Hello Group:
I built a single-ended 2A3 stereo amp about 10 years ago. A beautiful sounding amp! It went into storage for a few years and I recently fired it back up.
It still sounds fantastic, but in ONE CHANNEL ONLY there is a very low-level noise present when I am not playing music. It is very faint but still annoying, barely detectable from my listening position (95 dB efficient speakers) I really need to get up to the speaker to hear it clearly.
The sound is totally random, it comes and goes as it pleases. Its hard to describe, but it sounds like a kind of random sputtering or the crumpling of a sheet of paper.
Changing tubes does not change anything. And it is not my peripheral gear...the amp does it even when everything else is off. It is not the driver circuitry (a paralleled 6SL7 per channel) because it still does it when the driver tube is completely removed! It is not the AC balance pot across the filament, as I recently replaced the old noisy one and hit the replacement with Pro-Gold cleaner. Rotating the pot does not induce any noise, nor stop it.
I changed the 5U4 rectifier to see if it was sputtering...no change. A scope on the LCLC power supply (independent supply for each channel) shows virtually no ripple or sputtering on the B+ line, my tiny 4 mV of ripple is right about at the ambient RF noise level.
I cleaned the tube sockets and the tube pins with Pro-Gold, and yes, some black oxidation came off but still NO change. Wiggling the tubes while in the sockets does not start, nor stop the noise! I even changed the "cathode" resistor on the bad channel, thinking that it might be a noisy resistor. Nope!
It's driving me crazy!
This is a dirt simple amplifier circuit. There can only be few possible places for this intermittent noise to come from:
1) Bad tube socket...but wiggling the tubes does not affect the noise!
2) Some power-supply related noise...perhaps the electrolytics in that channel or the polypro bypass caps have developed some kind of internal leakage or intermittent micro arcing?
3) Unlikely, but some anomaly in the filament or output transformers? *The main power transformer is shared for both channels, so it can't be that.
I'm going loony! Any advice, suggestions?
Thanks!
I built a single-ended 2A3 stereo amp about 10 years ago. A beautiful sounding amp! It went into storage for a few years and I recently fired it back up.
It still sounds fantastic, but in ONE CHANNEL ONLY there is a very low-level noise present when I am not playing music. It is very faint but still annoying, barely detectable from my listening position (95 dB efficient speakers) I really need to get up to the speaker to hear it clearly.
The sound is totally random, it comes and goes as it pleases. Its hard to describe, but it sounds like a kind of random sputtering or the crumpling of a sheet of paper.
Changing tubes does not change anything. And it is not my peripheral gear...the amp does it even when everything else is off. It is not the driver circuitry (a paralleled 6SL7 per channel) because it still does it when the driver tube is completely removed! It is not the AC balance pot across the filament, as I recently replaced the old noisy one and hit the replacement with Pro-Gold cleaner. Rotating the pot does not induce any noise, nor stop it.
I changed the 5U4 rectifier to see if it was sputtering...no change. A scope on the LCLC power supply (independent supply for each channel) shows virtually no ripple or sputtering on the B+ line, my tiny 4 mV of ripple is right about at the ambient RF noise level.
I cleaned the tube sockets and the tube pins with Pro-Gold, and yes, some black oxidation came off but still NO change. Wiggling the tubes while in the sockets does not start, nor stop the noise! I even changed the "cathode" resistor on the bad channel, thinking that it might be a noisy resistor. Nope!
It's driving me crazy!
This is a dirt simple amplifier circuit. There can only be few possible places for this intermittent noise to come from:
1) Bad tube socket...but wiggling the tubes does not affect the noise!
2) Some power-supply related noise...perhaps the electrolytics in that channel or the polypro bypass caps have developed some kind of internal leakage or intermittent micro arcing?
3) Unlikely, but some anomaly in the filament or output transformers? *The main power transformer is shared for both channels, so it can't be that.
I'm going loony! Any advice, suggestions?
Thanks!
Post the schematic and some photos. It could be a bad resistor, perhaps the output tube's grid resistor.
Last edited:
Dang...I knew I had to have overlooked something- the GRID leak resistor! It's right there in plain sight, too - DUH! I need to change that out to eliminate one more possibility!
Could also be the coupling cap between driver and triode final tube...or the plate resistor of the driver stage!
Now I've got my thinking cap on!
Thanks!
Last edited:
If you're going to use the process of elimination leave the capacitors for last, since the resistors
are more likely to cause such a problem. Start with the grid resistor first.
Thank you, and yes I agree the noise must be originating somewhere in the grid circuitry, as a scope on the power supply shows it to be stable and exceptionally low noise. No "sputtering" could be seen on the B+.
Thank you...good idea. I'll reflow all the joints in that channel that I have not yet recently done.
A quick test would be to remove (or lift one end) of the grid coupling capacitor.
If noise is still present, then go for the grid resistor.
For sure! Caig ProGold - considered one of the very best in USA. Non conductive, non-flammable.
Hi, parallel 6SL7 driving 2a3 is pretty simple. Are you using a mosfet follower?
If no driver tube is there, and you still get this noise, then I can't imagine it being the grid stopper.
Ian
No follower, the parallel 6SL7 is low enough impedance to give me flat response out to 20 kHz, then it rolls off.
Even when the driver tube is removed, there is still the plate resistor for the driver tube, driver-to-final tube coupling cap and grid leak (not stopper) resistor on the 2A3 still connected to the B+. A current leakage path could still exist even when the 6SL7 is unplugged. Tomorrow I will open it up and see if I can determine which of these parts is at fault.
You don't have a grid stopper on your 2a3 grid? Try a 1k ohm metal film grid stopper, close to the pin. 🙂
As an aside:
I found that a driver circuit with output impedance around 2k ohm or less works nicely for single ended 2a3. High output impedance tubes/valves (like 12ax7 and 6sn7) just didn't sound as good....
But hey. if you like the sound then that is fine.
As an aside:
I found that a driver circuit with output impedance around 2k ohm or less works nicely for single ended 2a3. High output impedance tubes/valves (like 12ax7 and 6sn7) just didn't sound as good....
But hey. if you like the sound then that is fine.
Lets do some calculations for 2a3....
Cin = Cgk + Cgp*(A+1)
Cgk = Grid to cathode capacitance = 7.5pF + 0.7pF (stray) = 8.2pF
Cgk = Grid to plate capacitance = 16.5pF + 0.8pF (stray) = 17.2pF
A = Stage Gain = 4.2
Cin = 8.2pF + (17.2pF *5.2) = 97.6pf (let's round it up to 100pf)
Now let's calculate the frequency where this circuit starts to roll off since it works like an RC low pass filter:
Nominal output impedance of 6sl7 is 45k Ohm. Your circuit is using two in parallel, so we will make it 22.5k Ohm.
f = 1/ (2*pi*R*C) = 1/ (2*pi*22500*(1x10 to the power of -10))
f = 7.07 kHz 🙁
My maths say your circuit starts rolling off at around 7Khz.
If you use a source with say... 2k output impedance, then:
f = 79.5 kHz 🙂
There is more to consider as well, but this is a good place to start.
Cin = Cgk + Cgp*(A+1)
Cgk = Grid to cathode capacitance = 7.5pF + 0.7pF (stray) = 8.2pF
Cgk = Grid to plate capacitance = 16.5pF + 0.8pF (stray) = 17.2pF
A = Stage Gain = 4.2
Cin = 8.2pF + (17.2pF *5.2) = 97.6pf (let's round it up to 100pf)
Now let's calculate the frequency where this circuit starts to roll off since it works like an RC low pass filter:
Nominal output impedance of 6sl7 is 45k Ohm. Your circuit is using two in parallel, so we will make it 22.5k Ohm.
f = 1/ (2*pi*R*C) = 1/ (2*pi*22500*(1x10 to the power of -10))
f = 7.07 kHz 🙁
My maths say your circuit starts rolling off at around 7Khz.
If you use a source with say... 2k output impedance, then:
f = 79.5 kHz 🙂
There is more to consider as well, but this is a good place to start.
Last edited:
I’m not doubting your math, but my SPICE models and more importantly...measurement on test bench shows just 1 dB down at 20 kHz. Possibly why is that?
We’ve gone off topic here, but since we have - I ran simulations again for my 2A3 driver circuit (parallel 6SL7). John Broskie's Tube Cad program shows an output Z of 13.5 k, not 22.5k - which is what another poster calculated.
My original measurement notes show my amplifier was only -0.25 dB down at 20 kHz.
Considering my ears are only good to 12 kHz, I think I’m ok. The amp sounds great, regardless...
My original measurement notes show my amplifier was only -0.25 dB down at 20 kHz.
Considering my ears are only good to 12 kHz, I think I’m ok. The amp sounds great, regardless...
Attachments
Last edited:
Hi
Thanks for the screenshot. It tells a lot.
Your 6SL7 output impedance is lower since you use a lower value plate resistor (a single 37.5k ohm from your screenshot).
Since you tie both plates together, and use a single 37.5k ohm resistor, the rough calculated output impedance is 13'865.55 ohms (assuming the cathode is bypassed).
Broskie's Program probably looks up the internal resistance according to the actual bias of the tube. I personally use LT spice for simulations.
This nicely corresponds to a HF roll-off starting around 11.8 kHz. So I do not doubt that you find your amplifier nice to you. I can't quite hear 14 kHz, but my daughter still can. 🙂
Maybe a 1k ohm grid stopper would help, but it is still very odd that you get this sound even with the driver valve/tube removed.
I can't for the life of me think that a resistor with no current can "sound". The Grid bias resistor "RL" or "R load" in your screenshot really 470K Ohm?
This is a high value, but it needs to be due to the "wimpy" 6sl7 driver. Maybe swap it out for another and see if this makes a difference.
Best regards
Ian
Thanks for the screenshot. It tells a lot.
Your 6SL7 output impedance is lower since you use a lower value plate resistor (a single 37.5k ohm from your screenshot).
Since you tie both plates together, and use a single 37.5k ohm resistor, the rough calculated output impedance is 13'865.55 ohms (assuming the cathode is bypassed).
Broskie's Program probably looks up the internal resistance according to the actual bias of the tube. I personally use LT spice for simulations.
This nicely corresponds to a HF roll-off starting around 11.8 kHz. So I do not doubt that you find your amplifier nice to you. I can't quite hear 14 kHz, but my daughter still can. 🙂
Maybe a 1k ohm grid stopper would help, but it is still very odd that you get this sound even with the driver valve/tube removed.
I can't for the life of me think that a resistor with no current can "sound". The Grid bias resistor "RL" or "R load" in your screenshot really 470K Ohm?
This is a high value, but it needs to be due to the "wimpy" 6sl7 driver. Maybe swap it out for another and see if this makes a difference.
Best regards
Ian
Last edited:
My measurements are not inaccurate. I'm a retired electrical engineer. The amp is basically flat to 20 kHz, regardless of mathematical predictions.
I attribute the increased bandwidth to a lower-than specified shunt capacitance in the 2A3, hence less Miller effect. The specified grid capacitance for the 2A3 is likely a maximum value - not an average one - as the max value would be the most critical one for circuit designers.
Regards,
I attribute the increased bandwidth to a lower-than specified shunt capacitance in the 2A3, hence less Miller effect. The specified grid capacitance for the 2A3 is likely a maximum value - not an average one - as the max value would be the most critical one for circuit designers.
Regards,
There is no "load resistor" in the actual circuit...it is just there for simulation purposes. The 2A3 grid never goes positive, so it presents a near infinite load to the driver, so the 470k grid leak resistor becomes the major load - if we IGNORE the effects of grid capacitance and Miller effect, which of course will affect HF bandwidth.
I would check amp for grid stoppers too.
A valve makes a nice little antenna.
I had a fault on a hybrid amp and the chip amp output part oscillated.
The valve picked it up nicely and amplified it around the loop again !
Could be just low level radio pick up.
A valve makes a nice little antenna.
I had a fault on a hybrid amp and the chip amp output part oscillated.
The valve picked it up nicely and amplified it around the loop again !
Could be just low level radio pick up.