Hello all,
I'm attempting to teach myself how to use LTSpice to model the output stages of tube audio amplifiers and I've run into a stumbling block. I can't seem to get an output stage with a cathode feedback winding on the transformer to work properly.
The version of the output stage without feedback gives about 55 watts across the dummy load of 4 ohms with 80 volts peak between the 6146 grids. This is inline with commercially manufactured amplifiers using the same operating points. The THD is about 4%, which seems reasonable enough compared with a datasheet and quoted values from commercial amplifiers. The THD as calculated from the LTSpice simulation is obviously going to be quite optimistic.
When I connect a hypothetical feedback winding consisting of 10% of the primary winding (inductors L4 and L5) to the cathodes of the output tubes, the output stage works... to a point. With that same 80 volts peak between the 6146 grids, I get about 2.18 watts across the same 4 ohm dummy load. I also get the dramatic reduction in distortion expected when compared to the output stage without feedback working at the 2.18 watt mark.
When I increase the voltage on the cathode feedback output stage, I get rather nasty looking crossover distortion and the grids of the 6146s start drawing current. My understanding from the Walker and Williamson article on cathode feedback windings, and another article regarding an early 100 watt 6550 amplifier using such a feedback winding, is that this isn't supposed to happen.
Do I have some fundamental misunderstanding about how the cathode feedback winding works? Are the deficiencies of my rather crude model of the output transformer to blame for the behavior? I know that a simple collection of inductors can't account for core behavior, and that I'm not accounting for leakage inductance or parasitic capacitance, but I still didn't expect this result. I'm at a loss here...
Attached are schematics for both output stages, an example of the crossover distortion that results when a higher voltage drives the output stage with the cathode feedback winding, and a zipped file containing both LTSpice schematics.
I'm attempting to teach myself how to use LTSpice to model the output stages of tube audio amplifiers and I've run into a stumbling block. I can't seem to get an output stage with a cathode feedback winding on the transformer to work properly.
The version of the output stage without feedback gives about 55 watts across the dummy load of 4 ohms with 80 volts peak between the 6146 grids. This is inline with commercially manufactured amplifiers using the same operating points. The THD is about 4%, which seems reasonable enough compared with a datasheet and quoted values from commercial amplifiers. The THD as calculated from the LTSpice simulation is obviously going to be quite optimistic.
When I connect a hypothetical feedback winding consisting of 10% of the primary winding (inductors L4 and L5) to the cathodes of the output tubes, the output stage works... to a point. With that same 80 volts peak between the 6146 grids, I get about 2.18 watts across the same 4 ohm dummy load. I also get the dramatic reduction in distortion expected when compared to the output stage without feedback working at the 2.18 watt mark.
When I increase the voltage on the cathode feedback output stage, I get rather nasty looking crossover distortion and the grids of the 6146s start drawing current. My understanding from the Walker and Williamson article on cathode feedback windings, and another article regarding an early 100 watt 6550 amplifier using such a feedback winding, is that this isn't supposed to happen.
Do I have some fundamental misunderstanding about how the cathode feedback winding works? Are the deficiencies of my rather crude model of the output transformer to blame for the behavior? I know that a simple collection of inductors can't account for core behavior, and that I'm not accounting for leakage inductance or parasitic capacitance, but I still didn't expect this result. I'm at a loss here...
Attached are schematics for both output stages, an example of the crossover distortion that results when a higher voltage drives the output stage with the cathode feedback winding, and a zipped file containing both LTSpice schematics.
Attachments
Cathode feedback decreases sensitivity. You have to increase drive voltage from 40V to 125V to fully drive output tubes. Even then only getting 22W out.
Without cathode NFB the cathode is at ground potential (Uk = 0 V) and screen grid voltage (Ug2) is constantly at 180 V.
When using cathode NFB, the cathode is no more at gnd potential.
At the moment of positive drive cycle at g1, the Uk is at it's maximum value and therefore the value of Ug2 is at it's minimum.
With your circuit with cathode nfb, the maximum input signal (at the point where grid current begins) seems to be 125 Vpeak and max. output power is some 22 W.
At this input level, and at the positive drive cycle, the cathode reaches +85 V, which means that at this moment the Ug2 is 180 V -85 V = 95 V. This low Ug2 limits anode current very effectively. And it also cut 85 V from the anode voltage.
Therefore the maximum output is limited. This is the inherent feature of the cathode nfb.
The reduction of output power (due to cnfb) would be less dramatic if the tube type used would have Ug2 near the +Ub level, like EL34.
With 6146 the cathode nfb-% should be less than 10.
When using cathode NFB, the cathode is no more at gnd potential.
At the moment of positive drive cycle at g1, the Uk is at it's maximum value and therefore the value of Ug2 is at it's minimum.
With your circuit with cathode nfb, the maximum input signal (at the point where grid current begins) seems to be 125 Vpeak and max. output power is some 22 W.
At this input level, and at the positive drive cycle, the cathode reaches +85 V, which means that at this moment the Ug2 is 180 V -85 V = 95 V. This low Ug2 limits anode current very effectively. And it also cut 85 V from the anode voltage.
Therefore the maximum output is limited. This is the inherent feature of the cathode nfb.
The reduction of output power (due to cnfb) would be less dramatic if the tube type used would have Ug2 near the +Ub level, like EL34.
With 6146 the cathode nfb-% should be less than 10.
Last edited:
Thank you for your replies. I knew I was missing something simple. I hadn't considered that it would be the cathode to screen voltage that governs the amount of output power and drawing of grid current and not cathode to plate to voltage. The plate exists more or less to attract the electrons that were accelerated toward G2. I was thinking only about the reduction in effective plate voltage.
So it naturally follows then that if you can get around the ungodly drive requirements, that a cathode feedback winding when used with a triode output stage would afford the least loss of output power due to the feedback, no?
So it naturally follows then that if you can get around the ungodly drive requirements, that a cathode feedback winding when used with a triode output stage would afford the least loss of output power due to the feedback, no?
Is there anyone currently making, or willing to make, a transformer with a tertiary winding?
It took me about a decade to scrounge up a NOS pair of Merit A-3130 with a 10% tertiary winding.
Win W5JAG
It took me about a decade to scrounge up a NOS pair of Merit A-3130 with a 10% tertiary winding.
Win W5JAG
Yes. Toroidy of Poland has a line of 5 different push pull toroidal OPTs with tertiary windings. I have no clue if they are any good, but I intend to find out. The price is right; similar to Edcors in cost.
These ones (look for 'CFB' suffix on medel number)are expensive but the Vanderveen designed transformers I've used so far have been really good.
Dave Slagle makes 'em.
My guess is that most audio transformer winders would make them on request if they don't have a stock model already.
Dave Slagle makes 'em.
My guess is that most audio transformer winders would make them on request if they don't have a stock model already.
- Status
- Not open for further replies.