I can't decide between Analog Discovery 2 and external soundcard + software for diy tube amplifier testing (THD+N, IMD, frequency response etc.). On one hand sound interface is 4x cheaper, on the other, AD2 seems more functional and more accurate test equipment.
Or should I save money for something more serious? Like HP 3561A or 3562A Dynamic Signal Analyzer.
As always I really appreciate your thoughts guys. Thanks!
Or should I save money for something more serious? Like HP 3561A or 3562A Dynamic Signal Analyzer.
As always I really appreciate your thoughts guys. Thanks!
Last edited:
What testing have you done before?
What are you hoping to test, and what equipment have you got already?
What do you want to do with any test results you can make?
What are you hoping to test, and what equipment have you got already?
What do you want to do with any test results you can make?
Ok, I have all the essential equipment for a basic tube work like, HW and LW variable power supplies, multimeter, oscilloscope, function generator.
At this point I would like to characterize my tube amps and to see what parameters could be improved further. Like THD+N (scope FFT is not good enough + noise floor is very high), IMD, amplitude response (gain vs frequency) I could sweep with func. gen. and look at the scope, but it would be much better to obtain measured points.
It would be interesting to measure mains hum with it's harmonics, because it is always hard to trace that.
At this point I would like to characterize my tube amps and to see what parameters could be improved further. Like THD+N (scope FFT is not good enough + noise floor is very high), IMD, amplitude response (gain vs frequency) I could sweep with func. gen. and look at the scope, but it would be much better to obtain measured points.
It would be interesting to measure mains hum with it's harmonics, because it is always hard to trace that.
Last edited:
I don't know AD2.
I do know that even a relatively cheap modern soundcard can do what you seem to be aiming for, as nearly all those functions are in the software, and the soundcard won't likely limit you until you get below about 2Hz or above about 90kHz, and you have a scope and function gen for those esoteric regions. Your test accuracy would be pretty much up to your multimeter. Most of the effort needed would be in setting up a load and signal connections, adapting a scope probe, and trying all the available software functions and confirming the results are valid.
Having a spectrum analyser like the HP's is just one link in the chain I think you want to make. It would need a controllable function gen, and you would need to install HPIB connections and a control app(s) for the gen and analyser - maybe there are excellent apps around, or maybe you can configure an app for what you want. I know that the soundcard based software is up-to-date made for all kinds of audio testing (and one of the best is free), so I reckon you would be hard pressed to meet what is already out there for soundcards.
I do know that even a relatively cheap modern soundcard can do what you seem to be aiming for, as nearly all those functions are in the software, and the soundcard won't likely limit you until you get below about 2Hz or above about 90kHz, and you have a scope and function gen for those esoteric regions. Your test accuracy would be pretty much up to your multimeter. Most of the effort needed would be in setting up a load and signal connections, adapting a scope probe, and trying all the available software functions and confirming the results are valid.
Having a spectrum analyser like the HP's is just one link in the chain I think you want to make. It would need a controllable function gen, and you would need to install HPIB connections and a control app(s) for the gen and analyser - maybe there are excellent apps around, or maybe you can configure an app for what you want. I know that the soundcard based software is up-to-date made for all kinds of audio testing (and one of the best is free), so I reckon you would be hard pressed to meet what is already out there for soundcards.
It makes a lot of sense. Thank you Mr. trobbins!
I've decided to go with focusrite sound interface.
Now is there a reason why do I want 2i2 vs solo? Apart from two channels and XLR instrument input connector I can't see any difference. Or is it a better idea to have two separate channels for measurements?
I've decided to go with focusrite sound interface.
Now is there a reason why do I want 2i2 vs solo? Apart from two channels and XLR instrument input connector I can't see any difference. Or is it a better idea to have two separate channels for measurements?
Given it is a test instrument, if one channel fails then you have a spare to use with the 2i2! My old EMU0404 USB has one channel with known low level noise spikes on its frequency spectrum plot, so I use that channel for 'everyday' testing, as I know what those spikes are and they don't interfere with the information in the rest of the spectrum plot, and if I really need an aesthetically clean plot then I use the better channel 🙂
You may want to give yourself as much 'heads up' before going down this testing path, as it is DIY and you will have to work at a few things. Luckily there are many threads on various forums where others have battled through their issues, and so imho it is worth your while getting acquainted with threads such as those related to the sound interface you are looking at.
For valve amplifier testing I would suggest using a 100x scope probe (ebay cheapy has been fine for me) as that allows connection to many amplifier locations with negligible risk of damaging your sound interface. The EMU0404 was straightforward as it presents a 1Meg unbalanced 1/4" socket input, which I then use a metal 1/4" to RCA, and a metal RCA to BNC, to interconnect to the probe. I had to remove an internal smt compensation cap in the probe BNC end to allow the probe compensation trimmer to give a flat response for. Even then you get a frequency spectrum roll off at the low end due to the soundcard (ie. below about 10-20Hz), and above about 50kHz with a probe.
Often that roll off is of no concern, although you can set up a calibration file to achieve a flat 2Hz to 90kHz if you want, although you need to be aware of what the calibration compensation is doing when interpreting results. As I understand it, the Focusrite's unbalanced inputs are not exactly 1 Meg, so the attenuation wont be exactly 100:1, but imho that is of no consequence, as calibrating an input or output from a soundcard interface requires an external meter if you want absolute measurement result values.
You may need to use a battery powered USB isolator, or use a laptop, in order to suppress mains earth loop hum gremlins. And you may need to use quality screened cable or coax cable, along with metal shielding around the interface to suppress nearby hum and noise sources, or to confirm that signals originate from the amp under test and not from your measurement rig or test bench environment.
It is best if the amp under test has its input signal ground, and speaker output ground, directly linked inside the amp. Measuring amps such as class D require special setups.
One good DIYforum link you may have found is: Howto - Distortion Measurements with REW
The latest software options can achieve powerful FFT setups, which may require a relatively modern PC or laptop. Most software is fine for Windows, but you need to check carefully if you run an Apple or Linux OS.
You may want to give yourself as much 'heads up' before going down this testing path, as it is DIY and you will have to work at a few things. Luckily there are many threads on various forums where others have battled through their issues, and so imho it is worth your while getting acquainted with threads such as those related to the sound interface you are looking at.
For valve amplifier testing I would suggest using a 100x scope probe (ebay cheapy has been fine for me) as that allows connection to many amplifier locations with negligible risk of damaging your sound interface. The EMU0404 was straightforward as it presents a 1Meg unbalanced 1/4" socket input, which I then use a metal 1/4" to RCA, and a metal RCA to BNC, to interconnect to the probe. I had to remove an internal smt compensation cap in the probe BNC end to allow the probe compensation trimmer to give a flat response for. Even then you get a frequency spectrum roll off at the low end due to the soundcard (ie. below about 10-20Hz), and above about 50kHz with a probe.
Often that roll off is of no concern, although you can set up a calibration file to achieve a flat 2Hz to 90kHz if you want, although you need to be aware of what the calibration compensation is doing when interpreting results. As I understand it, the Focusrite's unbalanced inputs are not exactly 1 Meg, so the attenuation wont be exactly 100:1, but imho that is of no consequence, as calibrating an input or output from a soundcard interface requires an external meter if you want absolute measurement result values.
You may need to use a battery powered USB isolator, or use a laptop, in order to suppress mains earth loop hum gremlins. And you may need to use quality screened cable or coax cable, along with metal shielding around the interface to suppress nearby hum and noise sources, or to confirm that signals originate from the amp under test and not from your measurement rig or test bench environment.
It is best if the amp under test has its input signal ground, and speaker output ground, directly linked inside the amp. Measuring amps such as class D require special setups.
One good DIYforum link you may have found is: Howto - Distortion Measurements with REW
The latest software options can achieve powerful FFT setups, which may require a relatively modern PC or laptop. Most software is fine for Windows, but you need to check carefully if you run an Apple or Linux OS.
Last edited: