rongon - Instead of a mosfet source follower, how about a 15K:600ohm line transformer? Would a line transformer really be necessary if a plate choke was used, say 150H?
The transformer is not a replacement for a source follower. A MOSFET source follower is a unity gain buffer (high Z in, low Z out), but a transformer with a high Z primary and low Z secondary will be a stepdown device (it will reduce signal voltage from primary to secondary).
A 15k:600 ohm output transformer will introduce a voltage stepdown of 5:1, so will reduce the gain by 5X (1/5 or 0.2X). Since the '71A only has a gain of 2.5X to 3X, that stepdown will mean the final line stage will have gain of about 3/5 or 0.6X. (The source follower's gain will be about 0.95X, so the gain of the '71A will only be reduced from 3X down to 2.85X, which is not consequential.) Usually people will use a triode with higher mu (amplification factor) to get past the stepdown, but now we're replacing the tube you wanted to use in the first place.
Everything is a compromise.
A 15k:600 ohm output transformer will introduce a voltage stepdown of 5:1, so will reduce the gain by 5X (1/5 or 0.2X). Since the '71A only has a gain of 2.5X to 3X, that stepdown will mean the final line stage will have gain of about 3/5 or 0.6X. (The source follower's gain will be about 0.95X, so the gain of the '71A will only be reduced from 3X down to 2.85X, which is not consequential.) Usually people will use a triode with higher mu (amplification factor) to get past the stepdown, but now we're replacing the tube you wanted to use in the first place.
Everything is a compromise.
Double drats. I am trying to avoid any silicon, but maybe being too much of a purist. Any other alternatives to the mosfet source follower?
The '71A needs 5V 0.25A for its filament supply. I'll bet a typical 5V 2A SMPS wall-wart would work. You need the extra current for the inrush at power on, as the filament starts from cold. I don't know if that would sound better than this or that other way of supplying 5V DC for the filament...
I have a battery solution for the filament DC supply. Just need to solve for the output impedance mismatch with the 10K input impedance of the SS power amp.
Cathode follower.
Same thing as a source follower but using a tube. Doesn't work as well as a MOSFET, requires another heater supply, and you'll need to worry about the heater-to-cathode voltage at power on (which can poison the cathode of the follower over time, leading to premature tube failure).
You won't need a big honkin' tube to drive a cable and the power amp input. Even a 12AU7 or a 12AT7 would do fine. You could use a DHT as a cathode follower, but now things would be getting crazy.
I think you could try the '71A by itself and see if you like it. Its rp is low enough to be in the ballpark. Just not ideal. If you're looking to add tube juiciness to the sound, then that might be exactly what you're looking for.
Another way of getting low output impedance would be to configure the '71A as a mu follower.
https://www.valvewizard.co.uk/mufollower.html
That can also easily be made with a depletion mode MOSFET (e.g. DN2540) on top of the triode, like so:
![Gyrator-Output-impedance-1[1].jpeg](https://www.diyaudio.com/community/attachments/gyrator-output-impedance-1-1-jpeg.1334491/ "Gyrator-Output-impedance-1[1].jpeg")
https://www.valvewizard.co.uk/mufollower.html
That can also easily be made with a depletion mode MOSFET (e.g. DN2540) on top of the triode, like so:
***
Caution: I recommend 4 diodes across the preamps coupling cap to ground. A signal diode in series with one 5V zener, and another signal diode in series with another 5V zener, connect them so one limits + voltage, and one limits - voltage; +/- 5.6V.
If you do not do this, be prepared when that 1uF coupling cap sends 150V to the solid state amplifier input. OUCH!
The 1uF is essentially a dead short until it Charges at power-up, and is essentially a dead short until it Discharges at power down.
Post # 1, says the preamp has to drive a solid state amplifier with a 10k input impedance.
A 0.1 uF coupling cap will severely roll off the low frequencies; -3dB @ 159Hz.
So, use 1uF and 10k Ohms:
-3dB @ 15.9 Hz
-1dB @ 31.8Hz
My power mains voltage is 120VAC, it varies +/- 3VAC; +/- 2.5%
if your power mains are that good, you do not have to use the OD3. But you need a larger resistor to make B+ 150V that your deisgn calls out (150V, right or wrong, I did not design that stage).
If your power varies + / - 7%, like someone commented about his mains power, in a Tubes/Valves post, you should use the OD3.
The plate impedance, rp, is 1750 Ohms. That in parallel with the plate load resistor, is going to have to drive the 10k input impedance of the solid state amplifier.
Your schematics still do not show the plate load resistor from the plate to B+.
Yes, the signal does 'pass through' the MOSFET.
Attention to details works.
In-attention to details is a disaster waiting to happen.
Caution: I recommend 4 diodes across the preamps coupling cap to ground. A signal diode in series with one 5V zener, and another signal diode in series with another 5V zener, connect them so one limits + voltage, and one limits - voltage; +/- 5.6V.
If you do not do this, be prepared when that 1uF coupling cap sends 150V to the solid state amplifier input. OUCH!
The 1uF is essentially a dead short until it Charges at power-up, and is essentially a dead short until it Discharges at power down.
Post # 1, says the preamp has to drive a solid state amplifier with a 10k input impedance.
A 0.1 uF coupling cap will severely roll off the low frequencies; -3dB @ 159Hz.
So, use 1uF and 10k Ohms:
-3dB @ 15.9 Hz
-1dB @ 31.8Hz
My power mains voltage is 120VAC, it varies +/- 3VAC; +/- 2.5%
if your power mains are that good, you do not have to use the OD3. But you need a larger resistor to make B+ 150V that your deisgn calls out (150V, right or wrong, I did not design that stage).
If your power varies + / - 7%, like someone commented about his mains power, in a Tubes/Valves post, you should use the OD3.
The plate impedance, rp, is 1750 Ohms. That in parallel with the plate load resistor, is going to have to drive the 10k input impedance of the solid state amplifier.
Your schematics still do not show the plate load resistor from the plate to B+.
Yes, the signal does 'pass through' the MOSFET.
Attention to details works.
In-attention to details is a disaster waiting to happen.
Last edited:
The triode does the voltage amplification. The MOSFET is a constant current source (CCS), which provides a very high impedance to the plate of the triode.
Figure the impedance of the CCS is about 500k ohms.
That makes the load line almost flat, lying horizontally across the plate curves.
What that does is make the triode operate with maximum gain and linearity (lowest distortion). Not only that, but the output from the MOSFET source is very low impedance, so it won't load down driving reasonably short cables and the 10k input impedance of your amplifier.
I ran a simulation and got 1V rms out at 0.035% THD, all second harmonic, 3X gain. That's with no negative feedback (other than the intrinsic feedback of a triode and the source follower action of the plate load CCS). If you're looking for hi-fi, I think this would be the easiest way to go. It looks really good to me.
Another good thing is that the CCS only needs to drop 50V across itself. If the voltage at the 71A plate is about 110V, that means your B+ only needs to be 160V DC or so. You could do that with a single 0D3 VR tube.
Some people claim to dislike the sound of a mu follower made this way. I've tried it and I like it very much. I use it as the output stage of my phono preamp (a 6DJ8 with a MOSFET CCS plate load just like the one in the drawing in post 28).
I would make the output cap 2.2uF at minimum. If this was my project, I'd use a good quality 4.7uF polypropylene film cap there. A 10uF film cap would be a reasonable choice too, if you can get your hands on a couple.
On the other hand, if -1dB at about 32Hz is OK with you, then fine. The low bass will be lacking, though. Low E on a string bass or bass guitar is at about 40Hz, so that should play through with no attenuation. The only problem is that hum could possibly be picked up by the cable from the preamp with a 1uF output cap to the power amp. A larger cap would minimize that possibility.
Last edited:
I also simulated a 71A line amp with a plate choke instead of the MOSFET CCS.
Gain is lower, THD is higher but still very good.
The big difference is that with a plate choke, the output impedance rises at high frequencies, so the line stage would be very sensitive to cable capacitance. In other words, if you used it with a long cable run to the power amp, the highs would get rolled off.
The Zout looks to be about 2k ohms, which is pretty much the same as with a resistor plate load.
I went back to the Mu Follower version and loaded it with 5nF of capacitance on the output. The Mu Follower didn't roll off hardly at all.
The Mu Follower version's Zout is about 150 ohms, which is low enough that it won't have any problem working into a 10k ohm load.
This is what I've come up with:
6A3sUMMER raises some excellent points. I suggest heeding his advice.
Gain is lower, THD is higher but still very good.
The big difference is that with a plate choke, the output impedance rises at high frequencies, so the line stage would be very sensitive to cable capacitance. In other words, if you used it with a long cable run to the power amp, the highs would get rolled off.
The Zout looks to be about 2k ohms, which is pretty much the same as with a resistor plate load.
I went back to the Mu Follower version and loaded it with 5nF of capacitance on the output. The Mu Follower didn't roll off hardly at all.
The Mu Follower version's Zout is about 150 ohms, which is low enough that it won't have any problem working into a 10k ohm load.
This is what I've come up with:
6A3sUMMER raises some excellent points. I suggest heeding his advice.
Last edited:
rongon and 6a3sUMMER, thank you very much. very helpful.
The explanation of the MOSFET and horizontal load line is great. My B+ is 165V, so if the MOSFET drops 50V, this gives 115V which should be sufficient. Since I already have a Duelund 1uF cap, I will try it before potentially moving to a higher value. I listen to violin, cello and piano music. 27.5hz is the lowest on a piano. The 1uF cap is down 2db at 27hz, so I can live with this for now.
rongon (others are welcome to chime in) - my goal is to get DHT tonality / 3-dimentionality. Will the MOSFET counter what I am trying to achieve sonically? I do understand that the MOSFET will give lowest distortion, but will the DHT tonality still be prevalent?
The explanation of the MOSFET and horizontal load line is great. My B+ is 165V, so if the MOSFET drops 50V, this gives 115V which should be sufficient. Since I already have a Duelund 1uF cap, I will try it before potentially moving to a higher value. I listen to violin, cello and piano music. 27.5hz is the lowest on a piano. The 1uF cap is down 2db at 27hz, so I can live with this for now.
rongon (others are welcome to chime in) - my goal is to get DHT tonality / 3-dimentionality. Will the MOSFET counter what I am trying to achieve sonically? I do understand that the MOSFET will give lowest distortion, but will the DHT tonality still be prevalent?
Re: DHT tonality --
I sort of know what you mean, but this subject can get very controversial. Some people don't hear any advantage from DHTs at all, others will only listen to music amplified through DHTs. Blind ABX testing has shown that people can't reliably differentiate between the sound from a solid state amp and a tube amp (let alone an amp made with indirectly heated tubes vs a DHT amp), although I found one study that showed subjects could tell the difference between a 10 watt push-pull tube amp and a 100 watt solid state amp -- but again, that's controversial.
There is no way for anyone to know for certain what exactly you are trying to achieve.
I say build this thing the way you think it will sound best to you, and hear for yourself.
I'm writing the following out of genuine concern, at the risk of making myself sound like an obnoxious know-it-all:
Please be super careful with attention to construction and wiring details, especially AC wiring with the correct choices of fuses, safety earth connection, chassis construction, etc. Safety first and foremost! If you're unsure of something, please do ask for help. Lots of people around here will be happy to help.
I sort of know what you mean, but this subject can get very controversial. Some people don't hear any advantage from DHTs at all, others will only listen to music amplified through DHTs. Blind ABX testing has shown that people can't reliably differentiate between the sound from a solid state amp and a tube amp (let alone an amp made with indirectly heated tubes vs a DHT amp), although I found one study that showed subjects could tell the difference between a 10 watt push-pull tube amp and a 100 watt solid state amp -- but again, that's controversial.
There is no way for anyone to know for certain what exactly you are trying to achieve.
I say build this thing the way you think it will sound best to you, and hear for yourself.
I'm writing the following out of genuine concern, at the risk of making myself sound like an obnoxious know-it-all:
Please be super careful with attention to construction and wiring details, especially AC wiring with the correct choices of fuses, safety earth connection, chassis construction, etc. Safety first and foremost! If you're unsure of something, please do ask for help. Lots of people around here will be happy to help.
Yes, a cascode CCS is better.
But the single MOSFET CCS is easier to implement and will be very much good enough, as the rp of a 71A is low, about 1500 ohms. For a 71A, a plate load equivalent to 200k or so is pretty much as effective as a plate load equivalent to 2M ohms.
Single depletion mode MOSFET CCS:
Cascode CCS using depletion mode MOSFETs:
But the single MOSFET CCS is easier to implement and will be very much good enough, as the rp of a 71A is low, about 1500 ohms. For a 71A, a plate load equivalent to 200k or so is pretty much as effective as a plate load equivalent to 2M ohms.
Single depletion mode MOSFET CCS:
Cascode CCS using depletion mode MOSFETs:
To view the linearity of a triode tube view the plate curves. Many early triodes are extremely linear. View the 76 or 6P5 tube.
6SN7 is good
845 is good
But that is all before curve tracers. Many do not understand the old longhand ways.
Before the era of curve tracers, it took a lot of precision mirrored meters, lots of variable power supplies, lots of tubes, lots of paper, and then smart brains to average the reasonable tubes, and reject the "outsider" tubes, and a nice set of french curves.
Most of what it took was more hours of work than most people can imagine.
845 is good
But that is all before curve tracers. Many do not understand the old longhand ways.
Before the era of curve tracers, it took a lot of precision mirrored meters, lots of variable power supplies, lots of tubes, lots of paper, and then smart brains to average the reasonable tubes, and reject the "outsider" tubes, and a nice set of french curves.
Most of what it took was more hours of work than most people can imagine.
Here is the updated schematic. I made some changes based upon parts I already have. The coupling cap is 1uF (I can always upgrade to a larger value if the bass / low-end is noticeably insufficient). Notice the 12V battery fixed bias. Grid leak is changed to 72K (enough?).
Thoughts?
Thoughts?
Attachments
I'm not familiar with that way of applying fixed bias from a battery. I usually see it implemented like this (parts in the yellow box):
Note that R4 will need to be adjusted on test (AOT) for each MOSFET to obtain the desired plate current drawn through the 71A. You could put a 100 ohm trimpot in series with a 68 ohm resistor in place of R4, and adjust by measuring the voltage across that 68 ohm resistor (E/R =I). If you measure 1V across that 68 ohm resistor that means there's 14.7mA (0,0147A) drawn across it (1/68 = 0.0147).
Also note the addition of 'stopper' resistors before the 71A grid and after the DN2540 source.
The grid stopper on the 71A suppresses RFI pickup and possible oscillation.
The build out resistor in series after the DN2540 source inhibits oscillation when driving a capacitive load such as a long stretch of interconnect cable or an amplifier input with high input capacitance.
Those are just 'good practice' measures to stay out of trouble. 😉
--
Note that R4 will need to be adjusted on test (AOT) for each MOSFET to obtain the desired plate current drawn through the 71A. You could put a 100 ohm trimpot in series with a 68 ohm resistor in place of R4, and adjust by measuring the voltage across that 68 ohm resistor (E/R =I). If you measure 1V across that 68 ohm resistor that means there's 14.7mA (0,0147A) drawn across it (1/68 = 0.0147).
Also note the addition of 'stopper' resistors before the 71A grid and after the DN2540 source.
The grid stopper on the 71A suppresses RFI pickup and possible oscillation.
The build out resistor in series after the DN2540 source inhibits oscillation when driving a capacitive load such as a long stretch of interconnect cable or an amplifier input with high input capacitance.
Those are just 'good practice' measures to stay out of trouble. 😉
--
Last edited: