Pre/DI
- Thread starter analag
- Start date
Help Support GroupDIY Audio Forum:
pstamler
Well-known member
A very elegant design, Analag.
Peace,
Paul
Peace,
Paul
KingKai
Well-known member
Very Interesting.
I think something is amiss in the DI input. I looks like it's grounded out through R with no connection to the diff pair.
Peace - Out, Irv
I think something is amiss in the DI input. I looks like it's grounded out through R with no connection to the diff pair.
Peace - Out, Irv
Use your imagination....and also read what I said about the 1/4 inch jack that is going to break the transformer/150k connection. R then become the grid leak resistor and the DI goes to the junction. So the transformer and it's 150k resistor gets switched out and R and it's DI lead gets swithched in. I hope I gave a clear enough explanation.
analag
analag
KingKai
Well-known member
What !!! You want me to think!!!! For myself even!!!
Why of all the nerve......... well OK then.
It's all good.
Peace - Out, Irv
By the way, Thanx for sharing Analag. :thumb:
Why of all the nerve......... well OK then.
It's all good.
Peace - Out, Irv
By the way, Thanx for sharing Analag. :thumb:
Fun with SPICE.
Now get real.
What happens if the supply voltages vary? If V2 -25V varies, nothing. If V1 +250V varies, the voltage at the grid of U2a varies the same amount.
Do that and run distortion runs. The teeny residual THD of the White Cathode Follower is very sensitive to the voltage at U2a grid. It needs to stay not-quite-half the supply voltage to get everything to cancel. Not that exact cancellation is always our goal; but if small changes in supply voltage give audible shifts in distortion, the thing sounds different every day.
Of course you could regulate V1. But then you'd probably regulate V2, and then D1 Q1 are pointless: they could be replaced with a resistor.
And if you do replace Q1 with about 11K, the whole shebang auto-balances for large changes in supply voltage. With the high-Mu 6N1p, you could trim the 11K to get half supply voltage at the U2a cathode and be nearly at the THD null. Or trim off the null for a certain "sound". And it would stay there with simple unregulated supplies. (They do need to track together and drift slowly.)
The 6N1p idles at 7mA and swings to about 15mA peak, no? So 8mA peak output, 20mA on the secondary, 12V peak or 8Vrms or +20dBm in 600 ohms, OK.
The ~3K6 load is about the upper end of where a 6N1p WCF makes sense. 6N1p plate resistance is ~7K, the WCF switches from push-pull to push-push when the load is about equal to Rp/2. A half-6N1p cathode follower will do nearly as well, with one less tube (not one less bottle, unfortunately). A 12AU7 will also work nicely, at half the heater power. A 6BX7 draws a lot more heater, but could drive 1:1 into 600 ohms, effectly giving 2.5:1 more voltage gain or NFB (and a generic transformer).
Now get real.
What happens if the supply voltages vary? If V2 -25V varies, nothing. If V1 +250V varies, the voltage at the grid of U2a varies the same amount.
Do that and run distortion runs. The teeny residual THD of the White Cathode Follower is very sensitive to the voltage at U2a grid. It needs to stay not-quite-half the supply voltage to get everything to cancel. Not that exact cancellation is always our goal; but if small changes in supply voltage give audible shifts in distortion, the thing sounds different every day.
Of course you could regulate V1. But then you'd probably regulate V2, and then D1 Q1 are pointless: they could be replaced with a resistor.
And if you do replace Q1 with about 11K, the whole shebang auto-balances for large changes in supply voltage. With the high-Mu 6N1p, you could trim the 11K to get half supply voltage at the U2a cathode and be nearly at the THD null. Or trim off the null for a certain "sound". And it would stay there with simple unregulated supplies. (They do need to track together and drift slowly.)
The 6N1p idles at 7mA and swings to about 15mA peak, no? So 8mA peak output, 20mA on the secondary, 12V peak or 8Vrms or +20dBm in 600 ohms, OK.
The ~3K6 load is about the upper end of where a 6N1p WCF makes sense. 6N1p plate resistance is ~7K, the WCF switches from push-pull to push-push when the load is about equal to Rp/2. A half-6N1p cathode follower will do nearly as well, with one less tube (not one less bottle, unfortunately). A 12AU7 will also work nicely, at half the heater power. A 6BX7 draws a lot more heater, but could drive 1:1 into 600 ohms, effectly giving 2.5:1 more voltage gain or NFB (and a generic transformer).
skipwave
Well-known member
This is an interesting Frankenstein, the pieces are familiar but the method of cojoining them is unique. Thanks for sharing, I'm sure I will learn a lot in the resulting discussion.
Is the second half of V1 grounded-grid? I've never seen that done thru a pot. I'm struggling to understand how the -25v is biasing that triode.
Is the second half of V1 grounded-grid? I've never seen that done thru a pot. I'm struggling to understand how the -25v is biasing that triode.
It looks to me like the bias is done at the cathode, just like the first half of the tube. The output of the white cathode follower feedsback to the grid of V1b which is also shunted to ground by the pot, to get variable feedback.
adam
This is a tube opamp, the grid will only be grounded when the wiper is all the way up, ie. no global feedback. In this case U1b will act as a nonlinear cathode resistor, thus correcting some of the imperpections imposed by the cathode follower U1a.
Biased by the CCS.
PRR points out the impact of a varying B+ supply. I expect to use a regulated supply throughout the pre.
I have used this particular design many times over, finding it to be clean and quite extended. For the sake of simplicity I could use a "soft" resistance at the cathode junction of U1a and U1b but a "hard" resistance allow me to set the current through the 12AT7 without loosing too of the very high resistance that the tube see's at it's cathodes.
Make the emitter resistance variable then adjust until you get exactly 125V at the plate of U1b, or as PRR suggest at the junction of U2a and U2b.
A 6N1P is a medium Mu tube not a high Mu, yes there are better choices out there, I happen to have many of these lying around and they sound magnificent in any application. The 6BX7 sounds interesting as it allows the use of a 1:1 trannies thus preserving the gain and letting me use an XSM600:600 which is both super cheap and super sweet.
analag
I have used this particular design many times over, finding it to be clean and quite extended. For the sake of simplicity I could use a "soft" resistance at the cathode junction of U1a and U1b but a "hard" resistance allow me to set the current through the 12AT7 without loosing too of the very high resistance that the tube see's at it's cathodes.
Make the emitter resistance variable then adjust until you get exactly 125V at the plate of U1b, or as PRR suggest at the junction of U2a and U2b.
A 6N1P is a medium Mu tube not a high Mu, yes there are better choices out there, I happen to have many of these lying around and they sound magnificent in any application. The 6BX7 sounds interesting as it allows the use of a 1:1 trannies thus preserving the gain and letting me use an XSM600:600 which is both super cheap and super sweet.
analag
Im not sure i unterstand what you are saying here, and maybe im saying the same thing, but.......
We have two options for the tail R here:
True CCS and low negative voltage.
Simulated CCS ie. large resistor and thus large negative voltage.
Small value resistor and low voltage is a no-no.
The signal will be attenuated and S/N ratio will suffer.
As far as using this for a pre....i dont know....where´s the gain?
> Is the second half of V1 grounded-grid?
The whole "Frankenstein" is a perfectly simple
differential amplifier (U1)
plus a fancy cathode follower (U2).
You "can" see U1 as a cathode follower driving a grounded grid stage.
That's not as useful as understanding it as a Differential Amplifer.
But if you must: what is the difference between a Grounded Grid with a
hard-grounded grid, or a GG with a <10K resistor in the grid return?
For audio: very little. The grid is ~200Meg||~30pFd, so 10K is "nothing".
> where's the gain?
Mostly in the 1:10 input transformer.
U1 will have roughly half the 12AT7 handbook gain, say 20-25.
U2 has gain over 0.9. Then 2.5:1= 0.4 in the output transformer.
10*25*0.4= 100 overall.
NFB reduces the V1 gain from 20-25 toward a target of 11,
so probably 5 to 7, around 25 overall.
The 1:10 transformer is necessary for lowest noise,
because we've "stacked" the noise of two cathodes.
Gm of 12AT7 at 1.2mA is about 1,500uMho. Rk= ~660. Rn= ~2K.
For gain and noise calculations, Gm is effectively between
U1a grid and U1b grid, so Gm= 750uMho, Rk=1,300, Rn= 4K.
The source noise resistance is 200 ohms times 1:10^2 or 20K.
4K tube noise added to 20K source noise is "small",
but with a 1:5 transformer it would be ~3dB NF.
> Small value resistor and low voltage is a no-no.
The signal will be attenuated and S/N ratio will suffer.
Oh, it would not be good but not so bad.
If U1a's Rk is 1.3K, and we use an 11K cathode resistor,
attenuation is like 0.9 or -1dB.
The voltage across the long-tail only has to be much-much greater than
Vp/Mu, where Vp is around 125V and Mu is around 60.
So we need much-much more than 2V. 25V is much-much more than 2V.
We also want the voltage on the long-tail to be
much-much more than the common mode input voltage.
Hmmmm... the max output is 20V peak.
At lowest gain, we need 8V peak at U1 grids.
And 25V is not much-much more than 8V.
Then a CCS or a deeper voltage is called for.
The whole "Frankenstein" is a perfectly simple
differential amplifier (U1)
plus a fancy cathode follower (U2).
You "can" see U1 as a cathode follower driving a grounded grid stage.
That's not as useful as understanding it as a Differential Amplifer.
But if you must: what is the difference between a Grounded Grid with a
hard-grounded grid, or a GG with a <10K resistor in the grid return?
For audio: very little. The grid is ~200Meg||~30pFd, so 10K is "nothing".
> where's the gain?
Mostly in the 1:10 input transformer.
U1 will have roughly half the 12AT7 handbook gain, say 20-25.
U2 has gain over 0.9. Then 2.5:1= 0.4 in the output transformer.
10*25*0.4= 100 overall.
NFB reduces the V1 gain from 20-25 toward a target of 11,
so probably 5 to 7, around 25 overall.
The 1:10 transformer is necessary for lowest noise,
because we've "stacked" the noise of two cathodes.
Gm of 12AT7 at 1.2mA is about 1,500uMho. Rk= ~660. Rn= ~2K.
For gain and noise calculations, Gm is effectively between
U1a grid and U1b grid, so Gm= 750uMho, Rk=1,300, Rn= 4K.
The source noise resistance is 200 ohms times 1:10^2 or 20K.
4K tube noise added to 20K source noise is "small",
but with a 1:5 transformer it would be ~3dB NF.
> Small value resistor and low voltage is a no-no.
The signal will be attenuated and S/N ratio will suffer.
Oh, it would not be good but not so bad.
If U1a's Rk is 1.3K, and we use an 11K cathode resistor,
attenuation is like 0.9 or -1dB.
The voltage across the long-tail only has to be much-much greater than
Vp/Mu, where Vp is around 125V and Mu is around 60.
So we need much-much more than 2V. 25V is much-much more than 2V.
We also want the voltage on the long-tail to be
much-much more than the common mode input voltage.
Hmmmm... the max output is 20V peak.
At lowest gain, we need 8V peak at U1 grids.
And 25V is not much-much more than 8V.
Then a CCS or a deeper voltage is called for.
skipwave
Well-known member
[quote author="PRR"]You "can" see U1 as a cathode follower driving a grounded grid stage.
That's not as useful as understanding it as a Differential Amplifer. [/quote]
Ah, thanks for helping me see the forest through the trees. I was trying to dissect and in the process missing how the limbs worked together to move the body. In this circuit, CF + GG = DiffAmp (or TubeOpamp). Imagining the little triangle symbol in place of V1 actually helped me understand, even though it made me feel kind of dirty. :wink:
That's not as useful as understanding it as a Differential Amplifer. [/quote]
Ah, thanks for helping me see the forest through the trees. I was trying to dissect and in the process missing how the limbs worked together to move the body. In this circuit, CF + GG = DiffAmp (or TubeOpamp). Imagining the little triangle symbol in place of V1 actually helped me understand, even though it made me feel kind of dirty. :wink:
