Cheap as Chips OPA1688 Low-THD Muscle Amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Founder of XSA-Labs
Joined 2012
Paid Member
I normally make discrete SE Class A amps, but I have found that the OPA1688 sounds quite nice. This project started out in this thread, where I explored using a single OPA1688 with an output cap coupler to reduce the complexities of active PSU rail shut-off to prevent DC imbalance on the headphone outputs. I hand-etched a simple single OPA1688 and gave it a non-polar output cap bank and it sounds very nice. Of course, there is global feedback and it is not Class A. But it there is something nice about the simplicity and low power consumption for 9v battery use. So it got me to thinking of how to boost the output power by paralleling the OPA1688. These are capable of 75mA current drive and can handle rails up to +/-18v. Here is the single OPA1688 amp that I built:
691261d1531291477-op-amp-cmoy-amp-opa1366-build-completed-top-jpg


691262d1531291477-op-amp-cmoy-amp-opa1366-build-completed-bottom-jpg


The OPA1688, is not your typical LTP input stage opamp as the measurements show that it has a dominant H2 (second harmonic character with monotonically decreasing higher orders). Here is 2vpp into 47ohms:
691432d1531370292-op-amp-cmoy-amp-opa1688-hpa-2vpp-47r-fft-png


Internally, the OAP1688's are a 3-stage amp design and capable of remarkable performance. The circuit uses the much-liked LME49720 as the voltage amp stage with the OPA1688 as a power buffer and the global feedback loop is from the output of the summed buffers back to the voltage amp. Load balancing 1R resistors are used at the output of the buffers and 220R summation resistors are used at the inputs to the buffers. With 6 buffers, this will require three dual OPA1688IDR IC's. With SOIC8 package and full use of SMT parts where possible, the total footprint can be quire small. A total of 6 x OPA1366's will be needed with a single dual channel LME49720 for the voltage amp stage. The cost of the 7 chips is about $10 and will produce an amp capable of about 1.3wrms into 50ohms and maybe 3.5wrms into 16ohms. All at exceedingly low THD levels. I modeled it LTSpice with the following schematic:
691428d1531366246-cheap-chips-opa1688-low-thd-muscle-amp-muscle-amp-sch-png


Here is the predicted FFT for 2vpp into 50ohms:
691429d1531366246-cheap-chips-opa1688-low-thd-muscle-amp-muscle-amp-2vpp-50ohms-fft-png


The Fourier components and THD are way down there in the noise.
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 1.001e+00 1.000e+00 -0.03° 0.00°
2 2.000e+03 7.524e-09 7.513e-09 -24.92° -24.89°
3 3.000e+03 8.023e-08 8.012e-08 -89.87° -89.84°
4 4.000e+03 1.141e-09 1.139e-09 -40.43° -40.40°
5 5.000e+03 9.318e-10 9.305e-10 -8.94° -8.90°
6 6.000e+03 1.251e-09 1.249e-09 -18.57° -18.54°
7 7.000e+03 1.290e-09 1.289e-09 -0.13° -0.10°
8 8.000e+03 1.568e-09 1.566e-09 -10.49° -10.46°
9 9.000e+03 1.659e-09 1.656e-09 -0.22° -0.19°
Total Harmonic Distortion: 0.000008%(0.000000%)

Here is the predicted frequency response and phase for the above case:
691430d1531366246-cheap-chips-opa1688-low-thd-muscle-amp-muscle-amp-2vpp-50ohms-phase-png


Here is the FFT for 8vpp into 50ohms:
691431d1531366246-cheap-chips-opa1688-low-thd-muscle-amp-muscle-amp-8vpp-50ohms-fft-png


And the corresponding Fourier components and THD:
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 4.006e+00 1.000e+00 -0.03° 0.00°
2 2.000e+03 9.238e-08 2.306e-08 -8.09° -8.06°
3 3.000e+03 4.755e-06 1.187e-06 -90.26° -90.22°
4 4.000e+03 7.334e-09 1.831e-09 -24.03° -24.00°
5 5.000e+03 1.277e-07 3.189e-08 -89.48° -89.45°
6 6.000e+03 3.201e-09 7.992e-10 -29.19° -29.16°
7 7.000e+03 4.595e-09 1.147e-09 -54.64° -54.61°
8 8.000e+03 3.582e-09 8.941e-10 -17.89° -17.86°
9 9.000e+03 3.472e-09 8.668e-10 -1.11° -1.08°
Total Harmonic Distortion: 0.000119%(0.000000%)

So now, just need to build a prototype and test it.

Update July 13, 2018: 3d render of 8 OPA1366 Muscle HPA stable driving 8vpp into 8ohms with low distortion - another superb layout by JPS64:
691597d1531488043-cheap-chips-opa1688-low-thd-muscle-amp-opa1688-muscle-amp-3d-render-view2-png


Here is double buffer prototype:
692991d1532195158-cheap-chips-opa1688-low-thd-muscle-amp-img_2376-jpgopa1688-parallel-buffer-2x-p3-jpg


692989d1532195158-cheap-chips-opa1688-low-thd-muscle-amp-img_2376-jpgopa1688-parallel-buffer-2x-p1-jpg
 

Attachments

  • Muscle-amp-sch.png
    Muscle-amp-sch.png
    47.8 KB · Views: 4,233
  • Muscle-amp-2vpp-50ohms-FFT.png
    Muscle-amp-2vpp-50ohms-FFT.png
    18 KB · Views: 3,952
  • Muscle-amp-2vpp-50ohms-phase.png
    Muscle-amp-2vpp-50ohms-phase.png
    22.6 KB · Views: 3,859
  • Muscle-amp-8vpp-50ohms-FFT.png
    Muscle-amp-8vpp-50ohms-FFT.png
    18.9 KB · Views: 4,102
Last edited:
Founder of XSA-Labs
Joined 2012
Paid Member
This will be a desktop amp, but in any case, it is Class AB and the OPA1688 is capable of driving 75mA max each and has a quiescent current of 1.6mA ea. Really designed for portable applications. At max draw under music the amp can provide 450mA of current swing. At idle it is 14 x 1.6mA (if I use the 1688 for the voltage gain stage as well) so that’s 22.4mA idle. Even in portable use this is not too bad as the Pocket Class A amp draws 120mA and works for hours on 600mAhr batteries. So at idle and not playing music, it’s good for 26hrs. Music on depends on type of music and how loud one plays. Probably still well over 12hrs per charge of the dual 9v Li-ions.
 
Founder of XSA-Labs
Joined 2012
Paid Member
Normally I go with 100uF 25v and 10uF X7R and 0.1uF X7R at each opamp IC at each rail. But with this many amps so close together and actually working in unison, and now possibly 8 opamps output, will be using a pair of 1000uF 25v per 4 IC’s (8 opamps) and will be using the WSON8 package with a power pad on the bottom for heatsink connection and each amp side having its own power pins, will have to let the amps share a common 1000uF for compactness. Each power supply pin has its own 0.1uF X7R cap though.

So for 4 IC’s (8 opamps) there are 2x 1000uF 25v, and qnty. 16 x 0.1uF 100v X7R.
 
Last edited:
Founder of XSA-Labs
Joined 2012
Paid Member
Thank you JPS64 for a nicely done layout! Looks superb. Can't believe how small the WSON8 packages are, they are about the size of a pair of 0805 resistors. Nicely packed in there for a headphone amp that is stable putting 8vpp into an 8ohm load with low distortion!

attachment.php
 

Attachments

  • OPA1688-Muscle-Amp-3d-render-view1.png
    OPA1688-Muscle-Amp-3d-render-view1.png
    405.7 KB · Views: 1,760
Last edited:
Founder of XSA-Labs
Joined 2012
Paid Member
Another view...

attachment.php


Here is draft schematic:
attachment.php


Btw, the amp can be all DC-coupled as the input/output coupling caps are optional. Just replace with a wire jumper. OP1688 is specified to have less than 1mV dc offset in spec sheet. You just need to make sure your PSU is reliable, or have a means to detect DC imbalance and shutdown. Or, if you happen to own HE6 planars, you don't care, as they can take 6wrms. :p
 

Attachments

  • OPA1688-Muscle-Amp-3d-render-view2.png
    OPA1688-Muscle-Amp-3d-render-view2.png
    564.2 KB · Views: 4,070
  • OPA1688-Muscle-Amp-draft-schematic-v1.png
    OPA1688-Muscle-Amp-draft-schematic-v1.png
    272.1 KB · Views: 3,162
Last edited:
Founder of XSA-Labs
Joined 2012
Paid Member
Here is an interesting surprise, I replaced the LME47720, the supposedly "king of the low THD" opamps in TI's arsenal, with the lowly OPA1688 (which is also serving as the buffer).

Here is the schematic:
attachment.php


If we give the power supply rails lots of caps so that the impedance is infinitessimally low, here is what we can get for 8Vpp into 8ohms:
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 4.006e+00 1.000e+00 -0.05° 0.00°
2 2.000e+03 1.442e-09 3.599e-10 -0.08° -0.04°
3 3.000e+03 2.163e-09 5.399e-10 0.07° 0.12°
4 4.000e+03 2.883e-09 7.199e-10 0.19° 0.23°
5 5.000e+03 3.604e-09 8.998e-10 0.29° 0.33°
6 6.000e+03 4.325e-09 1.080e-09 0.38° 0.42°
7 7.000e+03 5.046e-09 1.260e-09 0.47° 0.51°
8 8.000e+03 5.766e-09 1.439e-09 0.55° 0.60°
9 9.000e+03 6.487e-09 1.619e-09 0.63° 0.68°
Total Harmonic Distortion: 0.000000%(0.000000%)

I think that's PPB (part per billion) HD components.

The graph is unusual too (I have never simulated anything that has zero features before):
attachment.php


So I took it to 28Vpp into 32ohms, that's 2.5wrms nd enough juice to drive the most inefficient planar headphones, here are the FFT components:
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 1.402e+01 1.000e+00 -0.05° 0.00°
2 2.000e+03 8.877e-10 6.332e-11 -0.14° -0.10°
3 3.000e+03 1.331e-09 9.497e-11 -0.33° -0.29°
4 4.000e+03 1.775e-09 1.266e-10 -0.50° -0.45°
5 5.000e+03 2.219e-09 1.583e-10 -0.65° -0.61°
6 6.000e+03 2.662e-09 1.899e-10 -0.81° -0.76°
7 7.000e+03 3.106e-09 2.215e-10 -0.96° -0.91°
8 8.000e+03 3.550e-09 2.532e-10 -1.10° -1.06°
9 9.000e+03 3.993e-09 2.848e-10 -1.25° -1.20°
Total Harmonic Distortion: 0.000000%(0.000000%)

Here is the time series:
attachment.php


Here is the FFT:
attachment.php


Now, that's a "Muscle Amp".
 

Attachments

  • OPA1688-Muscle-Amp-8-parallel-all1688-FFT-schematic-v4.png
    OPA1688-Muscle-Amp-8-parallel-all1688-FFT-schematic-v4.png
    53.8 KB · Views: 2,032
  • OPA1688-Muscle-Amp-8-parallel-all1688-FFT-8vpp-8ohm.png
    OPA1688-Muscle-Amp-8-parallel-all1688-FFT-8vpp-8ohm.png
    25.2 KB · Views: 1,598
  • OPA1688-Muscle-Amp-8-parallel-all1688-Time-Series-24vpp-32ohm.png
    OPA1688-Muscle-Amp-8-parallel-all1688-Time-Series-24vpp-32ohm.png
    18.7 KB · Views: 1,612
  • OPA1688-Muscle-Amp-8-parallel-all1688-FFT-28vpp-32ohm.png
    OPA1688-Muscle-Amp-8-parallel-all1688-FFT-28vpp-32ohm.png
    19.6 KB · Views: 1,620
Last edited:
Founder of XSA-Labs
Joined 2012
Paid Member
These simulations are assuming perfect power supply of course with limitless current and zero impedance. If we add series resistance of the power supply as 12mOhm (ESR of OSCON cap), 8vpp into 50ohms has a bit more distortion, but still very low. So it looks like having lots of low ESR caps on the rails here will help.

8Vpp into 50ohms:
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 4.006e+00 1.000e+00 -0.04° 0.00°
2 2.000e+03 1.784e-07 4.454e-08 -178.41° -178.36°
3 3.000e+03 2.205e-09 5.504e-10 -1.72° -1.67°
4 4.000e+03 6.898e-08 1.722e-08 -175.87° -175.83°
5 5.000e+03 3.674e-09 9.171e-10 -0.80° -0.76°
6 6.000e+03 4.180e-08 1.044e-08 -173.21° -173.17°
7 7.000e+03 5.142e-09 1.284e-09 -0.33° -0.28°
8 8.000e+03 2.836e-08 7.080e-09 -170.04° -170.00°
9 9.000e+03 6.611e-09 1.650e-09 -0.00° 0.04°
Total Harmonic Distortion: 0.000005%(0.000000%)

Here is 28 Vpp into 50ohms (or 2wrms) - distortion looks almost identical:
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 1.402e+01 1.000e+00 -0.04° 0.00°
2 2.000e+03 6.241e-07 4.451e-08 -178.41° -178.36°
3 3.000e+03 8.300e-09 5.920e-10 -0.35° -0.31°
4 4.000e+03 2.407e-07 1.717e-08 -175.86° -175.81°
5 5.000e+03 1.383e-08 9.866e-10 -0.13° -0.08°
6 6.000e+03 1.452e-07 1.035e-08 -173.14° -173.10°
7 7.000e+03 1.936e-08 1.381e-09 0.00° 0.05°
8 8.000e+03 9.774e-08 6.972e-09 -169.83° -169.78°
9 9.000e+03 2.489e-08 1.776e-09 0.10° 0.14°
Total Harmonic Distortion: 0.000005%(0.000000%)

And here is series resistance of 0.15ohms for the power supplies and 28Vpp into 50ohms:
[Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 1.402e+01 1.000e+00 -0.04° 0.00°
2 2.000e+03 7.864e-06 5.609e-07 -178.42° -178.37°
3 3.000e+03 8.375e-09 5.974e-10 -5.85° -5.80°
4 4.000e+03 3.135e-06 2.236e-07 -176.02° -175.97°
5 5.000e+03 1.392e-08 9.926e-10 -3.43° -3.39°
6 6.000e+03 2.004e-06 1.430e-07 -173.77° -173.73°
7 7.000e+03 1.946e-08 1.388e-09 -2.37° -2.32°
8 8.000e+03 1.473e-06 1.051e-07 -171.53° -171.49°
9 9.000e+03 2.501e-08 1.784e-09 -1.75° -1.70°
Total Harmonic Distortion: 0.000063%(0.000000%)
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.