The QA analyzer side seems to pick up beaucoup of IM products -- probably summing from the little SMPS.
This backs up what I said earlier.... isnt accurate and wouldnt use it below -100dB.
And, a notch filter is a must-have to be useful under -100dB.
A notch filter is required IMO.
A flexible fixed freq or variable freq notch would be a great addition..... this has been kicking around for awhile now. A balanced input with atten et al is a very nice added feature but you really Must have a notch filter on the thing if you want accurate harmonic readings/measurements when approaching -100dB.
It is a great value, that is for sure.
THx-RNMarsh
Last edited:
Hi all,
I’ve been monitoring this thread for awhile as I recently purchased a QA400.
I’m certainly not in the same league as most of you, as I mainly repair consumer electronics, including vintage audio equipment. So, if I show some ignorance here, please take that into consideration.
I occasionally repair some class D mobile amps and read somewhere that out-of-band noise produced by the signal source can be a big issue with the PWM in these amps.
Noting the 100 kHz noise so many have witnessed from the QA400 gen output, I take it that an audible band pass filter would be necessary between the QA400 gen output and the device under test? If so, any recommendations on a decent commercial filter?
Joe
I’ve been monitoring this thread for awhile as I recently purchased a QA400.
I’m certainly not in the same league as most of you, as I mainly repair consumer electronics, including vintage audio equipment. So, if I show some ignorance here, please take that into consideration.
I occasionally repair some class D mobile amps and read somewhere that out-of-band noise produced by the signal source can be a big issue with the PWM in these amps.
Noting the 100 kHz noise so many have witnessed from the QA400 gen output, I take it that an audible band pass filter would be necessary between the QA400 gen output and the device under test? If so, any recommendations on a decent commercial filter?
Joe
This document from TI has useful guidelines to use when measuring class D amplifiers: http://www.ti.com/lit/an/sloa068/sloa068.pdf
IIRC the AP site has similar but more reading is involved.
IIRC the AP site has similar but more reading is involved.
Here are 2 signals -- one from a KH4402, the other from the SG5010 -- you can see a strong IM product at f2-f1 and f2+f1:
An externally hosted image should be here but it was not working when we last tested it.
Yes, the same thing happens at the higher BW -- also, when the signals become more separated the IM product is reduced.
I have been experimenting with notch filters and was wondering if the QA400 would require an anti-aliasing filter so I have carried out some measurements. Yes it does alias signals above 1/2 the sampling frequency and unsurprisingly has a response similar to those shown in the Cirrus Logic CS4272 data sheet.
It's no big issue it's just something to be aware of.
My prototype notch filter tunes from 20 Hz to 20 kHz and gives greater than 40 dB attenuation of the fundamental without requiring any adjustments other than frequency and doesn't attenuate the 2nd harmonic and upwards. It works OK but is rather noisy being based on a state variable circuit so I am investigating alternatives.
It's no big issue it's just something to be aware of.
My prototype notch filter tunes from 20 Hz to 20 kHz and gives greater than 40 dB attenuation of the fundamental without requiring any adjustments other than frequency and doesn't attenuate the 2nd harmonic and upwards. It works OK but is rather noisy being based on a state variable circuit so I am investigating alternatives.
Hi PChi,
Tell us more about your notch filter, and how critical the tuning is. Also, how to keep the 2nd harmonic from being too much attenuated is always an interesting issue. On top of that, getting continuous tuning over the full audio band is an interesting challenge. I'm interested in learning more about what you have done.
The notches I have used in the past were twin-T based, and built at just two specific frequencies; 1kHz and 20kHz. The ones I have done use negative feedback (instead of positive feedback) to keep attenuation of close-in harmonics to less than 1dB or so. Of course, the negative feedback approach creates a tradeoff between notch depth and close-in harmonic attenuation.
The anti-alias issue is of interest. I wonder how much of the grass seen with the QA400 is due to aliasing of HF noise down to below 20kHz.
Cheers,
Bob
Can you elaborate on the twin-T feedback approach? I don't recall having seen this.
Thanks,
Samuel
Thanks for your interest. The notch filter is based upon the design by Ian Hickman described in the August 1999 issue of Electronics World. It is related to the Double-notch filter described by B. J. Sokol.
The frequency tuning is critical. I have used 9 switched ranges using a number of Dual In Line switches and crude 7 bit 'DACs' using Thumbwheel BCD coded switches. There is also a dual ganged potentiometer for fine frequency trim.
Using the FFT means that the notch doesn't have to be that deep so the fiddly gain and quadrature trims in the original article can be eliminated leaving only frequency adjustment. Theoreticaly the notch depth doesn't depend on component matching.
The measured attenuation of the second harmonic is within about 0.2 dB of unity gain along with all the others (1 kHz notch).
The schematic is just a crude sketch so in no fit state to attach. I reused a PCB and mounted switches using hot melt glue so the prototype is ugly. I intend to design a PCB (the main reason for the project) so need to draw a decent schematic. It's similar to one I posted before in the long running Low Distortion Oscillator thread.
The frequency tuning is critical. I have used 9 switched ranges using a number of Dual In Line switches and crude 7 bit 'DACs' using Thumbwheel BCD coded switches. There is also a dual ganged potentiometer for fine frequency trim.
Using the FFT means that the notch doesn't have to be that deep so the fiddly gain and quadrature trims in the original article can be eliminated leaving only frequency adjustment. Theoreticaly the notch depth doesn't depend on component matching.
The measured attenuation of the second harmonic is within about 0.2 dB of unity gain along with all the others (1 kHz notch).
The schematic is just a crude sketch so in no fit state to attach. I reused a PCB and mounted switches using hot melt glue so the prototype is ugly. I intend to design a PCB (the main reason for the project) so need to draw a decent schematic. It's similar to one I posted before in the long running Low Distortion Oscillator thread.
It is my understanding that putting negative feedback around notch filters to reduce attenuation at frequencies outside the notch is a very old technique. I think I first learned of it in the 70s. I believe it was used on many distortion analyzers (not necessarily ones using twin T).
Of course, the positive feedback approach for the twin-T just bootstraps the portion of the twin T that otherwise would be grounded. That reduces loss outside the notch in that approach.
The negative feedback approach keeps the twin T as an ordinary twin T. The output of the twin T is fed to an amplifier and the output of the amplifier is typically fed back to the input of the twin T, often being subtracted by an amplifier in front of the twin T. If the amplifier has a gain of 20dB, there will be 15-20dB of loop gain at frequencies outside the notch, flattening the frequency response in those regions. At the same time, in the notch, that 20dB of amplifier gain reduces the depth of the notch by 20dB. That's the tradeoff.
Quite awhile back I did SPICE comparisons of twin T filters with both positive and negative feedback to obtain the same amount of "error loss" at the second harmonic. I don't recall one being dramatically better than the other, but I seem to recall there was some stability issue with the positive feedback version that caused me to lean toward using the negative feedback version. But my recollection is fuzzy.
I first used the negative feedback version in the late 70's before I built my own THD analyzer. I could not afford much back then and made most of my own test equipment.
Cheers,
Bob
DSO-2820 x QA400
Hello to all!
I would like an orientation of more experienced friends.
I was determined to buy the QA400, but recently discovered the VT DSO-2820 (16 Bits).
VT 200MHz 8 16bit PC USB Oscilloscope Spectrum Analyzer AWG Signal Generator MSO | eBay
I found a very interesting product and talking to the supplier on the use of it as an audio Analyzer he replied:
"DSO-2820 are better than a sound card in the sense that they can measure DC and frequencies well above audio frequency range.
Moreover, they are pre-calibrated. They can be used to measure parameters like audio power and frequency response.
However, they are not as good as a professional sound card for measurement of THD, THD+N and SNR."
"The THD of VT DSO-2820 itself is about -60dB, therefore it can only be used to measure those THDs greater than -60dB."
In the opinion of you, can I invest in the DSO-2820 or will be better to invest in QA400?
Hello to all!
I would like an orientation of more experienced friends.
I was determined to buy the QA400, but recently discovered the VT DSO-2820 (16 Bits).
VT 200MHz 8 16bit PC USB Oscilloscope Spectrum Analyzer AWG Signal Generator MSO | eBay
I found a very interesting product and talking to the supplier on the use of it as an audio Analyzer he replied:
"DSO-2820 are better than a sound card in the sense that they can measure DC and frequencies well above audio frequency range.
Moreover, they are pre-calibrated. They can be used to measure parameters like audio power and frequency response.
However, they are not as good as a professional sound card for measurement of THD, THD+N and SNR."
"The THD of VT DSO-2820 itself is about -60dB, therefore it can only be used to measure those THDs greater than -60dB."
In the opinion of you, can I invest in the DSO-2820 or will be better to invest in QA400?
I think you answered your own question, since THDs greater than -60dB are generally fairly high level. The QA is more than 100dB, close to 110dB depending on certain details of implementation and measurement set up.
So, if ur measuring THD down into the 0.001% range you'd want either a top of the line sound card, the QA, or a high end stand alone instrument (of which several have been mentioned in this thread).
Hope that helps.
_-_-
So, if ur measuring THD down into the 0.001% range you'd want either a top of the line sound card, the QA, or a high end stand alone instrument (of which several have been mentioned in this thread).
Hope that helps.
_-_-