The reason I asked you to measure rise times in 1X mode is because the bandwidth is greatly reduced in that mode, so rise times should be long compared to 10X mode. You should be aware that if you want the full bandwidth of the scope, you can't use the probe in 1X mode.

Typically, the bandwidth and rise times for scopes are better than the spec.

On eBay you can find low cost scope probes that are considerably better than their advertised specs. For example, these are probably more like 200 MHz bandwidth rather than 100 MHz:

http://www.ebay.com/itm/Two-x1-x10-...020?pt=LH_DefaultDomain_0&hash=item334f669144

This probe is specified to have a bandwidth of 4 MHz in 1X mode. The bandwidth in 10X mode is much more than 100 MHz.

See this thread for more info about these probes:

http://www.eevblog.com/forum/produc...t-100mhz-oscilloscope-probes-hands-on-review/

I've attached two images. The first shows the output from a Rigol function generator set to generate a 20 MHz square wave as displayed on a Tektronix TDS210, which is a 60 MHz scope. The Rigol spec says the rise time should be 5 nS.

The second image shows the same signal as displayed on an Agilent DSO5034A 300 MHz scope. The measured rise time is 4 nS.

These two images were captured using the low cost probe referenced above which I bought on eBay.

One way you can generate a square wave with a fast rise time is to get a 74AC74 flip-flop (the AC series is Fairchild's "advanced CMOS" family). Connect the flip-flop to divide by two and feed it with the output of one of your clock generators. The rise time of this part is just a little more than 1 nS when properly wired and bypassed.

Your scope should be able to display a 20 MHz square wave at least as good looking as the TDS210 image.

I only have upper 20s MHz oscillators, so I can't really compare. But, with a 28MHz oscillator, I get nothing like a square wave. On 10X. The duty cycle of the output is pretty obviously 50% by the way.

Also, I discovered my probes are indeed 100MHz, so that can't be the issue. What good would a crystal oscillator be if it didn't truly output a decent square wave.

 

Keep in mind that if the mathematics didn't do a good job of modelling the real world, they wouldn't be used.

It's good at modelling the rise time, but in practice the square wave you'll get won't have the extra "humps" at the top of the waveform.

 

No. What you see in practice should follow the mathematical modelling. That is what simulators are suppose to do, if they are doing it right.

 

No. What you see in practice should follow the mathematical modelling. That is what simulators are suppose to do, if they are doing it right.

Then why does "The Electricians" 60MHz scope view a 20MHz square wave so well? It doesn't have those extra harmonics in there, it just looks like a tall trapezoid. Also, what would cause there to be a voltage swing at the crest of the square wave? It doesn't feel right what you're trying to describe. A square wave can be thought of as a sum of odd harmonics, BUT the actual sum itself doesn't have ANY ripple at the top, and therefor there's nothing for to attenuate, once the internal cap is taken care of, you'll see a stable DC "high" voltage.

 

I don't have the answer to everything, only almost everything. Maybe others can chime in.

The overshoot and undershoot after the initial rise is usually caused by impedance mismatch and reflections.

Signal drivers are sometimes rise-time limited, i.e. high frequency suppressed, specifically to prevent these kind of problems. Thus restricting the rise time and high-frequency rejection are directly related.

Why the trapezoid is flat on the top I am not sure. We would have to mathematically decompose the waveform in order to get a better understanding.

 

Then why does "The Electricians" 60MHz scope view a 20MHz square wave so well? It doesn't have those extra harmonics in there, it just looks like a tall trapezoid. Also, what would cause there to be a voltage swing at the crest of the square wave? It doesn't feel right what you're trying to describe. A square wave can be thought of as a sum of odd harmonics, BUT the actual sum itself doesn't have ANY ripple at the top, and therefor there's nothing for to attenuate, once the internal cap is taken care of, you'll see a stable DC "high" voltage.

You can find all over the web and in textbooks, images such as MrChips has posted. Those waveforms are indeed what you would get if you include the first 3 (or whatever) harmonics of a square wave and NONE of the higher harmonics.

But, that's not what happens in the real world. In the real world, the frequency response of a scope is that of a low pass filter. The response of a scope is not a "brick wall" filter such that after you get past 100 MHz, there is NO response. Actually, scope responses usually roll off rather gradually and so there is a certain amount of 3rd, 4th, 5th, and so on included. Furthermore, the mathematical curves MrChips posted include no phase shift in the higher harmonics. That phase shift is important and along with the gradual rolloff causes the "humps" in MrChips' theoretical curves to pretty much vanish.

If you compare the image from the 60 MHz scope to the one from the 300 MHz scope, you will see that there is some overshoot and ringing in the 300 MHz image. This is what happens when you use the short ground lead and clip that comes with these probes and look at a fast rising edge. The loop inductance formed by the ground clip "rings". This appears as "humps" in the image, but these "humps" have a different cause than the Fourier "humps" of MrChips' images. We don't see much ringing in the 60 MHz scope response because the lower frequency response of the 60 MHz scope attenuates the "ring" frequency.

 

What you use for a probe and also input termination makes a huge difference when you start looking at these faster rise time higher frequency signals.

The first image is a 40mhz square wave from a clock oscillator on a 100mhz Rigol 1052E using the supplied 10x probe and 1Meg internal termination (the only thing the Rigol has.

The second image is the same square wave on the Rigol but this time a BNC t-connector was used such that a 50 ohm termination could be added externally. This is about the best you can get on this 100mhz scope. The signal was fed through a short piece of coax instead of the 1x10x probe. (edit: oops, this signal is actually using a tinylogic buffer chip with a 700ps risetime...the 40mhz squarewave from the oscillator looked similar though...with not as much overshoot because it has a slower rise)

The third image is the same square wave in a 200mhz Agilent MSOX3024A with internal 50 ohm termination. The signal was fed through a short piece of coax instead of the 1x10x probe.

 

What you use for a probe and also input termination makes a huge difference when you start looking at these faster rise time higher frequency signals.

The first image is a 40mhz square wave from a clock oscillator on a 100mhz Rigol 1052E using the supplied 10x probe and 1Meg internal termination (the only thing the Rigol has.

The second image is the same square wave on the Rigol but this time a BNC t-connector was used such that a 50 ohm termination could be added externally. This is about the best you can get on this 100mhz scope. The signal was fed through a short piece of coax instead of the 1x10x probe. (edit: oops, this signal is actually using a tinylogic buffer chip with a 700ps risetime...the 40mhz squarewave from the oscillator looked similar though...with not as much overshoot because it has a slower rise)

The third image is the same square wave in a 200mhz Agilent MSOX3024A with internal 50 ohm termination. The signal was fed through a short piece of coax instead of the 1x10x probe.

The Agilent MSOX3024A must have come with some probes. What do you see if you use the Agilent probes with the Rigol and with the Agilent, rather than coax. In other words, how does the Rigol 10x probe compare with the Agilent 10x probe, on both the Rigol and the Agilent scope?

In this thread:

http://www.eevblog.com/forum/produc...t-100mhz-oscilloscope-probes-hands-on-review/

you can see where I obtained a 1.05 nS rise through the low cost 10x probe bought on eBay. It is possible to get such a fast rise through a 10x probe into a high impedance input on a scope. Perhaps the Rigol probe is the limiting factor that makes your 40 MHz clock look almost like a sine wave rather than a square wave.

 

Here are a few (horrible quality) pictures of the waveforms I'm getting. Each of these are different crystal oscillators between 26 and 28 MHz, with literally the shortest connections and such I could manage, no breadboard at all (I figured maybe that was causing some of the issues). Those are literally the best I could get.

Maybe they're just broken/defective? I DID get them from a grab-bag. However, I have a 2+ MHz oscillator from that same bag and it seems to do fine, and wouldn't at least ONE of them work ok? I don't really have much else to test with. Would a 555 timer be a useful?

The last one is the 1MHz crystal oscillator I bought from Jameco once upon a time. It's an example of a good capture.

 

The 3 clock outputs look good to me. The first looks like it has a totem pole output (active pullup); 2 and 3 look like they have a resistor pull up.

If you clean up those outputs by passing them through a fast logic gate, or flip-flop from a TTL logic family, you should get a pretty good square wave. The faster the logic family, the better. Use 74S if you have it, or 74AC is even better.

Don't use the ground clip that came with the probes. Pull off the probe tip and wind a piece of wire around the ground barrel which is near the tip; twist it tight and use that as your ground, soldered directly to the ground of your oscillator package. Then touch the probe tip to the oscillator output. This should get rid of a lot of the ringing.

 

The 3 clock outputs look good to me. The first looks like it has a totem pole output (active pullup); 2 and 3 look like they have a resistor pull up.

If you clean up those outputs by passing them through a fast logic gate, or flip-flop from a TTL logic family, you should get a pretty good square wave. The faster the logic family, the better. Use 74S if you have it, or 74AC is even better.

Don't use the ground clip that came with the probes. Pull off the probe tip and wind a piece of wire around the ground barrel which is near the tip; twist it tight and use that as your ground, soldered directly to the ground of your oscillator package. Then touch the probe tip to the oscillator output. This should get rid of a lot of the ringing.

So, why do your oscillators give you a cleaner square wave then?
I have a lot of 4000 series ICs, but I don't really have any TTL. Would 4000's work just as well to test?
EDIT: I just checked out 4000 series ICs, apparently they're not high-speed devices by any means.

 

The waveforms I posted on this thread came from a Rigol function generator, not from a clock oscillator. And, I didn't use the ground lead/clip with the probe. I pulled the spring hook off the probe and poked the probe tip into the center conductor of the BNC output connector on the Rigol. I used a piece of copper braid to make a very short connection from the ground barrel of the probe to the shell of the BNC connector.

If you'll use the technique I described in post #30 to avoid the use of the ground lead/clip, you'll get a cleaner waveform with fast rising/falling edges (less ringing).

Try this on the oscillator that you used for the first image and you should see a cleaner square wave.

 

The waveforms I posted on this thread came from a Rigol function generator, not from a clock oscillator. And, I didn't use the ground lead/clip with the probe. I pulled the spring hook off the probe and poked the probe tip into the center conductor of the BNC output connector on the Rigol. I used a piece of copper braid to make a very short connection from the ground barrel of the probe to the shell of the BNC connector.

If you'll use the technique I described in post #30 to avoid the use of the ground lead/clip, you'll get a cleaner waveform with fast rising/falling edges (less ringing).

Try this on the oscillator that you used for the first image and you should see a cleaner square wave.

This is an excellent point: at sufficiently high frequencies, the small inductance of even a short ground lead can have a very significant effect. The need for SHORT probe ground connections when measuring signals with high frequency content is hard to over emphasize - it is something that gives trouble to many people when they start this kind of work.

For higher frequency work, some manufacturers produce a variety of probes with very short ground springs, or probes mating with small sockets which can be soldered to the device under test.

 

The low cost probes from eBay I referenced indeed come with a special ground spring but I wasn't sure if the OP's probes had those, so I described how to make your own short ground connection. You're absolutely right; for fast edges it's necessary to avoid the excessive ringing. Maybe Austin will get a chance to try it out and report.

 

The low cost probes from eBay I referenced indeed come with a special ground spring but I wasn't sure if the OP's probes had those, so I described how to make your own short ground connection. You're absolutely right; for fast edges it's necessary to avoid the excessive ringing. Maybe Austin will get a chance to try it out and report.

I plan on giving this a shot soon, but I'm a bit hesitant to pull my probes apart, even if it is a pretty low-risk precedure and the probes are cheap, because I don't want to be without use of both channels for even a short period of time right now, too much to do in too little time.

I'll post my results, and I'll give an update when I get my function generator and/or new crystal oscillators to fiddle with.

 

The components of a typical probe kit are shown in the first image.

The second image shows a close up of the spring hook type tip.

The third image shows what the probe looks with the spring hook removed.

The spring hook tip is easily removed by just pulling it off of the probe. Unless you have a very unusual probe, the tip is removable and nothing is harmed by removing it. To reinstall it, just push it back on.

 

The components of a typical probe kit are shown in the first image.

The second image shows a close up of the spring hook type tip.

The third image shows what the probe looks with the spring hook removed.

The spring hook tip is easily removed by just pulling it off of the probe. Unless you have a very unusual probe, the tip is removable and nothing is harmed by removing it. To reinstall it, just push it back on.

Oh my gosh. Ha!
Yeah, that's actually pretty laughable. I didn't think it was that easy. Literally nothing to it. I thought you'd have to unscrew it, and then there'd be springs and stuff... but nope, I just pulled it a little bit and it slides off. My probe's are pretty much identical to others posted here.
I'll give it a shot tomorrow!

 

If I am testing a breadboard circuit I use a short piece of single strand hookup wire attached to the end of the spring-loaded hook. Then I use the free end of the hookup wire as my probe. Never insert the pin probe into the holes of the breadboard. If you break off the pin your probe is trash.

I reserve using the exposed pin for trouble shooting IC pins and pads on a PCB.