Hi all,
inspired by the Nelson Pass' Zen v9 and Tubelab's UNSET, I woud like to propose you a schematic and ask some suggestions on how to choose the right components and/or working points. It is a single ended, single gain stage amp with triode curves, a DF of 2.4 and 26 Wrms at 4%THD.
The basic concept is that the upper n-Mosfet is driven from its source by the lower p-Mosfet source follower.
The gate of the upper n-Mosfet receives a feedback from its drain through a voltage divider to linearize it and triodize its curves: the higher the ratio, the trioder and more linear the curves and the lower Zout.
I attach here the schematic hoping to receive some help on the choice of the lower p-Mosfet with a reasonable SOA.
The CCS should be right as I've used the same power supply voltage and CCS circuit as the Zen v9.
It is supplied at 46 V B+ and biased at 2.6 A (1.3 A each IRFP240), being able to deliver (on simulations) 26 Wrms with the following THD:
And 1 Wrms with the following:
The DF is 2.4 on 8 Ohm (Zout is 3.3 Ohm) and I'm sure that it can be done way more than this with better working points and components.
Thank you in advance
Roberto
inspired by the Nelson Pass' Zen v9 and Tubelab's UNSET, I woud like to propose you a schematic and ask some suggestions on how to choose the right components and/or working points. It is a single ended, single gain stage amp with triode curves, a DF of 2.4 and 26 Wrms at 4%THD.
The basic concept is that the upper n-Mosfet is driven from its source by the lower p-Mosfet source follower.
The gate of the upper n-Mosfet receives a feedback from its drain through a voltage divider to linearize it and triodize its curves: the higher the ratio, the trioder and more linear the curves and the lower Zout.
I attach here the schematic hoping to receive some help on the choice of the lower p-Mosfet with a reasonable SOA.
The CCS should be right as I've used the same power supply voltage and CCS circuit as the Zen v9.
It is supplied at 46 V B+ and biased at 2.6 A (1.3 A each IRFP240), being able to deliver (on simulations) 26 Wrms with the following THD:
Code:
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 2.024e+01 1.000e+00 0.53° 0.00°
2 2.000e+03 6.636e-01 3.278e-02 87.57° 87.04°
3 3.000e+03 4.175e-01 2.063e-02 7.60° 7.07°
4 4.000e+03 2.578e-02 1.274e-03 140.12° 139.60°
5 5.000e+03 1.271e-01 6.279e-03 -174.20° -174.73°
6 6.000e+03 4.537e-02 2.241e-03 -92.40° -92.93°
7 7.000e+03 4.576e-02 2.261e-03 -5.12° -5.64°
8 8.000e+03 1.662e-02 8.213e-04 72.36° 71.84°
9 9.000e+03 1.007e-02 4.973e-04 153.44° 152.91°
Total Harmonic Distortion: 3.940018%(3.940620%)And 1 Wrms with the following:
Code:
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 3.920e+00 1.000e+00 0.63° 0.00°
2 2.000e+03 1.866e-02 4.759e-03 90.42° 89.79°
3 3.000e+03 8.302e-04 2.118e-04 8.75° 8.12°
4 4.000e+03 1.684e-04 4.297e-05 -164.53° -165.16°
5 5.000e+03 1.285e-04 3.279e-05 -179.18° -179.82°
6 6.000e+03 1.046e-04 2.667e-05 -179.83° -180.46°
7 7.000e+03 8.966e-05 2.287e-05 -179.76° -180.39°
8 8.000e+03 7.845e-05 2.001e-05 -179.79° -180.42°
9 9.000e+03 6.973e-05 1.779e-05 -179.81° -180.44°
Total Harmonic Distortion: 0.476465%(0.476493%)The DF is 2.4 on 8 Ohm (Zout is 3.3 Ohm) and I'm sure that it can be done way more than this with better working points and components.
Thank you in advance
Roberto
Thank you Nelson, I've done a copy paste what you have made public in FIG.2 at this page: https://www.passdiy.com/gallery/amplifiers/zen-variations-9
Your C4 is my C2 and your Q5 is my Q1. Have I missed something?
I forgot to mention one thing about V1, the reference voltage of the voltage divider that brings voltage feedback to the gates of M1 and M3, the only two devices that provide gain: its role is to move the triodized curves to the left, increasing the capability of the amp to swing down to low voltages (local feedback alone would triodize the curves with the typical triode issue: being impossible to swing down to almost zero volt, so reducing dramatically the power we can get from the system, so its efficiency), so reaching a total efficiency (including the dissipation of the CCS and of the source follower at the bottom) of around 25%. Including just M1 and M3 in the count, the efficiency goes up to 46.5%.
Not bad for a single ended class A single stage amp.
Q1 dissipates 30mW and can be a TO92
M5 dissipates around 8W and can be TO220
M1 and M3 dissipate 27W each must be TO247. Is it better to have three of them in parallel to make them run cooler?
M2 dissipates 55W: needs to be two/three TO247 paralleled?
Thanks!
Not bad for a single ended class A single stage amp.
Q1 dissipates 30mW and can be a TO92
M5 dissipates around 8W and can be TO220
M1 and M3 dissipate 27W each must be TO247. Is it better to have three of them in parallel to make them run cooler?
M2 dissipates 55W: needs to be two/three TO247 paralleled?
Thanks!
Last edited:
Upgrade with p-Mosfet source follower FQB11P06, great improvement!
No need to decouple the input of the amp, and the distortion dropped significantly:
This is at 1 Wrms:
And this is at 22Wrms:
No need to decouple the input of the amp, and the distortion dropped significantly:
This is at 1 Wrms:
Code:
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 4.013e+00 1.000e+00 0.16° 0.00°
2 2.000e+03 4.407e-04 1.098e-04 -52.21° -52.37°
3 3.000e+03 1.272e-03 3.170e-04 1.28° 1.12°
4 4.000e+03 6.230e-05 1.552e-05 -131.72° -131.88°
5 5.000e+03 4.057e-05 1.011e-05 -177.83° -177.99°
6 6.000e+03 3.037e-05 7.566e-06 179.79° 179.63°
7 7.000e+03 2.627e-05 6.546e-06 -179.73° -179.89°
8 8.000e+03 2.300e-05 5.731e-06 -179.81° -179.97°
9 9.000e+03 2.045e-05 5.095e-06 -179.84° -180.00°
Total Harmonic Distortion: 0.033623%(0.033645%)And this is at 22Wrms:
Code:
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 1.880e+01 1.000e+00 0.09° 0.00°
2 2.000e+03 1.647e-01 8.762e-03 79.16° 79.07°
3 3.000e+03 2.392e-01 1.273e-02 9.12° 9.02°
4 4.000e+03 6.283e-02 3.343e-03 -113.93° -114.02°
5 5.000e+03 3.586e-02 1.908e-03 -144.04° -144.13°
6 6.000e+03 1.141e-02 6.068e-04 38.13° 38.04°
7 7.000e+03 5.335e-03 2.838e-04 23.39° 23.30°
8 8.000e+03 3.610e-03 1.921e-04 -27.25° -27.34°
9 9.000e+03 6.002e-03 3.193e-04 84.47° 84.38°
Total Harmonic Distortion: 1.594128%(1.594790%)I searched some p-mosfet that will work within SOA (so with a dedicated area for DC operation) but that will have the lowest Crss at this particular working point:
FQPF5P20RDTU ( https://www.mouser.it/datasheet/2/308/1/FQPF5P20RDTU_D-2313999.pdf ) around 200 pF
FQP8P10 ( https://docs.rs-online.com/89f0/0900766b812cfa38.pdf ) around 350 pF
FQPF11P06 ( https://www.mouser.it/datasheet/2/308/1/FQPF11P06_D-2313712.pdf ) around 250 pF
All of them have a too high Crss to install a potentiometer just before. Is there anything better around for the purpose?
Of course everything would be fine with two stages in series: more local feedback can be applied, so more linearity and lower Zout.
FQPF5P20RDTU ( https://www.mouser.it/datasheet/2/308/1/FQPF5P20RDTU_D-2313999.pdf ) around 200 pF
FQP8P10 ( https://docs.rs-online.com/89f0/0900766b812cfa38.pdf ) around 350 pF
FQPF11P06 ( https://www.mouser.it/datasheet/2/308/1/FQPF11P06_D-2313712.pdf ) around 250 pF
All of them have a too high Crss to install a potentiometer just before. Is there anything better around for the purpose?
Of course everything would be fine with two stages in series: more local feedback can be applied, so more linearity and lower Zout.
I'm going to optimize what by now is just an idea (all semiconductors need to be correctly chosen), but with this new addendum I've transformed the series cascode of the Zen V9 into a Shunt Cascode (see Rod Coleman's work): the nfet is triodized by the drain-gate local feedback, source driven by the p-mosfet, then CCS loaded, and shunt cascoded by the PNP BJT.
Still a single gain stage, with the gain proportional to R3.
After that there will be source followers to drive the speaker.
This is the not-optimized result with 1Vp as input and 31Vp as output:
Still a single gain stage, with the gain proportional to R3.
After that there will be source followers to drive the speaker.
This is the not-optimized result with 1Vp as input and 31Vp as output:
Code:
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 3.116e+01 1.000e+00 -0.00° 0.00°
2 2.000e+03 4.734e-01 1.519e-02 89.96° 89.96°
3 3.000e+03 1.629e-02 5.228e-04 0.00° 0.00°
4 4.000e+03 2.916e-04 9.358e-06 -86.14° -86.13°
5 5.000e+03 5.222e-05 1.676e-06 179.65° 179.65°
6 6.000e+03 3.826e-06 1.228e-07 -61.16° -61.15°
7 7.000e+03 5.664e-07 1.817e-08 0.81° 0.81°
8 8.000e+03 1.946e-07 6.243e-09 -23.90° -23.90°
9 9.000e+03 1.936e-07 6.214e-09 -1.17° -1.17°
Total Harmonic Distortion: 1.519981%(1.519981%)After having abandoned the idea for few months, I've recently spent some evenings finding how to improve the efficiency of the amp while keeping the same concept and I came out with this (power amp taken from here: https://www.tubecad.com/2018/01/blog0408.htm) :
For sure the IRFP240 will need to be more in parallel, because there will be need to dissipate 80W for something less than 40Wrms at its output, as expected from a inductor loaded class A power amp.
I also need to check that the driver/preamp fits the requirements for every component, and suggestions are very well accepted here as well.
The only applied feedback is the local feedback on the U440 to triodize its curves.
No global feedback applied.
I see from the curves on simulations (LTSpice file attached) that the results seems promising, but I will check on a single stage inductor loaded too, to see how it goes.
Thanks in advance for any feedback I will receive.
Roberto
For sure the IRFP240 will need to be more in parallel, because there will be need to dissipate 80W for something less than 40Wrms at its output, as expected from a inductor loaded class A power amp.
I also need to check that the driver/preamp fits the requirements for every component, and suggestions are very well accepted here as well.
The only applied feedback is the local feedback on the U440 to triodize its curves.
No global feedback applied.
I see from the curves on simulations (LTSpice file attached) that the results seems promising, but I will check on a single stage inductor loaded too, to see how it goes.
Thanks in advance for any feedback I will receive.
Roberto