Attachments
Last edited:
I want a controlled, specific, notch depth, either 20 or 40 dB. Then using it in front of conventional instruments keeps them useful. Otherwise you have lost context in terms of the fundamental.
It all depends on what you are looking to see. For distortion and other audio band info the QA400 is a great deal. As a general purpose scope the virtens is good. Check out Picoscope before you commit however. I have found in the past the Picoscope software to be far better than any other.
I would loike to beter understand how Virtens gets 8 bits to become 16 bits. Doinbg the same trick with a QA400 should get us to 32 bits! Maybe with a one hour sample rate. . .
I remember the QA400 working to DC. Does anyone know?
I have no doubt you already know this. For those following along.
I suppose this would be useful if you just want to glance and know what you have.
The distortion can be calculated from any notch depth. For a -60db notch, the measure distortion divided by 1000. -80dB divide by 10000 and so on. For anything in between then the measured distortion divided by 10 raised to the absolute (|-dB| / 20).
I think notch depth would have to be controlled by an automatic process. But then why not have the process calculate the distortion as well.
Easy to do by mixing a portion of the input signal with the output of the notch.
Samuel
I see, thanks. The addition in front of the filter would need to be done passively, as adding an active stage in front of it somewhat mitigates the distortion advantage of a passive implementation (although the same consideration applies to the positive feedback approach, where the bootstrapping opamp also has a , although more subtle, distortion contribution).
Samuel
Demian, one solution is the concept of Bob's Distortion Magnifier.
If you have a deep notch from one of your analyzers, say up to -80 or -100dB, then add fundamental to the residual at say -40dB, so that gives you an absolute ratio context. The actual depth of the notch now no longer is relevant, as long as it is appreciably below whatever you add later.
Bob's DM uses a different metheod; he cancels the fundamental in a high CMRR diff amp, THEN adds the fundamental in again. Same difference.
In the end you have increased your analyser's resolution by 40dB. I think 😎
Edit: I see Samuel suggest the same....
Jan
Last edited:
They can use higher bit depth if they lower the sampling rate:
"If this option is selected, the effective bit resolution increases from 8 bits to up to 16 bits as the sampling frequency goes down. "
I have a TiePie HS3 which uses the same trick. If you lower sampling rate to say 1MHz you get 16 bit resolution, at 25MHz I get 8 bits resolution, with 10, 12 and 14 bits at intermediate speeds.
Jan
A US version of a Quantasylum Differential Probe
should be available in the near future. Battery powered; X/10, X/100 attenuation. Will be called the QA190. Some info on it is available on the Quantasylum.net website
Charles
They also appear to be coming out with some really neat /cost effective RF signal generators, for those of you so inclined.
should be available in the near future. Battery powered; X/10, X/100 attenuation. Will be called the QA190. Some info on it is available on the Quantasylum.net website
Charles
They also appear to be coming out with some really neat /cost effective RF signal generators, for those of you so inclined.
The idea of isolating the QA400 with the ADUM4160 is a non-starter -- the chip is not nearly fast enough -- the transmission rate is in the user manual.
You are exactly right.
I addressed this same problem in my distortion magnifier. In that design, I subtract appropriately scaled versions of the input and output of the amplifier DUT, with amplitude and phase adjust, to achieve a deep null of the fundamental, greater than 60dB. I then basically add back in the appropriate amount of the original signal to get a fundamental that is, in fact, 20 or 40dB down. This remaining fundamental can act as a pilot tone onto which a conventional THD analyzer can lock. The residual of the THD analyzer can then be fed to a spectrum analyzer to get really down there, especially since the use of the spectrum analyzer largely takes away the analyzer noise from the picture.
Of course, the DM can be used directly with a spectrum analyzer to provide the relative value of the fundamental, as you describe. The DM was written up in Vol 0 of Jan Didden's Linear Audio. I first used the DM approach over 30 years ago to help me measure the low distortion levels of my MOSFET power amplifier. I think I even touched on the technique in the JAES article on the amplifier.
I some situations the DM approach reduces the effects of THD in the oscillator source.
Cheers,
Bob
Yes, the use of an up-front op amp will reduce the theoretical advantage to the distortion level of the op amp, so at minimum use of a very good op amp better than the DUT is important. But, of course, the same caveats apply to the limitations op amp distortion place on even the very best THD analyzers.
Doing the mix passively can be accomplished with a simple resistive combiner up front, but then the input impedance of the twin-T circuit is low and somewhat frequency-dependent. This might be OK if the source was a power amplifier being measured.
Cheers,
Bob
Many thanks for the help folks!
My goal is to make measurements of audio in aplificadores of headphones as recommended by NwAvGuy.
NwAvGuy's Heaphone Amp Measurement Recommendations | InnerFidelity
So, I'm going to choose the QA400.
Thanks again for your help.
That's a difficult question to answer. It's really a matter of choice. The Emu can be used with any sound card software. The QA400 only works with the QA400 SW. The two are comparable in performance. The generator is a pulsed sine gen and is not useful on it's own with other sound cards and FFT SW or analyzers. QA did at my request include an option for a continuous sine output but it won't work with the QA analyzer SW. QA included it as a pop up panel. It steals the focus locking up the other controls. Not quite what I had in mind. It's well worth the $200.00 you'll pay for it. But you get what you paid for, a $200.00 instrument. I don't think the price can be beat. Most people who bought one are happy with it and it didn't take long after it's introduction for third party DIY addons to start showing up on the net. It does have it's issue but no more than any other sound card. There are some tricks that can be done to vastly improve it's performance like any other sound card. Adding a notch filter to it's input. But this only works with a external continuous output generator. The pulse duration of the QA gen is too short to tune a notch filter into. Distortion analyzers require a continuous signal as well. The QA software is designed to work with the internal gen and will work with any external gen. I can't remember if it was me or someone else here who asked for the extra gain function. It is very useful for getting the ADC off it's floor by adding gain at the input. The extra gain function rescales the software to match the added gain. This cleans up a lot of the grass but it increases the input sensitivity which mean you can't put in as large a signal. If you're using a notch filter the extra gain really helps. The people at QuantAsylum are great to deal with. Someone posted they blew theirs up and QA fixed it for free and shipped it back to them. You won't find that kind of CS anywhere. Well, Acer is not far behind.
I Hope this helps you make an informed choice.
Sorry didn't make my question clear, I was asking if the performance of QA400 better than E-MU 0404 USB?
I not very particular about software, but I see QA400 can run on ARTA.
Demian is there a way to add a a numerically pure sine on the digital side? Maybe software generated. This would allow the SW to see the full signal without adding to the test signal. The notch depth would be irrelevant. This would trick the SW into reporting true value.
QA could add this into there SW. The SW would measure the fundamental level and normalize it to FS without adding or changing harmonic content.
Last edited: