Drive Mr. Miller, drive…
Whilst designing triode driver stages for a 300B SE amplifier, I looked into the input capacitance of high-mu triode (or triode-strapped pentodes) that are suitable candidates here. We know Miller capacitance is the big drawback of the triode driver stage, so I kind of liked the idea of testing how much this capacitance could be for the lovely Russian triode-strapped pentodes like 6e5p, 6e6p, 6j11p, 6j5p, 6j52p, etc.
The test rig
I used my valve curve tracer to provide fixed bias to the DUT. The anode was loaded with a classic CCS to maximise gain close to μ. I used my waveform generator to drive the grid with a square wave signal from the low impedance (Zo=50Ω) output. I used a 1kΩ 1% series resistor to form the RC circuit with the valve input capacitance. We know that the easier way to measure the τ=RC is by looking at the transient from 0% to 100% which is 5τ. Therefore if we measure 5τ with the oscilloscope we can then derive capacitance value as C=τ / (5R).
What I firstly measured was the parasitic capacitance of my valve tester which includes internal wiring and valve socket. I found 5τ=1,000μs. So input capacitance is 200pF. This value is the one I then subtracted from all my measures when testing triodes in action.
Testing input capacitance
My first victim was the 6C45P triode:
The yellow trace is the grid voltage and the blue is the anode voltage. The valve was biased at about 150V/20mA. The gain is approximately μ=Va/Vg=45.2. The 5τ=3,080μs so input capacitance is 416pF (excluding socket and parasitic capacitance totalling 200pF). This is clearly a hefty load that our friend Mr. Miller is asking us to bear! If we consider that Ci=Cm/(μ+1) then the input capacitance of this triode is actually Ci=416pF/(46.2)=9pF. This is right on the minimum value of the datasheet.
Looking now on a strapped triode like the 6e6p-e, the total input capacitance @ Ia=20mA μ=33.9 is Cm=236pF. The input capacitance is therefore Ci=6.76pF. This is lower from the Ci value published on the datasheet which is only reflecting tetrode operation, not triode.
I found this test very insightful and keen to test other triode-strapped pentode valves to see how they perform under Miller’s kingdom. Naturally if we could remove Miller from his kingdom by fixing the anode voltage like in a folded cascode or with a gyrator load, then the preamp will have a much easier life. This could determine whether your amp could lose a lot of treble due to the lack of driving capability of your source (or preamp). Worth considering when designing or setting up your system.
Hi Ale,
Unfortunately I don’t have a waveform signal generator. Is there another way to measure C miller? Did you by chance had the chance to measure 4P1L’s in triode mode? From the data sheet i can’t figure it out. I am assuming it is somewhere at 80-100pF ( comparing with other pentodes datasheets that provide Cag1 for triode connection, for example D3a has Cag1 of 0.035pF and triode connected is 2.7pF, which is a factor of 77. Therefore 4p1l with Cag of 0.1pF would have about 8 pF triode connected or Cmiller=8×9+8.9+stray=about 90pF)
i am trying to design a headphone amp with 4p1l final stage and need to determine the driver’s minimum requirements.
Thanks.
Best,
Radu
Hi Radu, if you can build a square-wave oscillator, that will do.
If my notes are correct, the total input capacitance I measured on a 4P1L in triode mode was about 176pF. With a mu of 9.3 in this DUT, let’s say 9 average then with a Cgk=8.5pF then Cag is about 16.7pF.
cheers
Ale
Thanks Ale. I was hoping for a lower value
Can you explain more about how you measure. I am nor sure my Rigol DS1102E has all the features I need in order to measure T. Do I need a square wave at about 300kHz?
Thank you so much for being always helpful.
Best,
Radu
PS. Now I realize that having the driver 4P1l in my PSE amp at 36mA was a good choice. The input for PSE would be about 350pF and multiplied by 3 (IT 2:3.5) is about 1nF. This would be about 1.75mArms at 20Khz.
I have same oscilloscope. A 1kHz square wave oscillator will do fine. Plenty of circuits out there. You will probably need a good low impedance output that can drive the 300-600pF of load you will measure.
Have a look at the recent post I added on the 4P1L driver. Running it at 36mA was a good choice to manage the capacitance as you well say.
Cheers, Ale
If C = t/(5R) and your valve tester gave a rise time of 1000us, and R = 1k, then C = 200nF, no?
Is not the rise time, is the total charge time. Rise time in electronics is used to measure voltage change from x% to y%. The total charge time in the RC circuit is 5*tau, which is easy to measure with the scope.
Here is some more information about this:
http://www.electronics-tutorials.ws/rc/rc_3.html
hope this helps
Ale