Oscilloscope training class (long)

[Chris Wilson]

I think that the analog CRT could be bigger with an electrostatic deflection without any sacrifice of a speed, but deflection angle of an electrostatic deflection system is usually small. Correction of non-linearities resulting from larger angle becomes more difficult considering faster sweeps. <SNIP>

http://www.ebay.co.uk/itm/OSCILLOSCOPE-COUNTER-TIMER-MULTIMETER-TEKTRONIX-2236-/150757466948?pt=UK_BOI_Electrical_Test_Measurement_Equipment_ET&hash=item2319d86344

He obviously doesn't know this

Thanks for another detailed and helpful reply, I have to say one of the nice things about my USB scope is I can be working on a vehicle and have the 17 inch laptop display show the traces, or even a 24 inch monitor or one could use a projected image. It's a big time saver being able to watch traces without coming out of the footwell, or walking upp to a tiny screen.

[tekfan]

Somewhere I read a story about Tektronix. Sometime in the late 80's they were designing transient digitizers that used a scan converter tube. They made digitizers that could capture 4 GHz transients. The russians then invited the guys from Tek to show them their transient digitizer. While the Tek digitizer fit into a 19'' rack the russian version had a scan converter tube that was 10 feet long and worked well into the 20 GHz range.

This tube is very similar to the Tektronix T-7912 scan converter CRT, which was capable of recording 2.5 gigahertz signals. A 'write' electron gun with traveling wave deflection system deposits charge patterns on a thin silicon wafer. These patterns are read by a low velocity 'read' electron beam from a gun mounted in the opposite end of the tube, scanning the back surface of the silicon wafer.
[From Peter A. Keller, The Cathode-Ray Tube, Palisades Press, 1991]
Total length : 57.5 cm (22.64 in)

A nice site for old tubes: http://www.lampes-et-tubes.info/p/sc.htm

[alm]

I think that the analog CRT could be bigger with an electrostatic deflection without any sacrifice of a speed, but deflection angle of an electrostatic deflection system is usually small. Correction of non-linearities resulting from larger angle becomes more difficult considering faster sweeps.

The voltage swing was also limited by the output transistors avaliable at that time. You may be able to do something about the plate geometry to increase deflection, but I'm not sure if this can be done without any reduction in bandwidth.

That means if the scope screen would be bigger (keeping the deflection angle constant), then the CRT neck would become impractically long (it is already very long even with "small" scope screens). Scale that length in proportion of your desired screen size and think whether or not you would want it to your bench. TV CRT's used 90 or 110° deflection angle, thus it makes shorter tube necks possible.

There's definitely a correlation between deflection angle and bandwidth. For example the 1 GHz Tektronix 519 (ca. 1959 or so, the input directly drove the vertical plates, without any amplifications/attenuation) had a very small CRT surface, only 4x2 divs or so. The low-bandwith Tektronix 5000 series, designed for biomedical experiments, had a very large CRT. Bandwidth was ultimately limited by the depth of a standard 19" rack as used by the US army (how many people do atomic bomb tests in their backyard?). I believe the USSR had a scope with a larger bandwidth, which just used an extremely long CRT. No idea if this idea ever made it into production.

[WBB]

Thanks for the excellent videos. They have been very educational, easy to follow & understand. Excellent work that is very much appreciated.

A few questions about the coax video. Would other faults show up? eg, a hole in the shield but not a complete break? A compressed shield but not touching the center conductor and creating a short, yet? Also, is it possible to calculate the distance to a short?

Sorry for all the questions. All your videos have been great but the coax one is a subject that I am specifically interested in! Again, thanks for the great work.

[w2aew]

Thanks for the excellent videos. They have been very educational, easy to follow & understand. Excellent work that is very much appreciated.

A few questions about the coax video. Would other faults show up? eg, a hole in the shield but not a complete break? A compressed shield but not touching the center conductor and creating a short, yet? Also, is it possible to calculate the distance to a short?

Sorry for all the questions. All your videos have been great but the coax one is a subject that I am specifically interested in! Again, thanks for the great work.

The answer is - it depends....

- a hole in the shield without a complete break, or a compressed shield may or may not show up.  It all depends on how much of an impedance discontinuity it creates.  Impedance discontinuities cause reflections, just like the mis-termination of the end of the line (like the open circuit example I used, or the shorted example I showed).  The worse the impedance discontinuity of the defect, the easier it will be to detect.  Once observed, you can measure the round-trip delay and calculate the approximate position of the defect.

- Yes, you can determine the distance to a short (only the first short!).  You can see this at the end of the video where I dialed the pot down to zero.  You can measure the width of the pulse that's shown, and that will be the roundtrip delay from the scope to the short and back.  Calculate the distance the same way as I did for the open.

For those that missed it, here is the video that WBB is referring to:

Use a scope to measure the length and impedance of coax

[WBB]

Thanks for the response!. Somehow I figured there would be an "it depends" in there, such is life! It does sound promising though.

Some of the stuff is easy to find. Shields like to break near connectors, or if the cable passes through trees I can pretty much bet good money a squirrel has gotten my coax mixed up with his acorns. Never knew aluminum was so tasty.

The worst for me inevitably turns out to be a couple shotgun pellets shorting out the cable. Not exactly the easiest thing to find and they NEVER radiate when you're looking for them!

What would be the maximin rise time that would be preferred in a function generator for this?

(I don't own a fc or I would have tried to answer those questions myself!)

[w2aew]

Thanks for the response!. Somehow I figured there would be an "it depends" in there, such is life! It does sound promising though.

Some of the stuff is easy to find. Shields like to break near connectors, or if the cable passes through trees I can pretty much bet good money a squirrel has gotten my coax mixed up with his acorns. Never knew aluminum was so tasty.

The worst for me inevitably turns out to be a couple shotgun pellets shorting out the cable. Not exactly the easiest thing to find and they NEVER radiate when you're looking for them!

What would be the maximin rise time that would be preferred in a function generator for this?

(I don't own a fc or I would have tried to answer those questions myself!)

Ideally, you'd want the risetime to be a few times faster than the propagation delay through the cable of interest, so that you get a clean "plateau" between the incident edge and the reflected edge.  For typical coax with a 0.66 velocity factor, the speed of a signal is 7.79 inches per nanosecond.  Thus, a 10ns rising edge of the pulse will be spread out over about 6.5 feet of cable (i.e. the voltage 6.5 feet from the generator is just starting to move at the same time that the rising edge has finished moving at the generator).  If the pulse's risetime is slower than the roundtrip delay, then the reflection will be seen during the edge, making it much harder to accurately measure the delay.

For basic coaxial work like was shown in the video - for 10, 20, 30 feet or more of cable, a 10ns risetime is good enough. 

You might check the risetime of calibrator signal on your scope - it might be fast enough to use for this application.

[vk6zgo]

I remember using a TU-5 Pulser for this purpose,which used a tunnel diode.
In use,you connected it to the 100 volt calibrator on a 545B or similar,& the TU-5 produced a very much lower level,(but very much faster rise time) pulse which was excellent for checking the 'scope,& for TDR purposes.

I had one hanging around for years,but I don't have a 545B,or anything similar,so no high voltage calibrator.
I thought of making a vacuum tube multivibrator to produce a 100volt square wave,but "put it in the too hard basket".
Eventually,I gave it to a friend who collected 545s.

[BravoV]

- Yes, you can determine the distance to a short (only the first short!).  You can see this at the end of the video where I dialed the pot down to zero.  You can measure the width of the pulse that's shown, and that will be the roundtrip delay from the scope to the short and back.  Calculate the distance the same way as I did for the open.

For those that missed it, here is the video that WBB is referring to:

Use a scope to measure the length and impedance of coax

Great tutorial video as always W2aew, thank you Sir !

This video inspires me to learn & tinker with this poor man TDR thingy, and since you did mention high speed rise signal as a reference in that video, I assume I can also use my vintage Tek 2901 time mark generator that generates those really sharp spikes for this purpose too right ?

Example of the time mark signals generated by that 2901 at various speed at my Tek 2465B.

[WBB]

Ideally, you'd want the risetime to be a few times faster than the propagation delay through the cable of interest, so that you get a clean "plateau" between the incident edge and the reflected edge.  For typical coax with a 0.66 velocity factor, the speed of a signal is 7.79 inches per nanosecond.  Thus, a 10ns rising edge of the pulse will be spread out over about 6.5 feet of cable (i.e. the voltage 6.5 feet from the generator is just starting to move at the same time that the rising edge has finished moving at the generator).  If the pulse's risetime is slower than the roundtrip delay, then the reflection will be seen during the edge, making it much harder to accurately measure the delay.

For basic coaxial work like was shown in the video - for 10, 20, 30 feet or more of cable, a 10ns risetime is good enough. 

You might check the risetime of calibrator signal on your scope - it might be fast enough to use for this application.

Thanks again! Normally if the problem is in the 1st 30' or so I'll find it by other means, usually when I'm not really looking!  It's those few hundred feet that follow that can be a royal pain.  Thanks for the tip about the Cal signal as well. A function gen was the next thing that I wanted to get anyway, but the Cal signal should give me a bit of a starting point to help determine what I should get. Again, much appreciated!

[sdscotto]

While I can't find it at the moment, I remember one of Jim Williams Application Notes on building a ~400ps pulse generator for testing oscilloscopes and probes.

IIRC is was AN47, but doesn't appear to be on the Linear web site anymore. 

I'm dumb as a stump compared to most of you real engineers, but still got a kick out of reading his stuff, much like Alan's recent tutorials.

[alm]

IIRC is was AN47, but doesn't appear to be on the Linear web site anymore. 

If I go to linear.com and search for AN47, the first link is a PDF of the correct appnote. The later circuit with a charge line, as described in LT AN79, 94 and 122, might be better, since it generates something approaching a step, as opposed to a semi-Dirac pulse.

[sdscotto]

I played with some old coax after watching Alan's latest video and used my 2901, too.   Here are the rise time measurements on mine:

[sdscotto]

IIRC is was AN47, but doesn't appear to be on the Linear web site anymore. 

If I go to linear.com and search for AN47, the first link is a PDF of the correct appnote. The later circuit with a charge line, as described in LT AN79, 94 and 122, might be better, since it generates something approaching a step, as opposed to a semi-Dirac pulse.

Yeah, my bad, I just went to this link and saw it missing in the first section. 
http://www.linear.com/designtools/app_notes.php

[w2aew]

Those time mark generators are cool.  Wish I had one myself!  You can give one of them a try.  Since it isn't putting out a pulse, but rather an "impulse", you can measure the delay between the peak of the incident impulse to the reflected impulse (which might appear on the trailing edge of the incident impulse).  Just be sure to make the repetition interval of your marks much longer than the expected round trip delay so that you don't mix up a reflected impulse for a repeated incident one.