Bear --- I don't know. Maybe wrong thread.
However, it is clear to everyone for over a decade that you need a notch filter in front of an ADC if you want to use it as part of a test equipment with FFT. So on to the 'with and without notch filter' for the QA400 now that I have something reasonably accurate to calibrate it against.
Here's A-P's before and after harmonic data of their source generator (year 2003 literature).
View attachment ADC.pdf
Thx-RNMarsh
However, it is clear to everyone for over a decade that you need a notch filter in front of an ADC if you want to use it as part of a test equipment with FFT. So on to the 'with and without notch filter' for the QA400 now that I have something reasonably accurate to calibrate it against.
Here's A-P's before and after harmonic data of their source generator (year 2003 literature).
View attachment ADC.pdf
Thx-RNMarsh
Last edited:
Well, if you don't know...
...thought you were leading up to making a point.
Re: the AP w/wo notch... it seems like they are saying if you hit the ADC with enough signal it produces extra harmonics. That simple. If I have time I would like to try to whip up a passive >36dB notch to throw on the front of the QA.
As I noted before it seems that the VP-7725 *has* an auto notch feature running all the time.
Wish I had time to do this stuff...
...thought you were leading up to making a point.
Re: the AP w/wo notch... it seems like they are saying if you hit the ADC with enough signal it produces extra harmonics. That simple. If I have time I would like to try to whip up a passive >36dB notch to throw on the front of the QA.
As I noted before it seems that the VP-7725 *has* an auto notch feature running all the time.
Wish I had time to do this stuff...
Here's the deal as I see it so far;
1. The lower the ADC and supporting IC's operating supply voltage, the smaller the linear portion of its dynamic range.
2. As you approach the upper input limit, the device will produce extraneous harmonics on top of the DUT's harmonics and beyond. How soon this happens depends on the device and the operating voltage.
3. At the low end of the operating range, the error is in amplitude.... it does not produce extraneous harmonics, it just isn't accurate in the levels displayed... by as much as 10+ dB's in variation in level. This area is where DSP can have a correction look-up table is my suggestion. But other-wise just know your device's error and mentally correct.
So to minimize 1-3, above - you need to be sure that the ADC is operated only in the middle of the linear range at all times via notch filter and clean amplification.
Thx-RNMarsh
1. The lower the ADC and supporting IC's operating supply voltage, the smaller the linear portion of its dynamic range.
2. As you approach the upper input limit, the device will produce extraneous harmonics on top of the DUT's harmonics and beyond. How soon this happens depends on the device and the operating voltage.
3. At the low end of the operating range, the error is in amplitude.... it does not produce extraneous harmonics, it just isn't accurate in the levels displayed... by as much as 10+ dB's in variation in level. This area is where DSP can have a correction look-up table is my suggestion. But other-wise just know your device's error and mentally correct.
So to minimize 1-3, above - you need to be sure that the ADC is operated only in the middle of the linear range at all times via notch filter and clean amplification.
Thx-RNMarsh
Last edited:
For people who do not know (like I) what a VP7725 (exactly) is
http://www.kikusui.co.jp/common/product/pdf/levear/levear_aa_vp77xxx.pdf
http://www.kikusui.co.jp/common/product/pdf/levear/levear_aa_vp77xxx.pdf
I have an EMU 0404 and using it's signal source and ARTA I got around 0.0025% THD. THD+N noise performance was not as good good though, like 0.015%
It's drivers won't work with Windows 7 unfortunately, so couldn't use when I switched from my Mac to a PC laptop.
I'm just about to purchase the QA400. Looks like this the best option out there for THD measurements at $200.00?
It's drivers won't work with Windows 7 unfortunately, so couldn't use when I switched from my Mac to a PC laptop.
I'm just about to purchase the QA400. Looks like this the best option out there for THD measurements at $200.00?
The 0404 and ARTA should be a good combo..... however - just note that the low cost 24 bit ADC's should be used with a notch filter or at least don't push signals into them above -10dB or you will get fake data results.
Would you try your test again but with input at -10dB and see what the THD and THD+N is compared to the test you just told us about?
What I like about the QA400 is that everything is integrated in one package. But that package is more limited in apps than ARTA.
-RM
Last edited:
I have done this throughout the past year. The HP339A and the HP334 do not go down low enough to measure the generator such as Victor's .... they show the residual of their respective analyzers. This can be improved somewhat with various mods. The better analyzers I have are accurate to somewhere below the -140dB levels in THD. But now you are talking about new price around USD of $30K.
There are some good deals in used test equipment but still expect to pay at least $1000-3000+ USD. Nothing is accurate below -100dB so far as I can tell. Except the very expensive units. Even with a notch filter, you are then dealing with low level non-linearity's of the ADC system et al. I have to test that situation out more thoroughly though. But, if the residuals sent to the ADC can be amplified without added distortion up to a mid voltage level where the ADC is linear, then that would be OK - still don't know how low it could measure THD with accuracy. Maybe -110 -120dB ?
Thx-RNMarsh
Last edited:
I was thinking more along the lines of using your modified 339A as the source and the QA-400, AP, ShibaSoku, etc. individually to see how they measure the source. Should they be all about the same? I would think so. I will probably never own an AP, SibaSoku, or the other analyzer you have.
I would like to see if the QA-400 is close and accurate because that is in my price range. Add in an interface and a notch...should be a nice package.
Dave
I would like to see if the QA-400 is close and accurate because that is in my price range. Add in an interface and a notch...should be a nice package.
Dave
yes, and that is eactly one of my suggestions.... Put an FFT system on the notched Monitor output of an HP-339A or similar. You get an auto-null and wideband voltmeter to 300v input and a variable osc/generator that can be easily better than -100dB in THD+N direct-metered/read. And the output level at the Monitor port is 1 volt (fs or full scale). This allows level adjustment to the ADC from the Monitor port to see harmonics at a level with the ADC/FFT system which is not too high or too low.
THx-RNMarsh
THx-RNMarsh
Last edited:
I'm with Dave. I am not looking for measurements below -100dB. I just want to test vintage audio gear. I have become very comfortable with my QA400's software and it's operation. I learn something every time I use it. It is perfect for my hobby needs. I also use my E-MU 204 and ARTA STEPS for response measurements. I also use my HP339 (not modified) as a source for some measurements.
I have some confidence in the accuracy of the measurement results. So it would be nice to know how the QA400 (and E-MU 204) measurements stack up against the AP and other analyzers at those levels.
One thing I have to get my head around is how to relate the final (say QA400) FFT numbers to the fundamental. In my case I use the Boonton 1130 as the pre-QA400 autoranging/condinioning but there is no direct indication of the attenuation/amplification that the Boonton applies at any given input signal.
I can hack the autoranger control signals and drive a display directly in dB but
I'm not sure i want to spend that time.
Similar issue if you use a pre-notch - how much does the notch actually suppress? If you knew that, you could add that number to the FFT.
Jan
Yes that would work but only for a specific situation. As you do your measurement, and vary the input signal to the test unit, the Boonton (or any other similar) will continually switch the input attenuator to keep the input signal to the notch within the dynamic range. I would like to have a continuous indication of the momentary ranging attenuation/gain.
Of course the controller in the Boonton will be 'aware' of the ranging control signals and include the settings in the level- and THD displays.
In a two channel unit like the QA400 one could use the 2nd channel to look at the ranging output and add that number to the THD/FFT indication.
Jan
Of course the controller in the Boonton will be 'aware' of the ranging control signals and include the settings in the level- and THD displays.
In a two channel unit like the QA400 one could use the 2nd channel to look at the ranging output and add that number to the THD/FFT indication.
Jan
Thats what my cal trick was for- My Boonton 1121. I checked the manual and readback of the "gain" to the distortion output is not available. If we could find someone to reverse engineer the firmware and add some of this is would be great. Meanwhile using a BNC cable and injecting a known voltage with it does work well. The generator accuracy of the Boonton is as good or better than the QA400. The distortion of the generator is dominated by 2nd harmonic at about -105 or so. The analyzer has a similar issue. I know that can be improved but my early attempts were not successful.