
That causes intermodulation of 2x line frequency and a test tone, or musical note.

You have already noted that the transconductance changes when the DHT filaments are DC powered. That alludes to changing transconductance when AC (Time Varying DC) is applied.

Measurement test:
Apply a sine wave test tone to the amplifier (pick a frequency, use one at mid frequencies or high frequencies). Look at the output of the amp with a spectrum analyzer.
You will see sidebands below and above the test tone.
Example, 60 Hz power line frequency, 120 alternations; 1000 Hz test tone . . .
The spectrum analyzer sees: 880Hz sideband, 1000Hz, 1120Hz sideband.

Now, apply DC to the filaments, and repeat the test.
Great, the 880Hz sideband, and the 1120Hz sideband are gone, and only the desired 1000Hz tone is left.

A third effect of AC DHT filaments is when the amp is at or near clipping due to the grid to DHT filament voltage at or near 0V.
I will not even discuss this one in this thread, because no Hi Fi amp should be driven to clipping for good musical playback.

Guitars amps are another animal. Clipping, often a great thing.
And somehow, guitar amps keep showing up on Tubes / Valves Threads, instead of on the Instruments & Amps Threads.
Just my observation.
 
Last edited:
kmair,

45 tubes sound great.
One thing I have noticed about most 45 tubes . . .
The grid wires extend both above and below the plate structure.
That eliminates or at least reduces the "diode effect" when the tube is near cut off.
Many other tubes do not do that.
Instead, the filament wires conduct directly to the plate where there is no intervening grid wires.

Do you think that is part of the intrinsic linearity of the 45 tube?

Kind of reminds me of the difference between overhung and underhung voice coils,
versus driving either one beyond its linear range.
 
6A3sUMMER,

All good points.... and valid. You can also do a similar test with DACs and see some interesting artifacts, etc. Studio grade convertors perform better, but cost a lot more. In these types of tests, you should also be looking at the signal level for those sideband signals. If they're down 80-90dB, then there's a good chance they won't be audible... but there's a lot of variables in a "system" which might highlight them or not.

The 45 amps I designed and built back in 2008 have respectable ratings:
- 25Hz - 50KHz within 1dB at 1-watt output
- 2.25 watts RMS before clipping
- 0.35% THD at 1-watt (1KHz - less at higher frequencies)
- Hum and noise output around 150 microvolts (16-ohm output) - ~88.5dB below 1-watt.

And these specs are with AC filaments and a zero-feedback design. Granted, hand-selected and tested tubes from a pretty large stock, but still, they perform very well. I guess in some ways, I just don't like the concept of a DC filament supply on the 45 triode. Most others, as noted above, simply won't yield acceptable results with an AC filament supply.

RE the post on the 45 triode itself. I have a large number of these and different manufacturers as well. They include: RCA, Tung-Sol, Ken-Rad, Zenith, Philco, Sylvania, GE, Cunningham and possibly others. Most are ST glass but I have half a dozen older globe style as well.

I've done extensive measurements on these... but in general, ALL have the same basic internal construction. The differences lie mainly in internal supports, i.e., mica discs, vibration decoupling to the glass envelope, robustness of the wire framing, etc. There's also a lot of rebranding done during those days. I actually have a pair of RCA 45s in original boxes with identical date codes. Yet, one is made by RCA and the other is made by Sylvania. IIRC, Most of Zenith's tubes were made by Sylvania and marked for them (Zenith).

In any case, my overall favorites are the Sylvania 45s, but with specific internals. The worst are always the Globes, period. As much as people have raved about them, they're mostly terrible in operation. As there's no supports in the envelope, they "sing" when mechanically excited, even low-level playback is apparent. Perhaps that inherent mechanical feedback (from the sound from the speakers) is what some people perceive as a wonderful tube sound... I don't.

One thing is certain however, the 45 (as long as they're good ones) have a very linear transfer function. Perhaps it is a side effect of the grid structure extending beyond the anode, but the filament also does this.

I've also noticed that pretty much everyone who has listened to a well done 45 SET amp likes them... who knew.
 
  • Like
Reactions: 1 user