Don't know that, I know its a pro sound stalwart circuit block especially in esteemed mic preamps.
Apart from that as I mentioned ringing when serious is heard as hardness and unpleasantness in the mid-treble;
I've got a very simple and straightforward argument to confute yours: if you haven't understood yet, I judge opamps mainly (not only..of course) by sonic color. Every opamp has a strongly specific signature in terms of tonal character, if you can hear that. I do, and very strongly... so much that it's the main parameter in my judgment of sound (all the basic hi-fi stuff being there, at least).
So even if it was ringing of oscillating, my criterion would remain as solid as ever... because oscillation won't ever change the tonal character of an opamp. And tonality is what's below the very (crucial, in audio) idea of musicality; at least for an 85%
That said, the DAC where I've been making most comparisons was designed for the LT1364CS8 (was soldered on the PCB)... a 70 MHz 1000V/us opamp. So you see, slower "for audio" opamps are not likely to oscillate in my application
I must also remark that you overlooked my kind invitation to check things in a realistic application (more specifically, a CS4398 output summer/filter/buffer application, as showed in its datasheet) completely.
Last edited:
OK, thanks for clarifying. I've a kind of timidity in front of any form of power.
well, of course I meant identical parts....2*1363 > 1364, etc etc
hehe ok...got it, A-GD don't mind lying to their customers anyway
Alright, in that case, there is a small sonic advantage in favor of the single version - that's all!
Some worthwile opamps like the LME49723 and LME49725 (I prefer the former to all the other LM(E); I have yet to try the latter but I have reason to expect it to perform, too), simply do not exist as singles... Should we avoid them for this reason? Nah! Give them a try
Edit: it's a different story for the OPA2132P and the OPA132UA. The latter is a revised/improved version of the original dual, and it hears. In fact, the OPA132 now only exists SMD, the DIP being marked "obsolete"..
Last edited:
ah well, I've redone the browndog soldering twice, reseated the AD797BN on their adapters...got sound
two things are dead obvious when I go dual on the AK4396 output as LPF(each opamp does the +/- polarities):
1) I get strange interferences bleeps from time to time, they go away when I go 2*singles.
BUT my soundcard is known to have a lousy ground layout: Audiotrak Prodigy HD2 Gold Sound Card
and the browdog's are said to have "Superior electrical and thermal isolation between channels vs. integrated dual op-amp": Single-to-dual Op-AMP Adapter - DIP version (p/n 021001)
I also had the same problem w/ a M-Audio Audiophile USB soundcard(on its own wallwart), it's got three NJM5532 SOIC8 chips on the headphones output...they were doing the EXACT same intermittent interferences bleeps.
2) the stereo image is shrunk down to death on dual chips
the middle channel is MUCH wider on singles chips...I've just rolled a few dual chips, it's DEAD obvious.
most ppl who go dual IC > burson also experience this DRASTIC change, as the burson's are made of two single discrete chips.
anyway, the burson is nice as final buffer after the 4*AD797BN LPF from the AK4396....but it kills the amazing AD797B SS, and it's really reverbed to death. The instruments separation is crazy and the SS is uber-wide, but it really doesn't sound dry at all...it'd be like having Dolby Virtual Speakers or Dolby Headphones enabled on a permanent basis
LT1028CS8 is boring to me, the wahwah guitar on the Isaac Hayes - Shaft title song is very dull on it...OTOH it's amazing on LT1363CS8
plus the 1028CS8 seems to kill the AD797B SS quite a bit, the 1363 leaves it pretty much intact...it's like an improved burson w/o the reverb. The 1364 is a notorious favorite for my soundcard.
but well, in a few days I'll be bored of the 1363 sound...as usual.
Last edited:
I must also remark that you overlooked my kind invitation to check things in a realistic application (more specifically, a CS4398 output summer/filter/buffer application, as showed in its datasheet) completely.
My example is just as "typical", and is the basis of many a low pass filter/buffer/or preamp input stage. The kind of stage that the signal you are listening to has passed through many times in all probability.
I understand you hear differences... I do too.
I can also understand that you judge solely by sound quality alone... and this is where I part company... sound quality is an over riding factor, but it's essential to back up what you hear with measurements of some kind. I would never be happy with a device that was oscillating but sounded "good".
Another way of looking at it
You fit a device and conclude it's the top performer sonically. That's all you need to know. Device "A" is best.
I come along and find it's undercompensated (say) and exhibits ringing etc. Perhaps it's driving a capacitive load too and really needs a little resistance adding to the output terminal to isolate the device from this effect. You are blissfully unaware of this... simply enjoying the "sound"
You now fit another device "B" and it's lacking sonically.
Again I come along and find it's actually pretty good from the technical view, compensation etc all spot on. You don't like the sound...
I now tweak the circuit (perhaps deliberately undercompensating etc) so that it now measures more like device "A"... technically inferior. You like it
(Perhaps)That's what I am getting at Andrea... you might (if you did tests etc) find you could improve (to you) the sound much further (perhaps with several of the better opamps) than you will ever get just swapping devices.
Yeah, 100 K of source and feedback resistance... typical.
Me neither. I always do my best to check that everything's working fine, without a strict necessity to be provided with an oscilloscope (unless you want to send me one for free, since they're so cheap
).Not exactly... I'm not the kind that "listening is all that matters". Of course I take my provisions first... i.e. check the datasheets accurately for electrical compatibility with a given application.Another way of looking at it
There's nothing "best" in this world. All that exists is diversity. I state what I like best.
Not really... I've thought many times about ringing due to capacitive loading. Fortunately though, even a delicate opamp in this regard like the LT1028 work(ed) great in my circuit without changing anything.... so... I'm fine with it, and feel free to enjoy the (smooth) sound.
I see nothing wrong with my stance here.
This is pure astraction, unless you can substantiate with practical facts.I now tweak the circuit (perhaps deliberately undercompensating etc) so that it now measures more like device "A"... technically inferior. You like it
Not really... I don't think. Proof is that my favorite chips all sound better (for my taste) than the stock opamp, which is widely appreciated for its sonic qualities, that is, the LT1364. This one has its charm, but its apparent limits too.
The same goes for the NE5534 which my other DAC came with. It's so very easy to improve on it (the LT1357/1358 are good examples of trouble-free upgrades to the NE5534/5532 in most cases.....and it's not just me), even though the circuit is supposed to be optimised for the NE5534, and not for my favorites...
Besides, as I told you before, an OPA627 always sounds like an OPA627, in every place I tried it... no way...
And, conversely, an OPA2132 (or also an OPA2134) always sounds good... In my subjective appreciation of things, naturally.
Last edited:
It's nice indeed & I plan to use mine again somewhere soon. Maybe in the AD1852 NG98.
One thing puzzles me... since the CS4398 needs (see circuit in the datasheet) an asymmetrical load on its + and - ouputs, while the AD1852 does not, how did they manage to make a DAC with switchable D/A chips, when the output filtering stage remains the same?
Either the filter is asymmetrical, and so it goes better with the CS4398, or it is symmetrical, thus it goes better with the AD1852...
Since looking at the circuit board as far as I can see the resistor & cap values are not asymmetrical, my guess is that they chose the compromise that unfavors the CS4398...
So, if it were NG98, I'd buy it with the AD1852.
Either the filter is asymmetrical, and so it goes better with the CS4398, or it is symmetrical, thus it goes better with the AD1852...
Since looking at the circuit board as far as I can see the resistor & cap values are not asymmetrical, my guess is that they chose the compromise that unfavors the CS4398...
So, if it were NG98, I'd buy it with the AD1852.
Actually, there was a case where the LT1358 (25 MHz and 600 V/uS; 120 nA of Ib) didn't seem to fit the circuit quite right. It sounded dry and somehow unnatural, differently from my typical experience with it.
It was in a gain stage (set to a very low gain) of a headphone amp. The original opamp was an OPA2134UA. I remember well that the LT1358 didn't satisfy...so I went with something else. The OPA2107 sounded fine there; not that it was an improvement over the OPA2134 in every way...
I believe that the cognitive power of the senses is severely underestimated..
It was in a gain stage (set to a very low gain) of a headphone amp. The original opamp was an OPA2134UA. I remember well that the LT1358 didn't satisfy...so I went with something else. The OPA2107 sounded fine there; not that it was an improvement over the OPA2134 in every way...
I believe that the cognitive power of the senses is severely underestimated..
Last edited:
Much thanks all. Lively discussion. Lots of chips I had never heard of, but like I said, out of touch. No op-amps in a 64 Morgan you know. I am not adverse to SMT on adapters, but for what I have planned right now, just traditional rework seems in order. Besides, I am not fond of sockets. I guess I need to brush up on my skills. Back when I got certified, SMT had not been invented. I bet my board etching methods (sharpie) need an update.
For now, singles vs twins is kind of moot. I can easily see for some applications both crosstalk and dissipation would have their issues. In a crossover for one. Crosstalk could show up as phase delay added in.
Point well taken Mooly, even "drop in" is not. Always need to run some tests. Before and after as no one ever said the OEM did it correctly in the first place!
Fire up my good old Tektronix. I splurged on that (thanks e-bay) after putting up with a Ramsey for years. I also swapped my old Heath VTVM and signal generator for old HP's. They don't actually work that much better, but using them is more fun.
Plan on picking up some Grado's this week for analytic use. My old cans are very good at masking problems. Great for listening, not for critical evaluation.
I am of the school that an amp should either: Just amplify. Straight wire with gain, OR, does something one considers pleasing. Intentional modification of the transfer function as a decision where you believe you know better than the recording engineer. As I get older, I tend to lean to the latter. There are many great engineers, but a lot of bad ones too. Trumpets recorded with condensors without due post production compensation for example.
I am also of the belief that by now, integrated op amps can out-do by a long shot the basic old Jensen design. There is enough space on a chip to do all kinds of laser trimming and compensation one could never afford in discrete. I have recently heard a descrete vs a hum-drum 2143. The chip won. This was in a external DAC.
While covering all the ground, I don't see much discussion of biasing into class A. Did this go by the wayside as the parts got better?
For now, singles vs twins is kind of moot. I can easily see for some applications both crosstalk and dissipation would have their issues. In a crossover for one. Crosstalk could show up as phase delay added in.
Point well taken Mooly, even "drop in" is not. Always need to run some tests. Before and after as no one ever said the OEM did it correctly in the first place!
Fire up my good old Tektronix. I splurged on that (thanks e-bay) after putting up with a Ramsey for years. I also swapped my old Heath VTVM and signal generator for old HP's. They don't actually work that much better, but using them is more fun.
Plan on picking up some Grado's this week for analytic use. My old cans are very good at masking problems. Great for listening, not for critical evaluation.
I am of the school that an amp should either: Just amplify. Straight wire with gain, OR, does something one considers pleasing. Intentional modification of the transfer function as a decision where you believe you know better than the recording engineer. As I get older, I tend to lean to the latter. There are many great engineers, but a lot of bad ones too. Trumpets recorded with condensors without due post production compensation for example.
I am also of the belief that by now, integrated op amps can out-do by a long shot the basic old Jensen design. There is enough space on a chip to do all kinds of laser trimming and compensation one could never afford in discrete. I have recently heard a descrete vs a hum-drum 2143. The chip won. This was in a external DAC.
While covering all the ground, I don't see much discussion of biasing into class A. Did this go by the wayside as the parts got better?
My cmoy (which, leeperry, does sound better than the LDM+) likes a lot the OPA132UA, except that the pairing with my headphones isn't quite perfect: OPA132 + HD238, and the usual maths apply... the doubling of the 3 is easily appreciated as a slightly too luminous "sunny" sound. The AD845KN has been the opamp that fits my cmoy & my headphones the best, so far. I wanna try the 200mA-output OPA551 too though. It fared surprisingly well in Samuel Groner's testing. I used it a while ago...need to refresh my memory.