Agreed.
Let me give you a more down to earth example: being from the American continent, I presume you're familiar with brisket. How do you make a tender and juicy brisket? There's some science behind that, e.g. hydrolyzing the collagen, which is a function of temperature and time. But how the temperature is applied makes a difference: open flame, charcoal, oven, pot, etc., and also the application of a constant or varyng temperature. Now it's getting more complicated to science your way out of that. And I'm not even talking about seasoning!
Richard are you familiar with buffered IC headphone amps Neurochrome has been making? They use LME49600 for the buffer: HP-1: Ultra-High End Headphone Amplifier
My apologies, I should have said not good to me. You could order one of each of the two referenced dacs and see what you think. I'm pretty sure you will find they sound rather different from each other despite measuring measuring rather well and pretty similarly.
What I could do if someone wants to come for visit is demonstrate the effects of various power supply options including filter caps on Allo Katana dac. There are also a couple of slightly different output stages for it, and several interpolation filters that all sound different. In addition there are graphs of the filter frequency responses you can look at, and they are all quite flat in the passband.
EDIT: Having been working on dacs for awhile it is quite possible what seems interesting to me could seem very boring to someone else, so I wouldn't force it on anyone. There are certainly other effects to demonstrate as well, some of which can be done using an FFT as an adjustment reference.
Last edited:
Oh, you could probably measure anything that can be heard. I don't see why not in principle. The question is more to the effect of whether or not you actually do measure everything that can be audible.
Erm no, You can still science your way out of it. What you cannot exactly measure is a preference.
As I have said on many occasions I have active speaker, so I'd need two of each and I'm really not in the market for a DAC for another decade as other things that have gross errors to fix. You know things that are 1dB or more out of flatness. When I can get my in room power response where I want it then I might worry. I'd still be interested by what you consider good. Decay tails talk is still alien language to me as far as having any reference to the music I consider worth critical listening to.
Hello,
I didnt think of an old design... was thinking of the newer Opamps. but, you are right... it has an external comp pin... very rare these days.
Which gets me back to my earlier question,.... if you do it (lower the OLG) will the distortion still be extremely low? See, this is the Thing. Even when there is an external comp pin available and lower the LG and FB amount, the distortion will be higher. These ic opamp devices depend on high FB for their performance. And, it obscures the need for matched devices which is still not commonly done in IC. There is a process to make good compl now but it isnt needed, usually, with high FB topologies.
Many of the CFB-Mode of operation comes closer to the mark for my use. With them... high FB is not required to get extremely low distortion.
THx-RNMarsh
Last edited:
Time domain; measured?
Frequency domain; measured?
Brain domain; measured?
What about amplitude domain?
Transient settling domain (or whatever you want to call it)?
How do you measure what it is that makes one interpolation filter sound different from another? Do it with illegal signals that make them pre-ring? If not that then what?
Which time domain or frequency domain test do you use to measure the perceptual width of a sound panned to center in a multitone, multi-panned signal enviornment? What measurement shows whether a sound panned to middle is perceived as emanating from an approximate point source or a widely spread out elliptical source?
Last edited:
I'm not sure what you mean by amplitude domain. As for width, that would I imagine be very difficult to analyse since it would mean identifying how a single sound sent to both channels differs between channels thus creating the perception of width. Basically you seem to be asking how one could reverse engineer the mix?
Wrong narrative. He was talking about audio reproduction. It would be a closer down to earth example if you talked about recreating brisket that someone made. There are culinary version of doing so. It's called recipe.
Just what is the current state of the art matching NPN and PNP both discrete and monolithic?
If all I care about is a single parameter such as gain at a specific current I certainly can select discretes to match. Throw in bandwidth at that current and possibly can. Vbe.....!
As that is fundamentally a function of the speakers and room if a DAC widens a mono source then there is something pathologically wrong with it that will certainly show up in simple measurements. Inverting one channel and doing a null test for example.
Yes. I used this simple circuit concept a few decades ago for matching compliments. Later, when I got a characteristic curve tracer, I used that to match.I've heard of this CMOS thing. Supposedly they make some ICs in it.
View attachment Compl matcher circuit.pdf
THx-RNMarsh
Last edited:
No idea, you of can of course design around a lot of this. After all a matched pair of either starts out with no inherent seconds in its transfer function.
I'm guessing you are possibly a bit out of the loop on this but the latest
range of cutting edge discrete opamps use forms of two pole compensation
to get their cake and eat it, so to speak, WRT OLG and BW.
A good example is the Weiss opamp. I doubt there is a better performing
monolythic opamp, objectively speaking.
T
Not that one before..... certainly the thd is extremely low. What i dont like is that the different parameters are not all at same Z load. Its all over the map on load. Maybe to get the best number for that particular test? I would like to see all specs into one same Z load. In my case, in addition, sometimes measure all with the actual headphone I am going to listen with as a load.
THx-RNMarsh
Okay. Well, many dacs do it to some extent or another. With some dacs the effect is so extreme that perceptually the sound seems to emanate from two separate speakers and there is virtually no illusion of stereo even at the sweet spot listening location out in front of the speakers and in the exact center from left to right. It is so common that I asked how we are measuring it, not whether or not it can be measured. I take the latter point for granted that we could find a way to measure it if we were so inclined.
- Status
- Not open for further replies.