Bob Cordell said:
Hi Pavel,
Looks interesting, but a little more explanation would help.
Cheers,
Bob
Hi Bob,
I have started to play with frequency modulation again. The files created in Cool Edit.
Another one, for carrier = 1OkHz, modulation frequency = 4kHz, modulation index = 2
Regards,
Pavel
Attachments
Hi,
Is there any interest in a test signal generator software for generating single and multitone sines, sweeps, windowed bursts etc, also usable in synced measurements (that is, on *exact* FFT/DFT bin frequency multiples)? At arbitrary sample rate, DFT block size and common bit resolutions (16, 24, 32, 32-float), plus some simple noise shaping.
And its complement piece of software, a time domain averager, for the synced stuff?
Many of that is also doable with existing .WAV software, but sometimes very awkward (e.g., no batch processing) and/or with lacking precision (especially with the synced options, can your software of choice produce an exact 19538.0859375Hz + 20540.0390625Hz IMD Signal? And, more important, do you want to calculate these exact frequencies every time by yourself?).
I have all of these already running as several (dirty) prototypes, I'd need to clean them up, combine them in one program and provide a nice end-user interface to it (they'd be DOS16 or, more likely, WIN32 command line executables). That would need some effort, but I want to do it anyway, sooner or later... once I have enough motivation.
- Klaus
Is there any interest in a test signal generator software for generating single and multitone sines, sweeps, windowed bursts etc, also usable in synced measurements (that is, on *exact* FFT/DFT bin frequency multiples)? At arbitrary sample rate, DFT block size and common bit resolutions (16, 24, 32, 32-float), plus some simple noise shaping.
And its complement piece of software, a time domain averager, for the synced stuff?
Many of that is also doable with existing .WAV software, but sometimes very awkward (e.g., no batch processing) and/or with lacking precision (especially with the synced options, can your software of choice produce an exact 19538.0859375Hz + 20540.0390625Hz IMD Signal? And, more important, do you want to calculate these exact frequencies every time by yourself?).
I have all of these already running as several (dirty) prototypes, I'd need to clean them up, combine them in one program and provide a nice end-user interface to it (they'd be DOS16 or, more likely, WIN32 command line executables). That would need some effort, but I want to do it anyway, sooner or later... once I have enough motivation.
- Klaus
KSTR said:Hi,
Is there any interest in a test signal generator software for generating single and multitone sines, sweeps, windowed bursts etc, also usable in synced measurements (that is, on *exact* FFT/DFT bin frequency multiples)? At arbitrary sample rate, DFT block size and common bit resolutions (16, 24, 32, 32-float), plus some simple noise shaping.
And its complement piece of software, a time domain averager, for the synced stuff?
Many of that is also doable with existing .WAV software, but sometimes very awkward (e.g., no batch processing) and/or with lacking precision (especially with the synced options, can your software of choice produce an exact 19538.0859375Hz + 20540.0390625Hz IMD Signal? And, more important, do you want to calculate these exact frequencies every time by yourself?).
I have all of these already running as several (dirty) prototypes, I'd need to clean them up, combine them in one program and provide a nice end-user interface to it (they'd be DOS16 or, more likely, WIN32 command line executables). That would need some effort, but I want to do it anyway, sooner or later... once I have enough motivation.
- Klaus
This could be very nice to have, especially if it would work well with some of the better, but modestly-priced, soundcards, such as the Juli@. I often like to do tone burst tests on power amplifiers, for example, sometimes with as little as a one-cycle burst with a 1% duty cycle.
Not being particularly gifted with computers or having good luck with them, for me the user interface and turn-key nature of a piece of software means a great deal.
Cheers,
Bob
Hi Bob,
Actually generating bursts was what started it all... I wanted to have sets of LF sine-bursts (with 2-cycle raised-cosine envelopes) in 1/24th oct increments, with decreasing distance between bursts and different phase relations vs. time, also one channel inverted at times. Alltogether 97 rather complex burst sequences, plus a specifically created sweep, all to be burnt on CD for a close investigation of room modes or speaker alignments... totally impractical to shape that all manually with a conventional .WAV-editor. These generators are already available, unfortunately all the docs are in german as of now...
I think bursts are very important test signals, for example if one measures an amp with lower level signal in the PSU recovery phase right after a longer power burst one might get a big surprise... also I think bursts with alternating phase in the burst center (again with short raised-cosine slopes) are the only way to find out how much current a speaker will really draw, givin the back-EMF and energy storage issues.
Hhm, is the ESI Juli@ really that good as a measurement card?... there are many rumours on the WEB about it, so I tend to believe it...
- Klaus
Actually generating bursts was what started it all... I wanted to have sets of LF sine-bursts (with 2-cycle raised-cosine envelopes) in 1/24th oct increments, with decreasing distance between bursts and different phase relations vs. time, also one channel inverted at times. Alltogether 97 rather complex burst sequences, plus a specifically created sweep, all to be burnt on CD for a close investigation of room modes or speaker alignments... totally impractical to shape that all manually with a conventional .WAV-editor. These generators are already available, unfortunately all the docs are in german as of now...
I think bursts are very important test signals, for example if one measures an amp with lower level signal in the PSU recovery phase right after a longer power burst one might get a big surprise... also I think bursts with alternating phase in the burst center (again with short raised-cosine slopes) are the only way to find out how much current a speaker will really draw, givin the back-EMF and energy storage issues.
Hhm, is the ESI Juli@ really that good as a measurement card?... there are many rumours on the WEB about it, so I tend to believe it...
- Klaus
KSTR said:
Klaus,
I agree completely. I only wish I had the time and skill to do that kind of computer programming.
I've been using my Juli@ card with Audio Tester the last couple weeks and I am impressed with its performance thus far. I still have a lot to learn about Audio Tester's interface and capabilities.
The Juli@ card seems to deliver a lot of bang for the buck. I'm also interested in hearing about others' experience with the Juli@ and other audio software.
Cheers,
Bob
Jack,
Nope, it's 2.47989135830729E-06
you might have missed my point completly... I know those "precision" freq. numbers look ridiculous but they just do so because they are normalized to 1Hz, which is the normal way for us to specifiy frequencies, also the way we input them in a wave software.
But when written in ... base fft_bin_frequency (fft_bin_frequency=48000 /16384 in this example -- can't even be displayed correctly as a fraction in the decimal system) these are perfect integers (6673 and 7013, resp. and all IM products are at multiples of 171, "500.9765625Hz"), and those integer multiples are needed for any synced measurement -- otherwise it's just not synced and e.g. time domain averaging won't work. FFT's don't know about sample rates and frequencies, their base unit is "bins".
EDIT: Ooops, the numbers are for 19549.8046875Hz and 20545.8984375Hz, which where choosen (after my initial post) so that the integers become prime numbers.
- Klaus
Nope, it's 2.47989135830729E-06
you might have missed my point completly... I know those "precision" freq. numbers look ridiculous but they just do so because they are normalized to 1Hz, which is the normal way for us to specifiy frequencies, also the way we input them in a wave software.
But when written in ... base fft_bin_frequency (fft_bin_frequency=48000 /16384 in this example -- can't even be displayed correctly as a fraction in the decimal system) these are perfect integers (6673 and 7013, resp. and all IM products are at multiples of 171, "500.9765625Hz"), and those integer multiples are needed for any synced measurement -- otherwise it's just not synced and e.g. time domain averaging won't work. FFT's don't know about sample rates and frequencies, their base unit is "bins".
EDIT: Ooops, the numbers are for 19549.8046875Hz and 20545.8984375Hz, which where choosen (after my initial post) so that the integers become prime numbers.
- Klaus
KSTR said:Nope, it's 2.47989135830729E-06
you might have missed my point completly...
Not really, I had to learn all about head and toe problems with photosensitive materials back when I seriously got around to doing this. The eye is incredibly sensitive to very small changes in density at the extrema and the available ADC's were incapable of resolving the differences (this is probably 20 years ago) ...and sometimes you have to play music between the cracks...
so I appreciate your point.
An interesting dissertation
I've found this dissertation interesting, in particular Ch. 2 an 3, on various opamp compensation techniques. Ch. 6 on open loop gain testing strategies is interesting as well.
http://contentdm.lib.byu.edu/ETD/image/etd1701.pdf
I've found this dissertation interesting, in particular Ch. 2 an 3, on various opamp compensation techniques. Ch. 6 on open loop gain testing strategies is interesting as well.
http://contentdm.lib.byu.edu/ETD/image/etd1701.pdf
janneman said:
Doctor of Philosophy, abbreviated Ph.D. (American English) or PhD (British English) (Wikipedia), equivalent to Dutch 'doctoraat'.
The better part of getting a PhD is spent writing papers
Other part to fulfill the requirements is in most cases defending the work before a panel of experts.
Or is the content (quality) of this paper in relation to getting a PhD the thing that strikes you?
Remco