What is the goal to build an custom instrument, that performs same as commercially available QA400. It costs just $200, and far more accurate than all of those software toys due to internal reference.
I think that custom made instrument should to exceed the parameters of an average quality, commercially available units.
I think that capability to measure down to 0.001% THD is enough for the DIY purposes. This is -100dB. To deal with those new low distortions designs we need to go to 0.0003% THD floor -110.. dB. Additional 6dB for accuracy, and here is a grand total: -117dB.
I think that custom made instrument should to exceed the parameters of an average quality, commercially available units.
I think that capability to measure down to 0.001% THD is enough for the DIY purposes. This is -100dB. To deal with those new low distortions designs we need to go to 0.0003% THD floor -110.. dB. Additional 6dB for accuracy, and here is a grand total: -117dB.
Why not to add several relays, and switch like: CONSTANT IMPEDANCE <-> LOW NOISE ?
That could be a solution for the single ended output. Another solution could perhaps be a quasi-floating balanced output as described in figure 6 here:
Balanced Line Technology
I have not figured out a good solution for combining this with an attenuator though.
Adding the solution you suggest with high current drivers again makes the design grow.
At the moment the single ended output connectors (BNC) are grounded. The thought was that this might be OK if the balanced input was used to eliminate the ground loops. But perhaps the output should be improved as per your suggestion or something similar.
I don't think we need a notch filter that good. If the notch filter is combined with an FFT analyzer I assume that it will be sufficient to attenuate the test signal by e.g. 40-50dB, amplify the output and let the ADC convert the output of the filter and amplifier. Then the ADC will give a very small contribution to the distortion. And the distortion can be calculated, knowing what the test signal level should have been without the notch filter.
The front-end before the filter needs to have very low distortion. Not an easy task, but more realistic than getting an ADC with very low distortion (e.g. less than -130 to-140dB).
I don't own a QA400, but it seems to give good value for money. I have no intention or interest in bashing the QA400. It seems like a good design within the constraints of the cost and power consumption. But I don't think it is in the same league as what I am targeting and have just about achieved.
Adding some relays to select minimum or constant impedance is of course an option. But all additions make the design more complex and expensive, so I need to balance the cost/complexity/risk of additional functions against the added benefits.
Hello JensH,
The mine idea is not a bashing of the QA400, or any other product, but to give to DIY community an instrument, that performs close to best industry instruments, but still affordable to individuals. And modern components base making that dream possible to come true.
Panasonic VP 7723A 5Hz 100KHz 0 0005 Audio Analyzer | eBay
Just look at price, and how much people are interesting about it.
I believe, that you project can to reach even better price/performance/functionality ratio today.
The mine idea is not a bashing of the QA400, or any other product, but to give to DIY community an instrument, that performs close to best industry instruments, but still affordable to individuals. And modern components base making that dream possible to come true.
Panasonic VP 7723A 5Hz 100KHz 0 0005 Audio Analyzer | eBay
Just look at price, and how much people are interesting about it.
I believe, that you project can to reach even better price/performance/functionality ratio today.
The standard for industrial distortion analyzers is 10dB above the base line of the analyzer.
So if an instrument can do -100dB then it's good for measurements down to -90dB.
That 10dB is considered the region of uncertainty. I think it covers all the reasons for uncertainty in distortion measurement. Phase cancellation or addition of the harmonics and anything in between. if one wants to measure -110dBV then the base line of the analyzer needs to be -120dBV.
Most of the observed distortion with direct ADC measurement is from the analog front end and nothing to do with the accuracy of the ADC. it the analog front end which benefits from and notch filter. I think the non linearity of a good ADC are somewhat insignificant to that of the analog process.
So if an instrument can do -100dB then it's good for measurements down to -90dB.
That 10dB is considered the region of uncertainty. I think it covers all the reasons for uncertainty in distortion measurement. Phase cancellation or addition of the harmonics and anything in between. if one wants to measure -110dBV then the base line of the analyzer needs to be -120dBV.
Most of the observed distortion with direct ADC measurement is from the analog front end and nothing to do with the accuracy of the ADC. it the analog front end which benefits from and notch filter. I think the non linearity of a good ADC are somewhat insignificant to that of the analog process.
The main challenge to address with input buffers seems to be common-mode distortion. You can choose an inverting amplifier but you have to trade off noise for input impedance and thus distortion from the source. The non-inverting configurations will give lower noise of course but then you have the increased common-mode distortion.
You can bootstrap the rails but then you're losing supply headroom.
You can bootstrap the rails but then you're losing supply headroom.
The problem with notch filters is the distortion they can add. At -120 dB+ the nonlinearity of the notch circuit limits the system. Some notches are better than others. For this a broader band shallower notch is more useful but still complex. The input buffer is the bigger challenge and you are correct- common mode distortion (that can be measured at the input directly with the right hardware) is the key culprit.
There is also the next step-up model (7722) for sale.
Panasonic VP 7722A 10Hz 110kHz 0 0001 Audio Analyzer | eBay
Both use notch filters. Accuracy is spec'ed at 1dB max error of measured/indicated harmonic level.
[BTW - I have 2 of these.... one in California and one in Thailand ..... both have residual levels of .00003-5% using built-in osc]
THx-RNMarsh
Last edited:
Before adding complexity like parallel chips, maybe a good idea would be to wait a little: ESS-Tech is about to release new ADCs.
Also, AKM has a new DAC in preparation, the AK4497, with apparently higher performance than AK4490 and AK4495s.
No info on when both will come into production though.
Also, AKM has a new DAC in preparation, the AK4497, with apparently higher performance than AK4490 and AK4495s.
No info on when both will come into production though.
Last edited:
It depends what you are doing - if tube amplifier, it is more then enough, but what if you developing Audio-DAC?
0.0003% - is not a problem, even my "Entry-Level ADC" called Olivine measure Victor's Generator as: 0.00025% at -0.5dBFS (THD Floor ~-120dBFS) and 0.00017% at -3dBFS (THD Floor -125dBFS) with ~-145dBFS Noise Floor.
A problem is to explore harmonics at less then -130-140dBFS level, this is really problematic w/o notch filter.
Хаг самеах.
Last edited:
How much people you think, are playing around -130 140dB THD ? And I know, that there are guys, that want to try it, but can't, due to they are have not access to appropriate measuring equipment. But how many such enthusiasts you know ? I don't think a lot. And as we know, 0.001% THD is quite below sensitivity threshold of human's ear. There is no margin for the good things, but how much it will cost ? Good if no more than 8-10% of price PA or Prism analysers.
So here we need to or determinate reasonable margins for price/performance, or to make 2 separate instruments: for average DIY'er, and the PRO guys. Another way probably is to do it modular... Each one will be able to get what he need, or what he will able to pay for.
So here we need to or determinate reasonable margins for price/performance, or to make 2 separate instruments: for average DIY'er, and the PRO guys. Another way probably is to do it modular... Each one will be able to get what he need, or what he will able to pay for.
Yes, sure.
Who need just -90-100dB THD, uses on board sound card.
Who need more - uses expancive sound cards, or special external A/D.
About this A/D we are talking in this thread. Who are not interesting, may not participate.
We are not talcking here about human's perception. I'm pesonally use tube amplifiers, whoes THD is much higher 🙂
But I neead to measure less then 0.0003% when I developing DAC, for example.
Or exploring the infuence to THD of some components (relays, connectors, resistor/capacitors, etc.)
The distance, between DIY and Pro - is not so big, trust me 🙂
But there are many devices, with harmonics at -100-120dB level.
To make an accurate measurement, we need A/D with its own harmonics at least 10dB less.
And, as I wrote before - there is a very interesting field of a passive components investigation.
altor, while dreaming about mega device, you probably missed what's author told several posts above. Here is it:
So if $1000 or 10000GBP are so close for you, why you wasting you valuable time here ?
Here are simple, and fast solutions for you demands, priced just "very" close to DIY kits:
dScope Series III analogue and digital audio analyzers
AP High Performance Audio Analyzer & Audio Test Instruments : apx555
https://www.rohde-schwarz.com/en/product/upv-productstartpage_63493-7558.html
But if you able to bring a solution, for -130-140dB, that will be within affordable price range, it will be much appreciated !
We are interesting to make that instrument not only fine quality, but affordable to maximal number of potential users also. So:
and than:
Hm... Prism costs around 9000-10000 British pounds. PA priced same. We are talking there about something within $1000-1500 range.
So if $1000 or 10000GBP are so close for you, why you wasting you valuable time here ?
Here are simple, and fast solutions for you demands, priced just "very" close to DIY kits:
dScope Series III analogue and digital audio analyzers
AP High Performance Audio Analyzer & Audio Test Instruments : apx555
https://www.rohde-schwarz.com/en/product/upv-productstartpage_63493-7558.html
But if you able to bring a solution, for -130-140dB, that will be within affordable price range, it will be much appreciated !
Last edited:
But the things, and futures you are asking for, coming not only by skills of designer, but at expenses of money too.
But if -130 -140dB floor is so required... Than modular, upgradable construction, or 2 models at least is the way to go.
But if -130 -140dB floor is so required... Than modular, upgradable construction, or 2 models at least is the way to go.
Last edited:
Good question. Subject to vote 🙂 To me $500-800 is threshold for very special instruments, that can be avoided, and up to $1500-2000 for the instruments, that I can't to work if those do not persist (like good oscilloscope for example).
Hm, I did not put my ADC "Onyx" into production, because it seems to me that it's selling price (at which it make to me sense to manufacture it) is too high - $450-500, and made an "entry-level" ADC "Olivine", just for $160.
But if you say that $500-800 is OK, I will re-consider the "Onyx" 🙂