DIY Audio Analyzer with AK5397/AK5394A and AK4490

Jump to Latest
1

1audio

Member
Joined 2004
Paid Member
For what its worth the APx555 uses the same ADC for the low distortion measurements. There may be some trick using two and adding them in a way that may cancel distortion to a small degree but its unlikely to be better. the AK5394A is quite old as these things go and even the designers have not been able to improve on its distortion. The newer chips may have slightly better signal to noise but they sacrifice some distortion to get it.

AP does use a separate ADC to get higher sample rates in their boxes. However I think a USB scope is a better solution and cleaner in application. They usually have FFT software and in my experiece can be good to 50 MHz as a spectrum analyzer.

I'll try to get a block diagram of the interface stuff I'm thinking of which should help explain the ideas. My one reluctance to having a plethora of connectors is that you still need a lot of matching cables to get use from them.

Also think about the signal levels on each interface.
 
anatech

anatech

Administrator
Joined 2004
Paid Member
Hi Demian,
That's very good of you. Thanks.

Hi AR2,
I'm glad you got that one. It really is a miserable connector. I'm not sure about this, but the DIN speaker plugs and jacks a maybe even more miserable to work with. I'm talking about the wonderful pin and bar connector. Anything larger than 20 gauge is a risky operation!

-Chris
 
A

altor

Member
Joined 2003
altor.co
JensH said:
The AK5397 does support higher sample rates. But in this case the distortion is not quite as good as for the AK5394A. And the frequency response at the high sample rates is far from flat. Not ideal for test equipment.
Click to expand...

I can say the same about ARDA's AT1201 - I hope to use it, it works at 384kHz and I'm able to send the data to PC via XMOS USB Interface, but it's performance is not good - THD at high levels is bad (the same, like TI's PCM4222 - cannot be used for measurements al levels, higher then -20dBFS without notch filter). Also frequency response after 110-120kHz drops to -10-20dB (for what make 384kHz sample rate, if the results are no so better then at 192kHz?).

P.S. BTW, as AT1201 is not good for measurements, but it is fine for music digitizing (vinyl records, etc), in XRD and DSD formats.
 
Last edited:
3

31697B

Disabled Account
Joined 2012
The front-end of the AP-2722 uses a relay complex to do input scaling which then goes to a descrete circuit design. That input also used the MAT02. One of the reasons for an new model was the MAT (and other parts) were no longer made. The newest AP is about 6dB better in the THD+N department. But to be more useful than that requires a lower distortion source/generator than the 2722.

That is why i am pushing for new gen designs.... they can be found in this forum under the low distortion audio gen section.

THx-RNMarsh
 
Last edited:
Turbon

Turbon

Member
Joined 2011
Push on RNMarsh - the only way to evolve :). Otherwise we get stuck on some group decided thruths that hold today but not tomorrow. Keep an open mind an let things flow and one does not have to like it but let it flow and grab anything you seem to be of interest. Flow to slow - push it :D!

Regards
 
C

chris719

Member
Joined 2004
RNMarsh said:
The front-end of the AP-2722 uses a relay complex to do input scaling which then goes to a descrete circuit design. That input also used the MAT02. One of the reasons for an new model was the MAT (and other parts) were no longer made. The newest AP is about 6dB better in the THD+N department. But to be more useful than that requires a lower distortion source/generator than the 2722.

That is why i am pushing for new gen designs.... they can be found in this forum under the low distortion audio gen section.

THx-RNMarsh
Click to expand...

Do you have a schematic of the input stage? I'd be curious to see it. Interesting that they are using a MAT-02 and not a JFET in front of (presumably) an op-amp.
 
J

JensH

Member
Joined 2009
Paid Member
A bipolar opamp will have higher current noise though. By using the LSK389 the voltage noise of the JFET is kept at a very low level. Much lower than JFET opamps.
The input noise of the differential input amplifier I made for the Audio Analyzer is around 4 nV/rtHz including series resistors for overvoltage protection.
 
J

JensH

Member
Joined 2009
Paid Member
That is pretty low for a bipolar opamp. But then the voltage noise is 11 nV/rtHz, so the total noise is still relatively high.
I have chosen a hybrid JFET/bipolar opamp design to get a low voltage noise and a low current noise. The DC offset and drift will be worse than with a good bipolar opamp, but for an Audio Analyzer I think this is less critical.
 
D

davada

R.I.P.
Joined 2004
The Shibasoku 725 uses a matched Jfet pair as buffers in darlington with bipolars making hybrid pair. Works quite well. Low noise and extremely low distortion. The Jfets have a cascode bootstrap which also serves to null leakage current from the input protection diodes. The Shibasoku is capacitor coupled.

Noise is an issue for the input amplifier of distortion analyzers much the same as input referred noise is for amplifiers. It defines the noise floor of the analyzer. So yes it matters.

Whatever works for your project.
 
J

JensH

Member
Joined 2009
Paid Member
The project is still moving forward, but not quite as fast as I had hoped.
I am working different parts of the project. Mechanical design, establishing a business case together with a manufacturer/supplier and a new USB interface to reduce the cost and not rely on a third party design.
 
You must log in or register to reply here.
Top Bottom