Hi, I have recently been building a small amplifier using 6S7B tubes, and to understand better what's happening I have tested the input stage in isolation, with a load roughly the same as the following stage presents. Finding ra from datasheet come out at 22k. So I make Ra=2*ra, 47k nearest value. No cathode bypass capacitor is used. I understand that degenerating the cathode reduces gain, and causes the seen ra to increase to ra'.
What is puzzling me is that when I increase Ra, then THD increases (which I felt was the correct thing to do, please correct me if not) This seemed like the opposite if what I expected. In fact reducing Ra to 33k (which is less than 2*ra) improves THD.
Any help understanding what is happening, would be great!
What is puzzling me is that when I increase Ra, then THD increases (which I felt was the correct thing to do, please correct me if not) This seemed like the opposite if what I expected. In fact reducing Ra to 33k (which is less than 2*ra) improves THD.
Any help understanding what is happening, would be great!
I agree with you, increasing Ra should indeed decrease THD.
Only changing Ra does, however, also change the operation point of the valve and could put it into a non linear region. This will probably be seen at some "extreme" operation, like the output signal coming close to clipping.
So, at what output level did you test the THD? How much gain did the stage realize (with both loads). Can you share a (raw) schematic with actual DC voltages of the circuit?
Only changing Ra does, however, also change the operation point of the valve and could put it into a non linear region. This will probably be seen at some "extreme" operation, like the output signal coming close to clipping.
So, at what output level did you test the THD? How much gain did the stage realize (with both loads). Can you share a (raw) schematic with actual DC voltages of the circuit?
Last edited:
But would larger Ra not also increase signal level at the anode, which means larger excursion on the transfer curve, which is non-linear, thus more THD?
Jan
Thanks all! I'm glad I understand some of what I have been reading! DF96 as it stands at the moment, I only have a degenerated cathode, no plate to grid local FB is used. The triode has mu of around 60, Vi is 1V p-p (0.353Vrms). This is my and signal level, and remains constant. I've just realised I have taken gain and miscalculated (wrote peak volts in place of p-p....
Oops. Erik, YoU make a good point I had neglected to pay enough attention to! I had adjusted bias without considering Vgk (I attempted to keep constant Vgk rather than keep the Iq constant and alter Vgk) I need to 're check everything now....
Oops. Erik, YoU make a good point I had neglected to pay enough attention to! I had adjusted bias without considering Vgk (I attempted to keep constant Vgk rather than keep the Iq constant and alter Vgk) I need to 're check everything now....
So I have taken out the cathode resistor and replaced with 100 R and 2k potentiometer wired as variable R.
Stupid award goes to me
I made a mistake.
In my haste, and despite attempting to write accurate easily followed notes, when reading through I failed to notice the note Plate to grid FB from follower. (Written in red biro as well...)
And to top that off, I was taking the signal from the following stage. In that case I measured something like 0.1 - 0.15% THD
I couldn't understand why I couldn't attain the THD performance I had previously...
So thank you.
Now I try and maintain roughly 2/3 B+ at the anode, adjusting cathode resistance and leaving to stabilise OP and searching for the lowest THD.
Ra=33k, 47k and 68k all set roughly 2/3rds B+ with Rk= 450R, 1k, and 1k4 respectively. The first case is the worst performer, but it's only in the 3rd D.P. of percentage result. Gain is about 11, 12 and 13 times V/V respectively also.
All THD measurements give 0.25% Total THD, with 2nd and 4th harmonics being highest in level, while 3rd and 5th remain under -40dBm. Input signal is between +26.5 to +28.5dBm.
It's only an 8 bit DSO, and I am only trying to measure F1 to F5 amplitude, so maybe I'm just running into the limitations of its dynamic range.
I did run a FFT on the signal generator, set to 50 Ohm Output, DSO set DC coupled 50 Ohm, straight in with BNC, as an initial test.
I'm sure the level was lower than 0.25% but I think I need to verify that when I can.
I'm considering getting a picoscope...probably the cheapo 2204 10MHz as the software looks better than this scope
Stupid award goes to me
I made a mistake.
In my haste, and despite attempting to write accurate easily followed notes, when reading through I failed to notice the note Plate to grid FB from follower. (Written in red biro as well...)
And to top that off, I was taking the signal from the following stage. In that case I measured something like 0.1 - 0.15% THD
I couldn't understand why I couldn't attain the THD performance I had previously...
So thank you.
Now I try and maintain roughly 2/3 B+ at the anode, adjusting cathode resistance and leaving to stabilise OP and searching for the lowest THD.
Ra=33k, 47k and 68k all set roughly 2/3rds B+ with Rk= 450R, 1k, and 1k4 respectively. The first case is the worst performer, but it's only in the 3rd D.P. of percentage result. Gain is about 11, 12 and 13 times V/V respectively also.
All THD measurements give 0.25% Total THD, with 2nd and 4th harmonics being highest in level, while 3rd and 5th remain under -40dBm. Input signal is between +26.5 to +28.5dBm.
It's only an 8 bit DSO, and I am only trying to measure F1 to F5 amplitude, so maybe I'm just running into the limitations of its dynamic range.
I did run a FFT on the signal generator, set to 50 Ohm Output, DSO set DC coupled 50 Ohm, straight in with BNC, as an initial test.
I'm sure the level was lower than 0.25% but I think I need to verify that when I can.
I'm considering getting a picoscope...probably the cheapo 2204 10MHz as the software looks better than this scope
Last edited:
monodogenerator,
+26.5 to +28.5dBm driving what impedance?
A 50 Ohm generator needs the load to be 50 Ohms if you want the dBM setting to be correct.
After you get the unbypassed cathode results, you need to do the same thing with the cathode bypassed.
Compare the distortion for both unbypassed and bypassed circuits (but be sure to do that with the same plate signal volts). In that case, you will have to reduce the drive voltage to the bypassed circuit, to get the same plate signal volts (because of the increased stage gain when it is bypassed).
Apples to Apples.
+26.5 to +28.5dBm driving what impedance?
A 50 Ohm generator needs the load to be 50 Ohms if you want the dBM setting to be correct.
After you get the unbypassed cathode results, you need to do the same thing with the cathode bypassed.
Compare the distortion for both unbypassed and bypassed circuits (but be sure to do that with the same plate signal volts). In that case, you will have to reduce the drive voltage to the bypassed circuit, to get the same plate signal volts (because of the increased stage gain when it is bypassed).
Apples to Apples.
I guess from your moniker you play with higher voltages.
*At what bias point??*
Ah. Yes, 2/3 is a fine first-guess at a happy bias, and triodes are VERY tolerant of >2:1 changes of resistors, as long as you don't cramp them into a corner.
8 bits is 256. It is like measuring a 256 meter high building with a 1 meter stick and NOT being able to discern smaller differences. 1/256 is 0.004 which is 0.4%. I suspect that your 0.25% readings are just rounding error.
I think I wasn't clear, the signal generator can be set for high Z or 50Ohm. During testing the triode both signal generator and scope are set on high Z. During the control measurement of signal generator I set both devices to 50 Ohm since I was coupling directly 1:1 using BNC cable. The dBm measurement is read from the DSO, signal generator output set to volts. AC coupling used where appropriate (floating gnds or no gnd ref), DC coupling in all other measurements.
Bypassing the cathode, in all cases causes an increase in THD of two fold, mostly 3rd and 5th harmonics. (Without bypass, harmonic structure is all 2nd and 4th) I think I had a plate to grid FB from 2nd stage to 1st (follower output to input tube grid) but my notes or my reading of them was inadequate. So when I measured 0.15% THD and could never attain the same result I began to wonder.
Turns out a couple of things:
DSO needs at least 60 minutes warm up time before measurements are reliable when at the limits of DNR and noise. I hadn't given sufficient warm up time on occasion, leading to less repeatability. DSO does have ETS mode which I think boosts bit depth to 12 bits (or a 'virtual' 12 bits)
I think the practical limit of resolution is somewhere between 0.1 and 0.3% THD using this signal generator (its an Agilent but not necessarily a device designed for the absolute minimum THD output) The signal generator also needs a good hour to warn up and stabilise for the lowest THD output.
The tube circuit itself needs probably 10 minutes to stabilise (I guess temps, bypass cap currents, resistor TCs) but longer than that is better (waiting the longest for DC charge to ebb away from coupling caps after HT is applied to the circuit) However, IF the practical limit of measurement resolution is 0.25% and I can get repeatability then precision is ok. But what about the case when I measure 0.1% THD?(which I suspect is below the DSO limits of resolution)
Aside
I was pleasantly surprised at the THD measurement of 0.25%, averaged over 3 or more test runs, with enough warm up time, it is at least repeatable. FFT windowing options are few: VonHann, Flat top and Rectangular - Flat top seems the most appropriate for amplitude. I read off the harmonic dBm at 500Hz/div but it makes little difference using 100Hz/div, just more detents for each FFT bin when turning the digital pot. The peak value I read in either X axis scale are the same, just less detail of values either side of the peak. So that's either the limit is measurement, or the limit of THD performance (that I can achieve) for this Triode in a simple cathode degenerated circuit. Perhaps even both!
I hadn't paid this enough thought but it is true enough, or at least I measure marginally higher fundamental (say 32dBm vs 28dBm) with higher anode loads, gain is greater, hence volt swing, but I find in the limited range I've tested that thus is accompanied with higher levels of THD, and each case ends up very similar albeit with different gain.
0.25% at around 17V is very good performance. Almost stunningly good.
With ordinary mass-market triodes, and assuming 300V supply, I would expect 40V at 5%, so 17V at 2%. Unbypassed cathode cuts that in half, 1%. Some tubes are more linear, so your 0.25% may be true for your tube and bias. Still very good performance.
Hi PRR,
If memory serves I got a measured output of between 8.5 and 10V the depending the anode load, on the last series of tests I performed.
At the outset I thought 1% THD would be nice to achieve, so I was happy with a better result than that.
THD and audibility HAS been debated here time and time again, and to be honest, it still feels like "THD? How low can you go?", and I never really got a feel for what level is audible. So it ends up a race to the bottom (like all things it seems)
So, my train of thought, if 0.25% is very good performance (and I trust your opinion more than mine at this point),
Then that leads me to doubt the validity of my simple scope measurements, at least a little.
I'm thinking about a picoscope 2200 (of 3000 series if I save up) as the DSO is my work scope and im using lunchtimes to run tests (warming up circuit and instruments im the morning)
I can just about squeeze in 3 repeats to adjust bias, leave to settle for 10 then take an FFT.
Getting a USB scope would allow me to test for longer periods at home- and I can confirm just how accurate or inaccurate my current measurements are.
Then if I measure 0.25% I can smile!
If memory serves I got a measured output of between 8.5 and 10V the depending the anode load, on the last series of tests I performed.
At the outset I thought 1% THD would be nice to achieve, so I was happy with a better result than that.
THD and audibility HAS been debated here time and time again, and to be honest, it still feels like "THD? How low can you go?", and I never really got a feel for what level is audible. So it ends up a race to the bottom (like all things it seems)
So, my train of thought, if 0.25% is very good performance (and I trust your opinion more than mine at this point),
Then that leads me to doubt the validity of my simple scope measurements, at least a little.
I'm thinking about a picoscope 2200 (of 3000 series if I save up) as the DSO is my work scope and im using lunchtimes to run tests (warming up circuit and instruments im the morning)
I can just about squeeze in 3 repeats to adjust bias, leave to settle for 10 then take an FFT.
Getting a USB scope would allow me to test for longer periods at home- and I can confirm just how accurate or inaccurate my current measurements are.
Then if I measure 0.25% I can smile!
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.