I always was under the impression that those fully integrated FDAs are intended for rather low noise gains (like well below 20dB) as seen in DAC I/Vs, ADC drivers, active filters and format converters.
Thank you I was studying the datasheet for OPA1633 and it lists 10V/1KHz about -132dB (0.0000251%) TDH+N for a G=1, this should imply for a gain of x100, there should be 20dB increase in the distortion product thus indicating a distortion of 0.0002512% adding the residual distortion of your generator, your measurement is extremely accurate.
This means, accordingly to the datasheet, that in proper condition I should see no more than -63dB i.e. 0.0707946% at 1KHz @10mV. My measurements are way off I suspect something is definitely wrong with my breadboard. I am going to spin a board today I finished the design.
I took a look at the LME49990 it seems a fantastic opamp. Isn't it almost identical to the AD797? I compared noise and distortion they are basically very, very similar. I can still find some LME49990 it makes me so curious to try. However, in the end I don't know how any of these opamps sounds.
Last edited:
interesting observation. Why would FDA with 1.1nV of noise and 120dB of OLG and -130dB of distortion be suitable only for small gainstage? I am just trying to understand.
I am replacing the first gain on my current phono prototype, which was fully balanced and fully discrete; it worked fine except the output and input offset was not good enough for practical use and I had to dig too much on thermal performance for this round of design that I considered substituting it with an FDA opamp.
The OPA1633 datasheet seem to have used the older OPA1632 as a template and so didn't offer any noise figure for lower frequencies which is more or less customary for op-amps intended for audio use, 10 kHz is rather too high frequency, taking OPA1611 -2 datasheet for instance where en is given for 10 Hz, 100 Hz and 1 kHz, frequencies that matters for audio use.
Fortunately OPA1633 datasheet have graphs, fig. 6-16 & 6-17 p.9, showing how its noise contribution looks like between 10 Hz and 100 kHz, and indeed it's still a very nice performer.
Fortunately OPA1633 datasheet have graphs, fig. 6-16 & 6-17 p.9, showing how its noise contribution looks like between 10 Hz and 100 kHz, and indeed it's still a very nice performer.
Correct, the OPA1633 has very nice specs for FDA. AD797 is slightly better and can always use of them for balance operation; even though having few distortions improvement is very nice, in the end sound is the most important. Which opamp sound better? I know this is the $1M question, sound quality is very subjective and depends on the application.
According to TI, the OPA1633 is "the next generation OPA1632" and it it also 95% identical to THS2630 which has a much more extensive datasheet that we can refer to to get a better feeling of the capabilities.
THS2630 has some guaranteed DC specs (offset and drift) important for instrumentation and a bit more bandwidth and a bit less of quiescent current, OPA1633 "was optimized for audio" (whatever that means, likely more testing of audio range distortion) but most of the specs are identical (some graphs are actually identical to the pixel).
Same situation like with TPA6120A2 vs THS6012, or OPA1611 vs OPA211, etc.
@Stefanoo , it just a gut feeling of course. As far as I can see, the main application for those FDAs probably outnumbering all other uses, is instrumentation ADC driver, and they were optimized for that (low gain) task. I haven't come across an integrated FDA application schematic with more than 20dB noise gain.
BTW, OPA1633 has "only" 97dB OLG, and 1632 had even less, 78dB.
@Ultima Thule, current OPA1632 datasheet shows voltage and current noise densities down to 10Hz, maybe these were added in the August 2023 datasheet revision? But in general those (audio) FDAs spec sheets are lacking, compared to the usual standard at TI. For some reason, even for the THS2630 they don't show an open-loop gain plot, wtf.
THS2630 has some guaranteed DC specs (offset and drift) important for instrumentation and a bit more bandwidth and a bit less of quiescent current, OPA1633 "was optimized for audio" (whatever that means, likely more testing of audio range distortion) but most of the specs are identical (some graphs are actually identical to the pixel).
Same situation like with TPA6120A2 vs THS6012, or OPA1611 vs OPA211, etc.
@Stefanoo , it just a gut feeling of course. As far as I can see, the main application for those FDAs probably outnumbering all other uses, is instrumentation ADC driver, and they were optimized for that (low gain) task. I haven't come across an integrated FDA application schematic with more than 20dB noise gain.
BTW, OPA1633 has "only" 97dB OLG, and 1632 had even less, 78dB.
@Ultima Thule, current OPA1632 datasheet shows voltage and current noise densities down to 10Hz, maybe these were added in the August 2023 datasheet revision? But in general those (audio) FDAs spec sheets are lacking, compared to the usual standard at TI. For some reason, even for the THS2630 they don't show an open-loop gain plot, wtf.
@KSTR yes indeed, there are noise graphs for OPA1632 too, I wasn't explicit enough, my beef with both datasheets was how they are cherry-picking a noise figure to be used on page 1 as a bragging point, then when one look up in the electrical characteristics table one finds out they obtained that figure at a relatively very high frequency, 10 kHz, so on to the next task finding a noise graph if such exist that could give a better picture of its true nature, yet we also need to know source impedance, feedback resistor value etc to figure it all out.. small dirty tricks these opamp manufacturer are using, or maybe a matter of overlooking things, the apprentice who got the task to prepare one of their datasheets? :)
Good catches there with the other opamps!
Good catches there with the other opamps!
Below distortion AD797 SE Rf= 2K, Rs = 45R, Vcc +/-17V (BW 100-22KHz) to match abt 20dB in differential mode. I have observed that when driving a low impedance input, the distortion is highly masked by the distortion of the gen driving the short.
I have arranged the Rs to ground to set the proper gain and I inject with 1.8K ohm to the (-) input to the output of the unbalanced gen thus simulating a 40ohm cartridge connected to the input.
I believe this setup to produces in fact more realistic figures.
Question is how will the cartridge be able to drive the low "current stage" input impedance and produce low distortion? I think I will setup one of the test records I have and attempt to measure distortion. Unfortunately, it's hard to tell what the distortion level of the reference signal on the groove itself, so even doing this, the results could be unreliable.
In a few days I will receive the prototypes for the OPA1633 and the differential config using 2 AD797, using the same injection method I will post measurements.
Note: the results above are with the circuit wired in the breadboard.
I have arranged the Rs to ground to set the proper gain and I inject with 1.8K ohm to the (-) input to the output of the unbalanced gen thus simulating a 40ohm cartridge connected to the input.
I believe this setup to produces in fact more realistic figures.
Question is how will the cartridge be able to drive the low "current stage" input impedance and produce low distortion? I think I will setup one of the test records I have and attempt to measure distortion. Unfortunately, it's hard to tell what the distortion level of the reference signal on the groove itself, so even doing this, the results could be unreliable.
In a few days I will receive the prototypes for the OPA1633 and the differential config using 2 AD797, using the same injection method I will post measurements.
Note: the results above are with the circuit wired in the breadboard.
Last edited:
Forgotten to include the noise plot of the AD797. Please ignore the 60Hz pick up I cannot get fully rid of in this setting.
Rf = 2Kohm, Rs = 40ohm (thermal noise 0.8nV/sqrtHz)-> X41 gain.
AD797 0.9nV/sqrtHz
In order to get better resolution at low frequency, I will need to implement or buy a low noise preamp such as groaning, but this is it for now until I get the new prototypes with the OPA1633.
Rf = 2Kohm, Rs = 40ohm (thermal noise 0.8nV/sqrtHz)-> X41 gain.
AD797 0.9nV/sqrtHz
In order to get better resolution at low frequency, I will need to implement or buy a low noise preamp such as groaning, but this is it for now until I get the new prototypes with the OPA1633.