Cap ESR on vref isn't particularly important. Heck AKM encourage the addition of increased series resistance, both to the supply and ground, as a way of creating an RC network to filter the supplies. What's important here is the size of the cap in uF.
So far this thread just seems like snake oil solutions to a problem we don't even know (measurements please) actually exists. None of these cap modifications are going to do a thing if the original implementation was well conceived.
The fact an opamp change resulted in more of something you don't normally want, noise and/or other additional signals, the first thing you want to do is remove that opamp!
Yes. If the Nicicon KW is for Vref, just keep it. For the fun you can try UKA low leakage and try to decouple the pins with a 1 uF pps/acrylic or a std black gate if having it...works fine at least for ess chips. Keep the 0603 that has due to its size very low inductance. As says 5thElement it's not by chance...caps are just changing small things and you often worse the things by doing wrong choices. I m curious about the opa1612 I selected for the ad1862 dac from Miro1360...not implemented yet...so 11v advised if I understand....
Diggy,
+-11v would probably be wise to try.
These things can be handy for trying different rail voltages: LT1083 HIFI Linear Power Supply Dual Output High Adjustable Power Supply Board | eBay ...they sound quite decent for opamp rails. Also recommend using an R-core power transformer with separate windings for rail power (no center tap).
By the way, tried your idea of using Rubycon .22uf 805 size film caps for clock bypass. Like 'em! Thanks for the tip
+-11v would probably be wise to try.
These things can be handy for trying different rail voltages: LT1083 HIFI Linear Power Supply Dual Output High Adjustable Power Supply Board | eBay ...they sound quite decent for opamp rails. Also recommend using an R-core power transformer with separate windings for rail power (no center tap).
By the way, tried your idea of using Rubycon .22uf 805 size film caps for clock bypass. Like 'em! Thanks for the tip
Last edited:
Because it's not a fashionable part and doesn't allow you to differentiate your product. Since it's in the reference design, it's not too expensive, and everyone uses it, it's automatically bad.
Remove which opamp exactly, already tried lme49860 (the original in dac), opa1656 (loads of noise), ada4610, and now opa1612. I don't think the smd pad can take too many more desoldering.
I had a nice improvement by replacing the i/v resistor (that's my subjective opinion), and played with the cap directly at the output of tps7a47/tps7a33 regulator (polymer tan ultralow esr don't sound good in that position). So I naturally want to see if the Vref can made a noticeable difference. But I agree that lots of time parts change don't make much impact to the sound.
I think they had raised the I/V resistor value, so as to keep somewhat lower load on the original opamps.
The factory recommendation is half of it, 360ohm..
And already a huge output signal even with that..
But if the opamp is not capable to output the required full current with adeguate low distortion, than one has to use a higher value..
And, probably, lower the gain in the 4499. Would be not surprised if the factory setting would be "low gain"...
Ciao, George
The factory recommendation is half of it, 360ohm..
And already a huge output signal even with that..
But if the opamp is not capable to output the required full current with adeguate low distortion, than one has to use a higher value..
And, probably, lower the gain in the 4499. Would be not surprised if the factory setting would be "low gain"...
Ciao, George
[ mhz isn't anything SI, mHz is millihertz, MHz is megahertz. All SI units and
symbols are case-sensitive. For instance hertz is the unit, Hertz is the name of the guy it was named after, and the symbol is always Hz. ]
This noise could be digital noise being picked up, such as an artifact of measuring with inadequate probing technology, could be due to a lack of high frequency decoupling, could be a faulty unit, could be the circuit and chip don't work together. well
Band limited noise, as opposed to a single tone or white noise is usually due to something being close to instability, or white noise seen through a resonant circuit (LC parasitics). I can't see any obvious source of white noise here, I'd vote for instability. It might be exacerbated at some harmonic of 96kHz from the sample clock.
If this is only happening with certain opamps it suggests the circuit isn't designed for those opamps, either in terms of decoupling or positive feedback due to parasitics at higher frequencies. Often layout can be crucial and if the PCB has only been tested with one device there's a chance it is borderline for another.
It would be interesting to put an RF spectrum analyzer on these signals, and see if there are more wobbles in the spectrum indicative of issues
I thought they were using two 680R in parallel for 340R total in the IV stage.
680R on 11V supplies will clip.
I was meaning the opamp that gave you increased noise should be discarded. That is all.
The fact you claim the IV resistor had the biggest impact on the sound is quite worrying as it's one part that's going to do nothing to alter the sound. Sure using a tiny thick film will undoubtedly cause higher distortion than a more suitable part but this distortion is going to be of low order and still way below audible limits regardless.
Some OT removed or moved to - 