The other night I was tweaking my 2A3 amp, and came up with a method for very low HUM levels without having a HUM Pot in the single path.
The typical method’s you see for DHT cathode bias are:
1. R & C on the filament transformer center tap
2. R & C from a HUM Pot
3. R & C from the center of two 22R 3W resistors across the filament
A better method I found on a schematic from diy member kmaier is:
4. The cathode bias resistor on the filament transformer center tap and the capacitor on the HUM Pot
The new method I proposing:
5. The cathode bias resistor on the filament transformer center tap and the capacitor on from the center of 2 close (but not always exact) resistors.
Some test results:
1.4mV HUM with method #4
4.2mV with R & C on the filament transformer center tap
3.2mV with R on the filament transformer center tap and the Cap for the center of 2 exact matching 470R’s, .5w, carbon resistors
Why was the HUM so much higher with the two 470R’? Because I need 2 resistors that mimic the HUM Pot
So with the HUM Pot disconnected, I measured it and found it was at 49.4R and 51.5R,
Or 1 side was about 4% larger than the other.
Next I measures my supply of 470R’s and came up with a pair that had about a 4% delta
(450R and 469R in my case)
Using jumper wires I tested the amp with my 450/469 resistors in place of the HUM pot and ended up with the same 1.4mV
Bottom line…. Very low hum without a variable resistor in the single path!
The typical method’s you see for DHT cathode bias are:
1. R & C on the filament transformer center tap
2. R & C from a HUM Pot
3. R & C from the center of two 22R 3W resistors across the filament
A better method I found on a schematic from diy member kmaier is:
4. The cathode bias resistor on the filament transformer center tap and the capacitor on the HUM Pot
The new method I proposing:
5. The cathode bias resistor on the filament transformer center tap and the capacitor on from the center of 2 close (but not always exact) resistors.
Some test results:
1.4mV HUM with method #4
4.2mV with R & C on the filament transformer center tap
3.2mV with R on the filament transformer center tap and the Cap for the center of 2 exact matching 470R’s, .5w, carbon resistors
Why was the HUM so much higher with the two 470R’? Because I need 2 resistors that mimic the HUM Pot
So with the HUM Pot disconnected, I measured it and found it was at 49.4R and 51.5R,
Or 1 side was about 4% larger than the other.
Next I measures my supply of 470R’s and came up with a pair that had about a 4% delta
(450R and 469R in my case)
Using jumper wires I tested the amp with my 450/469 resistors in place of the HUM pot and ended up with the same 1.4mV
Bottom line…. Very low hum without a variable resistor in the single path!
Salute to Kmaier
I have applied kmaier's method to my PX4 amp with good results.
I am using AC heat and I found it difficulty to use DC heating as the voltage drop via the rectifiers and resistors would not be enough for a 4V AC to convert to 4V DC and the filament transformer only had 4V windings.
I was able to get around 1.5mV to 2mV of residual noise with PX4 but with PX25 (drawing 2A filament current), I could only able to get around 6mV of noise.
I understand your idea of eliminating the pot and use fix resistors, but I wonder would the drift in the tubes and transformers later make your fix resistors inappropiate ?
I would still like to keep a pot so that I can readjust it later.
I have applied kmaier's method to my PX4 amp with good results.
I am using AC heat and I found it difficulty to use DC heating as the voltage drop via the rectifiers and resistors would not be enough for a 4V AC to convert to 4V DC and the filament transformer only had 4V windings.
I was able to get around 1.5mV to 2mV of residual noise with PX4 but with PX25 (drawing 2A filament current), I could only able to get around 6mV of noise.
I understand your idea of eliminating the pot and use fix resistors, but I wonder would the drift in the tubes and transformers later make your fix resistors inappropiate ?
I would still like to keep a pot so that I can readjust it later.
Interesting idea.
Is there any difference in sound quality that you can make out with all four methods?
andy
Is there any difference in sound quality that you can make out with all four methods?
andy
Hi sgerus,
I think you are just fooling yourself😉!
With ca 235ohms in series with 470u you just partly bypass the cathode.
So you loose signalvoltage of the same percentage as the humreduction.
I think you are just fooling yourself😉!
With ca 235ohms in series with 470u you just partly bypass the cathode.
So you loose signalvoltage of the same percentage as the humreduction.
Andy:
I can’t really tell if the sound is any better, I was only interested in HUM.
Lars:
I will scope check and report back later. I do know that the DC voltage on the cathode bypass cap is exactly the same as with the 100R hum pot.
I can’t really tell if the sound is any better, I was only interested in HUM.
Lars:
I will scope check and report back later. I do know that the DC voltage on the cathode bypass cap is exactly the same as with the 100R hum pot.
The voltages shouldn´t be different as there is no current flow.
By your arrangement you will get lower gain and higher Zout.
So you actually hasn´t won anything, as the quote between output voltage and hum voltage stays the same.
Still it might sound better.....
By your arrangement you will get lower gain and higher Zout.
So you actually hasn´t won anything, as the quote between output voltage and hum voltage stays the same.
Still it might sound better.....
All,
I did a lot of testing during the prototype design (45s and 2A3s) which eventually resulted in the split balance technique. Sonics are also important in addition to low output noise (hum). Using good quality 45 tubes, I can get hum levels below 200 micro-volts (which is < 0.0002 volts at the 8-ohm tap) which results in a S/N ratio better than 80dB referenced to 1-watt. 80dB is my personal acceptance factor... anything less and I don't use the tube. Anything better and I smile more, have a few pairs that are around 90dB.
As pointed out, the high value resistors simply form a RC network and you lose gain, i.e., you don't have a low AC impedance to ground. The reason you still benefit for reduced hum levels is the RC network is a lower impedance at 60Hz than nothing at all. Still, you have limited cathode (AC) bypass which decreases at lower frequencies.
You're also not able to compensate for the imperfect balance of the cathode versus the grid and plate alignment, it's basically impossible to have the internal assembly perfectly balanced. There's too many variables.... the gauge linearity of the filament wire, the coating on the filament wire (and the consistency of the coating itself), the spiral grid assembly plus the plate assembly and the overall alignment. It might not seem like a lot, but we're talking electrons flowing... they're really small.
I use a pair of low value padding resistors for three reasons: 1- to reduce the resistance that ends up in series with the bypass capacitors (12 ohm resistors results in less than 5 ohms), 2- to widen the adjustment range of the 100-ohm pot (reduces the "knife-edge" null point), and 3- to lessen the (sonic) effect of the element of the pot versus the fixed resistors which should be of similar quality to the rest of the circuit.
Regards, KM
I did a lot of testing during the prototype design (45s and 2A3s) which eventually resulted in the split balance technique. Sonics are also important in addition to low output noise (hum). Using good quality 45 tubes, I can get hum levels below 200 micro-volts (which is < 0.0002 volts at the 8-ohm tap) which results in a S/N ratio better than 80dB referenced to 1-watt. 80dB is my personal acceptance factor... anything less and I don't use the tube. Anything better and I smile more, have a few pairs that are around 90dB.
As pointed out, the high value resistors simply form a RC network and you lose gain, i.e., you don't have a low AC impedance to ground. The reason you still benefit for reduced hum levels is the RC network is a lower impedance at 60Hz than nothing at all. Still, you have limited cathode (AC) bypass which decreases at lower frequencies.
You're also not able to compensate for the imperfect balance of the cathode versus the grid and plate alignment, it's basically impossible to have the internal assembly perfectly balanced. There's too many variables.... the gauge linearity of the filament wire, the coating on the filament wire (and the consistency of the coating itself), the spiral grid assembly plus the plate assembly and the overall alignment. It might not seem like a lot, but we're talking electrons flowing... they're really small.
I use a pair of low value padding resistors for three reasons: 1- to reduce the resistance that ends up in series with the bypass capacitors (12 ohm resistors results in less than 5 ohms), 2- to widen the adjustment range of the 100-ohm pot (reduces the "knife-edge" null point), and 3- to lessen the (sonic) effect of the element of the pot versus the fixed resistors which should be of similar quality to the rest of the circuit.
Regards, KM
Lars was right......
With the 100R hum pot 1 Volt in = 1 Watt into 8R at 1000Hz
With the two 470R's it's up to 1.2V = 1Watt's
SO IT LOOKS LIKE THIS EXERCISE WAS A FLOP!!!
Well, that's how we learn.
Looks like the hum pot's will go back in, this time in // with two 12R resistors for all the reasons KM pointed out.
Thanks Everyone
With the 100R hum pot 1 Volt in = 1 Watt into 8R at 1000Hz
With the two 470R's it's up to 1.2V = 1Watt's
SO IT LOOKS LIKE THIS EXERCISE WAS A FLOP!!!
Well, that's how we learn.
Looks like the hum pot's will go back in, this time in // with two 12R resistors for all the reasons KM pointed out.
Thanks Everyone
Interesting thread. Can I just make sure I've understood this? The final choice was a hum pot with 12R from each end to the wiper. The filament supply goes to each end of the pot. Where does the wiper go -
- to unbypassed cathode resistor to ground
- to bypassed cathode resistor to ground
- other?
I want to try this in PP so I have two filament supplies, two tubes, but one cathode resistor.
So how exactly do I implement the "best method"?
Many thanks!!
andy
- to unbypassed cathode resistor to ground
- to bypassed cathode resistor to ground
- other?
I want to try this in PP so I have two filament supplies, two tubes, but one cathode resistor.
So how exactly do I implement the "best method"?
Many thanks!!
andy
andyjevans said:
Andy,
Not sure what you're suggesting is possible... i.e., two filament supplies but one cathode resistor. Are you using directly-heated cathode output tubes in push-pull?
The above, if he really takes my suggestion(s), would have the hum balance pot configured as above but the wiper would only have the cathode bypass cap to ground (AC balance being adjustable) and the cathode bias resistor from the filament center-tap to ground (fixed DC balance via the filament winding).
Can you clarify your output stage configuration and tube types?
Regards, KM
Hello - thanks, I understand the single ended design now.
For push pull, separate filament supply for each tube - let's say 2a3 - coming from transformers with no centre tap. Actually this would be AC in the case of 2a3, but maybe DC in the case of a 300b.
Then join one end of the filament of tube A to one end of tube B and from that point, run a resister to ground. Resistor could be bypassed or not.
So, in your system do you use a hum pot on EACH 2a3 with a capacitor to ground on EACH 2a3, and then join the two ends of the filaments as above and run a resistor to ground from that point? Sounds like it could be one way to do it. Would that have any benefit?
Or is there some other way?
andy
For push pull, separate filament supply for each tube - let's say 2a3 - coming from transformers with no centre tap. Actually this would be AC in the case of 2a3, but maybe DC in the case of a 300b.
Then join one end of the filament of tube A to one end of tube B and from that point, run a resister to ground. Resistor could be bypassed or not.
So, in your system do you use a hum pot on EACH 2a3 with a capacitor to ground on EACH 2a3, and then join the two ends of the filaments as above and run a resistor to ground from that point? Sounds like it could be one way to do it. Would that have any benefit?
Or is there some other way?
andy
My design is only single-ended, not push pull. For what you are doing, I would implement independent supplies, cathode resistors and hum balance pots. Beyond this, you might be fine with a single supply (filaments in parallel) and hoping for the best from CMRR which could be acceptable.
Regards, KM
Regards, KM
- Status
- Not open for further replies.