I did a google search and I'm not finding much, other than expensive instruments like the Keysight U8903.
I half expected a google search to yield some sort of USB based box with PC software that would sweep and analyze audio filters etc.
Surely there must be a better way than dialing up various values on the signal generator and measuring the response point by point?
I recently picked up an Anritsu network analyser from Ebay. It could be a brand to look for.
Maybe consider scoring an old HP 3563A off ebay. They tend to go for $400 to $600 and cover DC to around 100kHz; that's more than enough span for characterizing audio circuits and note that the 3563A measures both phase and frequency response (ie - it is a proper vector network analyzer, just limited in frequency range).
If you want something more modern - albeit a bit more expensive - then perhaps consider the frequency response analyzer offerings from Cleverscope or Newtons4th.
N2PK
+1 - Whilst I don't have one, I think this is a good option.
It drives the LO port of the mixer with a sine wave which is not ideal as switching mixers would have more temperature-stable gain if driven from a square wave, but it would be good enough for most things. For <1MHz it would be quite feasible to build an active s-parameter test set using differential amplifiers instead of transformers, to work down to lower frequencies.
Another option would be to use a spectrum analyzer and input white noise from a PC or something to get the audio frequency response. Alternatively, get an audio spectrum analyzer with a tracking generator.
In my own lab I have an HP 8556A/8552B spectrum analyzer plugin stack for an old 141T mainframe. The 8556A plugins are somewhat uncommon, but have a tracking generator built in. That said, a massive 141T from the 1970's is going to be quite a bit bigger than what you have in mind, so I suggest going the white noise route. given that a PC sound card can be used as a spectrum analyzer, you may have everything you need already.
http://www.syscompdesign.com/Instruments_ep_42.htmlSurely there must be a better way than dialing up various values on the signal generator and measuring the response point by point?
This works pretty well if your DSO supports automatic measurements but an alternative is to use a sweep generator; with the oscilloscope triggered at the start of the sweep, gated/windowed measurements can be made visually.
White noise and a spectrum analyser is another way
White noise and a spectrum analyser is another way
If you go that route an oscilloscope with good FFT (say 64kpoints or more) and a function generator (swept or noise) are also suitable unless you want to measure >24dB differences. However a network analyser will also show phase shift which is something you might be interested in especially when dealing with audio.
White noise and a spectrum analyser is another way
If you go that route an oscilloscope with good FFT (say 64kpoints or more) and a function generator (swept or noise) are also suitable unless you want to measure >24dB differences. However a network analyser will also show phase shift which is something you might be interested in especially when dealing with audio.
If the stupid DSOs did not throw away the phase results when calculating FFTs, then they could also show phase shift.
Thanks for all the great leads. I may pick up the Digilent, looks like a fun toy/tool to play with.
Using the Tektronix-Excel add-in for my scope actually works reasonably well as long as I sweep the signal generator at a slow rate. This only takes seconds to configure and automates the tedious part. This method also has the advantage that I can run these tests into the xxx Mhz range if desired.
This is a LPF I just tested using the Excel data capture add-in (100 or so test points)
One obvious downside is the dynamic range amplitude wise, when sweeping a steep filter, I hit the noise floor pretty quickly
I started looking for some kind of low frequency SA again. Something that would be able to do from 0.01 to 100kHz minumim. I came across this piece of Chinese test gear
https://www.aliexpress.com/item/TDO3062BN-Digital-Storage-Oscilloscope-Low-frequency-Network-Analyzer/32855509772.htmlthe same unit sells on Ebay for ~$600 more than Aliexpress. Attached is a datasheet.
Anyone have experience with this type of unit? they also make two other versions, both with higher end frequency capability which I don't need, but experience with any of them would be helpful.
I've been trying to make a soundcard/software based SA work but so far the software just crashes when I try to get small spans and RBWs. Even though the spec says it should work. I'm using the software from Spyro Gumas (fatpigdog). Hopefully my emails to him will result in some enlightenment.
I started looking for some kind of low frequency SA again. Something that would be able to do from 0.01 to 100kHz minumim. I came across this piece of Chinese test gear
https://www.aliexpress.com/item/TDO3062BN-Digital-Storage-Oscilloscope-Low-frequency-Network-Analyzer/32855509772.html
A lot of vaguery in those specs. No mention of ADC bit depth, available input ranges, etc.
For basic specs, .08dB gain and 1º phase resolution is not very good.
For that kind of money, buying new I'd just step up into the Cleverscope CS320 VNA version.
In older used gear, the Anritsu MS's are excellent. HP 3577A's are becoming more affordable. If 100kHz is enough, HP 3562A is a great piece. But these are large items. Having worked with you before, we know that getting stuff to you can be "challenging"...
What software package are you using? I had very good results with ARTA. For drives, I always use the ASIO drives for the soundcard, not windows WDM. Arta has a list with soundcards that were tested with their software:
http://www.artalabs.hr/requirements.htmThe sound card approach is by far the cheapest option to get very good results. Disadvantages are that you have to take care of input levels and avoid ground loops. But you can achieve an THD+N below -105db with easy when using a decent sound card.
I had very good results with ARTA.
When I worked more in the audio frequency domain, I also got excellent results with ARTA. It is a very powerful software, and it does swept frequency, chirp, MLS, time-gated measurements and other analyses. The documentation is also very educational and worth a look on its own:
http://www.artalabs.hr/download/ARTA-user-manual.pdfJohn
I download ARTA and it work well except that it is limited to 0.1Hz at the low end. I need 0.01Hz. This was probably because for audio 0.1Hz is plenty low, but I'm using this to characterize a 10000:1 amplifier with a bandpass of 0.1Hz to 10Hz. So I need at least an order of magnitude below my low end to see the rolloff.
Well my swept VNA recommendations aren't going to work at/below 0.1Hz. You'll need an FFT-based approach with DC-coupled inputs. Choice of test signal might be interesting... what do you have in mind?