amplifierguru said:Hi bogicp,
"Do you have possibility to calibrate your hammer from time to time?"
Is it a cordless hammer (to minimise ground related problems) and even better solar powered for low impact.
Cheers,
Greg
Ah!
Sorry!
Can you forgive me for this stupid question?
Best regards,
-boggy
I found this on Bozo's website.
It is very interesting that there are things that are not shown on steady mode measurement
http://www.linkwitzlab.com/sys_test.htm
http://stereophile.com/interviews/503/index2.html
Is there anything similiar idea for audio power amps?
It is very interesting that there are things that are not shown on steady mode measurement
http://www.linkwitzlab.com/sys_test.htm
http://stereophile.com/interviews/503/index2.html
Is there anything similiar idea for audio power amps?
Lumanaw, you are mixing a short sequence with a pulse. A pulse would sound more or less like a pop. YMEC uses a MLS method, so do many other manufacturers. MLS does reveal nonlinearity, i.e. that it is there, but not the nature of it - by evidence of it changing the spectrum of the MLS. A simple way of explaining it would be to see it as a measurement of distortion specta - you know thet distortion is there, and what it does to the harmonic spread, but you can't tell from that data how it is created.
In theory, you could analyse an amp in this manner, but in practise we are talking rather high sampling rates, in the tens of MHz easily in order to reveal eg. stability margins. On the other hand, the low frequency end may also be of interest simultaneously, which means very long MLS is needed, and a lot of data is generated, which also requires a long time to calculate into a semi-useful result.
In theory, you could analyse an amp in this manner, but in practise we are talking rather high sampling rates, in the tens of MHz easily in order to reveal eg. stability margins. On the other hand, the low frequency end may also be of interest simultaneously, which means very long MLS is needed, and a lot of data is generated, which also requires a long time to calculate into a semi-useful result.
Hi, Ilimzn,
Sorry for the confusion. First of all, I just know that there existed such a smart measurement for room accoustics. I just feel that if such a thing existed for audio amp measurement, it can bring some benefit, I just don't know if such an "impulse response" for audio amplifier existed or not
If this kind of measurement existed for audio power amps, I hope it can give a clue of how an amp will sound. Why? Because ordinary measurement (THD, S/N, Damping, Slewrate, etc) cannot tell how an amp will sound.
If somekind of measurement can tell how an amp will sound, it will be very usefull. At least I can see in what way a 2 stages amp will differ from 5 stages amp. Or in what say a non-feedback power amp differs from feedback power amp. Those difference is not very clear in steady mode measurements. Maybe can make feedback transistor amp that sounds exactly like SET, based on how each differs clearly in somekind of measurement
Sorry for the confusion. First of all, I just know that there existed such a smart measurement for room accoustics. I just feel that if such a thing existed for audio amp measurement, it can bring some benefit, I just don't know if such an "impulse response" for audio amplifier existed or not
If this kind of measurement existed for audio power amps, I hope it can give a clue of how an amp will sound. Why? Because ordinary measurement (THD, S/N, Damping, Slewrate, etc) cannot tell how an amp will sound.
If somekind of measurement can tell how an amp will sound, it will be very usefull. At least I can see in what way a 2 stages amp will differ from 5 stages amp. Or in what say a non-feedback power amp differs from feedback power amp. Those difference is not very clear in steady mode measurements. Maybe can make feedback transistor amp that sounds exactly like SET, based on how each differs clearly in somekind of measurement
Hi lumanauw,
It is quite possible to use impulse on amplifiers (and tone controls etc), it's regularly done. But, as others have already pointed out, the information this produces is simply the frequency and phase respones of the DUT. If you believe that knowing the frequency/phase response can tell you how an amp will sound, then by all means use impulse/mls testing.
If you're hoping it will tell you something about how linear an amp is etc, what I know is it won't - this is proven by the fact that you can drive your amplifier stage into clipping, and the resulting impulse will still translate into pretty much the same output plot.
BTW, an impulse sounds more like a small little click, not a 'clap'. There's very little energy in it, thus the resulting frequency response plot usually has very terrible S/N ratio, particularly at low frequencies - a good reason to switch over to MLS for speaker testing.
Cheers!!
It is quite possible to use impulse on amplifiers (and tone controls etc), it's regularly done. But, as others have already pointed out, the information this produces is simply the frequency and phase respones of the DUT. If you believe that knowing the frequency/phase response can tell you how an amp will sound, then by all means use impulse/mls testing.
If you're hoping it will tell you something about how linear an amp is etc, what I know is it won't - this is proven by the fact that you can drive your amplifier stage into clipping, and the resulting impulse will still translate into pretty much the same output plot.
BTW, an impulse sounds more like a small little click, not a 'clap'. There's very little energy in it, thus the resulting frequency response plot usually has very terrible S/N ratio, particularly at low frequencies - a good reason to switch over to MLS for speaker testing.
Cheers!!
Mabuhay, Clem_o,
From your description, it doesn't look like what I have in mind. I am thinking about something powerfull, not single tone (but many distictive tones at once, maybe a part of pink noise) that can drive the amp into clipping/saturation in very short pulse.
This way it will always generate kinds of "high-order" artifacts that will arise. How one amp differs with others in what high-order artifacts they left when driven into clipping/saturation.
I always wanted to see how transistors and tubes differs in "clipping behavior" as many stated, but I have difficulties to see it.
I maybe wrong here but I think testing amps with square waves will show nothing audibly, since there is no square wave music in real life , especially if the square wave is a single tones.
It will be more meaningfull to test amps with powerfull sinusoidal group impulse (with some frequencies/magnitude at once) that will clip the amp, and see what artifacts they left.
From your description, it doesn't look like what I have in mind. I am thinking about something powerfull, not single tone (but many distictive tones at once, maybe a part of pink noise) that can drive the amp into clipping/saturation in very short pulse.
This way it will always generate kinds of "high-order" artifacts that will arise. How one amp differs with others in what high-order artifacts they left when driven into clipping/saturation.
I always wanted to see how transistors and tubes differs in "clipping behavior" as many stated, but I have difficulties to see it.
I maybe wrong here
but I think testing amps with square waves will show nothing audibly, since there is no square wave music in real life , especially if the square wave is a single tones. It will be more meaningfull to test amps with powerfull sinusoidal group impulse (with some frequencies/magnitude at once) that will clip the amp, and see what artifacts they left.
In this case, possibly Hann-wavelet is best thing that you can have for testing... but there isn't much more tools for analyzing response, only osciloscope and ears AFAIK.
You have link in thread with Hann wavelet test signal in our "video edition" with subtitles for frequency. Audio part of this "video" file, is uncompressed (not mp3, AC3, or whatever), spectral pure and numericaly generated wav file, noiseless, silence with real zeros...
You can test analog and digital amplifiers with this test signal...
Result analysis is another (problematic) story... there is possibilities for analyse distortion, stored energy (if exist), phase anomalies, etc.
But software for analysis don't exist AFAIK.
Hann wavelet is more closer to real music signal than any other test signal already known, and have all characteristic you listed IMHO.
link (again):
http://www.bozoel.com/products/bozotest/download.html
best regards
-boggy
You have link in thread with Hann wavelet test signal in our "video edition" with subtitles for frequency. Audio part of this "video" file, is uncompressed (not mp3, AC3, or whatever), spectral pure and numericaly generated wav file, noiseless, silence with real zeros...
You can test analog and digital amplifiers with this test signal...
Result analysis is another (problematic) story... there is possibilities for analyse distortion, stored energy (if exist), phase anomalies, etc.
But software for analysis don't exist AFAIK.
Hann wavelet is more closer to real music signal than any other test signal already known, and have all characteristic you listed IMHO.
link (again):
http://www.bozoel.com/products/bozotest/download.html
best regards
-boggy
Hi lumanauw - I see you've picked up some Filipino!
bogicp's suggestion is interesting. I know a little bit about wavelets, but have not investigated the properties of the Hann. It might be possible to write the software to do the analysis, perhaps initially in Matlab or similar... Will try asking some resident DSP gurus and see!
Cheers!!
Clem
bogicp's suggestion is interesting. I know a little bit about wavelets, but have not investigated the properties of the Hann. It might be possible to write the software to do the analysis, perhaps initially in Matlab or similar... Will try asking some resident DSP gurus and see!
Cheers!!
Clem
Hi, Clem_O,
If I feed one feedback transistor audio power amp with 5 sinusoidal at once, 50hz/0.5V, 300hz/1V, 1khz/3V, 8khz/1V, 20khz/0.5V in the form of short pulses (not steady sinusoidal).
Because the gain of the amp, the 3V signal/1khz, will be clipped (semi-square), while other 4 should be not in clipped mode.
The amp will reproduce all the 5 signals at once, because they are coming in the same time. 1 of the signal is clipped (1khz). Will the rest of the 4 signals still in perfect shape?
If they are still in perfect shape. If I can get the output of these 5 signals nulled towards input signal (I don't know how to nulled the already clipped 1khz ) what harmonic artifacts will appear, in the condition that 1 of them is clipped? Will there artifacts only around 1khz, or there will be artifacts also around the non-clipped 50hz, 300hz, 8khz, 20khz, due to the same transistor is working in clipping mode for 1khz?
If I feed one feedback transistor audio power amp with 5 sinusoidal at once, 50hz/0.5V, 300hz/1V, 1khz/3V, 8khz/1V, 20khz/0.5V in the form of short pulses (not steady sinusoidal).
Because the gain of the amp, the 3V signal/1khz, will be clipped (semi-square), while other 4 should be not in clipped mode.
The amp will reproduce all the 5 signals at once, because they are coming in the same time. 1 of the signal is clipped (1khz). Will the rest of the 4 signals still in perfect shape?
If they are still in perfect shape. If I can get the output of these 5 signals nulled towards input signal (I don't know how to nulled the already clipped 1khz
) what harmonic artifacts will appear, in the condition that 1 of them is clipped? Will there artifacts only around 1khz, or there will be artifacts also around the non-clipped 50hz, 300hz, 8khz, 20khz, due to the same transistor is working in clipping mode for 1khz?Hi david,
"The amp will reproduce all the 5 signals at once, because they are coming in the same time. 1 of the signal is clipped (1khz). Will the rest of the 4 signals still in perfect shape?"
No, spectrally speaking it can affect the higher-frequency sine waves if the clipped waveform's spectrum falls into the same frequency. Theoretically, a square-wave's spectrum will have only odd-order harmonics (so in this case, a 1KHz signal will ahve harmonics at 3, 5, 7, 9 KHz etc), however since this 'squaring' has been caused by clipping, signal can become assymetric producing a fairly different set of harmonics.
The other thing to consider is this: its possible two signals by themselves will not cause an amp to clip, but their combination will. Take a 50Hz signal that causes the amp to come very close to its maximum output swing (but still not clipped), and take another one, say at 1KHz, whose amplitude is small, perhaps causing the amp to get to only 1/2 its maximum swing. Put the two together, the amp will clip.
Cheers!
Clem
"The amp will reproduce all the 5 signals at once, because they are coming in the same time. 1 of the signal is clipped (1khz). Will the rest of the 4 signals still in perfect shape?"
No, spectrally speaking it can affect the higher-frequency sine waves if the clipped waveform's spectrum falls into the same frequency. Theoretically, a square-wave's spectrum will have only odd-order harmonics (so in this case, a 1KHz signal will ahve harmonics at 3, 5, 7, 9 KHz etc), however since this 'squaring' has been caused by clipping, signal can become assymetric producing a fairly different set of harmonics.
The other thing to consider is this: its possible two signals by themselves will not cause an amp to clip, but their combination will. Take a 50Hz signal that causes the amp to come very close to its maximum output swing (but still not clipped), and take another one, say at 1KHz, whose amplitude is small, perhaps causing the amp to get to only 1/2 its maximum swing. Put the two together, the amp will clip.
Cheers!
Clem
lumanauw said:
If they are still in perfect shape. If I can get the output of these 5 signals nulled towards input signal (I don't know how to nulled the already clipped 1khz
The artefacts (someone correct me please if my thinking is wrong) will be above 1KHz. The 50 and 300Hz signals should remain clean. The assmmetric nature of the clipping may produce artefacts at 8 and 20KHz, distorting these two signals...
Cheers!
Clem
Hi lumanauw, all
it is a common method to measure freq. response of amps.
When I use my audio card and an FFT software I both use
white noise and impulses to have the freq. resp.
with w.noise i have to average many FFT to have a clean result.
with impulses (dirac's delta fun) the graph is clean but you have to
decide the number of points of the time window (e.g. 4096)
and to burn a CD with an impulse every 4096 samples.
Then, in the measurement phase, you select the same window lenght, no windowing, and you have an extremely clean result.
This is an impulse measure, different from a steady state one and, maybe, some difference is possible since the system is non linear
(a bit).
Federico
IF (again IF) a single clipping signal will ruin all signals (even the small ones), then we have a problem.
Someone in the internet makes a "clipping indicator" with LED to know whether his amp (transistor feedback amp) clips or not. He found something interesting, that even in low level (not pushed hard), the transistor amp ALWAYS clips.
Someone in the internet makes a "clipping indicator" with LED to know whether his amp (transistor feedback amp) clips or not. He found something interesting, that even in low level (not pushed hard), the transistor amp ALWAYS clips.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.