It seems there have been a number of posts lately asking about adding a line stage/preamp to a system so that a tube may be added to the signal chain.
I'm always left asking why a line stage/preamp would even be necessary, since most 'power' amplifiers these days have plenty enough gain to be driven to full power and beyond by any digital source, and even most phono preamps these days provide >40dB gain so can develop 1V peak output from most any MM cartridge.
One place where I've found I'd like just a little bit of gain is for getting the subjective level of my phono setup to match that of my digital sources. Making a line stage with 3X (9dB) gain and low output impedance is not as easy as it sounds. So I've been experimenting...
I came up with a circuit that provides:
- 3X (9dB) gain
- super low THD (0.0003% at 1kHz, 1V rms out)
- low output impedance (about 250 ohms)
- uses a 12AX7 or 6N2P, or with a couple of parts value changes can use 6N1P, 6BQ7, 12AT7.
However, it's not a 'pure tube, zero negative feedback' type of thing. It uses:
- a MOSFET source follower
- local shunt voltage feedback
The triode has its plate load bootstrapped as described by MerlinB in his preamps book.
I simulated the circuit with three different 12AX7 models and a 6N2P model, all with nearly identical results, even for THD.
I wonder if this would end up sounding indistinguishable from an op-amp line stage set for just a few dB of gain.
Does anyone see any fouls, gaffs, mistaken assumptions made, other errors?
I'm always left asking why a line stage/preamp would even be necessary, since most 'power' amplifiers these days have plenty enough gain to be driven to full power and beyond by any digital source, and even most phono preamps these days provide >40dB gain so can develop 1V peak output from most any MM cartridge.
One place where I've found I'd like just a little bit of gain is for getting the subjective level of my phono setup to match that of my digital sources. Making a line stage with 3X (9dB) gain and low output impedance is not as easy as it sounds. So I've been experimenting...
I came up with a circuit that provides:
- 3X (9dB) gain
- super low THD (0.0003% at 1kHz, 1V rms out)
- low output impedance (about 250 ohms)
- uses a 12AX7 or 6N2P, or with a couple of parts value changes can use 6N1P, 6BQ7, 12AT7.
However, it's not a 'pure tube, zero negative feedback' type of thing. It uses:
- a MOSFET source follower
- local shunt voltage feedback
The triode has its plate load bootstrapped as described by MerlinB in his preamps book.
I simulated the circuit with three different 12AX7 models and a 6N2P model, all with nearly identical results, even for THD.
I wonder if this would end up sounding indistinguishable from an op-amp line stage set for just a few dB of gain.
Does anyone see any fouls, gaffs, mistaken assumptions made, other errors?
Attachments
Pete Millet made an AB1 buffer with a MOSFET and a CCS (for example 10M90S) instead of the 33K R9 resistor.
Regards, Gerrit
Regards, Gerrit
Remove C4 to reduce the gain.
Z out is surely 33k because that is the value of the load resistor that will govern the output impedance.
Z out is surely 33k because that is the value of the load resistor that will govern the output impedance.
Last edited:
See link below:
[Solved] Output Impedance of a source follower is:
The output impedance of a follower isn't very dependent on the source resistor (in this case). If gm is fairly large, the output impedance can be approximated as 1/gm.
[Solved] Output Impedance of a source follower is:
The output impedance of a follower isn't very dependent on the source resistor (in this case). If gm is fairly large, the output impedance can be approximated as 1/gm.
Yes, you could do that. It would improve the performance slightly.
I'm not sure what you mean by "an AB1 buffer". Class AB1 is a push-pull amplifier class where at higher voltage swings one output tube goes into cutoff while the other conducts.
33k is the load resistor of the MOSFET source follower (R9). The output impedance of the source follower is approximately 1/gm. Since the MOSFET has very high gm even with only 6mA Ids, output impedance is very low.
This is a feedback amplifier. The voltage divider R1 and R2 dictate the gain of the amplifier, in this case just shy of 3X (about 9dB).
The gain of U1 (the 12AX7) drives the feedback loop, so you want more open loop gain, not less. You could remove C4 to degenerate the 12AX7, which would normally reduce its gain and increase its output impedance. However, in this case that won't do much because the plate load of U1 is bootstrapped — the output is sent back via C3 to a tap midway between R4 and R5.
Since M1 is a source follower with gain of almost 1 (unity), the signal voltage at the junction of R4 and R5 is almost exactly the same as the output signal voltage from M1's source. That means there is no signal voltage dropped (or swung) across R5, which maintains a constant AC current across it. Thus R5's impedance to AC signals is very high. It's sort of a sideways mu-follower. U1 sees an extremely high impedance at its plate, flattening its load line and increasing its gain.
The source follower is driving both the load impedance of the driven device and the negative feedback voltage divider R1 and R2. R1 could be made to be a larger value, making for a lighter load on M1, but that would interact with the Miller effect on U1's grid, causing a bit of unwanted attenuation of frequencies above about 18kHz.
See Merlin Blencowe "Designing High Fidelity Tube Preamps" pg. 268, figs. 7.23, 7.24.
I'm sure you'll have fun with this design!
I don't believe in preamps these days - adding an unnecessary gain stage can only reduce transparency. I would think of it as the last part of your phono stage. That makes sense since you need some gain there. Adding gain to the line inputs makes no sense to me.
I don't believe in preamps these days - adding an unnecessary gain stage can only reduce transparency. I would think of it as the last part of your phono stage. That makes sense since you need some gain there. Adding gain to the line inputs makes no sense to me.
andyjevans said:
Hey Andy -- That's exactly what I was thinking! I was thinking of using something like this as the output stage on a phono preamp. Maybe use a 6DJ8 common cathode 1st stage > RIAA EQ > 6DJ8 common cathode 2nd stage > bootstrapped pair anode follower. (6DJ8 as an input triode has low enough Cmiller that AT MM carts wouldn't get that nasty bump at around 10kHz or so...)
A phono preamp using two 6DJ8s and an all-in-one-go RIAA EQ would yield gain of about 34dB (about 50X). I'd like to get 46dB (200X) or so gain for MM carts in my system, so I'd be looking to add an output stage with about 12dB (4X) gain, with very low THD and with low output impedance. Hence my interest in anode followers.
Also, the gain would be adjustable simply by increasing the value of R2. I could theoretically get to 60dB of gain by changing R2 to 22*R1 (3.3M). I doubt it would be quiet enough for most MC carts straight in, though. But it would make for a fun experiment.
Last edited:
That would work great, I'm sure. It would not have super-ultra-low THD, and it would not have quite as low output impedance as a source follower buffered version.
I've built something very much like that with a 6DJ8/ECC88. That was my first test of this kind of circuit. It has excellent PSRR, you can't hear any hum at all from it, even with AC heaters in a tiny chassis. The B+ is only 160VDC, and I only put 3mA plate current through the 6DJ8, so it's not optimal. But it sounds nice as a sort of 'tube-ifier'. It sounded great after my Hagerman Bugle preamp. Brought some 'life' to the sound. Yeah, it's just adding THD, but I like it subjectively. It makes records sound more fun.
This 12AX7 > source follower version is more optimized. Maybe too optimized.
PS - I have a question -- In your schematic, you have a grid stopper R402 on the ECC99. Would you be able to remove that if R401 was soldered directly to the grid pin on the tube socket? R401 would then be acting as a high value grid stopper, would it not?
--
Last edited:
Zout= 500, also low. Yes you can bypass R402. I am using alltimes a gridstopper. This stage does not sound as a tubisator. It sounds very clean.
The bootstrapping acts like a CCS load, so C4 can likely be removed because it's not really doing anything. Removing it will hardly affect the gain I think.
Thinking some more, C4 will have an affect at very low frequencies where the bootstrapping is less effective, although I still think it could be removed.
Removing C4 wouldn't affect the gain much at all, but it would increase the rp of U1, which could affect high frequency response just within the audio band if using a 12AX7 for U1. The higher rp of the degenerated 12AX7 would interact with the Cin of the source follower. You could use a ZVN5045A or some other super-low Crss MOSFET for M1 and then removing C4 would not matter at all.
If using a 12AT7 for U1 then all this should not matter since 12AT7 has 1/4 the rp of 12AX7.
You could also use a cheap red LED for R6 and leave C4 out.
Edit to add:
In simulation, removing C4 did cause a very slight attenuation of response at 20kHz, going from -0.04dB with C4 in circuit to -0.1dB with C4 removed. I'd call that negligible.
Last edited:
R3 is not essential. The grid can leak through R2 and R10. There's no particular limit on 12AX7 gridleak, anyway 1.5Meg is not a problem.
Thanks PRR, I see what you mean.
Here's the same circuit but using 12AT7. Less open loop gain, so higher closed loop THD, but still ridiculously low. R10 is now 470k since 12AT7 has more of a tendency for grid current than 12AX7.
12AT7 is cheaper than 12AX7, so that's a plus.
The same circuit could be used for 12AV7, which is very cheap. Lower open loop gain, so higher closed loop THD, but still low.
I also simulated the circuit with 6N1P and it came out looking very good.
Lots of options.
Here's the same circuit but using 12AT7. Less open loop gain, so higher closed loop THD, but still ridiculously low. R10 is now 470k since 12AT7 has more of a tendency for grid current than 12AX7.
12AT7 is cheaper than 12AX7, so that's a plus.
The same circuit could be used for 12AV7, which is very cheap. Lower open loop gain, so higher closed loop THD, but still low.
I also simulated the circuit with 6N1P and it came out looking very good.
Lots of options.
Attachments
Last edited:
I simulated the same basic circuit with 12AX7 or 12AY7 and a cathode follower using 12AU7 or even ECC88. They all had excellent results in simulation even with lower than optimal B+. It seems like a perfect circuit for a utility preamp. With the ECC88 or a MOSFET you could even drive two outputs. 12AU7 might struggle with two low impedance amps.
If you made the load on both stages roughly equal you would heave near-constant PSU current too... Personally though I am not keen on bootstrapping in a line driver, since the feedback fraction becomes dependant on the external load which is unknown.
Thanks Merlin, that's a good point!
That argues for as much open loop gain as you can get, right? Or do you mean the closed loop gain will change with dynamically changing load conditions, which would make this a variable gain amplifier?
However, what if the load is always going to be rather light/benign, for instance, a 100k ohm or 50k ohm volume pot, with maybe a half-meter of interconnect cable?
My intended use here is to add 9dB of gain to a phono preamp, to bring that up to the same subjective level as my DAC's output. This line stage would go between the output of the phono preamp and the input of my 'passive preamp' (input switch and Intact Audio autoformer volume control).
Is it a better solution to use a 12AT7 mu-follower (DN2540 plate load) DC-coupled to a MOSFET source follower, with NFB to the 12AT7 grid to reduce the gain to 9dB?
Merlinb said:
That argues for as much open loop gain as you can get, right? Or do you mean the closed loop gain will change with dynamically changing load conditions, which would make this a variable gain amplifier?
However, what if the load is always going to be rather light/benign, for instance, a 100k ohm or 50k ohm volume pot, with maybe a half-meter of interconnect cable?
My intended use here is to add 9dB of gain to a phono preamp, to bring that up to the same subjective level as my DAC's output. This line stage would go between the output of the phono preamp and the input of my 'passive preamp' (input switch and Intact Audio autoformer volume control).
Is it a better solution to use a 12AT7 mu-follower (DN2540 plate load) DC-coupled to a MOSFET source follower, with NFB to the 12AT7 grid to reduce the gain to 9dB?
Last edited:
Sort of, I'm more bothered by stability with unknown loads. You'll probably never have a problem, I'm just a fan of simple buffers that do as they're told.
Personally I think if you're a tube fan, stick to tubes! Gain stage, tube CF, loop feedback. Classic and simple. But if you like sand...