Author Topic: Using a fast pulse and DSO for vector network analysis  (Read 1716 times)

0 Members and 1 Guest are viewing this topic.

Offline _Wim_

  • Frequent Contributor
  • **
  • Posts: 677
  • Country: be
Re: Using a fast pulse and DSO for vector network analysis
« Reply #25 on: March 16, 2019, 08:56:18 pm »
Go for a Tek 1502 without the suffix. Higher bandwidth, shorter range, static-sensitive tunnel diode step generator. They will not turn on unless there is a functioning battery in them.

50ps risetime, system roundtrip bandwidth 140ps, so you can work out the resolution.

I've sold some by having a 300mm piece of microstrip, and letting people watch the trace wiggle as I lightly touch the microstrip with first and index fingers, and move them apart and together. That viscerally convinces them of the resolution and sensitivity. Plus the lovely old-world sampling display :)

I agree they are also very nice, but because of the static sensitive input that is difficult to repair, they are more risky to buy "As is". The 1502B and 1502C also do not start without a good battery, and many of the AS IS currently on Ebay show this "issue". Have a good look at the LCD for discoloration, I have seen some with damaged LCDs, which is also a difficult fix.

I bought one of these from Express_auctions, and it worked perfectly. They still have a few which seems to be ok:

https://www.ebay.de/itm/Tektronix-1502B-Metallic-Cable-Tester-Telecom-Datacom-B038148/292786089778?epid=78758240&hash=item442b692f32:g:U-AAAOSwbKdb0Kn9

I fully agree about the resolution and sensitivity, this is really a very nice piece of test equipment to test cheap stuff from china (or items with unknown specs). The cursor can be set to display ohms read-out also (very handy to see what is going on), and the measured reflection coefficient can be converted to VSWR by an easy formulae or a look-up table in the manual. Combine this with  a distance resolution of only 1mm (0.004feet), and you can see how much you can learn from this measurement.
« Last Edit: March 16, 2019, 08:57:55 pm by _Wim_ »
 

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 8402
  • Country: gb
    • Having fun doing more, with less
Re: Using a fast pulse and DSO for vector network analysis
« Reply #26 on: March 16, 2019, 09:24:35 pm »
Go for a Tek 1502 without the suffix. Higher bandwidth, shorter range, static-sensitive tunnel diode step generator. They will not turn on unless there is a functioning battery in them.

50ps risetime, system roundtrip bandwidth 140ps, so you can work out the resolution.

I've sold some by having a 300mm piece of microstrip, and letting people watch the trace wiggle as I lightly touch the microstrip with first and index fingers, and move them apart and together. That viscerally convinces them of the resolution and sensitivity. Plus the lovely old-world sampling display :)

I agree they are also very nice, but because of the static sensitive input that is difficult to repair, they are more risky to buy "As is".

I've had four (gulp, must be a fetish!), and they all worked after recapping and adding a battery. One example had clearly been dropped: the rear corner was broken with a 3cm*1cm piece dislodged and cracks up to 10cm long. That too worked and since all pieces were there, methylene chloride repaired the case extremely well. The CRT is shock mounted, to make it military-boot proof :)

I think the special "shorting BNC" connector on the front panel is a useful way of protecting against static in the cable. That wouldn't work if the UUT was powered, of course.

Quote
The 1502B and 1502C also do not start without a good battery, and many of the AS IS currently on Ebay show this "issue". Have a good look at the LCD for discoloration, I have seen some with damaged LCDs, which is also a difficult fix.

I bought one of these from Express_auctions, and it worked perfectly. They still have a few which seems to be ok:

https://www.ebay.de/itm/Tektronix-1502B-Metallic-Cable-Tester-Telecom-Datacom-B038148/292786089778?epid=78758240&hash=item442b692f32:g:U-AAAOSwbKdb0Kn9

I fully agree about the resolution and sensitivity, this is really a very nice piece of test equipment to test cheap stuff from china (or items with unknown specs). The cursor can be set to display ohms read-out also (very handy to see what is going on), and the measured reflection coefficient can be converted to VSWR by an easy formulae or a look-up table in the manual. Combine this with  a distance resolution of only 1mm (0.004feet), and you can see how much you can learn from this measurement.

I'm not so interested in the distance resolution of a single discontinuity per se, since the indicated distance depends on the precise velocity factor.

I'm more interested in being able to resolve two (or more) close discontinuities, as can be found in a connector or attenuator etc. For that, the system risetime is the critical parameter.
There are lies, damned lies, statistics - and ADC/DAC specs.
Gliding aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2286
  • Country: us
Re: Using a fast pulse and DSO for vector network analysis
« Reply #27 on: March 17, 2019, 01:45:00 am »
@rhb

How will you configure the correction for the Owon? Would it be possible to "import" corrections for other 12-bit scopes with high memory depth. This would be a great application for my 12-bit picoscope also.  ::)

Could the correction be something like recording the step response directly on the scope (without reflections) where your software uses some kind of convolution to correct for the scope ringing, phase and bandwidth limitations? Or is this a bit optimistic?

The process is as follows:

Window the recorded data so that you have a series of segments of the incident wave and a series of segments of the reflected wave.

Fourier transform each segment separately with adequate zero padding to prevent wrap around.

Compute the mean value of the phase by making a linear fit to the phase of each of the segment transforms and then averaging the phase delay.  Correct each of the segments to the mean value of the phase.  Then sum the segments at each frequency.  Every doubling of the number of segments gives an additional 6 dB of dynamic range in the result.

Divide the summed complex series for the reflected wave by the summed value for the incident wave.  Watch out for divide by zero or near zero.  Then convert the result to the magnitude and phase of the DUT just as you would get with a conventional VNA.

Corrections for the DSO and the pulse source must be done via SOL or SOLT with a cal kit.  The basic procedure is the same except for the cal kit you know the theoretical values for a cheap kit and measured values for an expensive kit.

My 11801 came yesterday, DOA.  But reseating the boards and cables in the main cage got it working again, though I had to do it twice.  It now runs the extended diagnostics and stops on an E5622 error which is bad NVRAM (dead battery).  I've got replacements on their way from the UK.  With a bit of luck I'll have one or more channels with a 23 pS rise time and 12.5 GHz BW.  I also got an extension cable, so I can work on the sampling heads if needed.

Now I need to sharpen the pulse from Leo Bodnar's pulser so I can test the two SD-22 sampling heads I bought.  The really hard part is being able to test an SD-32 head with 7 pS rise time and 50 GHz BW.

I'm considering a non-linear transmission line using a series of diodes for capacitors.  I've not dug into the math yet as I'm having trouble printing Michael Case's 1993 PhD dissertation at UC Santa Barbara.

Generally this is more suited to fabrication in an IC, but I'm going to try using a piece of single sided PCB with the diodes standing on end and a wire running from the input SMA to the output SMA.  The pulse sharpening is a consequence of the dispersion produced by the non-linear capacitance of the junction as a function of the applied voltage.  I'll drive it with the 100 pS impulse unit I got from Leo Bodnar.

The really scary part is the price of a 2.4 mm RF connector to connect to an SD-30 or SD-32 head.
 

Offline _Wim_

  • Frequent Contributor
  • **
  • Posts: 677
  • Country: be
Re: Using a fast pulse and DSO for vector network analysis
« Reply #28 on: March 17, 2019, 01:46:25 am »
I'm more interested in being able to resolve two (or more) close discontinuities, as can be found in a connector or attenuator etc. For that, the system risetime is the critical parameter.

Spatial resolution is according to Dr. Charles Mohr (=designer the 1502B/C for Tektronix ) <2cm

http://www.mohr-engineering.com/tektronix-1502-comparison.php

which is compared to the modern equivalent (<1cm) still very good.

This is also something I like about these units, it can also be used as a fast pulse generator with a rise time of 130ps.
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2286
  • Country: us
Re: Using a fast pulse and DSO for vector network analysis
« Reply #29 on: March 17, 2019, 01:52:37 am »
It is possible to throw together a sampling oscilloscope/TDR with a digital delay chip and a few high speed comparators.
This might provide bandwidth an order of magnitude or two cheaper than going for real time oscilloscopes.

I'm currently testing a 90 GS/s (equivalent time sampling) TDR with a bandwidth of maybe 8 GHz using a topology pulled out of the AD9500 datasheet.
(https://github.com/loxodes/tdr)

Very cool.   Getting the BW is *really* difficult and expensive.   I just got a Tek 11801 and two SD-22 heads.  I'm waiting now for NVRAM replacements to arrive from the UK.  I plan on studying the SD-22s very carefully.
 

Offline _Wim_

  • Frequent Contributor
  • **
  • Posts: 677
  • Country: be
Re: Using a fast pulse and DSO for vector network analysis
« Reply #30 on: March 17, 2019, 02:01:05 am »
Go for a Tek 1502 without the suffix. Higher bandwidth, shorter range...

Looking at the specs of the 1502B vs the orignal 1502, I would say that they are identical, or am I missing something?

 

Offline _Wim_

  • Frequent Contributor
  • **
  • Posts: 677
  • Country: be
Re: Using a fast pulse and DSO for vector network analysis
« Reply #31 on: March 17, 2019, 02:06:21 am »
The process is as follows:

Window the recorded data so that you have a series of segments of the incident wave and a series of segments of the reflected wave.

Fourier transform each segment separately with adequate zero padding to prevent wrap around.

Compute the mean value of the phase by making a linear fit to the phase of each of the segment transforms and then averaging the phase delay.  Correct each of the segments to the mean value of the phase.  Then sum the segments at each frequency.  Every doubling of the number of segments gives an additional 6 dB of dynamic range in the result.

Divide the summed complex series for the reflected wave by the summed value for the incident wave.  Watch out for divide by zero or near zero.  Then convert the result to the magnitude and phase of the DUT just as you would get with a conventional VNA.

Corrections for the DSO and the pulse source must be done via SOL or SOLT with a cal kit.  The basic procedure is the same except for the cal kit you know the theoretical values for a cheap kit and measured values for an expensive kit.

Thanks for the detailed explanation. Is the above a process you want to automate with your software?
The 11801 looks very nice, but a bit too expensive for something where my main use is "play" only. And indeed at these bandwidths all the accessories are $$$ and typically difficult to find cheaply on the used market also.
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2286
  • Country: us
Re: Using a fast pulse and DSO for vector network analysis
« Reply #32 on: March 17, 2019, 02:16:14 am »
A  sub 15 nS pulser can be produced using a single 74HC00 wired up to produce a single shot pulse and retriggered by a 555.  That is *very* cheap to build.  The connectors and cables for the test fixture will cost more.   I'm hoping that I'll have a bit of luck today and find the 7400 based single shot I built 30+ years ago.

I now have the benefit of 30 years of professional DSP experience and a budget that lets me buy anything I need.  My strongest interest is in low cost T&M kit. Yhings a person on a tight budget can afford.  There is a lot of cheap Chinese T&M gear and eval board modules available.  I'm looking for opportunities to improve the benches of those who cannot afford to spend much money.  If you have a F***Tech or similar you can sweep an HF filter.  But if you just bought a DSO and don't have the money for an F***Tech, a 74HC00 and a 555 will do until you can afford it.

http://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg1902941/#msg1902941

I too have an "unlimited" budget for the same reasons as you, but, for the reason in my .sig, I too enjoy doing more with less.

A very dear friend of mine, Leonard Pratt,  had a Diamond C and was inducted into the Aviation Hall of Fame for setting records by flying along weather fronts in the late 50's and early 60's.  Not sure of the date.  I just saw the newspaper clipping about it.  He also held an unlimited aerobatics endorsement in a sailplane from the FAA.  He had a funny story about doing an airshow where  Charlie Hilliard was performing.  Charlie was telling the organizers that he *had* to perform before Leonard because they both did the same stuff and Leonard was doing it without an engine.  So going second would be anticlimactic.  Sadly, Charlie was killed at Sun & Fun when he stepped on the brakes of his Hawker Hurricane too hard and flipped it.

@_Wim_  Yes, I plan to write software to do all that.  It will be command line driven with the data and cal files passed on the command line with output to stdout.  Classic Unix model.  With a simple gnuplot script it will produce publication quality figures.

FWIW I just got an 11801 for $185 delivered, a pair of SD-22s for $150 and an extension cable for $100.  All from eBay.  Replacement NVRAM chips cost $40 and are coming from the UK as there won't be any in the US until June.  So in about 10 days I'll be able to find out if the SD-22s still work.  But with the cable I should be able to fix them if I can get parts. 
 

Offline _Wim_

  • Frequent Contributor
  • **
  • Posts: 677
  • Country: be
Re: Using a fast pulse and DSO for vector network analysis
« Reply #33 on: March 17, 2019, 02:31:13 am »
FWIW I just got an 11801 for $185 delivered, a pair of SD-22s for $150 and an extension cable for $100.  All from eBay.  Replacement NVRAM chips cost $40 and are coming from the UK as there won't be any in the US until June.  So in about 10 days I'll be able to find out if the SD-22s still work.  But with the cable I should be able to fix them if I can get parts.

Yes, I have seen the US-prices, quite a bit better that around here in Europe. But as these are heavy and fragile beasts, shipping overseas is expensive and dangerous if the unit is not packed accordingly. 
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2286
  • Country: us
Re: Using a fast pulse and DSO for vector network analysis
« Reply #34 on: March 17, 2019, 02:41:12 am »
My 11801 arrived DOA from Arizona.  I'm in Arkansas.  Fortunately reseating the boards and cables in the main cage resolved the issue.  But I had to do it twice.  Shipping was $85.

Hopefully one of the projects previously mentioned will provide a lower cost and less delicate solution.
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 8649
  • Country: us
  • DavidH
Re: Using a fast pulse and DSO for vector network analysis
« Reply #35 on: March 17, 2019, 09:12:32 am »
Couldn't the dynamic range be improved by aggregating multiple measurements ? You don't need to do single shot capture and, by the same token, you don't need real-time capture. So by taking longer over the measurements (and a traditional VNA would have to do a sweep) you can improve the accuracy.

Sure, but a VNA has a greater inherent dynamic range because of the integrated measurement over a very narrow bandwidth.  An oscilloscope or other time domain instrument has a wide input bandwidth and short measurement time from essentially sampling.

Another problem is the non-linearity in the pulse and vertical amplifier chain.  The best instruments calibrate this out but it is not trivial to do.  HP/Agilent made some sampling based VNAs which were basically sampling oscilloscopes optimized to operate as VNAs and I think they may do this.

 

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 8402
  • Country: gb
    • Having fun doing more, with less
Re: Using a fast pulse and DSO for vector network analysis
« Reply #36 on: March 17, 2019, 09:32:03 am »
Go for a Tek 1502 without the suffix. Higher bandwidth, shorter range...

Looking at the specs of the 1502B vs the orignal 1502, I would say that they are identical, or am I missing something?

Not quite, but certainly better than I remembered.

Perhaps I have confused myself by looking at my 1502, which is certainly better than the 140ps spec. The corresponding 1502B spec is 200ps.
There are lies, damned lies, statistics - and ADC/DAC specs.
Gliding aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2286
  • Country: us
Re: Using a fast pulse and DSO for vector network analysis
« Reply #37 on: March 17, 2019, 10:22:44 am »
Couldn't the dynamic range be improved by aggregating multiple measurements ? You don't need to do single shot capture and, by the same token, you don't need real-time capture. So by taking longer over the measurements (and a traditional VNA would have to do a sweep) you can improve the accuracy.

Sure, but a VNA has a greater inherent dynamic range because of the integrated measurement over a very narrow bandwidth.  An oscilloscope or other time domain instrument has a wide input bandwidth and short measurement time from essentially sampling.

Another problem is the non-linearity in the pulse and vertical amplifier chain.  The best instruments calibrate this out but it is not trivial to do.  HP/Agilent made some sampling based VNAs which were basically sampling oscilloscopes optimized to operate as VNAs and I think they may do this.

An Owon XDS2102A will collect 20 Mpts of 12 bit data at 500 MSa/s.  That gives an upper bound of 72 dB of dynamic range.  20 Mpts at 500 MSa/s is 40 ms.  So with a 10 Mhz pulse repetition rate we have 400,000 waveforms in a single record.

If we take the increase in dynamic range as 18 bits (2**18 = 262,144) that's an additional 108 dB for a total of 180 dB of dynamic range with a 12 bit ADC and 156 dB with an 8 bit ADC. In reality, those are overly optimistic estimates.  But they justify my prior assertion that a 100 dB dynamic range is a reasonable expectation for DSOs with deep memory.  DSOs with shallower memory can always combine multiple records.  The 8753 series VNAs have a dynamic range of around 100 dB. So except for being limited by the BW of the DSO comparable performance should be achievable

The first order issue is jitter which is why the phase corrections I outlined previously.  Without those summing will not provide an increase in dynamic range.  As I outlined previously, one has to do the same calibrations as would be done with a VNA.  While I would not characterize it as trivial, the required mathematical operations are well known and not difficult.

Doing VNA and SA by TDR is treated in several application notes here:

https://kh6htv.com/pspl-app-notes/

Many of the 8753 series offer TDR as an option.  It's just a Fourier transform.  Doing the opposite is also just a Fourier transform.

Applying phase corrections to permit summing samples is routine in seismic processing.  Without it, it would be impossible to collect usable data in most circumstances. The signals are simply too weak relative to the ambient noise.  Achieving 100 dB or better improvements in S/N are routine.  There are a wide variety of types of noise and methods for suppressing them.  IMHO the biggest issue is the question of whether the software is correct.  I have demonstrated major errors in widely used commercial software.  The most difficult part is coming up with suitable test cases.
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 8649
  • Country: us
  • DavidH
Re: Using a fast pulse and DSO for vector network analysis
« Reply #38 on: March 17, 2019, 11:14:55 am »
Couldn't the dynamic range be improved by aggregating multiple measurements ? You don't need to do single shot capture and, by the same token, you don't need real-time capture. So by taking longer over the measurements (and a traditional VNA would have to do a sweep) you can improve the accuracy.

Sure, but a VNA has a greater inherent dynamic range because of the integrated measurement over a very narrow bandwidth.  An oscilloscope or other time domain instrument has a wide input bandwidth and short measurement time from essentially sampling.

Another problem is the non-linearity in the pulse and vertical amplifier chain.  The best instruments calibrate this out but it is not trivial to do.  HP/Agilent made some sampling based VNAs which were basically sampling oscilloscopes optimized to operate as VNAs and I think they may do this.

An Owon XDS2102A will collect 20 Mpts of 12 bit data at 500 MSa/s.

But the samples include 100MHz (actually 160MHz) worth of front end noise limiting dynamic range, the 100MHz front end does not even have close to 12 bits of linearity limiting spurious free dynamic range, and the transient response is also significantly worse than 12 bits which has the same effect as corrupting the pulse edge.
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2286
  • Country: us
Re: Using a fast pulse and DSO for vector network analysis
« Reply #39 on: March 17, 2019, 01:44:45 pm »
Never said it was easy.  Just that I understand how to do it.
 

Offline _Wim_

  • Frequent Contributor
  • **
  • Posts: 677
  • Country: be
Re: Using a fast pulse and DSO for vector network analysis
« Reply #40 on: March 17, 2019, 06:18:07 pm »
Perhaps I have confused myself by looking at my 1502, which is certainly better than the 140ps spec. The corresponding 1502B spec is 200ps.

I was more looking at the comparison made by Mohr for this (http://www.mohr-engineering.com/tektronix-1502-comparison.php), were for the 1502B 130ps is stated, but this is the rise-time between 20-80%, and 10-90% is indeed stated <200ps and 140ps for the 1502. 

It would be nice to measure the actual rise time, because knowing Tekronix, these numbers are probably still conservative. If I come across and affordable 11801, I now have a "reason" to buy one!  ;)

 

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 8402
  • Country: gb
    • Having fun doing more, with less
Re: Using a fast pulse and DSO for vector network analysis
« Reply #41 on: March 17, 2019, 08:37:00 pm »
Perhaps I have confused myself by looking at my 1502, which is certainly better than the 140ps spec. The corresponding 1502B spec is 200ps.

I was more looking at the comparison made by Mohr for this (http://www.mohr-engineering.com/tektronix-1502-comparison.php), were for the 1502B 130ps is stated, but this is the rise-time between 20-80%, and 10-90% is indeed stated <200ps and 140ps for the 1502. 

It would be nice to measure the actual rise time, because knowing Tekronix, these numbers are probably still conservative. If I come across and affordable 11801, I now have a "reason" to buy one!  ;)

It would indeed be pleasant to see definitive measurements, and preferably example traces of known discontinuities such as a couple of back-to-back BNC T-pieces in the middle of a cable.

Apart from that, I haven't paid too much attention to new TDRs, since I'm never going to buy one.
There are lies, damned lies, statistics - and ADC/DAC specs.
Gliding aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2286
  • Country: us
Re: Using a fast pulse and DSO for vector network analysis
« Reply #42 on: March 17, 2019, 11:23:31 pm »

It would indeed be pleasant to see definitive measurements, and preferably example traces of known discontinuities such as a couple of back-to-back BNC T-pieces in the middle of a cable.

Apart from that, I haven't paid too much attention to new TDRs, since I'm never going to buy one.

I'm planning to compare such measurements made with a DSO and Leo's pulser to the same measurement made with my 8753B once I get my APC-7 to N adapters and learn how to use the 8753B.

When that happens I'll be posting time and frequency domain plots of each measurement and the ratio of the two in both domains.

I'll post some screen shots of various connectors using a 200 MHz DSO and Leo's 100 ps impulse generator later today.  I've got a bin of reject devices. which I'm going to spray with red paint and keep for such comparisons.
 

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 8402
  • Country: gb
    • Having fun doing more, with less
Re: Using a fast pulse and DSO for vector network analysis
« Reply #43 on: March 18, 2019, 01:53:02 am »
I, and I expect others, will be interested to see those plots.

A poor man's TDR can be made with a noise source and a SDR dongle. The main disadvantages are the limited dynamic range and that it is effectively a scalar network analyser. You get less information, but nonetheless the information can be useful in terms of indicating presence/absence of problems.
There are lies, damned lies, statistics - and ADC/DAC specs.
Gliding aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 8649
  • Country: us
  • DavidH
Re: Using a fast pulse and DSO for vector network analysis
« Reply #44 on: March 18, 2019, 02:40:48 am »
Perhaps I have confused myself by looking at my 1502, which is certainly better than the 140ps spec. The corresponding 1502B spec is 200ps.

I was more looking at the comparison made by Mohr for this (http://www.mohr-engineering.com/tektronix-1502-comparison.php), were for the 1502B 130ps is stated, but this is the rise-time between 20-80%, and 10-90% is indeed stated <200ps and 140ps for the 1502. 

It would be nice to measure the actual rise time, because knowing Tekronix, these numbers are probably still conservative. If I come across and affordable 11801, I now have a "reason" to buy one!  ;)

The earlier 1502 used a tunnel diode and the faster edge gave it better resolution.  There is a trade-off between resolution and range for a given transition time so the fastest transitions were not always used.

I do not have an S-6 TDR sampling head for my 7T11 but I messed around using an S-4 and it is amazing what can be seen even with only a 500 picosecond edge.
 

Offline _Wim_

  • Frequent Contributor
  • **
  • Posts: 677
  • Country: be
Re: Using a fast pulse and DSO for vector network analysis
« Reply #45 on: March 18, 2019, 02:52:30 am »
it is amazing what can be seen even with only a 500 picosecond edge.

I can imagine. TDR is much more useful than you would initially think. I copied this into my Onenote during reading of a tektronix appnote (forgot exactly which one), as a reminder how to interprete the results from the 1502B

Edit: found the app note:
www.tek.com/dl/55W_14601_2.pdf
« Last Edit: March 18, 2019, 03:00:06 am by _Wim_ »
 

Offline _Wim_

  • Frequent Contributor
  • **
  • Posts: 677
  • Country: be
Re: Using a fast pulse and DSO for vector network analysis
« Reply #46 on: March 18, 2019, 03:07:34 am »
The earlier 1502 used a tunnel diode and the faster edge gave it better resolution.

I think the TekWiki is a bit incorrect when the following statement is made:

The other major difference between versions is the line charging method. The 1502 uses a fast (36 ps) tunnel diode pulser, the later models used a half sine wave to charge the line. The TD pulser, with its Dirac delta edge, gives much better short range sensitivity, although it is much easier to destroy. 


The 1503B/C do indeed use a half sine wave to charge the line, but the 1502B/C still use a step rise function.

 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2286
  • Country: us
Re: Using a fast pulse and DSO for vector network analysis
« Reply #47 on: March 18, 2019, 10:33:19 am »
I, and I expect others, will be interested to see those plots.

A poor man's TDR can be made with a noise source and a SDR dongle. The main disadvantages are the limited dynamic range and that it is effectively a scalar network analyser. You get less information, but nonetheless the information can be useful in terms of indicating presence/absence of problems.

The noise source does not allow you to align and sum so you are limited to the dynamic range of the ADC.  There's no opportunity for processing gain.  Thirty years ago a noise bridge was a popular method for tuning antennas.  But even a poor quality noise source and SDR is better.

I was out all day attending a Celtic music concert.  I'll do some connector reflection plots tomorrow if nothing intervenes.

The DSO & pulser comparison to the VNA will take a lot longer.  I've got to write and test software and learn to use the VNA.  And parse proprietary DSO data formats :-(
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 8649
  • Country: us
  • DavidH
Re: Using a fast pulse and DSO for vector network analysis
« Reply #48 on: March 18, 2019, 11:39:36 am »
The 1503B/C do indeed use a half sine wave to charge the line, but the 1502B/C still use a step rise function.

I have read the 1503 and 1503C service manual in detail and they do *not* use a step rise function in the way it is normally thought of.  Selected LC networks provide different half sine drives to a class-b or class-c output transistor which produces a variable width pulse instead of the step function of a tunnel diode.  This is the "long line" design with less resolution.

The various "short line" 1502 models use a fast edge with a long duration instead and I think they all use tunnel diodes but that is not clear because the later models do not describe the inside of the pulse generator hybrid in detail.  Tektronix had some hybrids by this point which could use transistors to generate better than 200 picosecond edges so this technology might have been used instead of tunnel diodes in the later 1502 models but the documentation is explicit that a step edge was used in the 1502 models.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf