A New EC-Composite LM3886 Amp

After few days of testing, modifications, testing, and improvements. Finally I got EC/Comp 2XLM3886 to work. The amp went thru all these hard tests, like clipping, capacitance load (1uF).

This is the final schematics.
 

Attachments

  • Fig157 P-LM3886 4 fianl sch .jpg
    Fig157 P-LM3886 4 fianl sch .jpg
    212.8 KB · Views: 198
Using the SMPS, the maxi power at 4 ohm is 75W.
In the figure, it is the compare with TI PA100 amp. Even using my primitive test gears and power supply, the amp has better THD+N than TI PA100.

The measured THD+N is approaching the limits of the Solo 3rd, these numbers might be mainly caused by the Solo 3rd itself.
 

Attachments

  • Fig158 2xLM3886 THDN .jpg
    Fig158 2xLM3886 THDN .jpg
    56.3 KB · Views: 79
Last edited:
Using 600W, 120/double 24VAC toroidal transformer, 140,000uF CRC smoothing, at +/-32VDC, the maximal output power is 95W at 4 ohm.

The THD+N at lower power level is much higher than TI figures because of noise. But above 30W, it has less THD+N than TI PA100.
 

Attachments

  • Fig159 2xLM3886 in case .jpg
    Fig159 2xLM3886 in case .jpg
    55.2 KB · Views: 98
  • Fig160 2xLM3886 THDN in case .jpg
    Fig160 2xLM3886 THDN in case .jpg
    87.2 KB · Views: 95
Can someone explain in a sentence of two what the error correction component of these amps is? Is it just negative feedback applied, or something else? I've seen the term a few times, but never understood what it means with regards to these types of chip amps.
 
This is my understanding. It could be wrong.

Monitoring V(b) at +in of LM3886 of EC amp, it is very low. I think it is only the error signal.
But for Comp amp, it is almost equal to the input signal.
 

Attachments

  • EC.png
    EC.png
    13.6 KB · Views: 81
  • EC wave.png
    EC wave.png
    155.5 KB · Views: 84
  • Comp.png
    Comp.png
    12.1 KB · Views: 71
  • Comp Wave.png
    Comp Wave.png
    169.4 KB · Views: 85
New toys.

Topping E70 DAC and Cosmos ADCiso.

Topping THD+N: OEM (XLR) 0.00008%, Audiosciencereview 0.000094%

ADCiso(B): THD+N: Audiosciencereview 0.000098-0.000138%

My setting up:
Topping RCA -> ADCiso 2.5mm mono: 0.0002%
Topping XLR -> ADCiso XLR mono: 0.000091%
On par with OEM and some reviewers.
 

Attachments

  • E70 to ADCiso RCA Mono.jpg
    E70 to ADCiso RCA Mono.jpg
    65.7 KB · Views: 70
  • E70 to ADCiso XLR Mono Floor.jpg
    E70 to ADCiso XLR Mono Floor.jpg
    62.6 KB · Views: 73
  • 20250121_161458.jpg
    20250121_161458.jpg
    425.1 KB · Views: 72
  • 20250121_222021.jpg
    20250121_222021.jpg
    351.1 KB · Views: 69
  • E70 to ADCiso XLR Mono.jpg
    E70 to ADCiso XLR Mono.jpg
    65.4 KB · Views: 71
Measurements:

On par with Modulus-286.

The biggest problem is that my tests show a very high background noise at 60Hz. This might be caused by the power supply and my SE buffer at the input( compare to Tom's balanced input buffer)
 

Attachments

  • Fig180 EC 8ohm 40w.jpg
    Fig180 EC 8ohm 40w.jpg
    68.9 KB · Views: 96
  • Fig179 EC 4ohm 40w.jpg
    Fig179 EC 4ohm 40w.jpg
    78.4 KB · Views: 80
  • Fig178 comp 8ohm 40w.jpg
    Fig178 comp 8ohm 40w.jpg
    76.2 KB · Views: 77
  • Fig177 comp 4ohm 40w.jpg
    Fig177 comp 4ohm 40w.jpg
    70.1 KB · Views: 78
  • Fig176 THD.jpg
    Fig176 THD.jpg
    61.8 KB · Views: 92
  • Fig176 Compare.jpg
    Fig176 Compare.jpg
    77.9 KB · Views: 93
I have been listening to the amp for months. I can't hear any difference.

They are very similar sound, pure and straight.

It is my personal subjective feeling that the EC mode sounds a little bit more rounded in high if you put your ears close to tweeter. It might to be its H2 dominance than the comp mode.
 
#45, single LM3886 EC amp measurement with old Focusrite Solo 3rd.
#49, EC/Comp 2xLM3886 amp measured with the new ADCiso.
It does not look very convincing.
Here is why:
The 3rd harmonic in the picture EC 1654 40W4Ohm in post #45 is 0.000086%. And you said that is was measured using Focusrite Solo 3.
The picture EC/Comp 2xLM3886 in post #49 shows the third harmonic to be 0.000063%.
The difference between these two is 0.000023%
This is impossible as this means that Focusrite and the difference in the amps itself have contributed 0.000023%.
BUT this audio interface alone has much higher 3rd harmonic, around 30dB higher.
How would you explain that?
 
@Gonkar : "The difference between these two is 0.000023% ... Focusrite and the difference in the amps itself have contributed 0.000023%"

This isn't how harmonic distortion adds up.

For example you can have measurement equipment with 1% 3rd harmonic distortion and a device-under test (DUT) with 1% 3rd harmonic and together they can measure 0%! Why? Because if the phase of the 3rd harmonic of one of the devices is 180 degrees relative to the other, that frequency component will cancel out and you will measure 0.

This is why as a rule-of-thumb the measurement equipment must be an order of magnitude better than the DUT, otherwise you can't separate the measurement of the DUT from that of the measurement equipment.
 
For example you can have measurement equipment with 1% 3rd harmonic distortion and a device-under test (DUT) with 1% 3rd harmonic and together they can measure 0%! Why? Because if the phase of the 3rd harmonic of one of the devices is 180 degrees relative to the other, that frequency component will cancel out and you will measure 0.
I'm sorry but I can agree with that assumption.
First, why there should be phase difference.
Second, the odd harmonics don't cancel each other. In some cases even harmonics can be canceled but not the odd ones. The even and odd harmonics represent different non-linearity.
This is why as a rule-of-thumb the measurement equipment must be an order of magnitude better than the DUT, otherwise you can't separate the measurement of the DUT from that of the measurement equipment.
I completely agree with that but not because of the argument in the first part of your post.
 
Second, the odd harmonics don't cancel each other.
Not correct. When you have odd harmonics from, say, a symmetic expanding transfer function then a symmetric but inversly compressing transfer function can cancel it. This no different than with even order harmonics.
1738524976705.png


Code:
Fourier components of V(out1)
Harmonic	Frequency	 Fourier 	Normalized	 Phase  	Normalized
 Number 	  [Hz]   	Component	 Component	[degree]	Phase [deg]
    1   	 1.000e+3	 1.196e+0	 1.000e+0	  -89.96°	    0.00°
    2   	 2.000e+3	 6.426e-9	 5.373e-9	  -83.83°	    6.13°
    3   	 3.000e+3	 4.345e-1	 3.633e-1	   90.25°	  180.21°
    4   	 4.000e+3	 1.280e-8	 1.070e-8	  -86.22°	    3.74°
    5   	 5.000e+3	 2.326e-1	 1.945e-1	  -89.45°	    0.51°
    6   	 6.000e+3	 1.917e-8	 1.603e-8	  -86.72°	    3.24°
    7   	 7.000e+3	 9.247e-2	 7.733e-2	   91.18°	  181.14°
    8   	 8.000e+3	 2.553e-8	 2.135e-8	  -86.74°	    3.22°
    9   	 9.000e+3	 2.761e-2	 2.309e-2	  -87.35°	    2.61°
Total Harmonic Distortion:   41.994570%

Fourier components of V(out2)
Harmonic	Frequency	 Fourier 	Normalized	 Phase  	Normalized
 Number 	  [Hz]   	Component	 Component	[degree]	Phase [deg]
    1   	 1.000e+3	 5.999e-1	 1.000e+0	   90.03°	    0.00°
    2   	 2.000e+3	 6.432e-9	 1.072e-8	   96.64°	    6.61°
    3   	 3.000e+3	 1.146e-4	 1.910e-4	    0.73°	  -89.30°
    4   	 4.000e+3	 1.280e-8	 2.134e-8	   93.99°	    3.97°
    5   	 5.000e+3	 1.119e-4	 1.866e-4	 -178.00°	 -268.03°
    6   	 6.000e+3	 1.918e-8	 3.197e-8	   93.40°	    3.37°
    7   	 7.000e+3	 6.771e-5	 1.129e-4	    2.60°	  -87.43°
    8   	 8.000e+3	 2.554e-8	 4.257e-8	   93.33°	    3.30°
    9   	 9.000e+3	 2.613e-5	 4.355e-5	 -177.01°	 -267.04°
Total Harmonic Distortion:   0.028444%

EDIT: And as soon as you have some measurement experience you will see this (partial) cancelling of some of even or odd harmonics happens all the time, typically from the ADC having the inverse error that the DAC at specific levels and frequencies.