Opa2156 is really not that noisy. And when you parallel it the voltage noise will drop further.
-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Opamp Rolling, Does It Work?
- Thread starter jackenhack
- Start date
but i use 4 opa827 in parallel in objective 2 it works.
since lme49990 is discontinued. I believe lt1028/1128 can be good candidate for this type of application.
It is designed for low noise for its application, but is still ~12 nV/rtHz compared to other low-noise audio op-amps, e.g. the OPA1612 at ~1 nV/rtHz. Whether it is that noisy for you depends upon your application. And of course if you want to parallel several, you'll reduce the random noise by sqrt(N) for N in parallel on average, but seems a lot of work vs. just buying a lower-noise op-amp to begin with.
A lot of folk seem to like using op-amps rated for video or other things than audio but IME/IMO that is not always the best thing for the overall design. I've talked before about a guy who decided mW GaAs FETs must be awesome for audio and got a healthy intro into 1/f noise when I had to explain how a 1/f corner of 10~100 MHz did not a good audio transistor make.
YMMV - Don
In objective 2 you can't avoid the output opamp. Sure opa1612 isn't bad in this application but opa827 performs better.
About the noise performance, unless you are using 1kohm potentiometer,(which i am using) 10k ohm potentiometer will have much higher noise compared to opamp(again, in objective 2 application). 4 opa827 in parallel gives 300nV noise, which is about as good as we are getting. If you want to achieve 100nV using opa1612 that's fine.
Wondering if you could please attach a pic with the QA401 in loopback, just to see how the DAC/ADC tandem is working and compare with the O2.
Thank you,
Raul.
Not really. Two 4556 can already give 200ma of output current. Swapping to other opamps may improve performance but not likely to increase output power. Opa1622 is supposedly a better choice. But I experience some stability issue and over heating. So maybe they shouldn't be used on a dip8 converter. Other candidates are muses8920 which is almost as powerful and lower distortion. Opa827 was my favorite but expensive and doesn't have much current. Opa1656 is pretty good with 100ma. Try for yourself. But if you simply want more power, there aren't many choices.
syn08
Senior Member
LT1206
LT1210
OPA564
OPA567
And tons of more choices, from the usual suspects.
Opamp rolling is a stupid exercise, BTW.
syn08
Senior Member
Chopper op amps can in principle cancel completely 1/f noise. It's not unusual to see such opamps with ruler flat noise down to under 1Hz. I wouldn't use a chopper op amp in a MM cartridge head amp, but for line level most of the modern implementations like the OPAx189 should be perfectly fine. I don't see much of an incentive using them for audio though, you would pay for features that don't really matter (like near zero offset, etc...).
Those aren't even the same pin out. And opamp rolling isn't stupid if you can take measurements.
syn08
Senior Member
Sorry, I didn’t see a requirement to be pin compatible (with what?).
Op amp rolling is a stupid exercise, period. Replacing an op amp is a different story, it implies one knowing what he’s doing. Even so, if the original design was competent, it is unlikely replacing the op amp with a “better” one (whatever that means) will necessary improve the overall performance.
When I hear “exchanging the op amp provided a much cleaner bass” I am closing the window to avoid the BS stench.
Last edited:
I believe the person who asked required the opamp to be pin compatible with 4556 in DIP8 or soic8 in Objective 2, at least easily convertible.
Like I said, it's not stupid with measurements, rolling meaning swapping one by one. You don't know how one device would perform when such condition is not specified in data sheet. Like two opamps in parallel and loaded by a 32 or 16ohm headphone. And I agree without measurements, all the sonic improvements are basically bs.
Now onto whether you can improve a competent design. That depends on what competent means. O2 was designed for cost performance 4556 is basically the one of the cheapest opamp with 100ma current capability, using two opamps in each 4556 in parallel doubles the current to 200ma+ taking the advantage of the low cost. If it's opa827, that requires 4 pieces of opa827 which isn't cheap for even a single part tho you can achieve sub 0.0002% thd with 32ohm load and taking the advantage of fet input not dependent to source impedance and having ultra low offset. Also it's possible to use only one side of opa1622 for each channel and achieve better performance when not exceeding the output capability.
If you don't understand the circuit or the OpAmp specs you are basically engaging in a lucky-dip activity. Same with tube rolling.
But what happens when some conditions are not specified in the datasheet? And spice model doesn't include distortion.
First hunt down to a few parts and compare the performance of them in the circuit, how wrong is that?
Stable with 10nF load in simulation
Flat open loop output impedance
Low noise low distortion with 2kohm load at least
Better have performance with lower impedance load
100ma current output capacity
Most opamps don't specify performance with low impedance load like opa1622, tpa6120 etc But they also have their own issues. AGAIN what's so wrong about component rolling with measurements? Douglas Self has done it, basically everyone does it. Take measurements first then you can try to explain some phenomenon that causes the results.
syn08
Senior Member
Ok, next time check the catalogues yourself.
THS3120
OPA2674
THS3162
OPA551
Etc... most of these also have dual versions, but power dissipation could be a big issue.
THS3120
OPA2674
THS3162
OPA551
Etc... most of these also have dual versions, but power dissipation could be a big issue.
Still lucky-dip stuff. A more technical term is 'trial and error'. Another term is 'suck-it-and-see'.
It worked for Edison.
Last edited:
Or simply, experimenting. Tiral and error works too.Still lucky-dip stuff. A more technical term is 'trial and error'. Another term is 'suck-it-and-see'.
Ths3120 and opa2674 are CFA, which has to have specified Rf instead of directly connecting together.
Ths3162 doesn't exist.
opa551 is ok but has somewhat high voltage noise. With single to dual converting board it will work.
