I have a very cleaned up environment and extensive ac line isolation and filtering plus all chassis floating and no ground wires in common etc etc. The room lighting has been changed to low voltage dc LED. ETC. So I get very clean measurements, I have no rfi/emi issues, either. I used PC on battery power only during testing. Guarding is used on the triple shielded coax test cables. You should do as well, in your environment, Waly. .
This is the measured results from that exact setup --
You only see a small amount of ac related freq below the 400Hz HP filter (ps ripple) and the rest is signal harmonics from DUT and an internal 500Hz analyzer signal.
[signal freq is 980Hz]
BTW -- as has been noted here before by others, the 725D version is a completely new design. And, there are no schematics available that I know of for the "D" model. So, the 3K series R is a wild guess/assumption.
THx-RNMarsh
Last edited:
Jan,
Are these numbers specified at 1kHz or 20kHz? Are they THD only, or THD+N?
Cheers,
Bob
Those are spec numbers... actual measured are better, of course.
https://www.ap.com/analyzers-accessories/2700-series/2700-series-analog-performance/
and the 555 -- https://www.ap.com/analyzers-accessories/apx555/
THx-RNMarsh
Last edited:
The audio ADC market is small with only a few "SOTA" players- AKM and Cirrus seem to be all that are left. Its not profitable. In both cases the best distortion is still old parts- AK5394A and CS5381. Both claim -110 dB THD+N. There may be some SAR ADC's that can do a little better but with a lot of overhead to get an audio compatible stream (viz Frex's efforts).
The APx555 has a lot of analog processing before the ADC and is not quite relevant. They have a hot rodded version I have heard. Probably an extra db or 2. The APx 515 is the most relevant and most like a high power sound card with some analog processing.
The Dscope is similar to the APx 515 as far as I can tell. The SRS and the R&S are more like the APx555 with lots of analog processing prior to digitizing.
The APx555 has a lot of analog processing before the ADC and is not quite relevant. They have a hot rodded version I have heard. Probably an extra db or 2. The APx 515 is the most relevant and most like a high power sound card with some analog processing.
The Dscope is similar to the APx 515 as far as I can tell. The SRS and the R&S are more like the APx555 with lots of analog processing prior to digitizing.
https://www.gearslutz.com/board/gee...udio-interfaces-their-ad-da-chips-listed.html
some with the newest ESS chips are showing up in the market - although finding info on their ADC seems difficult
some with the newest ESS chips are showing up in the market - although finding info on their ADC seems difficult
Last edited:
It's an SVO The tuning is done with two sets of AD5543. Two in parallel per section. So a best case Mdac nominal 2.5K. The rest are 4.99K. You can see the noise drop running two Mdac parallel. The data sheet for the AD5543 says 12nV/rtHz for each dac. So whatever it is in parallel. Two of the op amps are 5nVrtHz and the other 1.1nV/rtHz so that and the 4k99 resistors.
Yes I believe the 555 has Bruce's 'composite opamps' with seperate supplies to the opamps in a pair.
The -117dB THD+N (-120dB typ) is at 1kHz in 22k BW. It is actually specced as '-117dB+1.0uV'.
The FFT has 1MHz BW! @ 24 bit. It's a mean machine, I wish I won the lottery. Just to be able to stare at it 😱
The SRS1 is more like a 515, performance-wise, I believe.
Jan
Last edited:
Like oscillators there is a trade off I would venture to say the feedback network is not based on 100 Ohms just as using 100 Ohm resistors in the oscillator is impractical due to the drive requirements. Considering their unique synchronous measurement mode there would be no reason at all to compromise the distortion to lower the input referred noise.
BTW this plot is quite different from the others that have been posted and could be believable if you are saying there is 100db processing gain.
Makes sense rss this all together and a 10-20nV floor would make sense and would fit the plot. At this point we are splitting hairs and should move on. It still would be useful to document how to interpret these plots in an unambiguous way for others that follow.
Last edited:
Yes. This much better and more like what I get using the 725C/D
This is at unity and should be used as a reference for all other input levels. Easy to see from the noise where the 725D has scaled the gain.
I can see that one of the input (summing?) amps has power output devices (TO220). So maybe it is driving low Z.
??
-RM
Short of a complete teardown we should probably let this lie, the instrument's noise is usually not the limiting factor in any case. What you posted looks reasonable.
Last edited:
Just an aside the QuantAsylum QA400 (thanks Demian) is one of the most annoying pieces of kit that I have played with in a while. I would sent it back to its creator ASAP.
That made me laugh!
I don't have a QA device. But have had the same feeling with some GPIB setups!
Whats the complaint.... it isn't universal and has limited software flexibility/capability - but seems straight forward for a sound card FFT replacement. Anyway, I use the QA401, FWIW
-RM
-RM
Last edited:
same instruments and gen as always. same setup..... just different 725D input scale or gen output voltage.
??
THx-RNMarsh
My point on today's audio gear (look at as an example at Schiit, dCs, DIY-gear ... and so one) that it look's like putting things together.
In other words:
- no RF or EMI related isolation
- Digital in and on Analog Chip out almost on same board
- they never used any magnetic or electric RF probes system in conjunct with RF spectrum analyzer
- and sometimes you are surprised to get more than 120dB's of headroom
- noting boxed or shielded even against USB or galvanic isolation's (look into an TEK scope and than you will see ... 😱)
- just put a EMI filter at the power does not help, while for RF/EMI the question rises from source and sink
- in SW is no such cross talk or inter modulation issue (only on buffer overflows)
- rising the clock from the old day to xx or xxx MHz made the clock & jitter question more as an issue
just my 2 cents
Hp
The limited software flexibility is an understatement. Not being able to use it to acquire data for external applications would exclude it from consideration for me. The 401 looks to have some improvements.
And why the power of 2 only FFT's no common sampling rates are but they ALL are small prime factors. This helps eliminate some of those factors that keep getting forgotten. Even the SDK for 1998's CoolEdit had an FFT package to support this for plug-in developers.
The virtual knobs are "cute" but are mainly annoying if you don't use a mouse.
It's also quite microphonic.
Last edited: