Have you all gone mad? What exactly is the back story to this questioning of whether the LM6172 works or not? Cog liked what he heard (and I don't think emphasised detail in particular). Moreover he's saying that his LME has that "holy crap" moment - and that is after experiencing the 6172. So we're not talking about excessive detail on the 6172.
The test is whether the detail has body or not? If it is thin then maybe it is oscillating. But nor is it an especially temperamental part. If in doubt then shove a cheap 1-10uF tant across the bypass cap (2.2u is what I think the DS recommends) The change will anyway be as big as the op amp itself but it would be a rare thing to be oscillating after that.
My experience of the LM6172 is that its beauty is subtle. I am surprised that the 627 is getting the praise it is by comparison, though I'm not in the least bit surprised that the LME is doing what the spec sheet says. If that chip is wearisome (and I am not a fan of the 4562 though this one is far better) then I would look to the power supply and bypassing.
But, frankly, if you don't model this stuff then opinions are worthless. Most opinions stem from a complete misunderstanding of what is actually going on, so we have a list of products that we think sound good, most of which are replete with flaws. We have free tools with which to do that modelling so there is no excuse. And up in this realm of perfection in op amps, we are at - or past - the limit of what can be regarded as DIY ,without someone expert handing out a recipe.
I'm far more worried about which volume control he swapped for which?
CT
The test is whether the detail has body or not? If it is thin then maybe it is oscillating. But nor is it an especially temperamental part. If in doubt then shove a cheap 1-10uF tant across the bypass cap (2.2u is what I think the DS recommends) The change will anyway be as big as the op amp itself but it would be a rare thing to be oscillating after that.
My experience of the LM6172 is that its beauty is subtle. I am surprised that the 627 is getting the praise it is by comparison, though I'm not in the least bit surprised that the LME is doing what the spec sheet says. If that chip is wearisome (and I am not a fan of the 4562 though this one is far better) then I would look to the power supply and bypassing.
But, frankly, if you don't model this stuff then opinions are worthless. Most opinions stem from a complete misunderstanding of what is actually going on, so we have a list of products that we think sound good, most of which are replete with flaws. We have free tools with which to do that modelling so there is no excuse. And up in this realm of perfection in op amps, we are at - or past - the limit of what can be regarded as DIY ,without someone expert handing out a recipe.
I'm far more worried about which volume control he swapped for which?
CT
I read what is posted.
This tells me that further investigation is required.
That it happens with one chip does not mean it never happens with another chip.
What I read when I look at a thread like this is that there are some basic problems with the whole premise of rolling op-amps in the first place. While it can be done and in a particular instance one op-amp may be the best choice in that instance to project that and to say that the same op-amp will work the same in every different implementation makes no sense, Unless each circuit is optimized for each different op-amp I think that the entire exercise is a futile effort and there can be no real scientific objective or real subjective conclusions drawn from just changing op-amps alone, you could miss a great device just because the circuit in question was optimized around a different device.
A better questions should have been originally what op-amp in this circuit is the best sounding. not what is the best sounding op-amp in all instances.
A better questions should have been originally what op-amp in this circuit is the best sounding. not what is the best sounding op-amp in all instances.
Quite true, there needs to be some tuning in each case, and there are many factors that influence listening results. Not all data sheet provide all necessary information either. Douglas Self had a good write up on OPamp selection, those can be used as a starting point for selection.
Last edited:
Oscillation at extreme hf isn't always obvious by listening alone...
http://www.diyaudio.com/forums/anal...u-have-checked-see-its-stable-havent-you.html
Sometime it can....
I made tone control using NE5532. I feel the sound is not right, but only me in my family can notice it. I check my tone control using oscilloscope and found small oscillation around 100kHz. Small cap on feedback stop the oscillation.
May be it can create intermodulation distortion...
Interesting, and another practical example of why its important to always confirm that all is well stability wise.
Just to switch topics briefly; I removed CS3310 and installed Texas Instruments PGA2311. (I am talking about my pre-amp, now, not my DAC)
PGA2311 is a drop-in replacement (pin and software compatible) with considerably better performance numbers. http://www.ti.com/general/docs/lit/getliterature.tsp?literatureNumber=sbos218a
I can confirm that the replacement resulted in audibly better sound. More open and transparent - less like a hi-fi system and more like "the real thing".
The difference in sound between various opamps in my DAC is now that much easier to discern.
By the way, I have gone back to the LM6172 for the past week and I have to say that it really is a pleasure to listen to. Perhaps not as transparent as the LME49710HA or as smooth as the OPA627AP, but a pleasure to listen to all the same.
Last edited:
I'm surprised. Cogitec. I'd have to look at the data sheets again, but while I knew about the PGA2311, the one I had been looking at was the CS3318 (I think) and I seem to remember having come to the conclusion that it was comprehensively better. I would have been looking at the obvious things like distortion and s/n, but also at the TC of the resistors. I think Jan Didden uses it in something and there's a very interesting article (that I think one can link to from his site) from a US manufacturer - whose name escapes at the moment but he used a lot of LM3886s in one design of his - on its predecessor, which I think was the 3310.
Anyway, I'm glad to see you appreciating the LM6172. I don't suppose it is as clear as the LME49710 - which does have awesome specs (130dB CMRR to 20kHz at +-17V!). One would wish it to be just slightly more tangible, but those sort of wishes quite often take you into the realm of the uncomfortable.
Incidentally, I do agree with whoever it was that said that rolling op amps is not great an idea, though your experience does seem to pretty much line up with mine. Has anyone actually had the time to read the whole of this thread so we can get an idea of what sort of consensus might have been reached?
Anyway, I'm glad to see you appreciating the LM6172. I don't suppose it is as clear as the LME49710 - which does have awesome specs (130dB CMRR to 20kHz at +-17V!). One would wish it to be just slightly more tangible, but those sort of wishes quite often take you into the realm of the uncomfortable.
Incidentally, I do agree with whoever it was that said that rolling op amps is not great an idea, though your experience does seem to pretty much line up with mine. Has anyone actually had the time to read the whole of this thread so we can get an idea of what sort of consensus might have been reached?
When I have some time I'll look at this again.
What I can say is that when Naim produced their high-priced Statement I immediately thought that they had chosen the CS3318 and had thought "Good for you". I have no idea whether that is right, but it seemed to correlate. (Yes, I do know that apparently it switches over to their own thing, but I haven't bothered to think why they might want to do that.)
Incidentally, I wouldn't underestimate distortion at 20k. One day we'll understand what's going on, but unfortunately the people that might are in their 80s now, or dead. Suffice to say that I am always amazed that these low level things seem to have an effect that is equal to, or greater, than the hideous errors in loudspeakers. (And, no, I don't subscribe to the source first credo.)
The CS3318 has a couple of nice features:
+/-9V power supplies. The +/-5 PGA2311 and CS3310 overload at 3.5V rms, which is a bit marginal for some CD players, which go way above the 2V convention just to win sales by being louder.
8 channel, so inputs can be preset to a normalised level instead of having to change volume setting when changing source
+/-9V power supplies. The +/-5 PGA2311 and CS3310 overload at 3.5V rms, which is a bit marginal for some CD players, which go way above the 2V convention just to win sales by being louder.
8 channel, so inputs can be preset to a normalised level instead of having to change volume setting when changing source
Yeah, I thought that too, to begin with. But when you think about actually doing it, it isn't enough. And if you have balanced inputs - and want to keep them that way for as long as possible - you're largely stuffed.
I know you can use more than one quite easily, but I seem to recall that even that started to look a touch difficult. And of course it gets quite pricey too. My recollection is that it all started to seem like a bit of a nightmare once one took the layout and different power supplies into account. Having said that, it probably still looked like one of the best solutions. (Though not as easy as I would have hoped. <g>)
CD Players generally have a maximum output (0dBfs) of around 2Vac.
Some are over 2.1Vac, but I have not seem a measurement report of more than 2.3Vac.
3.5Vac gives an overhead margin of 3.6dB
I generally allow >=10dB of margin, in pre-amps and such like, to keep the maximum audio signal well away from the onset of increased distortion. This also allows interference to pass through unclipped. RF attenuation works better on unclipped interference than it does with clipped interference.
To get 10dB of margin on a 2.3Vac Source you will need a supply of about 16Vdc from which an amplifier can output a maximum of ~5Vac
A supply of 9Vdc does not get close to allowing a 10dB margin.
There are some, for example my Marantz SA-CD which is quoted as 2.4 vrms on SA-CD (and 1.8 vrms red book). I bet its not unique or the highest.
I have seen Application Notes that envisage way past 2V! And that is seen as one of their qualities.
I'm still with Davidrsb here. Even rail to rail needs some saturation voltage so 3.5V is unlikely anyway.
The whole CD thing is so utterly marginal on specs that people hardly bother to follow them any more. Just get JRiver to show you the peak levels on a recording and you'll get the picture. It's absurd.
I'm still with Davidrsb here. Even rail to rail needs some saturation voltage so 3.5V is unlikely anyway.
The whole CD thing is so utterly marginal on specs that people hardly bother to follow them any more. Just get JRiver to show you the peak levels on a recording and you'll get the picture. It's absurd.