@Jay Richart
The proposed change from 39 Kohms to 10 KOhms is technically unsound. A simple solution to the problem is inserting a 2nd 1 Megohm resistor between the I/P and the already present 1 Megohm resistance. That change introduces 6 dB. of signal attenuation. A S/N performance degradation is the price you pay.
Do you really need such a high I/P impedance. A pair of 100 Kohm parts would yield the same 6 dB. of attenuation and be less noisy.
If they fit and you can afford them, Caddock TF020 resistors would introduce minimal noise. Vishay/Dale RN65 metal film parts are milspec and are (IMO) probably good enough, without the space & price penalties.
The proposed change from 39 Kohms to 10 KOhms is technically unsound. A simple solution to the problem is inserting a 2nd 1 Megohm resistor between the I/P and the already present 1 Megohm resistance. That change introduces 6 dB. of signal attenuation. A S/N performance degradation is the price you pay.
Do you really need such a high I/P impedance. A pair of 100 Kohm parts would yield the same 6 dB. of attenuation and be less noisy.
If they fit and you can afford them, Caddock TF020 resistors would introduce minimal noise. Vishay/Dale RN65 metal film parts are milspec and are (IMO) probably good enough, without the space & price penalties.
Jay, that looks like a fairly elaborate design/build. Why don't you post a complete diagram and tell us who designed it and perhaps post a link to the designer's description. Hard to tell what the design intent was, some of it doesn't make a lot of sense.
And then tell us how much gain you actually want.
Looks like an Audio Note kit. I would ditch the OPT's and configure the ECC99's as cathode followers.
jeff
jeff
Holy crap. So it is:
L5 The Mentor 2.0 Line Triple C-Core Pre-amplifier – Ank Audio Kits – The authority in high end audio kits
$5500 for a preamp kit? Wow.
In light of that information, my advice is to leave it alone, sell it if it doesn't meet your needs and buy something more suitable. Modifying it is sure to destroy its value. Sort of like buying an Aventador and then butchering it for more trunk space.
But if you insist, I would go with Jeff's advice in post #23.
Edit: Going back to the OP, a preamp of this caliber should NOT have noise unless you made a mistake in the build or you have some external noise issues. Go back and try to find the source of the problem(s) before you do anything else.
L5 The Mentor 2.0 Line Triple C-Core Pre-amplifier – Ank Audio Kits – The authority in high end audio kits
$5500 for a preamp kit? Wow.
In light of that information, my advice is to leave it alone, sell it if it doesn't meet your needs and buy something more suitable. Modifying it is sure to destroy its value. Sort of like buying an Aventador and then butchering it for more trunk space.
But if you insist, I would go with Jeff's advice in post #23.
Edit: Going back to the OP, a preamp of this caliber should NOT have noise unless you made a mistake in the build or you have some external noise issues. Go back and try to find the source of the problem(s) before you do anything else.
Last edited:
Yeah, I guessed there was a transformer at the output. If you want to keep the output transformer then you might use some feedback to reduce the gain from the plate of the ECC99 (AC coupled with a good capacitor) to the cathode of the 6SN7.
Don't be afraid of properly using some feedback....
How much? I would say the gain of ECC99 + transformer. Just as guideline: ECC99 gain at 1 KHz is about 23 and that transformer's step-down ratio is 4:1 to 5:1 typically. So overall it's around 5 which means 14 dB feedback.
I think that DC coupling is more effective. Even with DC coupling you can have a mixed bias, partly imposed by the driver cathode + cathode bias if you like. I think some recent Audio Note UK designs are like this but frankly I have never tried and don't know what's the idea behind.
For DC coupling, you only need a negative supply. You can get dual supply from a normal secondary (without center tap) making a voltage doubler and reference to ground in the middle (see Audio Note old school).
I have located the source of the noise. It is from 6SN7 section. I have even sent the preamp to one of the builder for ANK, the build itself should have no problem and I personally verified many many times..
Just a bit of note for how I find the noise: when I shorted the grid of 6SN7, the noise is still there. When I shorted the grid of ECC99, the noise is gone. The white noise type of noise is from both channel. I tried adding 1K resistor+ 100uF cap in parallel to the plate load(39K) which significantly reduced the gain and the noise is totally gone but it also significantly changed the sound..
since the noise is from 6SN7, I think the best strategy is to only reduce the gain of 6SN7 with minimal modifications—I really love the sound of the preamp, but the white noise is annoying.
I have checked everything on the line stage and filament supply, they are wired correctly and built correctly. All components are working well. I also have changed one of the choke but it end up it is not the source of the noise. I am very positive that the noise is from B+, but don’t know where. There are two more chokes, two caps, and one power transformer that I haven’t replaced with new ones.
Since reducing gain is gonna help, I feel it might be a good move to simply reduce the gain a bit and see.. Is it really a problem that I use 10K for the plate resistor?
Yes, a 6SN7 with a 10k ohm plate resistor will be running in a quite non-linear region. You'd also have to alter the bias arrangement. For a 6SN7, you'll want to use no less than 22k for the plate load resistor.
Why not add a global feedback loop from the second stage back to the first stage? That way you can adjust the gain reduction to taste.
I'm looking for an overview of two-stage triode circuits with global feedback. I think this is the closest I can bring up on short notice.
Two-Triode NFB Stages
This explains the so-called 'anode follower':
The Valve Wizard
--
Why not add a global feedback loop from the second stage back to the first stage? That way you can adjust the gain reduction to taste.
I'm looking for an overview of two-stage triode circuits with global feedback. I think this is the closest I can bring up on short notice.
Two-Triode NFB Stages
This explains the so-called 'anode follower':
The Valve Wizard
--
Or one tries the "straight wire test". Just to check if things are OK without the device altogether. Simple and cheap test introducing some straight wire magic. AFAIK we don't know what the power amplifier is and that may be a deciding factor. As often lack of necessary information is what causes long threads full of opinions.
Only possible with a high input impedance power amplifier and possibly a suitable value volume potentiometer (if the power amplifier lacks that). The long cables people need a preamp for are in practice often less than 1 meter and the output impedance of tube preamps often is higher than the sources driving the same preamp 🙂 It might just be approximately a factor 5000 cheaper and possibly a factor better in final results (which is what counts I guess). If the power amplifier has volume control the test will take a few minutes.
Only possible with a high input impedance power amplifier and possibly a suitable value volume potentiometer (if the power amplifier lacks that). The long cables people need a preamp for are in practice often less than 1 meter and the output impedance of tube preamps often is higher than the sources driving the same preamp 🙂 It might just be approximately a factor 5000 cheaper and possibly a factor better in final results (which is what counts I guess). If the power amplifier has volume control the test will take a few minutes.
Last edited:
What about adding a current source paralleled to its cathode resistor? In this case the bias is changed.. please take a look of my first post..
"Since reducing gain is gonna help, I feel it might be a good move to simply reduce the gain a bit and see..
Is it really a problem that I use 10K for the plate resistor?"
Yes, but instead you can split the 39k into two unequal parts that sum to 39k, and take the signal from
their junction. For example, a 27k from the plate, and a 12k from the 27k to the power supply. That
would reduce the output level without changing the static bias conditions of the tube. Unless the power
supply is poorly filtered, the noise from that stage would decrease accordingly.
Is it really a problem that I use 10K for the plate resistor?"
Yes, but instead you can split the 39k into two unequal parts that sum to 39k, and take the signal from
their junction. For example, a 27k from the plate, and a 12k from the 27k to the power supply. That
would reduce the output level without changing the static bias conditions of the tube. Unless the power
supply is poorly filtered, the noise from that stage would decrease accordingly.
Last edited:
@Rayma:
A voltage divider on the output of a stage won't change the S/N of that stage, will it? If he wants less volume he can just turn down the volume pot.
He is assuming that reducing the gain of the stage (not the output signal amplitude) will improve the S/N. I don't see any basis for it, but that's his implied assumption.
A voltage divider on the output of a stage won't change the S/N of that stage, will it? If he wants less volume he can just turn down the volume pot.
He is assuming that reducing the gain of the stage (not the output signal amplitude) will improve the S/N. I don't see any basis for it, but that's his implied assumption.
He says that the noise is from the first stage (he grounded its output and the noise was gone).
He also said the volume control (located before the first stage) did not affect the noise.
That means an attenuator at the output of the first stage will reduce the noise and gain.
This is what the composite plate resistor will do, serve as an attenuator without affecting
the DC bias conditions. The only drawback is possibly higher power supply hum in the output,
but that could be dealt with by adding an RC decoupling network for the first stage.
If he reduces the gain too much while trying to reduce the noise, the input tube could overload from
excessive input signal, if the volume control is increased too much to compensate for the loss of gain.
He also said the volume control (located before the first stage) did not affect the noise.
That means an attenuator at the output of the first stage will reduce the noise and gain.
This is what the composite plate resistor will do, serve as an attenuator without affecting
the DC bias conditions. The only drawback is possibly higher power supply hum in the output,
but that could be dealt with by adding an RC decoupling network for the first stage.
If he reduces the gain too much while trying to reduce the noise, the input tube could overload from
excessive input signal, if the volume control is increased too much to compensate for the loss of gain.
Last edited:
Hopefully after the coupling cap, i.e. grounded the grid of the second tube.
If those statements are both true, then the noise is created in the first stage or comes from the supply. So, for a given listening level, what you propose is to cut down the signal at the output of the stage, then increase the input signal proportionally, until the signal swamps the noise. Somehow I find that sub-optimal in a $5k piece of equipment🙂
Here is what I find absurd about this thread. Presumably the reason someone buys a $5k preamp is a sonic quality unobtainable at a more reasonable price. A brilliant design in other words. So now we try to re-design the circuit instead of fixing it. The only logical conclusion is that we all think the original designer is an idiot. Is that reasonable?
As Jean-Paul said, time to step back and re-examine what we're trying to do .
Thank you sir, I feel you exactly got my point and know what my issues are.
I will try what you said! And would you think using current source at the cathode and decrease the value of plate resistor could be a better plan?
I will try what you said! And would you think using current source at the cathode and decrease the value of plate resistor could be a better plan?