Hello 1audio,
Your feedback is very interesting...
But really how can they proclaim -120dB THD
if you only reach -90dB ! :-(
Could you explain the complete setup of this test,
is that done with EVM board or did you have build your one ?
Jensh,
I understand, follow all opinions is not possible if we want to finish a project...
Sometime, that can help to take advised decisions.
Very good luck for this design, i follow with interest the project progress.
Regards.
Frex
Your feedback is very interesting...
But really how can they proclaim -120dB THD
if you only reach -90dB ! :-(
Could you explain the complete setup of this test,
is that done with EVM board or did you have build your one ?
Jensh,
I understand, follow all opinions is not possible if we want to finish a project...
Sometime, that can help to take advised decisions.
Very good luck for this design, i follow with interest the project progress.
Regards.
Frex
I don't know what the backstory is, but I was testing their demo board. I tried every option ESS suggested but its performance was not good. There may be some complext tweaks that need to be run from obsuure software to get it to work right but its basic performance was not as good as the built in codec on a cheap motherboard. Too much trouble. And clearly the AKM performance is exceptional in Jens implementation.
I met with AKM Friday and reviewed the current product line. The newer ADC's may have almost as good distortion but the noise floor starts rising at 30 KHz. They do NOT have flat response above 100 KHz so the higher sample rates are not a benefit for measurements.
The AK4497 should be sampling by the end of this year and available early next year. It may be an improvement over the AK4490. Its in a 64 pin package (not easy to hand solder) with interleaved power and ground connections all over. I would not wait for it for this project. I don't see a reason to wait for any other ADC's either. The AK5394A was introduced in 2001! It seems to still have a pretty unique place in the market.
The National/TI end of life stuff is distressing but it makes sense in the larger TI context. The National line these are on is very different and much more manual that any TI line so they probably need to cut it off. I may need to make a lifetime buy of some key parts.
here is a pretty complete list: http://media.digikey.com/pdf/PCNs/Texas Instruments/PCN20150901004.pdf
I suspect buying a good size stock could be a great investment.
Unfortunately we are going to the MP3 age. There will not be a place for fine quality music reproduction, as well for the precise audio measurement instruments. -140dB analyser just needless to check the analogue output of mobile devices.
In the upcoming dark times, the our hope are guys from the Analog Devises, and Linear Technology only. AD797BRZ, LT1028, LT1128, probably MAX9632.
Last edited:
@ Tolik.....
Fortunately, this forum is one place where we don't need to irritate/frustrate/depress ourselves with the fact that the "sheeple", as led by "Lord" Apple & the worthless music abomination otherwise known as "Itunes" are force feeding MP3 as a valid music encoding standard.
From a business perspective, it's a brilliant ploy to peddle & charge $ for mp3 music ( which in my opinion has no musical value whatsoever ) Spiritually & ethically it is dishonest & corrupt.
Personally I was using mp3 on my work machines over 20years ago & I was willing to accept the qualitative compromises in that context, for reasons of data storage cost/minimisation...
I'm sure that JensH is well aware of this & his effort to bring this project to fruition is that much more commendable.
JensH, I hope you bring the project to completion, there are many individuals & educational organisations that don't have the budget for professional instrumentation at this level & there is a definite renessance in the entire DIY ecosphere & your project is a most worthwhile & significant audio building block !
good luck & "hats off to you" as they say ;-)
Fortunately, this forum is one place where we don't need to irritate/frustrate/depress ourselves with the fact that the "sheeple", as led by "Lord" Apple & the worthless music abomination otherwise known as "Itunes" are force feeding MP3 as a valid music encoding standard.
From a business perspective, it's a brilliant ploy to peddle & charge $ for mp3 music ( which in my opinion has no musical value whatsoever ) Spiritually & ethically it is dishonest & corrupt.
Personally I was using mp3 on my work machines over 20years ago & I was willing to accept the qualitative compromises in that context, for reasons of data storage cost/minimisation...
I'm sure that JensH is well aware of this & his effort to bring this project to fruition is that much more commendable.
JensH, I hope you bring the project to completion, there are many individuals & educational organisations that don't have the budget for professional instrumentation at this level & there is a definite renessance in the entire DIY ecosphere & your project is a most worthwhile & significant audio building block !
good luck & "hats off to you" as they say ;-)
Just reading from the first message:
In other words there where no LME parts involved and the OPA's 1611/12 have a lower noise than LME49710. In the other hand the OPA1632 is full symmetric ADC driver..
May question rises to this:
what really changed performance wise, from the OPA parts to the LME parts.. what we gain what we loose
Hp
In other words there where no LME parts involved and the OPA's 1611/12 have a lower noise than LME49710. In the other hand the OPA1632 is full symmetric ADC driver..
May question rises to this:
what really changed performance wise, from the OPA parts to the LME parts.. what we gain what we loose
Hp
Hi Jens,
Thank you for sharing your work and being open about both your progress as well as the setbacks. I hope you will share the full schematic soon? Or did I miss it somewhere?
Your 1kHz THD figure is outstanding. How good is the 20kHz THD at 192kHz sampling rate?
What is the analyzer input noise level in dBFS with shorted inputs, and what is the noise level when connected to the generator DAC output? Often there are THD vs noise level tradeoffs.
Here is a little contribution from me for your inspiration. These are some pictures I took of one of your "competitors" products
https://www.dropbox.com/sh/48k5fbm6x4wfl2t/AABoEW1XxyCUqKnjRd7IUWWaa?dl=0
Br,
OJG
Thank you for sharing your work and being open about both your progress as well as the setbacks. I hope you will share the full schematic soon? Or did I miss it somewhere?
Your 1kHz THD figure is outstanding. How good is the 20kHz THD at 192kHz sampling rate?
What is the analyzer input noise level in dBFS with shorted inputs, and what is the noise level when connected to the generator DAC output? Often there are THD vs noise level tradeoffs.
Here is a little contribution from me for your inspiration. These are some pictures I took of one of your "competitors" products
https://www.dropbox.com/sh/48k5fbm6x4wfl2t/AABoEW1XxyCUqKnjRd7IUWWaa?dl=0
Br,
OJG
Hi There,
I ve spent some hours reading the whole thread and this was very inspirational! lot of great effort Jens. I was already inspired by your thread on Minidsp USBStreamer.
this post to quickly give a feedback from some reading around the Benchmark ADC1, it uses the ak5394AVS and it seems they run it at 221khz straight into and AD1896 and they reach 120db snr but 107db thd. The input are made of ad797 and op27 (bias?).
data taken from here : Benchmark ADC1
anyway this is not a device for test purpose, but I m sure they did try to get the best quality out of it this combo.
BTW in all the thread you focus on the ADC. What conclusion about DAC ? does the AK looks to be the right choice or some other old fancy gears might still be considered ??
When it comes to SAR ADC, I ve been thinking about that too, if you combine the power of an accurate SAR DAC with all the filtering technics which are embded in a DeltaSigma ADC, you certainly endup with a topgear... but a dsp or digital filter is needed.
jocker, you seems to have a nice litle board with the 5394, would you be ready to sell a spare PCB in case I pursue my own project with this ADC?
please keep us posted
I ve spent some hours reading the whole thread and this was very inspirational! lot of great effort Jens. I was already inspired by your thread on Minidsp USBStreamer.
this post to quickly give a feedback from some reading around the Benchmark ADC1, it uses the ak5394AVS and it seems they run it at 221khz straight into and AD1896 and they reach 120db snr but 107db thd. The input are made of ad797 and op27 (bias?).
data taken from here : Benchmark ADC1
anyway this is not a device for test purpose, but I m sure they did try to get the best quality out of it this combo.
BTW in all the thread you focus on the ADC. What conclusion about DAC ? does the AK looks to be the right choice or some other old fancy gears might still be considered ??
When it comes to SAR ADC, I ve been thinking about that too, if you combine the power of an accurate SAR DAC with all the filtering technics which are embded in a DeltaSigma ADC, you certainly endup with a topgear... but a dsp or digital filter is needed.
jocker, you seems to have a nice litle board with the 5394, would you be ready to sell a spare PCB in case I pursue my own project with this ADC?
please keep us posted
Last edited:
@HpW
A number of things changed from the earlier design to the present one. So it is not easy to compare the results directly. But I never got the same low distortion with the older design (with OPA1612 and OPA1632) that I have got with the LME49990 based design.
And using the OPA1612 alone in the instrumentation amp input stage can not beat the LSK389/LME49990 combo.
@ojg
You didn't miss the schematic somewhere, because I haven't published it.
I will make some measurements and report later.
You didn't mention the name of this unknown "competitor". But looking closely at the pictures I could see a logo saying "AP". I think I have heard of them before
Interesting to see that they also use the AK5394A in their design.
@maxidcx
The performance of the Benchmark ADC seems to be quite good. It does show an increase in distortion at high levels though. This could perhaps be caused by a relatively high temperature? I read somewhere that it does become rather warm. But since the datails of the analog front end is unknown this is of course purely speculation.
The floor plan of their layout is a bit surprising, but they seem to almost avoid spurious tones, so it doesn't seem to be a problem.
The AK4490 can give a very low distortion, especially if the level is somewhere below -10dBFS. At higher levels the distortion increases. It does eem to generate some out of (audio) band noise, dependent on the input level. This may not be a real problem for measurements, but of course it doesn't look so nice.
Are there better DAC's out there? I don't know. The AK4490 is definitely not bad. Perhaps the coming AK4497 will be better, but at the moment this is unknown.
I may be willing to sell some spare AK5394A boards.
A number of things changed from the earlier design to the present one. So it is not easy to compare the results directly. But I never got the same low distortion with the older design (with OPA1612 and OPA1632) that I have got with the LME49990 based design.
And using the OPA1612 alone in the instrumentation amp input stage can not beat the LSK389/LME49990 combo.
@ojg
You didn't miss the schematic somewhere, because I haven't published it.
I will make some measurements and report later.
You didn't mention the name of this unknown "competitor". But looking closely at the pictures I could see a logo saying "AP". I think I have heard of them before
Interesting to see that they also use the AK5394A in their design.
@maxidcx
The performance of the Benchmark ADC seems to be quite good. It does show an increase in distortion at high levels though. This could perhaps be caused by a relatively high temperature? I read somewhere that it does become rather warm. But since the datails of the analog front end is unknown this is of course purely speculation.
The floor plan of their layout is a bit surprising, but they seem to almost avoid spurious tones, so it doesn't seem to be a problem.
The AK4490 can give a very low distortion, especially if the level is somewhere below -10dBFS. At higher levels the distortion increases. It does eem to generate some out of (audio) band noise, dependent on the input level. This may not be a real problem for measurements, but of course it doesn't look so nice.
Are there better DAC's out there? I don't know. The AK4490 is definitely not bad. Perhaps the coming AK4497 will be better, but at the moment this is unknown.
I may be willing to sell some spare AK5394A boards.
Yes and since you were discussing opamps earlier, it is also interesting to note that they use a discrete transistor input amplifier with SSM2212 and also LM4562 opamps.
Jens,
From what I have read in this thread and others, you have played with motherboard and discrete ADC and DAC boards...
My experience is that you need to be uber cautious on grounding and noise. Down to -100 to -110dbc I have been able to keep noise and especially ground induced noise under control in a motherboard / daughterboard system which has single ended I/o. But I would dread to shoot for too much more without going fully differential. Unfortunately real world devices have an odd habit of being single ended.
Just watch that complexity meter! Might be easier and better going for a "good but not so simple" approach. Btw. I noted the AP photos had "gone large" on the common pcb, and I presume ground plane, approach.
From what I have read in this thread and others, you have played with motherboard and discrete ADC and DAC boards...
My experience is that you need to be uber cautious on grounding and noise. Down to -100 to -110dbc I have been able to keep noise and especially ground induced noise under control in a motherboard / daughterboard system which has single ended I/o. But I would dread to shoot for too much more without going fully differential. Unfortunately real world devices have an odd habit of being single ended.
Just watch that complexity meter! Might be easier and better going for a "good but not so simple" approach. Btw. I noted the AP photos had "gone large" on the common pcb, and I presume ground plane, approach.
I think, if there is going to be several revisions of the instrument, there is good idea to organize the information somehow.
It is possible to start new thread, something like: DIY Audio Analyzer with AK5397 by JensH rev 1.
Than in the first message place all of the specs data, safety info (max input, PSU values) info about important mine features, brief announce about upcoming revision if planned.
Than to start new thread about rev 2.
What's new / changes. And than same as for rev 1.
It is possible to create a wiki page...
Such way, it will be simple to people here, to compare the designs, and to choose most appropriate solution for their needs.
P.S. I'm interesting about fine analyser too
It is possible to start new thread, something like: DIY Audio Analyzer with AK5397 by JensH rev 1.
Than in the first message place all of the specs data, safety info (max input, PSU values) info about important mine features, brief announce about upcoming revision if planned.
Than to start new thread about rev 2.
What's new / changes. And than same as for rev 1.
It is possible to create a wiki page...
Such way, it will be simple to people here, to compare the designs, and to choose most appropriate solution for their needs.
P.S. I'm interesting about fine analyser too
