[ATAT]

Whee. A real scope is such a good investment.. check craigslist, you can get stuff for REAL cheap. otherwise, ebay is your answer.

I heart my HP54021D, I'll never /need/ the 400Mhz resolution, but its nice to know that whatever I do, my scope can see it.

 

[grasshpr]

Quote:

Originally Posted by tangent If your goal is to make fine audio measurements, yes, you want at least 16 bits of resolution. But that's hardly the only reason for having a scope. Try testing for ringing with a square wave with your sound card. Try looking for signs of oscillation with your sound card.

Never had a problem with simulating a square wave, 10th order fourier series hits the spot!

Anyhow, I understand your rationale. I've only put it out there just as an alternative to the parallax osc. I've had a chance to work with it and found it not pleasant mainly because of its software interface and data comm protocols.

Just my 2 cents.

 

[tangent]

Quote:

Originally Posted by raduray Out of curiosity Tangent, what scope do you use?

A TDS 2012. Its little brother, the TDS 1002, is capable of nearly everything this one is, you just lose color and a few other niceties. Yes, $1000 is a lot to spend. But I'd be surprised if there were anyone participating in this thread that had spent less than that on audio gear. It's just a question of priorities.

If you search the archives, you'll see that I was borrowing a TDS 320 from work for a while, but being a full-size design, I had to return it to get my bench space back. I bought a Bitscope to replace it, and that's how I was soured on the idea of PC based scopes. I bit the bullet and bought the new Tek soon after the Bitscope died. These new LCD-based scopes are small enough to fit on very cramped work benches; not as small as a PC-based scope, but small enough.

Quote:

Originally Posted by grasshpr 10th order fourier series hits the spot!

I'm guessing you're referring to the idea that if you combine 10 equal-value harmonics, you approximate a square wave?

The highest end sound cards sample at 192 kHz. Nyquist says that the highest frequency you can get from that is f/2, or 96 kHz. With the method you propose, the highest frequency square wave you can approximate with your 10 harmonic method is 9 octaves below that, 188 Hz. Yes, friends, we're limited to the upper bass here. If you use a more prosaic sound card, you won't even be able to test into the midbass!

Square waves don't get useful for testing audio amplifiers until 100 kHz or so. 9 octaves above that is 51.2 MHz. You'll find that ~50 MHz generators are awfully expensive. That pretty much puts the kibosh on direct digital synthesis of square waves, from the DIYer's standpoint.

The wise DIYer uses analog circuitry for this, not digital trickery.

 

[grasshpr]

Quote:

Originally Posted by tangent I'm guessing you're referring to the idea that if you combine 10 equal-value harmonics, you approximate a square wave? The highest end sound cards sample at 192 kHz. Nyquist says that the highest frequency you can get from that is f/2, or 96 kHz. With the method you propose, the highest frequency square wave you can approximate with your 10 harmonic method is 9 octaves below that, 188 Hz. Yes, friends, we're limited to the upper bass here. If you use a more prosaic sound card, you won't even be able to test into the midbass! Square waves don't get useful for testing audio amplifiers until 100 kHz or so. 9 octaves above that is 51.2 MHz. You'll find that ~50 MHz generators are awfully expensive. That pretty much puts the kibosh on direct digital synthesis of square waves, from the DIYer's standpoint. The wise DIYer uses analog circuitry for this, not digital trickery.

Nyquist sampling theorem says that you need at least twice the sampling speed of the signal you are measuring, which doesn't have anything to do with outputing signals, which from my understanding was what this issue was about.

Now the 10 harmonics should be in scales of f, 2*f, 3*f,... which depends on the frequency you will be outputing (which only limits your output bandwidth to approximately 10khz given a input switching speed of 96khz). I'm assuming your talking about output, not input.

Anyhow, I've generated square waves before on my sound card and moderate frequencies, but definitely not at the speeds that you mentioned though. But we can't expect much from a sound card. Function generators or home build oscillators are still a much preferred method for square, triangle, sine, waves. Only benefit I can imagine with using a PC sound card (or PC-based function generators e.g., DSP siglab) in terms of output would be its ability to output white noise, or any type of nonstandard signal.

Also keep in mind, if u have any type of circuitry at all, whether analog or digital, its still some form of trickery because of the fact that sine waves do not occur naturally. You can however provide very accurate approximations to this function. Ultimately your desired sine wave is distorted based on the analog circuitry you provide and may not be as perfect as you desire. Digital electronics will also provide similar results as well.

 

[star882]

Quote:

Originally Posted by tangent Square waves don't get useful for testing audio amplifiers until 100 kHz or so. 9 octaves above that is 51.2 MHz. You'll find that ~50 MHz generators are awfully expensive. That pretty much puts the kibosh on direct digital synthesis of square waves, from the DIYer's standpoint. The wise DIYer uses analog circuitry for this, not digital trickery.

What about use the video card? The DACs on those can easily go up to 100MHz or more. Or maybe use a FPGA driving a high speed DAC.
Although a simple digital circuit (gate-based oscillator) to generate square waves can be made for less than $5 of parts...

For a scope, what about use a video capture card? The good ones can attain 10 bits and 6MHz or so, which is good enough for many circuits. Or modify a TV card by reconnecting the IF to the chipset to an instrumentation amp and flash the firmware to give out the raw ADC value. It's possible to attain 40MHz or more that way.

 

[tangent]

Quote:

Originally Posted by grasshpr Nyquist sampling theorem says that you need at least twice the sampling speed of the signal you are measuring, which doesn't have anything to do with outputing signals

It works the same way in reverse. It's trivial to prove this to yourself. A DAC with maximum signal generation frequency f can go from its highest output voltage to its lowest in 1/f time. (Just converting samples per second to seconds per sample here.) That's only half a cycle, though. Doubling that to 2/f and inverting, the fastest periodic wave is f/2.

Quote:

the 10 harmonics should be in scales of f, 2*f, 3*f,...

You're right, I got the harmonic part wrong.

The point is still the same, though: you can't use a sound card to generate good square waves that are nearly fast enough to be useful for audio amplifier testing.

Quote:

Only benefit I can imagine with using a PC sound card...would be its ability to output white noise, or any type of nonstandard signal.

Oh, I don't take so harsh a stance. They're also excellent audio frequency sine wave generators. It's common for modern sound cards to be able to produce pure sine waves with distortion products below -96 dBFS. To get that in a general purpose function generator, you have to spend many thousands of dollars: this is the price of their flexibility. A sound card has a simpler job, so it can do this precise job with less cost. (Purchasing volume helps, too.)

There's a middle ground, and that's dedicated audio test sets. Notice that they all only go up to 250 kHz or so. By sacrificing high frequency ability, they're able to get performance equal to or better than your sound card, and certainly better than a general purpose generator costing the same amount.

Quote:

if u have any type of circuitry at all, whether analog or digital, its still some form of trickery because of the fact that sine waves do not occur naturally.

Eh, I can see why you might take that philosophical position, but I can't agree with it. To me, it's like saying that pi doesn't occur in nature, because there are no perfect circles in the real world.

To me, pure sine waves do exist. It's just that they have to travel through distorting media before we can perceive them.

 

[raduray]

Well, I got one on ebay. Hope it's the real deal.
Calibrated, w/probes and 7-day return
http://cgi.ebay.com/ws/eBayISAPI.dll...tem=7619714208

If anyone is interested, here was my backup which is still available for another 90 minuts or so. 7-day return according to question I posed, but does not take paypal
http://cgi.ebay.com/ws/eBayISAPI.dll...tem=7620601116

 

[grasshpr]

Quote:

Originally Posted by tangent It works the same way in reverse. It's trivial to prove this to yourself. A DAC with maximum signal generation frequency f can go from its highest output voltage to its lowest in 1/f time. (Just converting samples per second to seconds per sample here.) That's only half a cycle, though. Doubling that to 2/f and inverting, the fastest periodic wave is f/2.

I see what your saying. Never thought of sampled output in terms of nyquist criterion.

Quote:

Originally Posted by tangent To me, pure sine waves do exist. It's just that they have to travel through distorting media before we can perceive them.

I've been brought up with the concept that nearly eveything introduced into a system will provide distortion of a sine wave, that's why I don't think they exist. Mathematically, we should be able to see it all the time, e.g., flexible beam motion, van der pol oscillators, etc. But because of the fact that real world conditions modifies these type of phenomena, I can't bring myself to believe that they exist in its pure form. But of course this is not we were talking about. Thanks for the debate though, tangent

 

[tangent]

Quote:

Originally Posted by grasshpr Mathematically, we should be able to see it all the time, e.g., flexible beam motion, van der pol oscillators, etc.

Here's one: elemental spectra. Get a pure element hot enough and it will give off light with a characteristic spectrum. The bands you see in a spectrogram are, in a very real sense, sine waves made visible. It's a little abstract, but they're there.

 

[grasshpr]

Quote:

Originally Posted by tangent Here's one: elemental spectra. Get a pure element hot enough and it will give off light with a characteristic spectrum. The bands you see in a spectrogram are, in a very real sense, sine waves made visible. It's a little abstract, but they're there.

Only problem here is that the atmospheric medium makes distortions... Much like how light gets refracted through our atmosphere. Though pure sine wave, it does not get observed as one... Nice example though!

 

[tangent]

Quote:

Originally Posted by grasshpr Only problem here is that the atmospheric medium makes distortions...

Really?

The atmosphere does bend the light, but this doesn't change its frequency, hence it doesn't make it less pure. You can pick up "noise" -- stray light in this case -- but it will likely be in a different frequency, so you'll still get a nice sharp single frequency line in a spectrogram. And don't forget space-based instruments.

I don't see how you "distort" light in the spectrographic sense.

 

[grasshpr]

Quote:

Originally Posted by tangent Really? The atmosphere does bend the light, but this doesn't change its frequency, hence it doesn't make it less pure. You can pick up "noise" -- stray light in this case -- but it will likely be in a different frequency, so you'll still get a nice sharp single frequency line in a spectrogram. And don't forget space-based instruments. I don't see how you "distort" light in the spectrographic sense.

You are right about why the frequency won't change, however, with atmospheric effects producing refraction patterns of the source light, you would start seeing distortions in the original amplitude. This is primarily caused by not being able to observe the light directly from a source and also what you mentioned before, stray light.

Also, this is the same reason why high precision space observation stations (e.g., hubble, JWST, and a host of planned mission from NASA) are located above the Earth's atmosphere. To be able to observe the coherency of the incident light signal in the proper time frame, its difficult to deal with time domain refraction effects (e.g., temperature, humidity, and pressure variations which can potentially be fluctuating within a few minutes). Integration time to observe light from distant stars sometimes falls in the scale of an hour or two depending on its distance away from Earth, such that these atmospheric effects can massively blur your image til they are no longer coherent.

 

[ezkcdude]

My first scope! Just won a Tek 465B on eBay for $142.50. That's pretty good, right?

 

[NeilR]

Quote:

Originally Posted by ezkcdude My first scope! Just won a Tek 465B on eBay for $142.50. That's pretty good, right?

No. You overpaid. You could of gotten one for $39.00

That one was even guaranteed to power up. (Not guaranteed to do anything useful, however

)

Just kidding. Some of these listings kill me. Yours is one of the relatively few that more or less unconditionally claims it actually works. You should get a good scope for well under $200 to your door. Much better than a 200KHZ bandwidth PC scope.

 

[ezkcdude]

Quote:

Originally Posted by NeilR No. You overpaid. You could of gotten one for $39.00 That one was even guaranteed to power up. (Not guaranteed to do anything useful, however ) Just kidding. Some of these listings kill me. Yours is one of the relatively few that more or less unconditionally claims it actually works. You should get a good scope for well under $200 to your door. Much better than a 200KHZ bandwidth PC scope.

You had me going there, for a microsecond. Oops, I mean nanosecond.