Current sampling of the OPT primary can be done (perfectly for triodes in class A1) with a small unbypassed resistor in the cathode path to signal ground. Resistive feedback from here to the preceding stage's cathode is positive feedback, so reduces output stage impedance, potentially slightly below zero. The trick is make the negative output impedance equal to OPT winding resistance. Might want to include a trimpot to keep things on the up and up.
All good fortune,
Chris
All good fortune,
Chris
rdf,
Thanks for letting me know how bad the Fisher 500C and 800C were.
I simply ask that anybody who has one of those 2 models, please contact me, and send me the output transformers.
I have built several amplifiers using that output transformer; and so have a number of other club members.
I can always use those output transformers in future projects. Thanks!
Thanks for letting me know how bad the Fisher 500C and 800C were.
I simply ask that anybody who has one of those 2 models, please contact me, and send me the output transformers.
I have built several amplifiers using that output transformer; and so have a number of other club members.
I can always use those output transformers in future projects. Thanks!
This thread is exploring the idea of no-feedback pentode amplifier as a simple way of speaker current drive. So, no-NFB is a feature here, not a bug. I am attempting to make a no-NFB pentode amplifier with best linearity, with distortion levels close to those of good triode amplifiers.
Yet some advise to the effect that no-feedback is bad, that feedback is always good, and it must be used no matter what, because not using it is bad engineering and religion.
Am I supposed to make a turnabout and end up with an Eico or Scott kind of amplifier? I believe no, because that would be a) boring, and, b) a deviation from the focus of this thread.
Yet some advise to the effect that no-feedback is bad, that feedback is always good, and it must be used no matter what, because not using it is bad engineering and religion.
Am I supposed to make a turnabout and end up with an Eico or Scott kind of amplifier? I believe no, because that would be a) boring, and, b) a deviation from the focus of this thread.
Many things in life are hard, but that's not the issue. The issue is that speaker suspension is a spring, and no matter how hard you try, you cannot make it otherwise. And as a spring, speaker suspension has deflection-dependent force changes, a source of nonlinearity
How is that? It is simple physics that two drivers at a distance of 1/2 lambda or less are 3 dB more efficient than single driver, with same driving power. The same is true for a single driver with twice the radiating area.
Certainly. But saturation is asymptotic, so we are talking about the area of the B-H curve where large changes in magnetizing current cause very small changes in magnetization, and, by extension, in residual pole piece magnetic permeability.
sser2,
Do not get discouraged. You will find something to fit the bill.
I wanted to build a two stage push pull amplifier that uses Beam Power output tubes (essentially the same as using Pentode output tubes).
And, it had to have no NFB; not local, not global.
So, I built one. But it could use lots of improvement.
Very limited performance specifications:
No NFB
Using a non-inductive 8 Ohm load:
2.6 Watts maximum before visible waveform distortion
At 1 Watt, 2nd harmonic distortion -58dBc (0.13%); 3rd harmonic distortion -52 dBc (0.25%)
Frequency response, 1kHz 0 dB, 20kHz -0.3 dB, 40 Hz - 0.9dB, 21.5Hz - 2dB
Damping factor 0.6, not a very good current source, but at least it is less than Unity. (Output impedance of the 8 Ohm tap is 13.5 Ohms).
I can slightly increase the output power, and increase the output impedance, but the frequency response will not be quite as good. with those modifications, I will also have to see what happens to the distortion.
Hum 700uV (more than I like)
With the various problems it has, I may, or I may not, continue to work on that amplifier. I may modify it into another amplifier circuit.
Do not get discouraged. You will find something to fit the bill.
I wanted to build a two stage push pull amplifier that uses Beam Power output tubes (essentially the same as using Pentode output tubes).
And, it had to have no NFB; not local, not global.
So, I built one. But it could use lots of improvement.
Very limited performance specifications:
No NFB
Using a non-inductive 8 Ohm load:
2.6 Watts maximum before visible waveform distortion
At 1 Watt, 2nd harmonic distortion -58dBc (0.13%); 3rd harmonic distortion -52 dBc (0.25%)
Frequency response, 1kHz 0 dB, 20kHz -0.3 dB, 40 Hz - 0.9dB, 21.5Hz - 2dB
Damping factor 0.6, not a very good current source, but at least it is less than Unity. (Output impedance of the 8 Ohm tap is 13.5 Ohms).
I can slightly increase the output power, and increase the output impedance, but the frequency response will not be quite as good. with those modifications, I will also have to see what happens to the distortion.
Hum 700uV (more than I like)
With the various problems it has, I may, or I may not, continue to work on that amplifier. I may modify it into another amplifier circuit.
Last edited:
It's possible to make a valve amplifier with arbitrarily large source impedance at the output plate - OPT junction, just using current-sensing negative feedback, but the OPT's distortion will still be sensitive to the high driving impedance. So maybe the question will end up being the best engineering balance between OPT distortion and speaker motor distortion? Both are esoteric except at extremes, but so is amplifier (valve transfer) distortion.
All good fortune,
Chris
All good fortune,
Chris
I have two OPTs from Fisher 500, the one with EL37 tubes. When moving 17 years ago, couldn't keep the radios, so salvaged the transformers.
Very informative.
I will be also using beam power tubes, but RF, not audio. The reason is they can run at high plate voltage (800 V), where distortion is minimal according to what can be deduced and extrapolated from data sheets. Like you, I will use PP for cancellation of second harmonic. At high voltage low current, plate resistance is very high, so it should be good current drive.
The key to low distortion is high power overhead. The amplifier is to be designed to continuously output 100-120 W in Class AB1, with power supply capable of producing 2-3 times the required current. Open loop THD at 100-120 W is 2-3% (again, according to data sheets). But the usable power (limited by input signal) is only about 5 W. At 5 W, it is still in the A part of AB1. I expect THD, mostly 3rd, in a fraction of percentage point at 5 W.
I also thought of a two-stage design, but now believe it can be done one-stage.
OPT distortion with pentode driver is an issue, but only at lower frequencies. I am thinking of three remedies. First, as posted earlier, I would like to limit low end to 120 Hz, which will relieve OPT from handling frequencies where it would distort most. Second, I will use amorphous core, which has very low hysteresis and thus low distortion of magnetizing current, so driving with high impedance source will not cause much transformer distortion. Third, there will be two output transformers, one for 120-1,000 Hz powering midrange driver, the other 1,000 Hz and up powering HF driver. Crossover will be at transformer level, not at speaker level.
I agree here - making things way harder than they need to be. Crossover at the transformer level? Before doing that I’d just bi-amp it. That’s a REALLY good way to keep low frequencies out of the core. Out of the tweeter core,anyway.
I'm not a transformer expert, but is it 'easier' to design an OPT for say 20Hz to 100Hz and another one from 100Hz to 20kHz, instead of a single one for 20Hz to 20kHz?
Has that ever been tried?
Jan
Has that ever been tried?
Jan
Of course it’s easier. Even easier still to design 20-100, 100-2K, 2K-200k. (Any traditional trafo will cover the mid range well, an oversized one will do the bass, and you can get as esoteric as you want for the tweeter). And these no-feedback High Zout amps tend to have few problems/issues where they only drive a single transducer over the range where the impedance is well-behaved.
NP, I've got your back. Just make sure to restructure that front end.
Quick addendum/sanity check; those Spice sims assumed no destination load. Driving the 500C stage into 1 Meg increased THD to 1.6%. In contrast a 6SF5GT measured yesterday in circuit under the same conditions swings 30 Vp-p at 0.03% only 2nd harmonic visible above the noise floor.
Re: low distortion power pentodes, do they even exist? I went with the 7591 because the spec sheet for PP distortion suggested low odd harmonics. The lowest distortion no-FB pentode found so far - based on personal Paint Tool models - is an 828 at 850 V plate driving a 10K primary. ~25 watts out at 4%.
Biamp is more complicated. Transformer-level XO is just extra HF transformer and one capacitor. Result is same as biamp.
Jan - sure thing, narrow band transformers are much easier to design as there are no conflicting requirements for simultaneous handling of frequency extremes. Mid-frequency transformer needs neither large inductance nor interleaving. A high frequency transformer can be small in size, with low number of turns and also w/o interleaving.
These are the pictures of my HF transformer in making. The transformer uses 40 mm OD 80% Ni toroid. 320 turn primary has inductance of 28 H. First, naked core; second, segmentation applied to the core; third, 16 segments of primary winding. That is then followed with electrostatic shield and secondary.
So what freq range would that have? Or is it for HF (measurement?) use?
I guess for LF (20-100Hz) you'd still need a mountain of metal.
Jan
I guess for LF (20-100Hz) you'd still need a mountain of metal.
Jan
A linear spring has a deflection dependent force. That does not cause distortion no matter the excursion. Again, just a matter of engineering to make it as linear as possible.
has nothing to do with the wavelength and size - that influences the directivity pattern. Two drivers add 3dB over one driver. Provided that the driver is SMALL compared to the wavelength. Large drivers are beaming more.
the VC can drive it out of that region and that is very nonlinear