Project XI : Audio Analyser

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hey all,

Welcome to Project XI.

You may be asking why another audio analysis tool. Well, this is a DIY site isn't it? During the design I'll be checking rigorously with an Rohde & Schwarz UPL.

Highlights:

Isolated USB with LTM2884
Isolated DC input with LTC4365 9-16V protection
XLR input with range selection switches
OPA1612 & OPA1632 input opamps
OP1611 VREF buffers from precision voltage source
AK5578 ADC in 8:2 summation mode with VREF 5.25V for S/(N+D) 118db/ SNR 127db
AK5597 DAC with OPA1632 opamps and relay switched output level
CM6632A with GX3306-J/A clocks
Multiple LT3042 regulators
Isolated SPDIF I/O
Isolated i2s I/O

Watch this space!
 

Attachments

  • projectVI.png
    projectVI.png
    263.7 KB · Views: 880
Last edited:
H
AK5578 ADC in 8:2 summation mode with VREF 5.25V for S/(N+D) 118db/ SNR 127db

Well, for the ultimate Analyzer:

1. Separated ADC/DAC chips to avoid channel cross talks
2. Then also full isolated and shielded channels
3. keep in mind that those ADC's have poor freq. response on certain SR
4. may support also DSD 512 (native) on ASIO
5. Input / output attenuation with symmetric & asymmetric input
(and keep attention of stray input currents for asymmetric input
and may use multi layers > 4 or measurement ground)
6. A 1khz & 10kHz Notch for measurements :D
7. ASIO driver support for Windows

just my 2 cents

Hp
 
Member
Joined 2004
Paid Member
I would strongly suggest that the digital, ADC and DAC are all isolated from each other. This prevents some ground loops and enables using the DAC output with a balanced/BTL mode as a floating source without concern about grounds and shorting an output. The biggest limitation of the RTX and the QA401 is the common ground between input and output.

The QA401 uses the AK5397 but shows its limitations. Still lots less expensive than the AK5394A (soon obsolete). Still RTX demonstrated that the AK5394A has the best measured THD performance and a noise floor that's pretty flat to 100 KHz of available solutions, unlike most audio ADC's.

Frex's efforts with the SAR ADC show another option that may prove to be really good, still too soon to know. And there are some new industrial ADC's that are very promising, e.g. http://www.ti.com/document-viewer/ADS8904B/datasheet/abstract#SBAS4417472 with -125 dB THD with 500 Ksps. (Not sure just how they measured that. . .)

All of this is far beyond anything needed for actual audio measurements of real products.
 
I'm also thinking of experimenting with AK557x, mostly for learning and experimenting but also for occasional measurement checks at home. As I also have an APx to use at work, it would be useful to develop at least some measurement capability to home (knowing I may not have access to AP forever).

There has been quite some DIY Analyzer threads but not many seem to share schematics. I would be happy if someone could point me to good reads on following topics:

-Input attenuation and protection (while maintaining good performance). I am not designing a product but a tinkering platform for myself so it does not need to be ultimately robust but some protection would be good. Main focus on input scaling.

-Input filter/buffer considerations; I assume folks do not use the datasheet/EVM filters.

-Solutions for the abovementioned isolation between blocks.

Thanks
 
Member
Joined 2004
Paid Member
Not much on DIY but schematics for AP, HP, Boonton and Amber are all floating around the interwebs and they are the best source for solutions for protection and ranging. Lightbulbs and reverse biased diode protection are almost essential to get to SOTA noise and distortion performance.

Jan's autoranger also is a source for ideas for input circuitry.

Fixed or tunable notch filters are essential to get to real SOTA distortion. The best ADC's are still well short of what can be done with passive preconditioning networks. I believe Edmond's Diana DiAna, a software Distortion Analyzer will be implementing a scheme for making sense of a prefilter with a soundcard.
 
Is the OPA1632 up to the task of driving the AK557X input?

I did make my own AK5572 board after those comments above (went for AK5572 instead of AK5394 as it was about to become obsolete). Haven't optimised the BOM really but turned out working quite well and consistently getting THD+N peaking at -113dB, which by datasheet spec is great, and THD peaking at around -117dB, and DR 121dBA. I use OPA1632 to drive the input.

Input clamping and such are not the best in this though, limiting single-ended input level.

ADC AK5572 – Very high performance audio AD-converter – nihtila.com (link to schematic at the end)

I will probably make another version of that as there has been some interest and at the moment this board is a pain in the bum to assemble.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.