This amp uses only 2 transistors, one J113 JFET and one BD139.
And only 4 resistors.
What is different is there is no feedback.
Because of this there is some distortion, THD 0.5% at +10dB into 32 Ohm.
The amp should probably work into headphones 16-600 Ohm.
But the design was only developed using 32 Ohm.
And only 4 resistors.
What is different is there is no feedback.
Because of this there is some distortion, THD 0.5% at +10dB into 32 Ohm.
The amp should probably work into headphones 16-600 Ohm.
But the design was only developed using 32 Ohm.
Your LTSpice is much better simulator than Multisym used by lineup. His results are overoptimistic, especially in case of MOSFET power devices.
If you would like to see real results with the Simplest amplifier, you might visit my thread
If you would like to see real results with the Simplest amplifier, you might visit my thread
0.2% THD I am willing to try but with 2% THD even I shall hear the distortion.
The thread you suggested I already watch but forgot about it. However, thanks for reminding me, I will look into it. With voltage gain it may also be worth a try (although my headphone amplifiers are already piling up😀)
The thread you suggested I already watch but forgot about it. However, thanks for reminding me, I will look into it. With voltage gain it may also be worth a try (although my headphone amplifiers are already piling up😀)
Last edited:
This is an improved circuit but without global feedback as the first one.
Distortion is lower.
Now we are at that level where distortion is difficult to hear.
This version uses three transistors instead of two.
Distortion is lower.
Now we are at that level where distortion is difficult to hear.
This version uses three transistors instead of two.
Thank you the new version Lineup 👍. However simulating does not change the THD at all, strangely enough. Maybe a setting somewhere or because of a component model difference between the simulator that Lineup uses and mine.
Perhaps I should just follow NeonDriver's suggestion in post #5 to try it out. I am very curious about this circuit so I will probably do that.
Perhaps I should just follow NeonDriver's suggestion in post #5 to try it out. I am very curious about this circuit so I will probably do that.
The input voltages not specified.
You must have been driving it near clipping or slightly into clipping.
With 100mV input I got .3% distortion.
And max input before clipping is around 500mV is 1.3% distortion
Could be slightly biased better, Voltage at output is under 6 volts and should be closer to 1/2 supply.
Class A output, so 30ma is little hot for a BD139.
With output DC at 6 volts it climbs even higher to 40ma. Class A so would have slightly more output.
Likely be easier to use a TO247 Darlington, biased to 50ma would have more output.
I think a typical phone or Mp3 player does about 600 to 800mV max output.
So the amp clips at about 540 mV
I attached the .asc LTSpice file for version 1 and 2 but should have specified the input voltage with a screenprint. The input voltage I used is also 100mV. It clips somewhere around 350mV in my case.
What simulator do you use? You seem to confirm the results from Lineup more than mine so I guess it is worth a try, not the most difficult task.
700mV input not possible, maybe not particular to headphone levels.
Would just like it to handle full phone output or possible less distortion at listening levels.
Anyhoo sticking to the 2x transistor theme I just tossed in a MJH6284 Darlington.
Figured why torture a TO-126 at 30ma when a TO-247 do it all day.
700mV input no problem now.
100mV input is now .04% THD instead of .3% and 400 mV is .8% instead of 1.4%
I use TinaTI. Most likely for you, FFT settings in LTspice.
Something I was gonna try like you did was drive a lower impedance. I see you tested with 32 ohm load.
And indeed TinaTI shows the Clipping levels match to your 350mV with lower impedance.
I dont know how to set LTspice FFT settings, I have seen numerous changes people do.
With a FFT it often has to do with the transient initial conditions.
TinaTi it is set to calculate the operating point. Likely as with lineup sim as well.
If I use Zero initial values , or use initial conditions in my sim.
It matches your test, in fact much worse close to 4% distortion
For FFT typically would calculate the operating point.
As with any transient the initial startup can have a small error in the first cycle.
This error will also add distortion to the FFT, so if I I am testing say 1000 Hz
I delay the sample start time for one cycle or 100u to remove the error.
For say 20K it only needs 5u or one full cycle.
Would just like it to handle full phone output or possible less distortion at listening levels.
Anyhoo sticking to the 2x transistor theme I just tossed in a MJH6284 Darlington.
Figured why torture a TO-126 at 30ma when a TO-247 do it all day.
700mV input no problem now.
100mV input is now .04% THD instead of .3% and 400 mV is .8% instead of 1.4%
I use TinaTI. Most likely for you, FFT settings in LTspice.
Something I was gonna try like you did was drive a lower impedance. I see you tested with 32 ohm load.
And indeed TinaTI shows the Clipping levels match to your 350mV with lower impedance.
I dont know how to set LTspice FFT settings, I have seen numerous changes people do.
With a FFT it often has to do with the transient initial conditions.
TinaTi it is set to calculate the operating point. Likely as with lineup sim as well.
If I use Zero initial values , or use initial conditions in my sim.
It matches your test, in fact much worse close to 4% distortion
For FFT typically would calculate the operating point.
As with any transient the initial startup can have a small error in the first cycle.
This error will also add distortion to the FFT, so if I I am testing say 1000 Hz
I delay the sample start time for one cycle or 100u to remove the error.
For say 20K it only needs 5u or one full cycle.
Thank you for the information. I have to study these FFT settings more but nothing changed since some other circuits I tested and built so far. TinaTi also uses Spice, not suprisingly.
This says the log:
This THD anomaly intrigues me. I will likely try to solve this but will built the circuit first. I was not planning to use a BD139 either. A TIP41C can handle more current, or a D44H11, etc.
Thanks again for the information 👍
This says the log:
[SIZE=4]Direct Newton iteration for .op point skipped.[/SIZE]Starting Gmin steppingGmin = 10Gmin = 1.07374Gmin = 0.115292Gmin = 0.0123794Gmin = 0.00132923Gmin = 0.000142725Gmin = 1.5325e-05Gmin = 1.6455e-06Gmin = 1.76685e-07Gmin = 1.89714e-08Gmin = 2.03704e-09Gmin = 2.18725e-10Gmin = 2.34854e-11Gmin = 2.52173e-12Gmin = 2.70769e-13Gmin = 0Gmin stepping succeeded in finding the operating point.N-Period=1Fourier components of V(out)DC component:-0.000377416Harmonic Frequency Fourier Normalized Phase Normalized Number [Hz] Component Component [degree] Phase [deg] 1 1.000e+3 2.777e-1 1.000e+0 -89.74° 0.00° 2 2.000e+3 2.071e-3 7.457e-3 -3.12° 86.62° 3 3.000e+3 5.798e-3 2.088e-2 73.55° 163.29° 4 4.000e+3 1.928e-4 6.943e-4 -117.07° -27.33° 5 5.000e+3 3.209e-4 1.155e-3 -8.23° 81.51° 6 6.000e+3 1.969e-4 7.090e-4 -86.78° 2.97° 7 7.000e+3 1.428e-4 5.143e-4 -45.96° 43.78° 8 8.000e+3 1.421e-4 5.116e-4 -103.72° -13.98° 9 9.000e+3 4.353e-4 1.567e-3 -22.15° 67.59°Partial Harmonic Distortion: 2.228923%Total Harmonic Distortion: 2.249799%Date: Mon Jul 7 12:21:33 2025Total elapsed time: 1.341 seconds.tnom = 27temp = 27method = modified traptotiter = 9716traniter = 9396tranpoints = 4016accept = 2418rejected = 1598matrix size = 13fillins = 0solver = NormalAvg thread counts: 1.1/1.3/1.3/1.1Matrix Compiler1: 710 bytes object code size 1.6/1.4/[1.3][SIZE=4]Matrix Compiler2: 1.00 KB object code size 1.8/2.2/[1.3][/SIZE]This THD anomaly intrigues me. I will likely try to solve this but will built the circuit first. I was not planning to use a BD139 either. A TIP41C can handle more current, or a D44H11, etc.
Thanks again for the information 👍