Buffer distortion data update

I have updated my buffer distortion measurements. The new data was taken in the META42 amp using the AD8620 op amp but with a higher inner loop gain than before (around 333), which benefits all but the best buffers. I have shown only the data for low resistive load since the distortion was always much lower than with a higher resistance load. I also boosted the level of my signal generator a bit from before which lowers the noise floor and gives lower THD. The first data was taken at 1v output but I measured several buffers ranging from 0.5v to 4v and the data was very close to the same. The measurement was made at 1KHz, but I have not seen much variation due to frequency.

With 31 ohm load, 1.0v output:
BUFFER,% THD,2nd, 3rd, 4th, 5 harmonic
EL2001,0.080-69-64-83-69
EL2001 x 2,0.069-74-65-87-70
EL2001 x 3,0.056-76-66-87-72
EL2002,0.040-71-70-85-76
EL2002 x 2,0.026-75-76-89-80
EL2002 x 4,0.021-79-83-91-84
HA5002, 0.043-83-69-89-74
HA5002 x 2,0.027-82-79-90-81
HA5002 x 4,0.021-82-86-90-84
BUF634 x 2,0.023-79-83-92-87
EL2008,0.023-79-78-90-83
EL2009,0.021-83-91-91-84

I also measured them with a real headphone load, a cheap pair of Panasonic headphones that came with a PCDP. The level was 0.5v, which was very loud. You can see the distortion is higher with a real load, but the buffers that did the best with a resistive load also performed better with the headphone load.

With cheap Panasonic headphones, 0.5v output:
BUFFER,% THD
EL2001,0.150
EL2001 x 2,0.104
EL2001 x 3,0.086
EL2002,0.066
EL2002 x 2,0.038
EL2002 x 4,0.027
HA5002, 0.067
HA5002 x 2,0.034
HA5002 x 4,0.027
BUF634 x 2,0.029
EL2008,0.027
EL2009,0.025

Thanks to Geom_tol for the loan of the EL2009 for these tests. It wins. It's also discontinued and now very expensive if you can even find them. Some of the other buffers like the HA5002 and the BUF634 aren't too far behind as long as you use multiple buffers, and they are still currently available. The BUF634's I used were in a TO220 package and had the WB pin tied to V-, the HA5002s were DIPs.

I should point out that only the EL2001 and EL2002 will plug into the META42 board directly.

Thanks for the follow-up, Kurt. Good stuff, as usual. What do you have the inner loop and outer loops set to in this test? I'm wondering just how high you went on the inner, in particular, as the last test stopped at about 200, IIRC.

I need to get a minilyzer .

Thanks again!

Not to criticize, but how come you didn't test HA-5002 in single configuration? I found EL2009 to have incredible control over load and the tightest bass you'll ever get but it is a bit dark. HA-5002 on the other hand is the most balanced chip I've heard.

Thanks Kurt!

It is interesting to note that the BUF634 comes out slightly better than the HA5002 in the distortion test, yet almost everyone who has heard both chips prefers the sound of the HA5002 over the BUF634.

Quote:



Quote:



Quote:



eh?

I've edited the post to address the first two issues.

aos, I didn't have data for a single 5002 because I was short of headers. I was disappointed in the data I took earlier with a single 5002 (with lower inner loop gain) so I used the headers I had to build dual and quad parts for testing. I found another header this morning so I used my last 5002 with the new header and added the data. Due to my header and chip shortage, it was not possible to do A/B comparisons of the sound of one vs two vs four 5002's.

I didn't add my previous disclaimer that there is more to how these sound than how low the distortion measures but it's probably worth repeating. It would be interesting to get more of people's inputs on a comparison of the sound of these buffers under similar conditions. I know it's been discussed elsewhere in various threads in some detail, but it would be nice to get a summary in one place.

Kurt/Morsel-

I have a multiloop amp on protoboard that I have tried both HA5002x2/channel and BUF634x2/channel in, and I actually do prefer the buf634s, for once. This was recent testing with a heavier PS (15.5v and 1A available). Loved them.

In fact, I was a bit bummed that the HA5002s weren't better (more affordable), but this was just me listening for a half hour or so. I have made no measurements.

Thanks kurtw! Apparently single chip does just fine on its own distortion-wise (I only listened to single chips so far). But I measured far greater power consumption than from EL2001 - 20mA for two of them which is higher than in datasheet (in line with max over full temperature range), compared to less than 3mA for two of EL2001 - which is even less than 1.8mA per chip in datasheet.