Jump to content


Please note: You can easily log in to MPN using your Facebook account!

Ever Try To Pass A Square Wave?


Recommended Posts

No, I don't mean as in #2! A square wave is the most difficult sound signal for any recording system to faithfully reproduce, especially high frequencies (say above 10kHz). I can record/reproduce a 12kHz square wave on my ATR-102 @ 15 IPS and it looks pretty good. There is some ringing just after the leading edge of the wave, but it looks very good... it still closely resembles a square wave. I haven't tried it through my DAW, but it should be interesting to see what it looks like on the scope. Anyone else tried this?

GY

 

Link to comment
Share on other sites



  • Replies 20
  • Created
  • Last Reply
Yeah, square waves can be useful as a quick, rough test of a system's frquency response. If the square wave passes through with no visible degradation, it tells you that the system is 'flat' from the square wave freq up to about 10x that freq. So good test freqs for audio are 50 to 2KHz. At 50, for example, it tests the response up to about 500 Hz. Types of degradation & what they mean: - slanted edges: called 'slew rate' limiting. It means the amp (or whatever) is not fast enough to respond to high freq. transients. - top and bottom is sloped: Freq. response problem: A) If the slope rises as it goes to the right, the high end freqs are attenuated, or lows are boosted. B) If the slope drops as it goes to the right, the low end is weak, or the highs are boosted. - top and bottom is scooped or has a bump in the center: Freq. response is weak (for scooped) or too strong (for a bump) in the mids. Keep in mind that low, mids, and highs, as I am using the terms, are relative to the test freq. low = 1 to 2x the freq mids = 3x to 7x the freq highs - 7x to 10x the freq
Link to comment
Share on other sites

A couple more things: - DON'T feed a square wave to any speaker you're fond of. - It's easy to demonstrate what I was talking about by generating a square wave in a DAW, then playing around with it by passing it through an EQ plug-in and watching the results. Flangers and choruses also do interesting-looking things to the wave.
Link to comment
Share on other sites

We use signal injection of square waves to test reverb units and some stompboxes while troubleshooting, but I can't see it as a useful tool as a test tone. Frequency response can be accurately determined using a spectrum analyzer if you have one. We mostly use sinewaves for audio T/S.
Down like a dollar comin up against a yen, doin pretty good for the shape I'm in
Link to comment
Share on other sites

I tried a square wave through my DAW. It starts to fall apart around 3 kHz. Yeah, there's no noise, no head alignment and other positives to recording to digital, but I believe this is why strings, cymbals and other signals above 5 kHz sound like shit (16 bit/44.1). That was looking at the waveform in the application. I'm sure that if I looked at the signal after the D>A at the analog output, it would look even worse. I grew up with analog. I grew up when manufacturers performed and published laboratory tests on the products they sold. Frequency response, wow and flutter, signal to noise, distortion, IMD. Then came the notion that we were becoming to "spec" oriented. We were told, "who cares what it looks like on a scope or analyzer, what does it sound like"? With digital, some of those tests became moot. I think we've become, or at least I have, to accepting of this digital stuff. I'll bet even these digital mixers can't do any better at representing a square wave at the output. My analog can! Is anyone out there using a 24 bit/192 system that can put a square wave through it and report what the signal looks like? Preferably above 5 kHz? I have a new mind set today. I think I need to seriously look at a 24 bit/192 kHz system or go back to analog. GY

GY

 

Link to comment
Share on other sites

Recently there were some tests with DSD/SACD in a Dutch ProAudio magazine. The square wave from DSD looked pretty good, but from the analog machine it looked better. Needless to say the 16/44.1 didn't look like a square wave :D
The alchemy of the masters moving molecules of air, we capture by moving particles of iron, so that the poetry of the ancients will echo into the future.
Link to comment
Share on other sites

Freq response, as an absolute, ain't all that, IMO. I started techno-geeking my brains out in about '79, hotrodding any and everything I could find. Over the next decade or so, I learned a great deal about the relationship between what looks good on a scope, and what sounds good when you're making a record. There [b]is no[/b] relationship between the two, basically. One of the biggest offenders in the analog domain, as far as passing a good looking square wave, is a transformer. Ringing, undershoot, overshoot, etc. In the late 70's and early 80's, it was "transformerless" madness. Yank 'em out whenever and wherever possible. One prime difference between the MCI 2" JH-16 and JH-24 series was that the JH-24 was transformerless. New and improved. I tested a JH-24 and was blown away at the freq response. Flat out to about 40K. The JH-16 started fading at 20K or so. Sqaurewaves looked way better on the JH-24. So I sold my JH-16 and grabbed a JH-24. Only one problem. After a couple of months, I noticed that my JH-16 had sounded better. Fatter. Ooops. I had sold it to Bill Schnee, who was mixing all manner of platinum records off it. Guys, the old Neve 1073 and most other vintage gear is [b]full[/b] of transformers. Squarewave performance of most transformers is pretty bad. But the 1073 happens to sound killer. Yank the transformer out, balance it electronically, and the squarewave response will improve. It will also not sound as good. People talk about wanting to hear some "iron" in the signal path. Iron = transformer. If somebody hooked a Berhinger mixer and a Neve 1073 up to a squarewave generator and without telling you what you were looking at on an O'scope, asked you to predict which [b]should[/b] sound better by the squarewave, the Berhinger would win every time. Digital and analog may well sound different, but looking at a squarewave on a 'scope is not going to tell you why, or what to do about it. Been there and done that 20 years ago. Regards, Brian T
Link to comment
Share on other sites

"Digital and analog may well sound different, but looking at a squarewave on a 'scope is not going to tell you why, or what to do about it. Been there and done that 20 years ago." If what you're saying were true, then, if a signal path can't pass a square wave at all, regardless of its frequency, it shouldn't matter. And, it does matter. I'm not talking about frequency response of a recording/playback device. I'm assuming the system is flat 20-20K. What's desirable is a device that can distinguish between a 15 kHz square wave and sine wave. Digital recording devices at 16 bit 44.1 see them both as something close to a sine wave. Not enough sampling points. Do the math. GY

GY

 

Link to comment
Share on other sites

You do realize that many fine analog tape decks, the virtual "Holy Grail" of audio to many people including me, are simply awful when it comes to a 15KHz square wave. Gruesome, in fact. Have a look for yourself sometime. A squarewave looks pretty darn close to a sine wave at 15KHz, assuming it still resembles anything even close to a symmetrical waveform at all. Like I said, infering that one thing sounds better than another [b]in the real world[/b] by simply looking at square wave response is just not realistic. I could show you nearly endless examples of gear with worse square wave response that sounds better than gear with good square wave response. Case in point. Most EQs and compressors that are vintage and considered excellent sonically, and constantly appreciate in value because of their sound; most of them will simply decimate high freq squarewave response as soon they are inserted into the circuit. The "after" picture is vastly worse than the before, as far as a squarewave goes. So why do people pay a fortune for that gear and use it whenever as possible? Hey, I'm not trying to bust your groove here. Just trying to share a little info that's the product of 25 years of making major label albums and a whole bunch of hours spent in technotrance mode, forgetting to eat and sleep. Squarewave away to your heart's desire with my best wishes. Maybe you'll discover something I missed. Regards, Brian T
Link to comment
Share on other sites

OK, I decided to do a test myself. I created a square wave in soundforge at 44.1/16 bit. This was 20 seconds long and at -1 dB. The waveform looked fine in Soundforge. Next, I saved it and imported it into Nuendo ... it sounds the same but zooming in as far as possible I see the waveform as anything but square! It's slanted and the peak has an indentation so it looks like a tooth: _ _ / \__/ \ / \ / \ / \ This wave form has never been anywhere except on my hard drive! What's going on? Another thing. Even at -1dB, with the faders on my O2R set at unity and the input via ADAT from a Hammerfall card, the clip LEDs light. How does -1dB clip a digital desk?
"That's what the internet is for. Slandering others anonymously." - Banky Edwards.
Link to comment
Share on other sites

[quote]Originally posted by Rog: [b]OK, I decided to do a test myself. I created a square wave in soundforge at 44.1/16 bit. This was 20 seconds long and at -1 dB. The waveform looked fine in Soundforge. Next, I saved it and imported it into Nuendo ... it sounds the same but zooming in as far as possible I see the waveform as anything but square! It's slanted and the peak has an indentation so it looks like a tooth: _ _ / \__/ \ / \ / \ / \ This wave form has never been anywhere except on my hard drive! What's going on? Another thing. Even at -1dB, with the faders on my O2R set at unity and the input via ADAT from a Hammerfall card, the clip LEDs light. How does -1dB clip a digital desk?[/b][/quote]You're on the right track, but to do the test accurately, you need to use an adjustable audio oscillator that is capable of producing a square wave. The signal needs to go through the A>D and D>A coverters to tell the story.

GY

 

Link to comment
Share on other sites

[quote]Originally posted by GY: [QBI'm not talking about frequency response of a recording/playback device. I'm assuming the system is flat 20-20K. What's desirable is a device that can distinguish between a 15 kHz square wave and sine wave. Digital recording devices at 16 bit 44.1 see them both as something close to a sine wave. Not enough sampling points. Do the math. GY[/QB][/quote]if i am remembring correctly, a square wave is a sine wave plus odd order harmonics(which are what cause the quicker rise and fall). given your test frequency of 15kHz, that would mean we are concerned with 45k, 75k, etc. i would expect that frequency response does matter.
Link to comment
Share on other sites

[quote]Originally posted by corey: [b] [quote]Originally posted by GY: [QBI'm not talking about frequency response of a recording/playback device. I'm assuming the system is flat 20-20K. What's desirable is a device that can distinguish between a 15 kHz square wave and sine wave. Digital recording devices at 16 bit 44.1 see them both as something close to a sine wave. Not enough sampling points. Do the math. GY[/b][/quote]if i am remembring correctly, a square wave is a sine wave plus odd order harmonics(which are what cause the quicker rise and fall). given your test frequency of 15kHz, that would mean we are concerned with 45k, 75k, etc. i would expect that frequency response does matter.[/QB][/quote]Your first part is correct. I'm focusing on how the 15 kHz looks upon reproduction.

GY

 

Link to comment
Share on other sites

then i am not understanding why it is relevant. if the only thing that is keeping it from looking like a square wave is frequencies far out of our hearing range, why does it matter? or, to phrase it another way, can you tell the difference between a 15k sine and square wave just by hearing them?
Link to comment
Share on other sites

[quote]Originally posted by corey: [b]then i am not understanding why it is relevant. if the only thing that is keeping it from looking like a square wave is frequencies far out of our hearing range, why does it matter? or, to phrase it another way, can you tell the difference between a 15k sine and square wave just by hearing them?[/b][/quote]I'm not sure it does matter. I can hear a difference in a 7k sine wave and square wave. Above that frequency, it is hard to discern. At 3k it is a very obvious distinction. So at what frequency do we not care about the square wave response of a recording/playback device?

GY

 

Link to comment
Share on other sites

i dont think we can pinpoint an exact spot, but i think your chosen examples reveal something. again, think of how a square wave is formed. the 3k has more harmonic info within our hearing range, so it is easier to hear their affect. again, some questions for you. how does that 3k square wave look post a/d? how does it sound?
Link to comment
Share on other sites

[quote]Originally posted by GY: [b] [quote]Originally posted by Rog: [b]OK, I decided to do a test myself. I created a square wave in soundforge at 44.1/16 bit. This was 20 seconds long and at -1 dB. The waveform looked fine in Soundforge. Next, I saved it and imported it into Nuendo ... it sounds the same but zooming in as far as possible I see the waveform as anything but square! It's slanted and the peak has an indentation so it looks like a tooth: _ _ / \__/ \ / \ / \ / \ This wave form has never been anywhere except on my hard drive! What's going on? Another thing. Even at -1dB, with the faders on my O2R set at unity and the input via ADAT from a Hammerfall card, the clip LEDs light. How does -1dB clip a digital desk?[/b][/quote]You're on the right track, but to do the test accurately, you need to use an adjustable audio oscillator that is capable of producing a square wave. The signal needs to go through the A>D and D>A coverters to tell the story.[/b][/quote]I'm not sure I understand. How does a square wave produced by software differ from one produced by an oscillator? Furthermore, how can I tell if the wave is modified when I send it out through convertors and back into Nuendo when Nuendo doesn't show a square wave to start with anyway?
"That's what the internet is for. Slandering others anonymously." - Banky Edwards.
Link to comment
Share on other sites

Corey, you are very much on the right track of explaining this; you just need to take it a little further. GY, you must realize that your square wave is nothing but the summation of a number of sine waves mathematically linked to the fundamental frequency. Like Corey said, your 15K square is really a 15K sine plus a 45K sine plus a 75K sine plus a 107K sine and so on. It is not reasonable to require any audio device to pass 45k, much less the needed 107K to actually make it "look" like a square wave. Let's do a quick little experiment. You say that the square wave through your DAW falls appart around 3KHz. First, I'll set up a few condidtions: --The anti-aliasing filter in the A/D limits the freq. response to 20KHz with a very steep filter. Take this to mean that no signal above 20K will pass through the converter. --A square wave doesn't really "look" like sqaure wave until you add at least the 7th order harmonic. That's 3rd order + 5th order + 7th order. The higher you go, the prettier the square wave will be. --If 20KHz is the 7th order harmonic of some fundamental frequency square wave, then increasing the fund. freq. of that square would quickly degrade the appreance of "squareness". So what is that frequency? 20KHz/7 = 2857Hz. So to be more precise, your square wave probebly fell appart right at 2847Hz. Asking your system to pass good looking squares above 2847Hz is asking your system to reproduce frequencies abouve the audible spectrum which is of very little benefit. So what are you worried about? I personnally am not concerned about the fact that my little PT rig won't pass a 30K signal becuase if it could, I wouldn't be able to hear it anyway. James
Link to comment
Share on other sites

I've had some spare time to re-read all the posts and have thought more about this. For some reason I was ignoring the fact (that you all tried to explain to me!) that without the third harmonic, you can't get a square wave, let alone the 5th and 7th. For a 12 k square wave, the third harmonic is 36 k and is well beyond the range of a 44.1 kHz sampling rate device. I get it now. Regardless, my ATR-102 was able to do a good job of representing a 12 kHz square wave. Pretty impressive. I still think that a 192 kHz sampling rate would be an audible improvement over 44.1. Thanks to all for your inputs...

GY

 

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...