Sounding of Operational Amplifiers, part II.
Part I can be found here. There was an interesting situation in the process of receiver's modification. Our part's supplier had been out of stock for OPA2134 and we have decided to offer OPA2132 as the replacement for JRC2068, those have almost the same specifications as OPA2134 and are also used in quite good High-End devices like Sony 777 SACD player. Even being good enough, those devices still had been tweaked by the different ways, for example here. We didn't pay enough attention that such devices, even being good sounding High-End components of AV/HT system, were a good target for the tweaking.
We had modified the board as it is shown here, for 4 channels and before sending of the modified board to the client, had to make some listening tests. We used the best known sources available at the moment - Ping Floyd -The Dark Side of the Moon, SACD multi-channel, song #7 - "Us and Them" and Steely Dan DVD-A (Advanced Resolution) - Two Against Nature - song #3 with the same name as for the album; as usually 47Ai through i-link to 49TXi with PQLS mode on. We switched the boards as it may be seen here, using the stereo mode for 2 channels. The difference was amazing. When the first board with combination of OPA627+OPA2134 produced the natural echo for the highs, which was giving the feeling of "flying sounds" and the saxophone sounding was extremely natural, the board with combination of OPA627+OPA2132 offered just quite good, solid sounding, but highs were as usual, like bursts with slight echo, and the saxophone sounded good, though without "the breathing reality ". We couldn't believe to our ears and switched the boards like 20 times. Our first suggestion was - OPA2134 (one of them or both) was oscillating. It was not, and if we will take a look at the schematics, it has no possibility to oscillate with such correction. Therefore we offered to the client to wait and checked the website of our supplier, OPA2134 had been available. We had ordered them and some other well known (by DIY people) good sounding OPAs to check our suggestions as well.
When they had arrived, we immediately replaced OPA2132 to OPA2134 and had sent the board to the client, because he was waiting for it longer then expected.
But we still had not been sure what was going on with the difference in sounding of OPAs and had decided to make some measurements to compare non-modified board and modifified. Finally in the process of comparison we made a suggestion about the difference in sounding, why the combination of OPA627+OPA2134 is the most naturally sounding combination.
Those tests had been made with 1 center channel driven (the whole 49TXi as AV processor, including DAC, OPAs and even DSPs) and disconnected other channels, in THX Music mode, with the same sources and songs, listed above. We didn't even take some pictures for OPA627 with OPA2132 (it had been done before) and OPA627+LM6172 due to the fact that anyhow the best sounding combination for OPA627 was OPA2134.
Here is the schematic diagram for the listening tests, songs and sources; we used 47Ai (757Ai) as the transport to 49TXi (AX10i) with i-link and PQLS on. Please note, that this is a subjective listening experience. Here is the table:
We had found 5534 sounding in any combination a bit "colder", more artificial, then both AD843 and OPA627. OPA2134, probably because of it's superior audio quality (SoundPlus) and ability to produce a bit higher level of even harmonics (as we can see in it's specification - PDF file), then the level of odd ones, was the best OPA for the post filter and volume adjustment stages with any of three listed OPAs in I/V (and for sure it is the best, as used in such High-End devices as Lexicon MC-12 and MC-8 everywhere). As expected, JRC5534+JRC2068 combination was the worst through the all tested ones. LM6172 with it's very high slew rate, used in post filter and volume adjustment stages, was just a good preciously sounding OPA at the level of OPA2132 or a bit less, but that's it ... may be it should be used in I/V stage, but we had much better candidates to be used here. AD843, claimed by many DIY specialists as one of the best sounding OPAs (if not the best one), was definitely a good one and very close to OPA627 in any combination; but still not as good, exactly as stated here. It had not been able to dethrone OPA627 for us. We even decided to check the suggestion of the very well known engineer in Russia Dmitry Andronnikov (aka Lynx Audio) that for the bass sound quality AD843 is better then OPA627 (he had used DIY boards with CS43122 and AD1862 for testing) due to it's more "attacking" sounding. So we used AD843K (K is the highest grade) in I/V stage for SW channel (loaded by M&K MX-350MKII subwoofer in THX mode), as shown here. Yes, it had more detailed, "punchy" bass, but even here, in SW channel, we still prefer flexible, "flowing" bass style, produced by OPA627. The results are our subjectivel preferences, someone else could prefer another style of sounding.
We had to find out why OPA627+OPA2134 combination was the best for us and why we started to be addicted to this quality of sound so fast. The suggestion is quite interesting - because such combination produced less odd harmonics. Here is the table of audibility of harmonics. As we can see, 3rd, 5th, 7th, 9th, 11th, 13th or even 17th harmonics (the results of measurements in RMAA 5.3 shown at the pictures when you click on each number, for the Direct mode) are more audible then even harmonics and produce artificial (just a bit, but audible), "digital" type of sounding. In the modified board odd harmonics were at least twice lower, and comparable (for example 3rd and 9th harmonics) to the levels of odd harmonics, produced by SB Audigy Platinum eX sound card, that had been used as the measuring device. It means that in the modified board those harmonics are very low, with the good sound card the measurements will show even less level, as it had been possible to get with so distorting and noisy measuring device as SB Audigy Platinum eX.
We used the graph of audibility for different harmonics and draw the results of measurements for modified and non-modified boards at 1 kHz THD measurements for -3dB below FS signal, using the table here (table 1 at page 6) for conversion of dB results in measurements into % of THD:
This table had been presented here by well-known engineer Steve Albini.
It is visible, that in the case of modified board (graph #1) only 7th harmonic is still audible (above the limit), but others - 11th, 13th, 17th, etc. are at the limit or below it, it means not audible. It the case of non-modified board (graph #2), all of them after 5th - 7th, (9th is almost at the limit), 11th, 13th, 17th, etc. are above the level of audibility, it means - audible. Is the difference of sounding in the level of higher harmonics (there is a nice article about the harmonics in op amps ) or in the TIM distortions (related to those slew rate and open loop gain), as suggested before by different tweakers ... or in higher harmonics and in the slew rate together? The above present results could let us make some suggestions; for those, who want to explore the problem with I/V a bit deeper, we recommend to read this perfect AES article. (PDF file)