DIY Audio Analyzer with AK5397/AK5394A and AK4490

Jump to Latest
#3,146
1. Completely separated ADC and DAC channels to avoid any traces (harmonics!) of cross-talk.
2. Higher input sensitivity (i.e. addition of 31.6 and 10mV) to make an LNA behind a twin T filter superfluous.
3. DAC output steps of 10db instead of 20dB.
4. Noise level as low as possible.
5. Digital sync. input and output (a la Lynx L22/44).
6. Service manual for re-calibration.
 
#3,147
1) optimise layout to lower 'noise' componets . Possible easy wins e.g. current product is compromised by the toroidal XFR location. If it's located outside the enclosure (or even its orientation changed) then there is a measurable improvement on measurements with inputs shorted.
2) as asked for the current product but never addressed, enough info to permit DIY repairs and calibration if required. Big pity that the current excellent product might be junked due to lack of such info in years to come.
3) ?
4) ?
5) ?
 
#3,148
1) Complete software compatibility with the previous version.
2) Of course any technical improvements that are reasonable.
3) Service manual and calibration instructions! YES!
4) Okay, make it a two box instrument to lower supply noise. At this level of performance you may have to, or a larger chassis with 19" rack mount options.
5) Probably will need a metal enclosure with compartments between circuits much like the test equipment I am used to working with.

Jens, you made an excellent product with amazing performance given the cost and enclosure style. Improving performance significantly will take a lot of extra effort I think, because factors other than a better DAC and ADC are involved. I don't know about the 20 Vrms request, I wouldn't increase instrument cost greatly fulfilling that. I'd rather have superb performance at normal signal levels as they are.
 
  • Like
Reactions: Edmond Stuart
#3,152
Hi Jens, To make it a real versatile audio development tool for speaker design and other types, include a powerfull DSP/FPGA and implement a fully digital cross-over functionality with 2 x 4 channels balanced audio out and same in digital domain with SPDIF opt / COAX output supporting 192K24 bit minimum.
Crossover functionality like the Yamaha D2040 for filter setting. Include digital Master volume control serving both analog and digital outputs. /Kurt
 
#3,153
  1. 32 bit 768KHz
  2. Auto-ranging
  3. Input impedance of 100K or more
  4. A single nice box
  5. Not much more than $500
I've been using Focusrite Scarlett gen 2 & 3 and they're alright.

Currently, the obvious upgrade path for me is IVX's excellent E1DA Cosmos ADC, in conjunction with a vanilla DAC, as described by PMA. The forthcoming Scaler would be an interesting addition. All of this go for $500 or less, hopefully, because the Scaler is not out yet. This is a formidable baseline to beat.

Oh, and before anyone screams, the #1 item is to hopefully allow me to retire my analog gears for frequency response measurements, open loop vs closed loop, that sort of things.
 
#3,154
Aside from a DAC/ADC technology uplift ...
a) Auto/Manual ranging w/ OVP
b) 50/75/47K-100K/1M Input-Z, w/ exposed telescoping/driven shields
c) Full digital I/O (S/PDIF, Toslink, STglass, AES3, IIS, USB), w/ SR+Bit indicators
d) Transformer coupling or THAT-style high CMMR : option?
e) Fully Isolated ADC/DAC topology
f) Battery option
😎
 
#3,156
lgo51 said:
Auto/Manual ranging
Click to expand...
No, please no auto ranging! At least not inside the analyzer hardware!

If really necessary, the gain changes must be controlled by the analysis software so that the software can keep track of the gain setting. That's how I implemented it in MATAA for harmonic distortion analysis with the RTX at stepped amplitude levels. However, other types of analysis MUST keep the gain fixed during analysis (many tests related to loudspeakers or filter transfer, etc.).
 
  • Like
Reactions: Edmond Stuart
#3,157
You could always follow AP's model and have auto-ranging on by default and offer the ability to select a specific input attenuation/gain in software.

I don't know what sort of protection you have on the inputs and outputs. If you don't have any it could be worthwhile adding some so a probe slip or incorrect connection doesn't kill the analyzer. I hooked up my HP8903A to the screen grid of a power tube once instead of the control grid. Blew a fuse in the analyzer and that was is.

Tom
 
#3,158
tomchr said:
You could always follow AP's model and have auto-ranging on by default and offer the ability to select a specific input attenuation/gain in software.
Click to expand...
True. However, AP has very fine control on their hardware and software, so they can always make sure the hardware and software work in tandem. This is different with the RTX, which is just a piece of hardware that will be used with whatever software users have / like / need. Most software packages (REW, Arta, etc.) do not even know about the option of variable gain at the RTX (and therefore also don't read the gain settings from the hardware). Still, things still work fine because users use their brain to get things right.

If you really want to implement auto ranging in the hardware, I can see two ways to deal with it cleanly:
  1. Autoranging should be off by default so that no software will run into trouble from gain switching. If a given software package can / needs to make use of the hardware autoranging of the "new RTX", it can explicitly tell the analyzer hardware to activate it.
  2. Use a hardware toggle switch on the face plate of the "new RTX" so users can turn it on or off to suit their needs and the needs of their software.
Anyway, I don't see the advantage of autoranging in hardware. Looking at the hardware + software measurement system as a whole, with hardware autoranging, the software will need to know about the (automatic) gain setting of the hardware. This means the software would have to continuously poll the gain setting of the hardware, and it would always know about gain changes after the change has happened in the hardware (i.e., too late). It would be a lot easier to let the software decide if autoranging is necessary or beneficial for a given test, and deal with the best gain setting as needed. That's how some (very few) software packages do it already with with the current RTX analyzer, and it works fine.
 
  • Like
Reactions: PierreQuiRoule and Edmond Stuart
#3,160
mbrennwa said:
However, AP has very fine control on their hardware and software, so they can always make sure the hardware and software work in tandem.
Click to expand...
True that. That $32k does buy something after all... 🙂

mbrennwa said:
Autoranging should be off by default
Click to expand...
I disagree on that actually. I would leave auto ranging on by default to protect the analyzer. Then have a front panel override.

I assumed that this analyzer required its own software. My bad. If you're using generic software (which is a brilliant move in my book) options for control are more limited.

Tom
 
You must log in or register to reply here.
Top Bottom