Roberto,
I have considered the idea that a cascode could substitute for the pentode in the input stage, but I have not had any time to pursue that in experiments.
When you say "buffered shunt feedback" did you make a mistake in referring to the series feedback arrangement that I am using or are you suggesting switching things up so that it is indeed a shunt feedback arrangement?
I personally like series feedback for the high input impedance it offers, but I have pursued buffered shunt feedback previously with p-channel fets. I built a pretty nice amp that way.
-Heath
I have considered the idea that a cascode could substitute for the pentode in the input stage, but I have not had any time to pursue that in experiments.
When you say "buffered shunt feedback" did you make a mistake in referring to the series feedback arrangement that I am using or are you suggesting switching things up so that it is indeed a shunt feedback arrangement?
I personally like series feedback for the high input impedance it offers, but I have pursued buffered shunt feedback previously with p-channel fets. I built a pretty nice amp that way.
-Heath
I made a mistake in referring to your actual system, indeed.
I will do some simulations with the LSK170+12A(T or U)7 cascode, I have bought a cheap SE amp kit thanks to a coupon, and I'm looking to different ways to improve it ( 6P13S SE cheap amp: improvement suggestions ).
Also my actual EL34 PP amp has a cascoded PI, and could be useful there too ( improvements on 12AX7 12AT7 EL34 schematic? )
I will do some simulations with the LSK170+12A(T or U)7 cascode, I have bought a cheap SE amp kit thanks to a coupon, and I'm looking to different ways to improve it ( 6P13S SE cheap amp: improvement suggestions ).
Also my actual EL34 PP amp has a cascoded PI, and could be useful there too ( improvements on 12AX7 12AT7 EL34 schematic? )
The LSK170 is not a good fit for signals at the input of a power amplifier. It was based on the 2SK170 - a JFET designed for Moving Coil input stages, <1mV input. Applying more than about 50mV pp to the gate will give very obvious distortion.
Using the 12A*7 at the bottom (or a triode-connected EF80), and a suitably rated bipolar transistor at the top will give much better performance.
Using the 12A*7 at the bottom (or a triode-connected EF80), and a suitably rated bipolar transistor at the top will give much better performance.
Thanks Rod Coleman, I've used that configuration because of Merlin Blancowe:
Which grade LSK389 to use for hybrid cascode a la Merlin B.
I also have the same configuration in one of my amps. Is it really that bad?
Which grade LSK389 to use for hybrid cascode a la Merlin B.
I also have the same configuration in one of my amps. Is it really that bad?
The circuit is not bad in itself. But look at Rongon's simulation: the input voltage is 10mV.
It may work well at that level. This is a circuit for phono preamps.
But connecting it to the output of a DAC will present it with ca. 2V rms, which is too much for K170.
To see why, look at Toshiba 2SK170 data sheet, Id vs. Vds curves for various Vgs values, and see how the Vgs plots become close together when th input increases
It may work well at that level. This is a circuit for phono preamps.
But connecting it to the output of a DAC will present it with ca. 2V rms, which is too much for K170.
To see why, look at Toshiba 2SK170 data sheet, Id vs. Vds curves for various Vgs values, and see how the Vgs plots become close together when th input increases
Hi bohrok2610 - will be very interested to see your results using SS's method. Curious, why did you not complete the amplifier? Seems you were most of the way there already, hope the new feedback topology bears fruit.
I created a similar set of source follower boards for a phono stage I recently finished with a SF output buffer. I will be repurposing them for this 801A A2 amplifier, but will change the SF package to TO-247 for the Cree chip (uses TO-220 for the CCS FET). Has a trimmer to set the CCS bias current with a jumper that can be removed and a set resistor placed when the value is determined. Happy to provide the gerber files if this is something others are interested in, nothing fancy but verified working in my phono stage.
Are you getting pretty good HF bandwidth in closed loop with the Edcor transformers SS?
Speaking of cooler weather, unfortunately the very cool Cleveland winter is coming fast and no sign of Sowter shipping the OPT for my amplifier. I work out of my garage typically, so I will be shutting down operations for the season, seems this experiment is to-be-continued come springtime.
without changing too much the pcb I would like to use for tests, so keeping the fet below & 12a*7 on top, can a J202 be a better solution? https://www.futurlec.com/Datasheet/Transistor/J201.pdf
Thanks, and sorry to hijack the thread on the cascode version vs original pentode configuration, but I think it can be an interesting variation.
It's adequate. The 10kHz square wave doesn't have quite as steep a rise as the push-pull amp I built for my brother. The speaker it is playing into right now has an extremely efficient tweeter so the high end comes out pretty hot, so it doesn't feel like it needs a flatter high end.
I haven't really measured what the -3dB point is. I'll do that after I get the bias servo all sorted out. I might connect that tonight. The output seems to behave properly so it is time to make it do its job.
Weather is pretty much opposite here in the Southern California desert. The garage is unbearable in the summer but quite nice with a light sweater in the winter. Perfect tube temperature.
Like I said, I'm very interested in how well a cascode could work as a substitute for the pentode.
I guess my excuse for not completing the amp was kids vs. top caps
I managed to squeeze out even lower THD without GFB (about 0.05%@1W, see Why we need to check on the harmonics and THD?). As the output impedance is also low there really is no need for GFB actually. So this is really a keeper. Besides my kids are grown ups already
But I have already made some initial tests with SS's feedback topology. So far the results have been disappointing. EF37A does not seem to be happy with CCS anode load.
Fantastic, looking forward to hearing how it works out. Cleveland is miserable in the summer too, there are 3-4 months out of the year with mild weather, unfortunately.
What issue are you seeing with the CCS loaded EF37A? I am using it in my amplifier as well. I was experiencing oscillation / ringing when using the IXYS IXTP08N100D2 / IXTP08N50D2 in a cascode CCS, the issue seemed to resolve when I switched to the IXCP10M90S / DN2540.
Here is a 1kHz square of the EF37A open loop with a CCS load and 470K resistor from anode to ground.
Last edited:
Thanks SpreadSpectrum,
by now I'm using a circuit I already know and have in hands to optimize it before applying feedback.
The circuit is an hybrid cascoded Schmitt phase inverter with CCS on the sources of the fets, dc coupled to a 12AT7 driver. There's shunt feedback including driver and output tubes, but the PI runs without feedback and it is the main source of THD (that's why it seems to me a good circuit to find a good starting point for the cascode).
My goal is to find the optimal components for that PI, then use that optimisation to a SE kit (of course with only one side of the PI) I already have bought, applying your feedback solution to improve the results even more.
> without changing too much the pcb I would like to use for tests, so keeping the fet below & 12a*7 on top, can a J202 be a better solution?
For SE operation, yes, the J202 will certainly allow a wider input swing before the distortion gets too bad. But 2.0V rms is still a challenge, because the input circuit should be designed to handle at least ±3V from the operating point, if 2V rms is to be applied without an attenuator.
The performance will be much worse than a suitable triode in the bottom of the cascode.
If you really want to use the JFETs, the differential circuit you posted is much more suitable, because the feedback prevents the gate→source voltage from increasing too far.
For SE operation, yes, the J202 will certainly allow a wider input swing before the distortion gets too bad. But 2.0V rms is still a challenge, because the input circuit should be designed to handle at least ±3V from the operating point, if 2V rms is to be applied without an attenuator.
The performance will be much worse than a suitable triode in the bottom of the cascode.
If you really want to use the JFETs, the differential circuit you posted is much more suitable, because the feedback prevents the gate→source voltage from increasing too far.
Thank you Rod, with the actual PP configuration there's no need for 2Vrms at its input, as with around 2Vpp it swings around 80Vpp that is plenty for the EL34s I have. This should simplify its working condition.
Thanks, so this solution can be used for this phase inverter, but not for the SE. I will then do what you suggest for the SE: cascoded triodes or triode+BJT as you suggested.
Thanks again, I will then try J202 in that circuit, but not in the SE.
A friend of mine made a comment that got the gears turning on how to simplify my 801A A2 design using SpreadSpectrum's feedback method, I'd be very interested to hear what you all think.
Originally, I was using a source follower buffer to drive the 801A grid into A2 with V+/V- supplies such that it could swing from +60 to -70V. What I am experimenting with (in LTSpice) is eliminating the source follower buffer, elevating the 801A filaments (~200V), and direct-coupling the 801A grid to the mu-output of the EF37A CCS. This does away with the V+/V- supplies, allowing the grid to swing positive/negative with respect to the filament, and eliminates a coupling capacitor. The output impedance of the DN2540 appears low enough to provide the grid current for A2 and does not seem to be adversely affecting the plate current of the EF37A. Could be different in the real world, I've attached my functional but not refined .asc model below, 6J7 has been substituted for the EF37A for proof-of-concept purposes. Thanks for looking.
Originally, I was using a source follower buffer to drive the 801A grid into A2 with V+/V- supplies such that it could swing from +60 to -70V. What I am experimenting with (in LTSpice) is eliminating the source follower buffer, elevating the 801A filaments (~200V), and direct-coupling the 801A grid to the mu-output of the EF37A CCS. This does away with the V+/V- supplies, allowing the grid to swing positive/negative with respect to the filament, and eliminates a coupling capacitor. The output impedance of the DN2540 appears low enough to provide the grid current for A2 and does not seem to be adversely affecting the plate current of the EF37A. Could be different in the real world, I've attached my functional but not refined .asc model below, 6J7 has been substituted for the EF37A for proof-of-concept purposes. Thanks for looking.
Why are you involved the V1 (200V) "elevating" PSU to output tube power supply?
The V6 PSU current "circle" closing from OPT to power tube's cathode.
The DC direct drive "circle" is from power tube's grid to cathode, and then via V1 to ground.
Two independent loop.
The later only (if it not static A2 type amp) conduct AC current, so V1 "last" capacitor is the grounding device (I usually use in this situation few 10uF "good" capacitor as power tube's cathode point AC "grounding").
The V6 PSU current "circle" closing from OPT to power tube's cathode.
The DC direct drive "circle" is from power tube's grid to cathode, and then via V1 to ground.
Two independent loop.
The later only (if it not static A2 type amp) conduct AC current, so V1 "last" capacitor is the grounding device (I usually use in this situation few 10uF "good" capacitor as power tube's cathode point AC "grounding").
Thanks SS, will just have to build it and see about the pentode bias stability.
euro21 - I think I am understanding your concern, the filament and regulator would get their AC ground reference through the final capacitor in the 200V supply regulator. I may have to ask Rod's opinion on using his filament regulators on top of an elevated supply, I have not seen an example but I am sure it has been attempted before. I will report back when I have some real-world results, thanks for the input.
euro21 - I think I am understanding your concern, the filament and regulator would get their AC ground reference through the final capacitor in the 200V supply regulator. I may have to ask Rod's opinion on using his filament regulators on top of an elevated supply, I have not seen an example but I am sure it has been attempted before. I will report back when I have some real-world results, thanks for the input.