LTspice giving very poor THD results (one opamp test)

[IvesH]

Hi,

I'm playing around with LTspice and even with one singe opamp as buffer I get very large THD. anybody can point me to what I'm doing wrong ?

Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 4.894e+00 1.000e+00 0.18° 0.00°
2 2.000e+03 8.335e-05 1.703e-05 -153.42° -153.61°
3 3.000e+03 6.513e-02 1.331e-02 -165.32° -165.50°
4 4.000e+03 2.624e-05 5.362e-06 9.14° 8.96°
5 5.000e+03 3.068e-02 6.269e-03 6.51° 6.33°
6 6.000e+03 1.186e-04 2.424e-05 -8.84° -9.02°
7 7.000e+03 6.043e-03 1.235e-03 -29.63° -29.81°
8 8.000e+03 1.056e-04 2.157e-05 125.99° 125.81°
9 9.000e+03 4.735e-03 9.676e-04 -174.96° -175.14°
10 1.000e+04 2.753e-05 5.624e-06 -139.51° -139.70°
Total Harmonic Distortion: 1.479406%(1.515575%)

 

[jan.didden]

There's a whole bunch of settings to get a high resolution FFT.
It is extensively explained in the sticky thread on LTspice - from beginner to advanced.

Jan

 

[IvesH]

Thx Jan

 

[bansuri]

increase the resolution
Tran: 10ms sample 1us output all data
measure cursor span from 1ms on
maybe 20V is too high, try with 15V

 

[IvesH]

Perfect, all good now
Set to:
.tran 0 10ms 0 1us
.OPTIONS plotwinsize=0
.OPTIONS numdgt=7.
Tried with and without compression, this did not make a difference.


Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 5.000e+00 1.000e+00 -0.01° 0.00°
2 2.000e+03 3.711e-06 7.421e-07 -10.61° -10.61°
3 3.000e+03 4.666e-08 9.333e-09 79.12° 79.13°
4 4.000e+03 9.307e-10 1.861e-10 70.92° 70.93°
5 5.000e+03 1.315e-09 2.629e-10 36.03° 36.03°
6 6.000e+03 1.491e-09 2.982e-10 31.10° 31.10°
7 7.000e+03 1.671e-09 3.343e-10 27.00° 27.01°
8 8.000e+03 1.858e-09 3.715e-10 23.67° 23.67°
9 9.000e+03 2.048e-09 4.096e-10 20.90° 20.91°
10 1.000e+04 2.242e-09 4.484e-10 18.57° 18.58°
Total Harmonic Distortion: 0.000074%(0.000000%)

 

[jan.didden]

That was quick - well done!

Jan

 

[peufeu]

Don't trust spice opamp models... Look inside the subcircuit model first.

If the opamp is modeled just with E, F and G's, plus some RC to get the open loop response phase and gain, and there's just the output transistors models inside, then distortion will be wildly optimistic.

 

[bordodynov]

Usually the models have input transistors, not output transistors. And output transistors can contribute a lot to distortion.

 

[peufeu]

The TPA6120 model I got has the input and output transistors, but the diamond buffer before them is just modeled as some E's so it gets very optimistic THD...

If NE5532 is just the default opamp model, then it is 100% linear, no THD modeled at all.

 

[Mark Tillotson]

maybe 20V is too high, try with 15V

Why? Its an NE5532.

 

[bansuri]

Why? Its an NE5532.

The idle current can be 16mA plus 2x3mA output (4k)
this makes 600mW dissipation for 15V rails or 880mW for 20V
The dice temp. will rise to 104°C or 130°C with 50°C ambient. (PDIP in fixture)
The high temp will not be beneficial, higher noise, lower gain, shorter life.

When a lower resistance is driven like a headphone the thing will fry, so cooling needs to be added.
That is why the

7.3 Recommended Operating Conditions MINMAXUNIT VCC +Supply voltage 5 15V VCC –Supply voltage –5–15V