An Improved Dual Class-A Output: The "DCAO2"

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
mediatechnology
Posts: 3600
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: An Improved Dual Class-A Output: The "DCAO2"

Post by mediatechnology » Mon Oct 10, 2016 8:11 pm

I have a construction thread for the Dual Class-A II DCAO2 and have been busy writing the instructions, ordering a test bill-of-materials and construction of a second unit.
See viewtopic.php?f=7&t=825

With a second unit built and tested I decided to do some additional measuring.
One of the questions I have is how much power is available for 'phones more difficult to drive than my 106 dB/mW Sony MDR-7506s.

There are combinations of impedance and low sensitivity that are pretty demanding.
Beyer and HiFi Man come to mind.
Some, such as the Beyer DT-880 Pro (available in 32, 250 and 600Ω) require lots of drive voltage because they are 600Ω, others such as the HiFi Man HE6 are in the 30-50Ω range with low 83.5 dB/mW sensitivity requiring almost 5W to produce 120 dB SPL.

Today I ran some maximum power levels for various load impedances.
The DCAO2 can be run in mono as fully-balanced output in a bridge-tied-load "BTL" configuration as well as stereo with single-ended "SE" outputs.

The following maximum power outputs were measured with +/-15V supplies just below clipping.
The DCAO2 is operating closed loop.


15Ω SE 7.5 Vrms 3.75W
15Ω BTL 12.1 Vrms 9.76W

30Ω SE 8.1 Vrms 2.1W
30Ω BTL 14.8 Vrms 7.3W

60Ω SE 8.7 Vrms 1.26W
60Ω BTL 16.1 Vrms 4.32W

120Ω SE 9 Vrms 675 mW
120Ω BTL 17.3 Vrms 2.49W

>120Ω the maximum output is 9 Vrms SE, BTL >17.3 <18 Vrms.

Measuring distortion presents additional challenges because at all but the higher power levels the DCAO2's distortion is less than the generator and analyzer.
In my next post I intend to show distortion measurements made using null testing where one channel of the DCAO2 is used to measure the other.
With null testing I'm able to see distortion products in the -130 (0.3 ppm) to -120 dB (1 ppm) range.

User avatar
mediatechnology
Posts: 3600
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

DCAO2 Distortion Measurement Using the Null Test Method

Post by mediatechnology » Thu Oct 13, 2016 3:13 pm

Measuring the distortion of the DCAO2 Dual is a bit of a challenge because its low and I'm too cheap to buy an AP or DScope.

The following FFT illustrates the problem.
The green trace is the input residual and the red trace the output.
The distortion contribution at 750 mW into 30Ω appears to measure only 0.0001% different.
It appears that the DCAO2 is actually subtracting the second harmonic.

Generator versus DCAO2 Distortion Signatures
Image
Dual Class-A II DCAO2 1 kHz THD at 750 mW into 30Ω. Input is green, output is red.

Null testing, by subtracting the input fundamental and the generator residual distortion from the output, allows us to see far lower levels and measure the actual distortion contribution.

By driving the inverting "-" input of the Dual Class-A II balanced input its possible to passively subtract the input from the output and measure the difference in distortion.

Originally I used the other channel's line receiver to perform subtraction which works equally well at 1 kHz but doesn't allow capacitive trimming to increase the HF null.

The null depth is adjustable by using the DCOA2's input trim.
Phase shift at frequency extremes limit the available null depth.
At 10 kHz the phase shift is still relatively low and can be compensated by a low value capacitor in parallel with one leg of the resistive "subtractor."
At the 20 Hz extreme the DCAO2's coupling capacitor requires bypassing to obtain a reasonable null.

This FFT is the nulled subtracted output showing the residual distortion:
Image
Dual Class-A II DCAO2 1 kHz THD at 750 mW into 30Ω. Null test. Output minus input.

The 1 kHz fundamental and generator distortion are nulled by approximately 70 dB.
Since the generator's distortion is in the -70 to -90 dB range and then rejected by 70 dB it's distortion is in the noise floor.
The remaining harmonics are those contributed by the DCAO2.

Once the fundamental and harmonics are nulled by 20-30 dB their contribution to the measured distortion is small.
When the fundamental is sufficiently low, the measured distortion products do not change appreciably with additional null depth.

The resistive summing network (1K) has a 6 dB loss.
To compensate for the network loss the measurement scale is adjusted upward by 6 dB to match the output reference level.

The second harmonic in the FFT above is very small, about -112 dB from the fundamental.
The third sits at about -110 dB; the fifth about -115 dB.

The power sum of the first four harmonics is -107 dB, 4.5 ppm or 0.00045%.

The 1 kHz THD of the DCAO2 is 0.00045% at 750 mW.

At 10 kHz 750 mW into 30Ω distortion rises as expected because there's lower open loop gain.
Image
Dual Class-A II DCAO2 10 kHz THD at 750 mW into 30Ω. Null test. Output minus input.

The third harmonic of 10 kHz is at -93 dB; the second at -113 dB.
The power sum of H2 and H3 are about -92.9 dB which is 22.6 ppm or 0.00226% THD.

At 100 mW power levels null testing finds the THD needle in the haystack.

DCAO2 10kHz FFT at 100 mW into 30Ω:
Image
Dual Class-A II DCAO2 10 kHz THD at 100 mW into 30Ω. Null test. Output minus input.

The third harmonic is not observed. H2 is about -118 dB or 1.3 ppm.
THD at 10 kHz 100 mW into 30 Ω is approximately 0.00013%.

DCAO2 1kHz FFT at 100 mW into 30Ω:
Image
Dual Class-A II DCAO2 1 kHz THD at 100 mW into 30Ω. Null test. Output minus input.

The third harmonic is a blip at about -110 dB. H2 is about -115 dB. The power sum is -109 dB or 3.5 ppm
THD at 10 kHz 100 mW into 30 Ω is approximately 0.00035%

DCAO2 Dual Class-A II Distortion at 100 mW into 30Ω with the output running open loop:

DCAO2 1 kHz 100mw Open Loop Output
Image
Dual Class-A II DCAO2 1 kHz THD at 100 mW into 30Ω with the output running open loop. Null test. Output minus input.

DCAO2 10 kHz 100mw Open Loop Output
Image
Dual Class-A II DCAO2 10 kHz THD at 100 mW into 30Ω with the output running open loop. Null test. Output minus input.

In the 10 kHz FFT above a small capacitor was placed in one leg of the summing point to compensate for phase shift and further reduce the null point.
The reduced null depth has no effect on the measured harmonic levels.

The DCAO2 third harmonic distortion at both 1 and 10 kHz with the output open loop is around -79 dB, 112 ppm, or 0.0112%. (100 mW 30Ω load.)

Conclusion

The 100 mW 1 kHz closed loop versus open loop numbers are 0.00035% versus 0.0112%.
At 750 mW 1 kHz closed loop the distortion is 0.00045%.

There is a big difference in closed loop versus open loop output impedance at 40 mΩ (at 20 kHz it's still only 70 mΩ) versus 2Ω open loop.

ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

Re: An Improved Dual Class-A Output: The "DCAO2"

Post by ricardo » Fri Oct 14, 2016 3:52 pm

mediatechnology wrote:The new LME49720s I recently bought are improved over the previous lots.
You mean no 1/f noise? :o

User avatar
mediatechnology
Posts: 3600
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: An Improved Dual Class-A Output: The "DCAO2"

Post by mediatechnology » Fri Oct 14, 2016 4:50 pm

You mean no 1/f noise? :o
Good to hear from you Richard.

Not that I can see at 12 dB gain or looking at the output Vos in the uV region.
But, I haven't tried it at high gain.

User avatar
mediatechnology
Posts: 3600
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: An Improved Dual Class-A Output: The "DCAO2"

Post by mediatechnology » Tue Oct 18, 2016 3:32 pm

I measured the Dual Class-A Output DCAO2 CCIF distortion using the null test method.

The 19 and 20 kHz carriers are mixed 1:1 and then buffered before feeding the DCAO2.
The source buffer (an NJM2114) introduces some additional distortion - it turns out to be minor - but that distortion is nulled by the test.

In the FFT below the two carriers are calibrated so that their peak levels are each at -6 dB.
The output power, 750 mW into 30Ω, is an rms measurement.

The DCAO2 is fed at its inverting input so that null can be obtained by resistive summation.
The input and output are summed with 1K resistors and a null is adjusted by the DCAO2's gain trim.
The carrier null is about -50 dB.

The null summation point then feeds the analyzer which is the FFT shown below.
The 18 and 21 kHz sidebands caused by intermodulation distortion within the DCAO2 are at their actual -100 dB level.
The op amps in the DCAO2 are LME49720s.

Image
DCAO2 CCIF Distortion 19/20 kHz 1:1 using the null test method at 750 mW into 30Ω.

The two sidebands are at -100 dB which works out to be 0.002% at 750 mW into 30Ω.

User avatar
JR.
Posts: 2643
Joined: Sat May 24, 2008 7:21 pm

Re: An Improved Dual Class-A Output: The "DCAO2"

Post by JR. » Tue Oct 18, 2016 4:38 pm

Nice, does your display reveal a spur at 1kHz? I can't tell from the frequency axis.


My recollection my old bench 19:20kHz IMD test rig ignored all the HF sidebands (harmonic distortion beating?) and measured the primary 1Khz IMD product from 20 kHz- 19kHz.

I doubt I could measure as low as you can either... but this was 30 odd years ago (very odd).

JR

User avatar
mediatechnology
Posts: 3600
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: An Improved Dual Class-A Output: The "DCAO2"

Post by mediatechnology » Tue Oct 18, 2016 4:53 pm

Thanks John!

This is the wideband FFT display of the DCAO2:

Image
DCAO2 CCIF Null Test 750 mW 30Ω Wideband Display

There's no detectable 1 kHz beat, only the third-order sidebands at 18 and 21 kHz. (2F1-F2 and 2F2-F1).

At 100 mW levels the products are in the noise floor so I had to use 750 mW.

User avatar
JR.
Posts: 2643
Joined: Sat May 24, 2008 7:21 pm

Re: An Improved Dual Class-A Output: The "DCAO2"

Post by JR. » Wed Oct 19, 2016 1:24 pm

Back in the day I was using the 1 kHz spur to design phono preamps with... Many designs delivered respectable THD because of the RIAA roll-off at HF but the 1kHz product from 19:20k IMD received 20 dB more gain than the test stimulus. My idea of a good test. :D

Have you used this on any of your modern RIAA paths? (Excuse me if I forgot).

JR

User avatar
mediatechnology
Posts: 3600
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: An Improved Dual Class-A Output: The "DCAO2"

Post by mediatechnology » Wed Oct 19, 2016 3:49 pm

That is an excellent suggestion.

I haven't tried it and have been looking for a good way to quantify distortion in an EQ'd environment.
I'm thinking it may reveal differences in the flat preamp's op amps.

Post Reply