Zarcon Dee Grissom's Idea Page
EV9
Updated 22DEC2010
M.C.C.HPA17r16 icon Home > IdeasHPA17r16 Intro.
HPA17r16 Operating Environment.
HPA17r16 RFI filters.
The Main Amplifier Board.
HPA17r16 Vmon & Sources.
HPA17r16 Tests & Afterthoughts.
M.C.C. HPA17r16 Tests and Afterthoughts.
When you have eliminated the impossible, whatever remains, however improbable, must be the truth. - Arthur Conan Doyle
Main Menu
Home

Ideas

Meet the computers

About Me

Links

Idea#017 Page 6 of 6

Test set up... and charts. 

My testing environment, vs space flight hardware testing procedures.
I am not NASA, LANL, LLNL, or Sandia National Laboratories, I wish! I do believe in testing equipment under the conditions that resembles the environment it will be used in, as closely as possible. I have significant resentment of certain manufacturers running tests with the maximum allowable supply voltage and input signal, and deep-freezing there amplifiers just to out-spec there competitors. These conditions far exceed the environment that any customer would use there product at. I don't know about you, I have never seen anybody driving there car to the local corner store with there trunk filled with liquid nitrogen sloshing around just to cool there car amplifier.

Cables, and dummy load.
Headphone SPICE model.
SPICE model
Modified from a speaker model.
Tangentsoft Based Headphone Dummy load
Very simple unit.
Dummy load guts
640x480
1024x768
I was unable to find a true inductive headphone/speaker dummy load, or even a headphone "Xz" chart to create one with at first. Dose it matter? There are A lot of people that use resistor based dummy loads. The dummy load presented at Tangentsoft's site was easy to build in a particular locally purchased breath mint container. I used a 39ohm and a 100ohm resistor set, as they closely resemble the range of headphones I expect to subject this amplifier to. The "no-load" setting on the dummy load gives a false impression of the test setup, as the impedance on the sound card line-in is 47k-ohms not infinite resistance (just being specific).

Scientific method; The test results must be reproducible!
What is "Margin of error"? Dose this amp outperform the sound card, and Are "Normalized Charts" cheating? "Normalized Charts" are a moot point, as RightMark's RMAA software dose not create them, and I am not going to waste my time trying. The important question is, how close dose the EUT (Equipment Under Test) measurements when inserted into the test loop, match the measurements of the test loop without the EUT in the chain.

As for margin of error, this can clearly be seen in the THD, IMD, Dynamic range, and crosstalk charts, when the four worst cases of the test loop baseline are overlaid.
Audigy 2 ZS THdAudigy 2 ZS IMD+N
Audigy 2 ZS Dynamic rangeAudigy 2 ZS Stereo crosstalk
The only thing different in the test conditions between these four runs, was when they were run. I ran a test, saved the results, and ran the test again without adjusting anything. As you can see, the lower frequency range of the crosstalk chart swings wildly. I may have touched one of the cables with my foot between the third and fifth run, changing the proximity of the wires inside my dummy load, altering the upper crosstalk results a bit on the fifth run. The other charts appear to have a gradually decreasing accuracy with decreasing frequency.

Limits of a sound card for the test setup?
The audigy2 ZS best THD+N is only 0.0015% (two zeros) at 16bit 44.1kHz sample rate according to Xbit Labs. I measured 0.0033% to 0.0031% THD+N with my card at 24bit 96kHz, using the above dummy load in the "no-load" setting. The OPA2134 can deliver 0.00008% THD (four zeros) under ideal conditions (excluding the unknown THD of the buffer), and the RFI filters may be removing allot of noise from the test loop.

I was expecting one of two results from the tests.
1: this amplifier will perform horrible, and do worse then my desired far-better then 0.02% THD+N.
or
2: This amplifier will produce results far better then the base-line results of just the sound card. Forcing me to throw out the test results, and request testing at a better equipped facility.

Test Conditions
My 22inch CRT monitors run at about 92F at 75F room temperature. The UPS unit the EUT is sitting on was running at about 94.3F, and produced a magnetic field similar to the degauss windings in my monitors. Room temperature was at 75.1F, and the EUT settled at 84.9F after two days being powered up on top of the UPS, and didn't deviate from that more then 0.5F for an additional three days before testing. The EUT was connected to the main 13.8VDC bus with the same wiring and cables that the EUT will use in its final mounting location on the mission control console. The main bus was at 13.635VDC during the test, which is about the normal voltage +/-20mVDC.
Test setup
Audigy 2 ZS settings

I set the volume settings on the sound card for the loop-back baseline test to RightMark's suggested -1.0dB, then used the volume knob on the amplifier to adjust for the gain of the amplifier. The closest I was able to get the Audigy loop-back baseline level to -1.0dB with the windows volume control thing, was -1.2dB Left and -1.4dB right. The much finer control of the volume knob on the amplifier allowed me to test the amplifier at -1.0dB left and -1.2dB right. The difference between the left and right channel is from something inside the sound card.

Test Results
I ran the test ten times for each load setting (baseline, 47kohm, 100ohms, and 39ohms), and compiled the worst results according to the THD+N field for the following charts. I am using "HPA17r16b" to indicate that this is the amplifier with the revised RFI (revision-1a) filters and +2dBV gain, and not the originally designed +23dBV gain, input, and output RFI filters.

RightMark RMAA screencap (638x355)
Tested at 24bit 96kHz sample rate.
Audigy2ZS
Baseline (47kohm)
Baseline Icn
HPA17r16b
47kohm
testloop Icn
HPA17r16b
100ohms
testloop Icn
HPA17r16b
39ohms
testloop Icn
Frequency response, dB [multitone] +0.01, -0.10 +0.01, -0.09 +0.01, -0.10 +0.01, -0.11
Noise level, dBA -90.2 -90.2 -90.1 -90.1
Dynamic range, dBA 90.0 90.1 90.1 90.1
Total Harmonic Distortion + noise (THD+N), % 0.0033 0.0033 0.0034 0.0032
Intermodulation Distortion + noise (IMD+N), % 0.0094 0.0098 0.0100 0.0099
Stereo crosstalk, dB -80.7 -81.4 -81.6 -82.6

RightMark RMAA Graphs
HPA17r16b Frequency responseHPA17r16b X-talk
HPA17r16b IMDHPA17r16b THD
HPA17r16b NoiseHPA17r16b D-range

Trying to understand the test results.
The amplifier increased the IMD+N by 0.0005% +/-0.0001 over the baseline results of the test loop. The Noise level and Dynamic range was between (-)90.0dBA to (-)90.2dBA for all tests. So the amplifier can not add more noise then what the program would round out to the given range. I got 0.0033% to 0.0031% THD+N with the test loop. I am yet to see anything beyond 0.0032% through 0.0031% at 39 ohms load, even after an additional 20 runs of the measurement. The THD+N is influenced by the noise level and some rounding of some kind must be used for the digits beyond what the program displays, which may account for the results. However for that to happen, the THD+N of the amplifier would have to be below 0.00004999%, implying a noise level well beyond -126.022dBA. That is extremely unlikely as the OPA2134 is specked at 0.00008% THD+N. I don't know if modest interstage matching and ensuring clean stable power to an op-amp, can make that much of a difference.

Given that the results of the amplifier and the test-loop matches the results of the test-loop by itself within the margin of error of the test-loop, the performance of this amplifier must be somewhere within the margin of error to the point that it does not cause some measurements to exceed the range of the margin of error.
E.G. Beyond the measurable limits of the test setup.

Conclusion
The first thing I noticed while performing the tests, was the Left channel was always 0.2dB louder then the right channel on every test run. Tho this is probably within the tolerance of most components on the sound card, and is present without the amplifier in the test loop, it gives me some doubts about the crosstalk figures. The imbalance is not caused by the amplifier, it is in the sound card.

I also noticed a difference in the noise floor between the "Noise Level" and the THD, IMD, and the Dynamic range graphs. The noise floor dose not change shape, it merely drops 4dB when the sound card is not sending out a signal. Again this may very well be within the margin of error for the sound card components, and is not caused by the amplifier.

There is a slight gradual increase in the crosstalk above 5kHz over the test loop baseline results. This is due to the crosstalk or magnetic coupling of the inductors on the input RFI filter. This may be mended by placing separate MuMetal shields over each channels inductors. As I need to remake this filter PCB for the revised filter layout anyways, I'll hold off on the shield till latter.

Even tho I have doubts about the test equipment, the amplifier performed respectfully under the crude test conditions. The revised input and output RFI filters perform better, they still need some work. That about sums up my thoughts on the test results.


Post prototype thoughts.

Op-Amps.
From the beginning, I was undecided over what op-amp to use in this amplifier. The OPA2132 is almost the same thing as the OPA2134. There are other op-amps as well that have slightly better THD ratings (OPA1642, OPA2227, OPA2228, etc). Ultimately I was restricted by the operating power voltage, and the 8-pin DIP for the prototype. If I decide to go with an op-amp that is not a Burr-Brown part, I will no longer have the nostalgia of having all Burr-Brown amplifier stages. This is still up in the air.

Future SMD and miniaturizing the boards.
I still have hopes of squeezing this amplifier into a floppy drive bay chassis. One limiting factor is getting the traces laid out similar to the original, and keeping some ground plain between them. Another is not sacrificing the stability of the ground plain, to squeeze everything onto a smaller board.
SMD Cad Smple

Alternative topologies.
I may still ditch the op-amp and buffer combination, in favor of an instrumentation-amplifier and a DSL line driver. I forget where I read about the idea, I am in strong favor of giving it a shot.

Adjustments after testing.
There are some things I intend to do, to further improve the prototype. I will not post the test results of the future upgrades, as I doubt most will care to go to this extent. One being soldering the SMD 220pf NPO caps directly onto the buffer power pins. Another being, Faraday caging the Burr-Brown op-amps [Example right].
Future V-mon Op-Amps
Future V-mon Op-Amps
640x480

T.B.C. some day.

<PG5 (Vmon & Sources)< - ^Ideas^ - >HOME>

Home > IdeasHPA17r16 Intro.
HPA17r16 Operating Environment.
HPA17r16 RFI filters.
The Main Amplifier Board.
HPA17r16 Vmon & Sources.
HPA17r16 Tests & Afterthoughts.

Valid XHTML 1.0
My Email link. copy and past. zarcondeegrissom@yahoo.com  
or is the status "coffee is good" on yahoo messenger