Home |
Search |
Today's Posts |
#1
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
One m-block is finished and I'm getting some worrisome SW results.
I've done quite a bit of searching but can't find a really satisfactory on-line, straight forward source for amp SW testing and result interpretation. Suggestions? Sometimes I nail Google search terms and sometimes not. Wink |
#2
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
On Mon, 11 Feb 2008 21:33:23 -0500, Wink wrote:
One m-block is finished and I'm getting some worrisome SW results. I've done quite a bit of searching but can't find a really satisfactory on-line, straight forward source for amp SW testing and result interpretation. Suggestions? Sometimes I nail Google search terms and sometimes not. Wink Generally speaking, the vertical edges of the square wave will reveal the HF response of the amp, while the flat tops will tell you about the LF response. The more tilted the verticals, the worse the HF, and the more droopy the flat tops, the worse the LF. Ringing will reveal the way in which the HF response rolls over from flat, and whether there is any amplifier instability. But really, this is all qualitative. Why not do a sweep which will let you do some measurements? Can you post a picture of the worrisome square wave? We might be able to give you a clue to what you are seeing. d -- Pearce Consulting http://www.pearce.uk.com |
#3
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
"Don Pearce" wrote in message
... On Mon, 11 Feb 2008 21:33:23 -0500, Wink wrote: One m-block is finished and I'm getting some worrisome SW results. I've done quite a bit of searching but can't find a really satisfactory on-line, straight forward source for amp SW testing and result interpretation. Suggestions? Sometimes I nail Google search terms and sometimes not. Wink Generally speaking, the vertical edges of the square wave will reveal the HF response of the amp, while the flat tops will tell you about the LF response. The more tilted the verticals, the worse the HF, and the more droopy the flat tops, the worse the LF. Ringing will reveal the way in which the HF response rolls over from flat, and whether there is any amplifier instability. But really, this is all qualitative. Why not do a sweep which will let you do some measurements? Can you post a picture of the worrisome square wave? We might be able to give you a clue to what you are seeing. d -- Pearce Consulting http://www.pearce.uk.com Second all the above. If you want a good all-in-one software test package, have a look at the Right Mark Audio Analyser. It's free, and works very well. S. -- http://audiopages.googlepages.com |
#4
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
|
#5
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
The pix are also at DIYaudio.com, though you have to be registered to
view photos, for some reason. http://www.diyaudio.com/forums/showt...05#post1426105 |
#6
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
On Tue, 12 Feb 2008 10:22:21 -0500, Wink wrote:
The pix are also at DIYaudio.com, though you have to be registered to view photos, for some reason. http://www.diyaudio.com/forums/showt...05#post1426105 Do you not have a bit of web space? Everyone does these days. My news server isn't binary and I don't want to register with anything. d -- Pearce Consulting http://www.pearce.uk.com |
#8
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
On Tue, 12 Feb 2008 15:47:09 GMT, (Don Pearce)
wrote: On Tue, 12 Feb 2008 10:22:21 -0500, Wink wrote: The pix are also at DIYaudio.com, though you have to be registered to view photos, for some reason. http://www.diyaudio.com/forums/showt...05#post1426105 Do you not have a bit of web space? Everyone does these days. My news server isn't binary and I don't want to register with anything. d No web space, I should have some, yes. Well, it sounds like you managed a look at the traces. How'd you manage the binary feed? This is an audio power amp, Musical concepts modded Hafler 220 in monoblock form. See please above thread "Hafler monoblock project....." if interested. Wink |
#9
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
"Don Pearce" wrote in message
... On Tue, 12 Feb 2008 10:18:46 -0500, Wink wrote: On Tue, 12 Feb 2008 07:57:26 GMT, (Don Pearce) wrote: On Mon, 11 Feb 2008 21:33:23 -0500, Wink wrote: One m-block is finished and I'm getting some worrisome SW results. I've done quite a bit of searching but can't find a really satisfactory on-line, straight forward source for amp SW testing and result interpretation. Suggestions? Sometimes I nail Google search terms and sometimes not. Wink Generally speaking, the vertical edges of the square wave will reveal the HF response of the amp, while the flat tops will tell you about the LF response. The more tilted the verticals, the worse the HF, and the more droopy the flat tops, the worse the LF. Ringing will reveal the way in which the HF response rolls over from flat, and whether there is any amplifier instability. But really, this is all qualitative. Why not do a sweep which will let you do some measurements? Can you post a picture of the worrisome square wave? We might be able to give you a clue to what you are seeing. d Don and Serge, I posted 2 traces at alt.binaries.pictures.misc. The 8 ohm load is a bunch of paralled 1/2 W, 10 ohm carbon resistors, the cap a good 2.2 ufd polyprop. I can post higher rez views later as my scope has manification capabilities. Looks like HF rolloff starts early. Not sure about overshoot ringing and damping when cap is added. Looks rather ominous. And thanks for your suggestions. Wink OK - just got a binary feed and had a look. Is this an audio amp? If so, I would say that is very good for a 10kHz square wave, and not too close to unstable with the cap on either. The usual frequency for square wave testing would be 400Hz, which will reveal performance at about 30Hz and 20kHz adequately at the same time. A 10kHz square wave reveals virtually nothing about the audio band performance. d -- Pearce Consulting http://www.pearce.uk.com Looks pretty good to me for a 10K squarewave into a substantial capacitative load. I don't think you'll have too much trouble into normal loads. In years gone by, before FFT testing became commonplace, I used squarewaves as a way of getting an overview of LF and HF performance, and HF instability, especially into capacitative loads. This was particularly relevant in the days of valve amplifiers, as the quality of the output transformer had a great influence on how much feedback could be safely applied, and how close to the edge of instability you were. With SS amps, and particularly now with even free software, you can do a test in seconds that pretty much characterises an amplifier's performance. Nevertheless, if you do the squarewave test again at a lower frequency as Don suggested, you can also get an assessment of the amplifier's LF performance. S. -- http://audiopages.googlepages.com |
#10
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
I've posted some more traces at alt.binaries.pictures.misc.
Results at 400, 100, and 20 hz. Getting a good shot of the scope is an art form. I'm using a Canon A70. A tripod may help. I need some web space! Wink |
#11
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
Don Pearce wrote
... But really, this is all qualitative. Why not do a sweep which will let you do some measurements? ... Depends on the purpose of the tests, I would have thought. A "square wave" test gives a view of the transient response, with time on the bottom axis, whereas a sweep and most of the analyses associated with it are in the frequency domain, with frequency on the bottom axis. Of course the two are related, such that each can be inferred from the other, completely for a linear system, but less so for a non-linear one. It is interesting and instructive to do them both and compare results. Also, a square wave can illustrate some faults in a more direct way. Unbalanced drive to a PP valve output stage springs to mind. Experienced techs can probably easily map between the two domains, but that's only because you know them both. Like Hegel said, enlightenment doesn't have an airport. Here's something: http://users.tpg.com.au/ldbutler/Waveforms.htm Also it may be worth checking RDH4, available for download from various sites. Ian |
#12
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
On Tue, 12 Feb 2008 17:47:27 -0500, Wink wrote:
I've posted some more traces at alt.binaries.pictures.misc. Results at 400, 100, and 20 hz. Getting a good shot of the scope is an art form. I'm using a Canon A70. A tripod may help. I need some web space! Wink Looking at the 20Hz square wave, I would estimate that you have a lower -3dB point somewhere around 3Hz. That will do very nicely and will allow you to drive a subwoofer directly should you feel the need. d -- Pearce Consulting http://www.pearce.uk.com |
#13
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
|
#14
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
http://users.tpg.com.au/ldbutler/Waveforms.htm Also it may be worth checking RDH4, available for download from various sites. Ian The Butler site is very much what I was looking for. Thank you -- Wink |
#15
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
"Wink" wrote in message
One m-block is finished and I'm getting some worrisome SW results. I've done quite a bit of searching but can't find a really satisfactory on-line, straight forward source for amp SW testing and result interpretation. Suggestions? Square wave testing is about majoring in minors. While it is a acute test for response at 10 Hz or 100 kHz, it has relatively low resolution in the all-important 20-20 kHz band, and pretty much misses the point in the even more important 100 Hz-10 kHz band. IOW, +/- 0.3 dB response from 100 Hz to 10 kHz is critical for sonically transparent performance, but a few dB either way at even 20 Hz and 20 kHz are minor inconveniences, at worst. Square wave testing with large capacitive loads also misses the point, because most speakers are inductive loads, particularly at the response extremes. |
#16
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
Arny Krueger wrote: "Wink" wrote in message One m-block is finished and I'm getting some worrisome SW results. I've done quite a bit of searching but can't find a really satisfactory on-line, straight forward source for amp SW testing and result interpretation. Suggestions? Square wave testing is about majoring in minors. While it is a acute test for response at 10 Hz or 100 kHz, it has relatively low resolution in the all-important 20-20 kHz band, and pretty much misses the point in the even more important 100 Hz-10 kHz band. IOW, +/- 0.3 dB response from 100 Hz to 10 kHz is critical for sonically transparent performance, but a few dB either way at even 20 Hz and 20 kHz are minor inconveniences, at worst. Square wave testing with large capacitive loads also misses the point, because most speakers are inductive loads, particularly at the response extremes. Wasn't the original intention of the 2 uF meant to be to simulate an electrostatic loudspeaker ? Graham |
#17
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
On Wed, 13 Feb 2008 12:03:15 -0500, "Arny Krueger"
wrote: "Wink" wrote in message One m-block is finished and I'm getting some worrisome SW results. I've done quite a bit of searching but can't find a really satisfactory on-line, straight forward source for amp SW testing and result interpretation. Suggestions? Square wave testing is about majoring in minors. While it is a acute test for response at 10 Hz or 100 kHz, it has relatively low resolution in the all-important 20-20 kHz band, and pretty much misses the point in the even more important 100 Hz-10 kHz band. IOW, +/- 0.3 dB response from 100 Hz to 10 kHz is critical for sonically transparent performance, but a few dB either way at even 20 Hz and 20 kHz are minor inconveniences, at worst. Square wave testing with large capacitive loads also misses the point, because most speakers are inductive loads, particularly at the response extremes. Very interesting Arny. Looking at SW performance of some superior tube amps at Stereophile suggests weak correlation between SW results and ultimate sonics. But this: "IOW, +/- 0.3 dB response from 100 Hz to 10 kHz is critical for sonically transparent performance" is interesting. I would think in room frequency response variations of even superb speakers would massively swamp your specified maximum deviation. By the way, though I'm familiar with you having read many posts in the audio groups, your name vaguely rings another bell. Did you write for TAS, Stereophile et al some time in the past/present? Wink |
#18
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
"Wink" wrote in message
On Wed, 13 Feb 2008 12:03:15 -0500, "Arny Krueger" wrote: "Wink" wrote in message One m-block is finished and I'm getting some worrisome SW results. I've done quite a bit of searching but can't find a really satisfactory on-line, straight forward source for amp SW testing and result interpretation. Suggestions? Square wave testing is about majoring in minors. While it is a acute test for response at 10 Hz or 100 kHz, it has relatively low resolution in the all-important 20-20 kHz band, and pretty much misses the point in the even more important 100 Hz-10 kHz band. IOW, +/- 0.3 dB response from 100 Hz to 10 kHz is critical for sonically transparent performance, but a few dB either way at even 20 Hz and 20 kHz are minor inconveniences, at worst. Square wave testing with large capacitive loads also misses the point, because most speakers are inductive loads, particularly at the response extremes. Very interesting Arny. Looking at SW performance of some superior tube amps at Stereophile suggests weak correlation between SW results and ultimate sonics. You can read far better than I. ;-) But this: "IOW, +/- 0.3 dB response from 100 Hz to 10 kHz is critical for sonically transparent performance" is interesting. I would think in room frequency response variations of even superb speakers would massively swamp your specified maximum deviation. Numerically, it does. However the ear can pick its way through the effects of the acoustics and perceive differences any larger than this as coloration. By the way, though I'm familiar with you having read many posts in the audio groups, your name vaguely rings another bell. Did you write for TAS, Stereophile et al some time in the past/present? http://stereophile.com/news/020705debate/ |
#19
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
"Arny Krueger" wrote in message news IOW, +/- 0.3 dB response from 100 Hz to 10 kHz is critical for sonically transparent performance... Pity there are almost no speakers that can manage that then! (I will admit however you don't want the amp possibly making things even worse) MrT. |
#20
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
"Mr.T" MrT@home wrote in message
"Arny Krueger" wrote in message news IOW, +/- 0.3 dB response from 100 Hz to 10 kHz is critical for sonically transparent performance... Pity there are almost no speakers that can manage that then! (I will admit however you don't want the amp possibly making things even worse) Yes, speakers are a pity, and yes, you don't want the amp making things worse. If it took any kind of serious effort to get +/- 0.3 dB response from 100 Hz to 10 kHz into a tough speaker load, then there might be some reason to compromise on that point. Even $80 stereo receivers generally do it. However. it is mission impossible for tubed amplifiers, particularly SETs. |
#21
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
On Feb 13, 2:24 pm, Eeyore
wrote: Square wave testing with large capacitive loads also misses the point, because most speakers are inductive loads, particularly at the response extremes. Wasn't the original intention of the 2 uF meant to be to simulate an electrostatic loudspeaker ? It might have been, in someone's simplified fantasy, way back when, when no one was paying much attention, maybe, in a dream. In reality, it's no better at simulating an electrostatic speaker than a pure 8-ohm resistor is. I have yet to find, in an awful lot of impedance measurements, an electrostatic loudspeaker system that shows any capacitive reactance at, say, 15-20 kHz, all of them show some amount of inductive reactance at the high end. I have also never seen an electrostatic speaker that exhibits enough capacitive reactance at any frequency to show a pahse angle exceeding about 55 degrees, and even that's pretty rare. |
#22
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
In article , Wink
wrote: I've posted some more traces at alt.binaries.pictures.misc. Results at 400, 100, and 20 hz. Getting a good shot of the scope is an art form. I'm using a Canon A70. A tripod may help. I need some web space! Wink It looks like all of these square waves are from an overdriven amp. In the 10KHz and 400Hz tests, the drive current appears to crap out right before becoming square. The 100Hz and 20Hz tests show what appears to be loss of output range from the boostrap or power supply capacitors draining. I don't know about the ringing. A square wave into a 2.2ufd cap is going to generate massive EM fields. That ring could EM coupling into the preamp or just a nearby chunk of metal interacting with your wires. Keeping associated wires as a balanced twisted set can help cut down on EM interaction. Twist the V-/GND/V+ power trio together, twist the speaker + and speaker - together, twist the Base and Emitter to the driver circuit, etc. -- I don't read Google's spam. Reply with another service. |
#23
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
On Fri, 15 Feb 2008 01:45:02 -0800, Kevin McMurtrie
wrote: In article , Wink wrote: I've posted some more traces at alt.binaries.pictures.misc. Results at 400, 100, and 20 hz. Getting a good shot of the scope is an art form. I'm using a Canon A70. A tripod may help. I need some web space! Wink It looks like all of these square waves are from an overdriven amp. In the 10KHz and 400Hz tests, the drive current appears to crap out right before becoming square. The 100Hz and 20Hz tests show what appears to be loss of output range from the boostrap or power supply capacitors draining. Are you looking at another set of traces? I can see absolutely no sign of any of that stuff. What I see is a perfectly normal bandwidth-limited square wave. There is no sign of slew rate limiting, which is what would happen if it runs out of drive current. I don't know about the ringing. A square wave into a 2.2ufd cap is going to generate massive EM fields. That ring could EM coupling into the preamp or just a nearby chunk of metal interacting with your wires. Keeping associated wires as a balanced twisted set can help cut down on EM interaction. Twist the V-/GND/V+ power trio together, twist the speaker + and speaker - together, twist the Base and Emitter to the driver circuit, etc. Not the slightest clue what you are talking about here. It is just ringing - that is what happens. And ringing is no way some sort of "special" signal that will generate massive EM fields. d -- Pearce Consulting http://www.pearce.uk.com |
#24
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
|
#25
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
On Fri, 15 Feb 2008 07:58:19 -0800, Kevin McMurtrie
wrote: In article , (Don Pearce) wrote: On Fri, 15 Feb 2008 01:45:02 -0800, Kevin McMurtrie wrote: In article , Wink wrote: I've posted some more traces at alt.binaries.pictures.misc. Results at 400, 100, and 20 hz. Getting a good shot of the scope is an art form. I'm using a Canon A70. A tripod may help. I need some web space! Wink It looks like all of these square waves are from an overdriven amp. In the 10KHz and 400Hz tests, the drive current appears to crap out right before becoming square. The 100Hz and 20Hz tests show what appears to be loss of output range from the boostrap or power supply capacitors draining. Are you looking at another set of traces? I can see absolutely no sign of any of that stuff. What I see is a perfectly normal bandwidth-limited square wave. There is no sign of slew rate limiting, which is what would happen if it runs out of drive current. It looks like the vertical slew rate slows down about 1/5 to 1/10 of a grid line before going horizontal. It's the soft curve in the 10KHz (no cap) test and illuminated vertical in the 400Hz test. The 20Hz tests look like they have a straight line going towards zero rather than the curve I'd expect with a normal frequency rolloff. That could be camera shake, though. That soft curve is nothing to do with limiting - it is simply what a square wave looks like when an amplifier has a finite frequency response. As for the 20Hz curves, no they don't - they have the characteristic curve - you just can't see enough of it for it to be obvious. If he had down the test at 5Hz, that would have been obvious. I don't know about the ringing. A square wave into a 2.2ufd cap is going to generate massive EM fields. That ring could EM coupling into the preamp or just a nearby chunk of metal interacting with your wires. Keeping associated wires as a balanced twisted set can help cut down on EM interaction. Twist the V-/GND/V+ power trio together, twist the speaker + and speaker - together, twist the Base and Emitter to the driver circuit, etc. Not the slightest clue what you are talking about here. It is just ringing - that is what happens. And ringing is no way some sort of "special" signal that will generate massive EM fields. d Driving a capacitor with a square wave makes plenty of EM fields. No it doesn't - it just makes current. For that to generate an EM field you need a means of coupling that to the outside world - an antenna is the usual way. d -- Pearce Consulting http://www.pearce.uk.com |
#26
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
On Feb 15, 4:45 am, Kevin McMurtrie wrote:
It looks like all of these square waves are from an overdriven amp. In the 10KHz and 400Hz tests, the drive current appears to crap out right before becoming square. Uh... No. What we see is precisely what we would expect for square wave of limited bandwidth: the "crap out" is simply the exponential curve of the waveform. The 100Hz and 20Hz tests show what appears to be loss of output range from the boostrap or power supply capacitors draining. Nope, simple band limiting again. I don't know about the ringing. A square wave into a 2.2ufd cap is going to generate massive EM fields. That ring could EM coupling into the preamp or just a nearby chunk of metal interacting with your wires. And where on earth did you come up with this? Precisely why do you think such would happen? (hint: it doesn't). |
#27
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
"Serge Auckland" wrote in message ... Second all the above. If you want a good all-in-one software test package, have a look at the Right Mark Audio Analyser. It's free, and works very well. If one were to want to test an amplifier, would not the amp's output signal require attenuation prior to connection to the soundcard's input? I'm in the midst of trying to figure out some frequency response problems in a tube amp (EL84 PP, ~15wpc) which I've recently refurbished/rebuilt. Can I just keep the amp's volume way down low? I think I should have no more than 2V input to the sound card or risk frying it. Any idea what size resistor (10R, 1k, 100k, etc.) I might need to put in series with the output? thx Dave |
#28
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
On Wed, 27 Feb 2008 23:46:25 GMT, "Dave" wrote:
If you want a good all-in-one software test package, have a look at the Right Mark Audio Analyser. It's free, and works very well. If one were to want to test an amplifier, would not the amp's output signal require attenuation prior to connection to the soundcard's input? I'm in the midst of trying to figure out some frequency response problems in a tube amp (EL84 PP, ~15wpc) which I've recently refurbished/rebuilt. Can I just keep the amp's volume way down low? I think I should have no more than 2V input to the sound card or risk frying it. Any idea what size resistor (10R, 1k, 100k, etc.) I might need to put in series with the output? You'll need several pieces to make an amplifier measurement. First is a load for the amplifier that can safely dissipate the power. For an EL84-sized amplifier, resistors of 50 watt rating should be fine; rig a way to keep them up in free air. Second is a way to measure output voltage. Third is an attenuator for the drive signal to the amplifier. This *may* be available in usable-enough form in software; depends on the D/A (soundcard or other) software. Fourth is an attenuator for the output signal - to - soundcard. Seems like a lot, but it's still just the easy to build bits of distortion measurement. Also keep in mind the A/D's bandwidth limitations. All good fortune, Chris Hornbeck "It's 90% boilerplate, 1% real work, 9% WTF?" -Les Cargill |
#29
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
"Chris Hornbeck" wrote in message
... On Wed, 27 Feb 2008 23:46:25 GMT, "Dave" wrote: If you want a good all-in-one software test package, have a look at the Right Mark Audio Analyser. It's free, and works very well. If one were to want to test an amplifier, would not the amp's output signal require attenuation prior to connection to the soundcard's input? I'm in the midst of trying to figure out some frequency response problems in a tube amp (EL84 PP, ~15wpc) which I've recently refurbished/rebuilt. Can I just keep the amp's volume way down low? I think I should have no more than 2V input to the sound card or risk frying it. Any idea what size resistor (10R, 1k, 100k, etc.) I might need to put in series with the output? You'll need several pieces to make an amplifier measurement. First is a load for the amplifier that can safely dissipate the power. For an EL84-sized amplifier, resistors of 50 watt rating should be fine; rig a way to keep them up in free air. Second is a way to measure output voltage. Third is an attenuator for the drive signal to the amplifier. This *may* be available in usable-enough form in software; depends on the D/A (soundcard or other) software. Fourth is an attenuator for the output signal - to - soundcard. Seems like a lot, but it's still just the easy to build bits of distortion measurement. Also keep in mind the A/D's bandwidth limitations. All good fortune, Chris Hornbeck "It's 90% boilerplate, 1% real work, 9% WTF?" -Les Cargill That's correct. You will need a dummy load and an attenuator. For flexibility, whether you're measuring power amplifiers or signal devices (CD Players, preamps etc) , a 20k pot will be fine. Regarding bandwidth of the A/D in the sound card, this used to be an issue, and still can for some semi- and true professional sound cards, but many even lowish-end cards now offer 96k or even 192k sampling, which should give you measurements with up to 100k bandwidth, which is quite sufficient, and is comparable with what conventional metering (like the Ferrograph RTS2 ) provides. Regarding my comment above about pro sound cards, my Digigram card only supports 44.1 and 48k, as do many of the USB cards in current use. However, with the exception of THD measurements at higher frequencies, a great many audio measurements can be done at 48k. S. -- http://audiopages.googlepages.com |
#30
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
"Chris Hornbeck" wrote in
message On Wed, 27 Feb 2008 23:46:25 GMT, "Dave" wrote: If you want a good all-in-one software test package, have a look at the Right Mark Audio Analyser. It's free, and works very well. If one were to want to test an amplifier, would not the amp's output signal require attenuation prior to connection to the soundcard's input? Yes. My favorite device for this purpose is a pair of 5K ohm 2 watt potentiometers, soldered up to appropriate cables and connectors for interfacing to the UUT and the input of the sound card. Yes, this means that you need to solder and maybe drill some holes in a little metal or plastic box to hold the parts. I'm in the midst of trying to figure out some frequency response problems in a tube amp (EL84 PP, ~15wpc) which I've recently refurbished/rebuilt. Can I just keep the amp's volume way down low? I think I should have no more than 2V input to the sound card or risk frying it. Any idea what size resistor (10R, 1k, 100k, etc.) I might need to put in series with the output? If you search google for "5K ohm 2 watt potentiometer" you will find any number of sources. About $11.00 each. You'll need several pieces to make an amplifier measurement. First is a load for the amplifier that can safely dissipate the power. For an EL84-sized amplifier, resistors of 50 watt rating should be fine; rig a way to keep them up in free air. The resistors will ideally be of a kind called "non inductive wirewound". You should be careful to get resistors that do not dramatically change their resistance when they heat up. The ones that Radio Shack sells are IME problematical. I am informed that the following are good: http://www.partsexpress.com/dayton-n...-resistors.cfm I have used the following and they are very good as well: http://www.mouser.com/Search/Product...bAFdVahA%3d%3d Second is a way to measure output voltage. The tricky issue here is the fact that a lot of digital voltmeters have lousy frequency response by design. Obviously, you want one that has good accuracy over the audio band. Check the spec sheets carefully, and/or buy some likely candidates with return priveleges. You can count on the output of a modern sound card in a computer being flatter from 100-10KHz than most DVMs. Third is an attenuator for the drive signal to the amplifier. This *may* be available in usable-enough form in software; depends on the D/A (soundcard or other) software. The 5 K resistors mentioned above are good for this, as well. What you are trying to avoid is the possibility that the output of your sound card picks up additional noise and/or distortion as you turn its output attenuator down. This is a pretty real possibility. Fourth is an attenuator for the output signal - to - soundcard. See above. Seems like a lot, but it's still just the easy to build bits of distortion measurement. Also keep in mind the A/D's bandwidth limitations. If you want to spend a little money, you can find audio interfaces that support sample rates up to 192 KHz, which gives you 96 KHz audio bandwidth. Check out eMu and M-Audio for good values. You can find Rightmark tests for most popular audio interfaces on line. They basically tell you the lowest distortion measurements you can make with any degree of accuracy. I say start out with the audio interface in your PC, and after you get some experience with it, figure out if you want to go further. |
#31
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
Maybe I didn't give enough information.
I am an experienced tech and have repaired all manner of devices. I have an old computer set up as a signal generator. I have and an equally aged 35MHz oscilloscope which can read output voltages. I have dummy load 8-ohm power resistors. All I want to do is be able to capture the amp's output with my soundcard for use by Right Mark Audio Analyzer. I only care about the audio spectrum, 20-20KHz, so do I really need 96KHz bandwidth to do what I want to do? I don't believe the 2W 5K attenuators are going to last long clamped onto the amp's 15W outputs, I guess I could keep the amp's volume very low, but is this giving me a meaningful representation of the output? I've done some poking around and some people think an op-amp based buffer is the way to go, which not only attenuates the amp's output signal, but provides a physical buffer between the amp and soundcard to guard against overvoltage getting to the card. I don't really want to build such a device, as I am awkward at best putting together circuit-boards, and so really like the idea of a big resistor instead. I'd just have to scope the output prior to connecting it to the soundcard and keep it down around 1V. Dave |
#32
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
"Dave" wrote in message news:ukXyj.58649$FO1.22188@edtnps82... I don't believe the 2W 5K attenuators are going to last long clamped onto the amp's 15W outputs, I guess I could keep the amp's volume very low, but is this giving me a meaningful representation of the output? Duh. sorry, the 5K resistors would be in parallel with the 8R dummy load resistors thus conducting a tiny percentage of the output signal. Didn't think that one through. |
#33
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
On Mar 3, 1:21 pm, "Dave" wrote:
Maybe I didn't give enough information. I am an experienced tech and have repaired all manner of devices. I have an old computer set up as a signal generator. And what signal are you generating with this computer? I have and an equally aged 35MHz oscilloscope which can read output voltages. I have dummy load 8-ohm power resistors. All I want to do is be able to capture the amp's output with my soundcard for use by Right Mark Audio Analyzer. I only care about the audio spectrum, 20-20KHz, so do I really need 96KHz bandwidth to do what I want to do? No, as you've stated it. I don't believe the 2W 5K attenuators are going to last long clamped onto the amp's 15W outputs, I might humbly suggest that if you were "an experienced tech," you'd see that your claim here is wrong. Assuming by "2W 5k" attenuator you might mean something like a 2w 5kohm potentiometer, it most certainly COULD handle the output of a 15 watt amplifier with PLENTY of margin. An amplifier producing 15 watts into an 8 ohm resistive load is generating about 7.75 volts AC RMS. assuming the 5kOhm pot is placed in parallel with the 8 ohm load, and assuming the load seen by the wiper on the pot is high compared with both the 8 ohm dummy load and the pot, the pot then presents a constant 5kOhm resistive load across the amplifier's output. 7.75 volts across 5kOhm results in 0.012 watts of dissipation. That's about 0.6% of the rated power of potentiometer. Let's look at it another way: how much power into 8 ohms would result in the power rating of the attenuator being exceeded? 2 watts across 5kOhm requires 100 volts RMS. An amplifier capable of putting out 100 volts RMS into an 8 ohm load would be producing 1250 watts in that same 8 ohm dummy load. I'd be far more concerned under pretty much ANY conditions about the longevity of your dummy load than of the 5 kOhm pot. More generally, the power dissipated across any resistance is proportional to the reciprocal of that resistance. A 5000 ohm resistor of ANY form will dissipate 8/5000 times the power of an 8 ohm resistor when connected across the same amplifier output. That's 0.16%. I guess I could keep the amp's volume very low, but is this giving me a meaningful representation of the output? No, because, as we've seen, there is absolutely no need to. I've done some poking around and some people think an op-amp based buffer is the way to go, which not only attenuates the amp's output signal, but provides a physical buffer between the amp and soundcard to guard against overvoltage getting to the card. Why do you think the op-amp in your buffer is going to be any better at handling the high voltage than the op-amp at the input to your sound card? I don't really want to build such a device, as I am awkward at best putting together circuit-boards, You're an "experienced tech," no? Assuming you build the 5kOhm potentiometer-based attenuator suggested in a box with a 2-conductor input jack and output jack, it's a task that requires 2 jacks, one pot, 4 pieces of hookup wire, drilling 3 holes in the box, and making 7 solder connections. and so really like the idea of a big resistor instead. I think you'd like the straightforward application of Ohm's law better. It gives, in this case, far cheaper and more accurate results. However, you have a MUCH more serious issue, having to do DIRECTLY with your use and interpretation of "square waves" that could be leading you far astray. If you're signal generator is, as you say an "old computer" and/or you are using something like a soundcard for capturing and looking at a square wave, then any square wave you either generate or capture is absolutely guaranteed to have overshoot and ringing. If it doesn't than something is deeply broken. A true square wave, one which has absolutely flat tops and perfectly vertical leading edges requires a whose bandwidth is infinite. Such a system, including the generation, transmission and capture, simply does not exist. In the case of computer based generation or capture, the bandwidths is quite well defined: a soundcard running at a 44.1 kHz sample rate is likely to have a bandwidth of 20 kHz, with NO input AT ALL (for all practical purposes) above that frequency. Now, take a THEORETICALLY PERFECT 1 kHz square wave and simply band-limit, and you WILL end up with a square wave with overshoot and ringing, because that's exactly what a perfectly band-limited square wave SHOULD look like. It SHOULD have about 18% of overshoot and should ring right about at 20 kHz or so. That's what a theoretically perfect 1 kHz square wave, brick-wall band- limited to 20 kHz is supposed to look like. |
#34
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
On Mar 3, 2:04 pm, wrote:
On Mar 3, 1:21 pm, "Dave" wrote: Assuming by "2W 5k" attenuator you might mean something like a 2w 5kohm potentiometer, it most certainly COULD handle the output of a 15 watt amplifier with PLENTY of margin. As can be seen by Dave's reply closely following mine, we're in agreement on this point. As to the other comments on the square wave testing, I may have confused his comments with another's in this thread. |
#35
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
Dave wrote:
Maybe I didn't give enough information. I am an experienced tech and have repaired all manner of devices. I have an old computer set up as a signal generator. I have and an equally aged 35MHz oscilloscope which can read output voltages. I have dummy load 8-ohm power resistors. All I want to do is be able to capture the amp's output with my soundcard for use by Right Mark Audio Analyzer. I only care about the audio spectrum, 20-20KHz, so do I really need 96KHz bandwidth to do what I want to do? I have two comments. First, the response of the amp and its stability will depend on the load. A pure resistive load such as you intend using is probably the easiest load it will be asked to drive. Secondly, if the amp has any negative feedback then you ought to be looking at its response well above 20KHz because that is where any anomalies will occur. If you are testing with square waves then such anomalies could manifest themselves as ringing at frequencies well above 20KHz so I would suggest 96KHz bandwidth is necessary. Cheers Ian |
#36
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
wrote in message ... On Mar 3, 1:21 pm, "Dave" wrote: And what signal are you generating with this computer? Square Wave and sine wave. Looking for distortion/clipping in a tube amplifier. I am new to tube amp technology, hence my questions. I might humbly suggest that if you were "an experienced tech," you'd see that your claim here is wrong. See my edit/comment to my own post, posted about 2 minutes after the first post. I've done some poking around and some people think an op-amp based buffer is the way to go, which not only attenuates the amp's output signal, but provides a physical buffer between the amp and soundcard to guard against overvoltage getting to the card. Why do you think the op-amp in your buffer is going to be any better at handling the high voltage than the op-amp at the input to your sound card? I don't. I do think I'd rather replace a $1 op-amp than a $80 sound card. I don't really want to build such a device, as I am awkward at best putting together circuit-boards, You're an "experienced tech," no? Assuming you build the 5kOhm potentiometer-based attenuator suggested in a box with a 2-conductor input jack and output jack, it's a task that requires 2 jacks, one pot, 4 pieces of hookup wire, drilling 3 holes in the box, and making 7 solder connections. Sorry if I appear rude, but what I SAID was that I didn't want to build the op-amp buffer, O Critical One, and that I'd RATHER use a pot, sort of like, NO, make that exactly like I explicitly stated in the line immediately following this one. and so really like the idea of a big resistor instead. I think you'd like the straightforward application of Ohm's law better. It gives, in this case, far cheaper and more accurate results. However, you have a MUCH more serious issue, having to do DIRECTLY with your use and interpretation of "square waves" that could be leading you far astray. That I do not doubt nor contest. If you're signal generator is, as you say an "old computer" and/or you are using something like a soundcard for capturing and looking at a square wave, then any square wave you either generate or capture is absolutely guaranteed to have overshoot and ringing. If it doesn't than something is deeply broken. I'm using a Tektronix T935 oscilloscope to capture the waveforms, not the soundcard. A true square wave, one which has absolutely flat tops and perfectly vertical leading edges requires a whose bandwidth is infinite. Such a system, including the generation, transmission and capture, simply does not exist. You're missing a key word, above... you say ".. leading edges requires a missing word whose bandwidth is infinite." In the case of computer based generation or capture, the bandwidths is quite well defined: a soundcard running at a 44.1 kHz sample rate is likely to have a bandwidth of 20 kHz, with NO input AT ALL (for all practical purposes) above that frequency. Now, take a THEORETICALLY PERFECT 1 kHz square wave and simply band-limit, and you WILL end up with a square wave with overshoot and ringing, because that's exactly what a perfectly band-limited square wave SHOULD look like. It SHOULD have about 18% of overshoot and should ring right about at 20 kHz or so. That's what a theoretically perfect 1 kHz square wave, brick-wall band- limited to 20 kHz is supposed to look like. I'm talking about a 400Hz square wave, captured on the oscilloscope, looking something like this: http://hmexc.dnsalias.com/400_Hz.jpg There are also some traces showing 80Hz, 4kHz, and 10kHz in that directory. The same square wave looks fine prior to entering the output stage, i.e. the output taken from the phase splitter looks, well, square. Thanks for your help, I could do without some sarcasm but hey, this is usenet after all and my comment about the 2W pots was shall we say, dumbass. Dave |
#37
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
"Dave" wrote in message news:ad%yj.58779$FO1.10408@edtnps82... I'm talking about a 400Hz square wave, captured on the oscilloscope, looking something like this: http://hmexc.dnsalias.com/400_Hz.jpg Make that http://hmexc.dnsalias.com:8080/400_Hz.jpg |
#38
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
On Mar 3, 5:46 pm, "Dave" wrote:
wrote in message However, you have a MUCH more serious issue, having to do DIRECTLY with your use and interpretation of "square waves" that could be leading you far astray. That I do not doubt nor contest. If you're signal generator is, as you say an "old computer" and/or you are using something like a soundcard for capturing and looking at a square wave, then any square wave you either generate or capture is absolutely guaranteed to have overshoot and ringing. If it doesn't than something is deeply broken. I'm using a Tektronix T935 oscilloscope to capture the waveforms, not the soundcard. Makes NO difference: if it's a soundcard producing the square waves, the effect will be precisely the same because the bandwidth limiting issue is the same. A true square wave, one which has absolutely flat tops and perfectly vertical leading edges requires a whose bandwidth is infinite. Such a system, including the generation, transmission and capture, simply does not exist. You're missing a key word, above... you say ".. leading edges requires a missing word whose bandwidth is infinite." "requires a complete system whose bandwidth is infinite." In the case of computer based generation or capture, the bandwidths is quite well defined: a soundcard running at a 44.1 kHz sample rate is likely to have a bandwidth of 20 kHz, with NO input AT ALL (for all practical purposes) above that frequency. Now, take a THEORETICALLY PERFECT 1 kHz square wave and simply band-limit, and you WILL end up with a square wave with overshoot and ringing, because that's exactly what a perfectly band-limited square wave SHOULD look like. It SHOULD have about 18% of overshoot and should ring right about at 20 kHz or so. That's what a theoretically perfect 1 kHz square wave, brick-wall band- limited to 20 kHz is supposed to look like. I'm talking about a 400Hz square wave, Makes NO difference what the frequency is, just used 1 kHz as an example. captured on the oscilloscope, looking something like this: http://hmexc.dnsalias.com/400_Hz.jpg Well, despite being out-of-focus, the envelope, ate any rate, looks like a classic example of a square wave showing brick-wall bandwidth limiting. There are also some traces showing 80Hz, 4kHz, and 10kHz in that directory. The same square wave looks fine prior to entering the output stage, i.e. the output taken from the phase splitter looks, well, square. I doubt VERY seriously that that is the case: nothing in an analog amplifier of the kind that you're describing could have a linear-phase in-band response with a brick wall low-pass filter that would lead to the symmetrical time response you see. If you're seeing "perfect" square waves at the phase splitter, you're either seeing something else entirely or you're not measuring right. Take a look at: www.cartchunk.org/audiotopics/400HzSq_Trunc.jpg I'd be willing to bet that it looks VERY MUCH like what you're seeing on your oscilloscope. That's what a 400Hz square wave would look like if it was band-limited to 20 kHz AND your scope was accurate enough to show it. Now, let's try this again: are you generating the square wave with your computer, or is the square wave coming of a CD or something like that? |
#39
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
wrote in message ... Well, despite being out-of-focus, the envelope, ate any rate, looks like a classic example of a square wave showing brick-wall bandwidth limiting. There are also some traces showing 80Hz, 4kHz, and 10kHz in that directory. The same square wave looks fine prior to entering the output stage, i.e. the output taken from the phase splitter looks, well, square. I doubt VERY seriously that that is the case: nothing in an analog amplifier of the kind that you're describing could have a linear-phase in-band response with a brick wall low-pass filter that would lead to the symmetrical time response you see. If you're seeing "perfect" square waves at the phase splitter, you're either seeing something else entirely or you're not measuring right. Take a look at: www.cartchunk.org/audiotopics/400HzSq_Trunc.jpg I'd be willing to bet that it looks VERY MUCH like what you're seeing on your oscilloscope. That's what a 400Hz square wave would look like if it was band-limited to 20 kHz AND your scope was accurate enough to show it. Now, let's try this again: are you generating the square wave with your computer, or is the square wave coming of a CD or something like that? I'm generating the square wave with my computer. Would writing a square wave to a DVD, then playing back using a CD/DVD player with a higher sample rate (i.e. 96KHz or 192KHz) than the sound card make any difference? I was erroneously under the impression that the soundcard, with it's 20khz of bandwidth, could accurately generate a perfect 400Hz square wave. I'm still not exactly clear on why this is not the case, but I'll take you at your word because you seem to know what you're talking about. I'm no EE. I'll also take another look at the traces from the phase splitter stage, right now the amp has the power supply somewhat disassembled... I suspect what you say is correct. I know I have traces of SINE waves looking perfect through the phase splitter but I couldn't find any pictures of square waves, so I may have mis-spoken. Thanks for your help. Dave |
#40
Posted to rec.audio.tech
|
|||
|
|||
Square wave testing and interpretation
On Mar 4, 10:35 am, "Dave" wrote:
wrote in message Well, despite being out-of-focus, the envelope, ate any rate, looks like a classic example of a square wave showing brick-wall bandwidth limiting. Take a look at: www.cartchunk.org/audiotopics/400HzSq_Trunc.jpg I'd be willing to bet that it looks VERY MUCH like what you're seeing on your oscilloscope. That's what a 400Hz square wave would look like if it was band-limited to 20 kHz AND your scope was accurate enough to show it. Now, let's try this again: are you generating the square wave with your computer, or is the square wave coming of a CD or something like that? I'm generating the square wave with my computer. Since your soundcard, like every soundcard in existance, has, in essence, a brick-wall low pass filter at a frequency slightly less than 1/2 the sample rate (or lower), then ANY attempt to generate a square wave will result in a waveform looking, in essence, like the one I posted. Would writing a square wave to a DVD, then playing back using a CD/DVD player with a higher sample rate (i.e. 96KHz or 192KHz) than the sound card make any difference? How are you "writing" the square wave? By what means is this square wave being generated? I was erroneously under the impression that the soundcard, with it's 20khz of bandwidth, could accurately generate a perfect 400Hz square wave. Yes, it CAN generate a perfect 400 Hz square wave that's band-limited to 20 kHz. That's what the waveform is that I posted, and seeing through the fuzziness of your photograph, that appears to be what you're getting. Once again, what you asked for is a "perfect" 400 Hz square wave bandwidth-limited to 20 kHz, and that looks like the pictures we both generated. Now, you want an absolutely PERFECT square wave with a rise time of 0 and no overshoot? Well, you CANNOT have it, not if you're band-limited to 20 kHz. In fact, you can't EVER have it if you have ANY bandlimiting at all. I'm still not exactly clear on why this is not the case, but I'll take you at your word because you seem to know what you're talking about. Here's a somewhat simple way of looking at it. A 400 Hz square wave consists of a fundamental and certain harmonics in certain proportions all added together with a certain time relationship. Assuming you buy the premise that the fundamental and each and every harmonic are sine waves of a certain frequency, amplitude and phase relationship, then you can represent a perfect 400 Hz square wave mathematically as: F(t) = sum sin(n*t/400)/n n = 1, 3, 5, 7, ... infinity That is, it is the sum of an infinite series of odd harmonics, starting at 1 and continuing to infinity, with each harmonic in phase and with an amplitude proportional to the reciprocal of it's harmonic number. You have the fundamental (n=1), 1/3 the amplitude of the third harmonic (n-3), 1/5 the amplitude of the 5th (n-5) and so on. A PERFECT square wave MUST have an INFINITE number of these harmonics. That means n must go all the way to infinity. infinity times 400 Hz is infinity. TO have a perfect square wave requires infinite bandwidth. Now, guess how I generated that picture of YOUR square wave? Simple: I just calculated that same equation above, but instead of letting n got to infinity, I limited it to 49. 49 time 400 Hz is 19600 Hz. I limited the bandwidth of the square wave to less than 20 kHz simply by "truncating" the series to a no higher than the 49th harmonic. The effect is EXACTLY the same as brick-wall filtering a "perfect" square wave to 20 kHz. Just like your soundcard does. Let me say this again, at the risk of overstating the point: a perfect 400 Hz square wave band-limited to 20 kHz will look exactly like the trace I generated and probably like the trace you photographed. You photographed a "perfect" 400 Hz square wave that was band-limited to 20 kHz. Okay? |
Reply |
|
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
FA: Heathkit G-2 Sine/Square Wave Generator 1948 LAST DAY! | Vacuum Tubes |