Re: 6SN7 Mu Follower Distortion Paper

Patrick Turner wrote:

Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
I have written up the work I did on investigating distortion in a 6SN7
mu follower and posted it at my website. You can read it here:

http://www.ianbell.ukfsn.org/data/mufollowerdistortion.pdf

Cheers

Ian
Its not clear how you would go about biasing the bottom triode with Eg =
-3V while keeping Ia at 8 mA.

Sorry, I don't see the problem - a 430R cathode resistor sits you pretty
much on that spot.

Anyway, maybe you'd get even lower THD if you had Ia = 4mA, and bias at
about -5V for both top and bottom tubes with identical Rk used to
develop the bias voltage. This way the working Ea for top and bottom
triodes will be equal and each about 137Vdc.
I already tried you idea of running both triodes at 5mA and -5V bias and
I get identical distortion figures. The 6CG7 seems pretty relaxed about
operating point.

You'll get

If you have Idc = 4mA, Ea = 137, the load line for the bottom triode
which "sees" a load approx 200k shows that the negative going anode
swing is from 137V down about 85pkV before grid current so the max Vrms
output should be in excess of 58Vrms, a lot more than what you have
described.

That is not surprising - if you raise the grid bias from -3.4V to -5V
you would expect higher output before the onset of grid bias distortion.
However, the lower standing current lowers the drive capability of the
output.

But your RL = 100k, and you don't need a huge Ia dc to get a wide Vswing
at the OP.

If you had RL = 25k, then sure, more Ia is better but the more you raise
Ia, the less swing you get...

I am sure you read my paper right through so I am sure you noticed I did the initial tests with 100K but also looked at how it fared into 25K and 10K loads.


I am not quite sure what point you are making. I did the tests at 8mA
and around 3V bias - grid current occurred at the expected input level.

The greater Iadc + lower Ea + low bias = low Vo max before THD exceeds
2%.


Yes but greater Ia does not necessarily mean lower Ae and bias.


Morgan Jones reckons the 6SN7 exhibits its lowest distortion at 8mA. He
does not qualify that statement though.

Perhaps it does. But say you had Ea = 200V, Ia = 8mA, and you draw a
load line for 200k.
The load is nearly a CCS or horizontal loadline and you get a much wider
Vswing without THD than if you allow Ea to be say 125V only.


About 54Vrms should be able to be applied to the 100k RL and the load
current change is a small 0.54mA rms.

Its awhile since I did experiments on µ-followers, but generally THD can
be a lot less than 1/2 what you'd get if the bottom triode had only a
load R to replace the top triode acting as a follower and buffer between
the bottom triode and the load. In fact, if the top triode was
abolished, the RLdc required would be 47k, and that in parallel with the
100k load gives a final RL = 32k. Expect THD to be 3 times what you have
with an active load as you show in your schematic which is approximately
10k x top triode gain of 18.5 = 185k which is about 20Ra.

Some people use a j-fet CCS to replace your R1 10k. This makes the
bottom triode see a very high value ac load and thus THD is about
reduced to near the minimum that is possible to get.

For measuring noise in such a circuit, the bottom grid should be taken
to 0V via a short lead.
That is exactly what I did.

This means the noise of R4 120k isn't injected
into the tube. Such resistor noise would be much higher than what the
grid input noise should be. You should have only about 2uV grid input
noise, and this is amplified about 18 times to become 36uV at the bottom
anode. DC must be used on heaters to achieve this. Additional noise is
generated in the C1 & R3 circuit because as F lowers, the LF noise of
R3 is applied to the top triode follower. I think 0.47uF plus 220k might
give better results for noise.
But its not a huge deal because the top triode should not contribute
much noise as it is a follower with gain just under 1.0.

Noise is generated in R1, 10k, and one has to be careful about leakage
currents between cathode and heaters if the dc difference is high.

All up, anything less than 50uV at the output would be OK.
That would be nice but as you can see from the measurements I have yet
to see better than 65uV and most tubes are in the 90 to 100uV region.

If that is the case, then the grid input noise is about what you are
getting at the anode divided by the gain,
so there's about 3uV to 5 uV of input noise, ( dc to 20kHz ) Good
samples of 6CG7 will yield 2uV.


Which reminds me, I would be interested to know exactly what set up do you use to measure noise.


Allowing BW to be wider results in more noise.

I like the µ-foll because its inherently low THD/IMD and it does the
business well in a preamp stage of any kind; ie, it has high Z in, low Z
out and wide BW and noise which is just want you want a voltage amp to
have. It also sounds very nice.

And its efficient, because rather than waste watts of power in anode and
cathode feed resistors with a traditional gain stage and CF, you are
able to series tubes and save 50% or otherwise wasted power in
resistors.


I agree. I like them too.

Cheers

Ian
.

Relevant Pages

  • Re: NFB101 Part 3
    ... There is no anode load. ... The Jolida 502 amp has a SRPP input stage using a 12AX7 with 470 ohms Rk ... bottom triode and 1k for the top triode. ...
    (rec.audio.tubes)
  • Re: NFB101 Part 3
    ... There is no anode load. ... bottom triode and 1k for the top triode. ... Expect gain to be approx 19. ...
    (rec.audio.tubes)
  • Re: 6SN7 Mu Follower Distortion Paper
    ... mu follower and posted it at my website. ... This way the working Ea for top and bottom ... load R to replace the top triode acting as a follower and buffer between ... For measuring noise in such a circuit, the bottom grid should be taken ...
    (rec.audio.tubes)
  • Re: SRPP
    ... The SRPP has equal Rk top and bottom and top grid is directly coupled to ... carrying load. ... titled Tube Power Buffers and he makes the point that using the tube ... There cannot ever be exactly equal signal currents in each triode, ...
    (rec.audio.tubes)
  • Re: 6SN7 Mu Follower Distortion Paper
    ... Its not clear how you would go about biasing the bottom triode with Eg = ... swing is from 137V down about 85pkV before grid current so the max Vrms ... For measuring noise in such a circuit, the bottom grid should be taken ...
    (rec.audio.tubes)
Loading