Discussion:
internet outages
(too old to reply)
Jude DaShiell
2018-12-22 15:20:01 UTC
Permalink
Has Linux got tools that can run while a computer runs that can poll
several sites and log internet outages? I figure a minute down time is a
failure and have experienced several of these where my wifi connection had
to be deactivated and reactivated to have the internet connection
restored. This is a new wifi router too. The log would be sent into
comcast along with payment requesting credits for the down times.



--
Dominik George
2018-12-22 16:00:02 UTC
Permalink
Here is a script that you can run from a cron job which will log
Internet status and store
#! /bin/bash
date >> ~/net-test.txt
ping -c 1 google.com >> ~/net-test.txt
#end of file
smokeping is a tool made for this.

-nik
m***@neidorff.com
2018-12-22 16:00:02 UTC
Permalink
Post by Jude DaShiell
Has Linux got tools that can run while a computer runs that can poll
several sites and log internet outages? I figure a minute down time is a
failure and have experienced several of these where my wifi connection had
to be deactivated and reactivated to have the internet connection
restored. This is a new wifi router too. The log would be sent into
comcast along with payment requesting credits for the down times.
--
From your description of the problem, it sounds like it is the router that is not doing its job
properly. Before you send your request for credit, make sure that it is a Comcast problem.

But, if you want to jost test when the net is down:

Here is a script that you can run from a cron job which will log Internet status and store
the results into a file in your home folder called net-test.txt:

#! /bin/bash

date >> ~/net-test.txt
ping -c 1 google.com >> ~/net-test.txt

#end of file
Jude DaShiell
2018-12-22 16:40:01 UTC
Permalink
Date: Sat, 22 Dec 2018 10:54:06
Subject: Re: internet outages
Resent-Date: Sat, 22 Dec 2018 15:54:23 +0000 (UTC)
Post by Jude DaShiell
Has Linux got tools that can run while a computer runs that can poll
several sites and log internet outages? I figure a minute down time is a
failure and have experienced several of these where my wifi connection had
to be deactivated and reactivated to have the internet connection
restored. This is a new wifi router too. The log would be sent into
comcast along with payment requesting credits for the down times.
--
From your description of the problem, it sounds like it is the router that is not doing its job
properly. Before you send your request for credit, make sure that it is a Comcast problem.
Here is a script that you can run from a cron job which will log Internet status and store
#! /bin/bash
date >> ~/net-test.txt
ping -c 1 google.com >> ~/net-test.txt
#end of file
Thanks for the script and the tool recommendation.
Has Linux got a tool to check up on the router and find out if the
router is doing its job?
Before I mail logs into comcast, I want to make sure I've done all due
dilligence on this end so if comcast isn't having a problem they don't
catch any undeserved heat. A router replacement can be done if that's
the source of these problems. On more than one of these outage
occassions I have used a stylus and rebooted the router to clear any
potential malware just in case.

--
Mark Neidorff
2018-12-22 18:40:02 UTC
Permalink
Post by Jude DaShiell
Thanks for the script and the tool recommendation.
Has Linux got a tool to check up on the router and find out if the
router is doing its job?
Before I mail logs into comcast, I want to make sure I've done all due
dilligence on this end so if comcast isn't having a problem they don't
catch any undeserved heat. A router replacement can be done if that's
the source of these problems. On more than one of these outage
occassions I have used a stylus and rebooted the router to clear any
potential malware just in case.
Rebooting the router will not clear any malware that is in NVRAM. That is Comcast's job to
diagnose and fix.

IMO, your due dilligence consists of testing things from the router to your PCs. So, the list
of suspects is: router, inside network wiring, insdie the building electrical wiring and the
pc. Most likely point of failure is the router in this case. You can also try jiggling ethernet
wires to see if you can find a problem there.

Mark
John Hasler
2018-12-22 17:30:10 UTC
Permalink
Post by Jude DaShiell
From your description of the problem, it sounds like it is the router
that is not doing its job properly.
This is a Comcast-supplied combination router-modem-ap? From everything
I've read and been told by people who know from experience the firmware
in those is always buggy and insecure (the supplier always has a
backdoor, for example). In my experience even simple DSL modems with no
wifi are unstable when you let them try to be routers. Comcast may be
willing to swap it for a different one (probably one they got back from
another unhappy customer but never tested) but it's not likely to be
better. Put the thing in bridge mode and put a real router behind it.
Otherwise you are entrusting the security of your LAN to Comcast.
--
John Hasler
***@newsguy.com
Elmwood, WI USA
john doe
2018-12-22 18:40:02 UTC
Permalink
Post by John Hasler
Post by Jude DaShiell
From your description of the problem, it sounds like it is the router
that is not doing its job properly.
This is a Comcast-supplied combination router-modem-ap? From everything
I've read and been told by people who know from experience the firmware
in those is always buggy and insecure (the supplier always has a
backdoor, for example). In my experience even simple DSL modems with no
wifi are unstable when you let them try to be routers. Comcast may be
willing to swap it for a different one (probably one they got back from
another unhappy customer but never tested) but it's not likely to be
better. Put the thing in bridge mode and put a real router behind it.
Otherwise you are entrusting the security of your LAN to Comcast.
Or if you can get a modem only and use your own router (1).


One easy way to determine if the wifi is at fault would be to be wiredly
connected to the router.


https://20somethingfinance.com/how-to-replace-a-comcast-modem-with-your-own/
--
John Doe
Jude DaShiell
2018-12-22 20:00:01 UTC
Permalink
Date: Sat, 22 Dec 2018 13:32:57
Subject: Re: internet outages
Resent-Date: Sat, 22 Dec 2018 18:33:16 +0000 (UTC)
Post by John Hasler
Post by Jude DaShiell
From your description of the problem, it sounds like it is the router
that is not doing its job properly.
This is a Comcast-supplied combination router-modem-ap? From everything
I've read and been told by people who know from experience the firmware
in those is always buggy and insecure (the supplier always has a
backdoor, for example). In my experience even simple DSL modems with no
wifi are unstable when you let them try to be routers. Comcast may be
willing to swap it for a different one (probably one they got back from
another unhappy customer but never tested) but it's not likely to be
better. Put the thing in bridge mode and put a real router behind it.
Otherwise you are entrusting the security of your LAN to Comcast.
Or if you can get a modem only and use your own router (1).
One easy way to determine if the wifi is at fault would be to be wiredly
connected to the router.
https://20somethingfinance.com/how-to-replace-a-comcast-modem-with-your-own/
Ah, direct ethernet connection. That's doable temporarily. Thanks for
the url too.


--
Nate Bargmann
2018-12-22 18:40:02 UTC
Permalink
This.

I have been very pleased with OpenWRT over the past several years. The
WISP wanted to setup their Ubiquity Nanobridge CPE as a router and were
willing to configure it as a bridge instead. That allows me remote SSH
access through my OpenWRT router. This really is the way to do it to
give yourself the most control.

- Nate
--
"The optimist proclaims that we live in the best of all
possible worlds. The pessimist fears this is true."

Web: http://www.n0nb.us GPG key: D55A8819 GitHub: N0NB
David Christensen
2018-12-22 20:20:01 UTC
Permalink
Post by Jude DaShiell
Has Linux got tools that can run while a computer runs that can poll
several sites and log internet outages? I figure a minute down time is a
failure and have experienced several of these where my wifi connection had
to be deactivated and reactivated to have the internet connection
restored. This is a new wifi router too. The log would be sent into
comcast along with payment requesting credits for the down times.
I suggest:

1. Test and verify your network cables:

https://www.idealnetworks.net/us/en/products/cable-testing/copper-testing/cable-verifiers-linkmaster.aspx

2. Buy and read "Networking for System Administrators" by Lucas:

https://mwl.io/nonfiction/networking#n4sa

3. Write a shell script that invokes some standard utility (ping(8),
wget(1), etc.), appends a timestamp and the utility output to a log
file, sleeps for 60 seconds, and then repeats. Put the script on two
hosts on opposite sides of the router, configure the router so that the
hosts can see each other, and have the hosts monitor each other.

4. Keep a plain text log file of all your host and network
administration activities. Timestamp your entries so that you can
correlate them to the monitoring logs.


David
john doe
2018-12-23 06:10:01 UTC
Permalink
Post by David Christensen
Post by Jude DaShiell
Has Linux got tools that can run while a computer runs that can poll
several sites and log internet outages?  I figure a minute down time is a
failure and have experienced several of these where my wifi connection had
to be deactivated and reactivated to have the internet connection
restored.  This is a new wifi router too.  The log would be sent into
comcast along with payment requesting credits for the down times.
https://www.idealnetworks.net/us/en/products/cable-testing/copper-testing/cable-verifiers-linkmaster.aspx
    https://mwl.io/nonfiction/networking#n4sa
3.  Write a shell script that invokes some standard utility (ping(8),
wget(1), etc.), appends a timestamp and the utility output to a log
file, sleeps for 60 seconds, and then repeats.  >
Along with redirecting the output to a file, the logger(1) utility can
also be used (1) to make entry in the system log.

http://man7.org/linux/man-pages/man1/logger.1.html
--
John Doe
David Christensen
2018-12-23 07:30:02 UTC
Permalink
Post by john doe
Post by David Christensen
Post by Jude DaShiell
Has Linux got tools that can run while a computer runs that can poll
several sites and log internet outages?  I figure a minute down time is a
failure and have experienced several of these where my wifi connection had
to be deactivated and reactivated to have the internet connection
restored.  This is a new wifi router too.  The log would be sent into
comcast along with payment requesting credits for the down times.
https://www.idealnetworks.net/us/en/products/cable-testing/copper-testing/cable-verifiers-linkmaster.aspx
    https://mwl.io/nonfiction/networking#n4sa
3.  Write a shell script that invokes some standard utility (ping(8),
wget(1), etc.), appends a timestamp and the utility output to a log
file, sleeps for 60 seconds, and then repeats.  >
Along with redirecting the output to a file, the logger(1) utility can
also be used (1) to make entry in the system log.
http://man7.org/linux/man-pages/man1/logger.1.html
Yes -- that would be a possible follow-on exercise to #3. Also see
rsyslog(8) and rsyslog.conf(5).


David
Curt
2018-12-23 10:10:02 UTC
Permalink
Post by David Christensen
Post by Jude DaShiell
Has Linux got tools that can run while a computer runs that can poll
several sites and log internet outages? I figure a minute down time is a
failure and have experienced several of these where my wifi connection had
to be deactivated and reactivated to have the internet connection
restored. This is a new wifi router too. The log would be sent into
comcast along with payment requesting credits for the down times.
https://www.idealnetworks.net/us/en/products/cable-testing/copper-testing/cable-verifiers-linkmaster.aspx
I think in my particular home situation I'd rather just replace my 10
dollar ethernet cable with a new one (or swap in a new one to see if it
makes some appreciable difference) than purchase an $80 device to test
whether it's failing or not (in which former case I'd then be out $90,
which is a real nice load of money).
David Christensen
2018-12-23 20:00:01 UTC
Permalink
Post by Curt
Post by David Christensen
Post by Jude DaShiell
Has Linux got tools that can run while a computer runs that can poll
several sites and log internet outages? I figure a minute down time is a
failure and have experienced several of these where my wifi connection had
to be deactivated and reactivated to have the internet connection
restored. This is a new wifi router too. The log would be sent into
comcast along with payment requesting credits for the down times.
https://www.idealnetworks.net/us/en/products/cable-testing/copper-testing/cable-verifiers-linkmaster.aspx
I think in my particular home situation I'd rather just replace my 10
dollar ethernet cable with a new one (or swap in a new one to see if it
makes some appreciable difference) than purchase an $80 device to test
whether it's failing or not (in which former case I'd then be out $90,
which is a real nice load of money).
If you own a continuity tester or multimeter, you could build a pair of
RJ-45 jacks with pigtails for a few dollars and use those to test your
cables.


David
r***@gmail.com
2018-12-23 20:20:01 UTC
Permalink
Post by David Christensen
Post by Curt
Post by David Christensen
https://www.idealnetworks.net/us/en/products/cable-testing/copper-testin
g/cable-verifiers-linkmaster.aspx
I think in my particular home situation I'd rather just replace my 10
dollar ethernet cable with a new one (or swap in a new one to see if it
makes some appreciable difference) than purchase an $80 device to test
whether it's failing or not (in which former case I'd then be out $90,
which is a real nice load of money).
If you own a continuity tester or multimeter, you could build a pair of
RJ-45 jacks with pigtails for a few dollars and use those to test your
cables.
Wow, for $80 I would have expected something that would measure dB loss and
maybe even a time domain reflectometer (well, maybe not that).

I have seen those continuity type testers on ebay for under $10 (the test sets
with two devices, both with RJ-45 jacks, you plug the ends of the cable into
those (one at each end of the cable), and then the "master device" tells you
whether the pairs are connected properly and there is continuity).
David Christensen
2018-12-23 21:30:02 UTC
Permalink
Post by r***@gmail.com
Post by David Christensen
Post by Curt
Post by David Christensen
https://www.idealnetworks.net/us/en/products/cable-testing/copper-testin
g/cable-verifiers-linkmaster.aspx
I think in my particular home situation I'd rather just replace my 10
dollar ethernet cable with a new one (or swap in a new one to see if it
makes some appreciable difference) than purchase an $80 device to test
whether it's failing or not (in which former case I'd then be out $90,
which is a real nice load of money).
If you own a continuity tester or multimeter, you could build a pair of
RJ-45 jacks with pigtails for a few dollars and use those to test your
cables.
Wow, for $80 I would have expected something that would measure dB loss and
maybe even a time domain reflectometer (well, maybe not that).
Please provide a URL for such test instrument(s).


One other point -- MRSP is not the same a street price:

https://www.homedepot.com/p/Ideal-LinkMaster-UTP-STP-Wiremapper-and-Tester-62-200/100091453
Post by r***@gmail.com
I have seen those continuity type testers on ebay for under $10 (the test sets
with two devices, both with RJ-45 jacks, you plug the ends of the cable into
those (one at each end of the cable), and then the "master device" tells you
whether the pairs are connected properly and there is continuity).
I am sure there are cheap clones and/or counterfeits of the LinkMaster
available. I prefer genuine, reputable brands.


David
r***@gmail.com
2018-12-23 21:50:01 UTC
Permalink
Post by David Christensen
Post by r***@gmail.com
Wow, for $80 I would have expected something that would measure dB loss
and maybe even a time domain reflectometer (well, maybe not that).
Please provide a URL for such test instrument(s).
Oh, I misread that, I thought you wanted a URL for the under $10 device. Try
googling for time domain reflectometer, I don't think you'll find one for even
near $80 (but I'd love to be wrong) -- maybe a cable loss tester at that
price.

Some URLs for under $10 testers (prices have changed ;-) (Prices are a little
over $3)

* [[https://www.ebay.com/itm/Pocket-LED-Ethernet-4-Port-RJ45-RJ11-Cat5-
Network-LAN-Cable-Tester-Keychain-
G6Z5/253214577684?epid=675204148&hash=item3af4c3d414:g:EIsAAOSwuLZY3IhA:rk:1:pf:0]]

From a review of that one:

<quote>
Excellent cable tester!

Plug in both ends and watch for the lights. It's as easy as that. Great for
testing Ethernet cables to rule them out as the issue.
Requires 1x 9v battery.
</quote>

* [[https://www.ebay.com/itm/WZ-468-RJ45-Ethernet-Network-RJ11-Cable-
Tester-Crimper-Punch-Down-Tool-Kit-
red/183476025283?epid=20024613296&hash=item2ab805f7c3:g:n0IAAOSw2JxboQgU:rk:2:pf:0]]
Post by David Christensen
https://www.homedepot.com/p/Ideal-LinkMaster-UTP-STP-Wiremapper-and-Tester-
62-200/100091453
Post by r***@gmail.com
I have seen those continuity type testers on ebay for under $10 (the test
sets with two devices, both with RJ-45 jacks, you plug the ends of the
cable into those (one at each end of the cable), and then the "master
device" tells you whether the pairs are connected properly and there is
continuity).
I am sure there are cheap clones and/or counterfeits of the LinkMaster
available. I prefer genuine, reputable brands.
David
Joe
2018-12-23 22:00:02 UTC
Permalink
On Sun, 23 Dec 2018 16:46:43 -0500
Post by r***@gmail.com
Oh, I misread that, I thought you wanted a URL for the under $10
device. Try googling for time domain reflectometer, I don't think
you'll find one for even near $80 (but I'd love to be wrong) -- maybe
a cable loss tester at that price.
Some URLs for under $10 testers (prices have changed ;-) (Prices are
a little over $3)
*
[[https://www.ebay.com/itm/Pocket-LED-Ethernet-4-Port-RJ45-RJ11-Cat5-
Network-LAN-Cable-Tester-Keychain-
G6Z5/253214577684?epid=675204148&hash=item3af4c3d414:g:EIsAAOSwuLZY3IhA:rk:1:pf:0]]
<quote>
Excellent cable tester!
Plug in both ends and watch for the lights. It's as easy as that.
Great for testing Ethernet cables to rule them out as the issue.
Requires 1x 9v battery.
</quote>
I have one, but they're not great at testing for intermittents while
you wiggle the cable. But one with lights constantly illuminated isn't
good for spotting incorrect wiring or shorts, so I'm currently making
one with switchable light cycle speed, which doesn't seem to be
commercially available.
--
Joe
Doug
2018-12-24 00:40:02 UTC
Permalink
Post by David Christensen
Post by r***@gmail.com
Post by David Christensen
Post by Curt
Post by David Christensen
https://www.idealnetworks.net/us/en/products/cable-testing/copper-testin
g/cable-verifiers-linkmaster.aspx
I think in my particular home situation I'd rather just replace my 10
dollar ethernet cable with a new one (or swap in a new one to see if it
makes some appreciable difference) than purchase an $80 device to test
whether it's failing or not (in which former case I'd then be out $90,
which is a real nice load of money).
If you own a continuity tester or multimeter, you could build a pair of
RJ-45 jacks with pigtails for a few dollars and use those to test your
cables.
Wow, for $80 I would have expected something that would measure dB loss and
maybe even a time domain reflectometer (well, maybe not that).
Please provide a URL for such test instrument(s).
https://www.homedepot.com/p/Ideal-LinkMaster-UTP-STP-Wiremapper-and-Tester-62-200/100091453
Post by r***@gmail.com
I have seen those continuity type testers on ebay for under $10 (the test sets
with two devices, both with RJ-45 jacks, you plug the ends of the cable into
those (one at each end of the cable), and then the "master device" tells you
whether the pairs are connected properly and there is continuity).
I am sure there are cheap clones and/or counterfeits of the LinkMaster
available. I prefer genuine, reputable brands.
David
RE: Time Domain Reflectometer:

Theoretically, you can build your own with a fast pulse generator and an
oscilloscope. The trick is, you need a REALLY FAST oscilloscope! The
pulse generator is easy, just a couple of transistors, maybe a diode.
The circuit is probably in every edition of the Radio Amateur's
Handbook. The scope is expensive. If you don't have at least a 1GHz
digital sampling scope, don't bother!

--doug, WA2SAY, retired RF engineer
Gene Heskett
2018-12-24 01:20:02 UTC
Permalink
Post by Doug
Post by David Christensen
Post by r***@gmail.com
Post by David Christensen
Post by Curt
Post by David Christensen
https://www.idealnetworks.net/us/en/products/cable-testing/coppe
r-testin
g/cable-verifiers-linkmaster.aspx
I think in my particular home situation I'd rather just replace
my 10 dollar ethernet cable with a new one (or swap in a new one
to see if it
makes some appreciable difference) than purchase an $80 device to
test whether it's failing or not (in which former case I'd then
be out $90, which is a real nice load of money).
If you own a continuity tester or multimeter, you could build a
pair of RJ-45 jacks with pigtails for a few dollars and use those
to test your cables.
Wow, for $80 I would have expected something that would measure dB loss and
maybe even a time domain reflectometer (well, maybe not that).
Please provide a URL for such test instrument(s).
https://www.homedepot.com/p/Ideal-LinkMaster-UTP-STP-Wiremapper-and-
Tester-62-200/100091453
Post by r***@gmail.com
I have seen those continuity type testers on ebay for under $10 (the test sets
with two devices, both with RJ-45 jacks, you plug the ends of the cable into
those (one at each end of the cable), and then the "master device" tells you
whether the pairs are connected properly and there is continuity).
I am sure there are cheap clones and/or counterfeits of the
LinkMaster available. I prefer genuine, reputable brands.
David
Theoretically, you can build your own with a fast pulse generator and
an oscilloscope. The trick is, you need a REALLY FAST oscilloscope!
The pulse generator is easy, just a couple of transistors, maybe a
diode. The circuit is probably in every edition of the Radio
Amateur's Handbook. The scope is expensive. If you don't have at least
a 1GHz digital sampling scope, don't bother!
--doug, WA2SAY, retired RF engineer
Doug, another retired tv CE here. And I won't make the statement that its
not worth the bother. In a pinch I've made one from a radio shack ttl
pulse generator and a 100 mhz scope, and was close enough that when I
sent a tower crew up to dis the transmission line & find a burnout, they
found the first fried teflon disk, and a bad bullet cobbled up
with "whisky cups" at the first joint they opened up. Plenty close
enough for the girls I go with. Part of the secret is of course the
teflon disked 3.125" line has a prop speed just above 98% of C, while
the sterate versions are only about 97.25. You see EVERTTHING with that,
even bullet dents from some kid with a Christmas 22. Still, I wasn't
able to count the 20' joints with that slow ttl signal and scope. That
takes the real thing, and even then you might have to wait till the
middle of the night to get rid of the local daytime AMers that can sure
screw up your display. The tek version uses a tunnel diode, which has a
femtosecond switch time, but is only about a half volt switch.

Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
r***@gmail.com
2018-12-24 01:30:04 UTC
Permalink
Post by Doug
Theoretically, you can build your own with a fast pulse generator and an
oscilloscope. The trick is, you need a REALLY FAST oscilloscope! The
pulse generator is easy, just a couple of transistors, maybe a diode.
The circuit is probably in every edition of the Radio Amateur's
Handbook. The scope is expensive. If you don't have at least a 1GHz
digital sampling scope, don't bother!
--doug, WA2SAY, retired RF engineer
Hmm, with CPU clocks hitting 4 GHz, I wonder how expensive an ADC converter to
work at corresponding speeds would be? (Just an idle question ;-)
Gene Heskett
2018-12-24 02:10:02 UTC
Permalink
Post by r***@gmail.com
Post by Doug
Theoretically, you can build your own with a fast pulse generator
and an oscilloscope. The trick is, you need a REALLY FAST
oscilloscope! The pulse generator is easy, just a couple of
transistors, maybe a diode. The circuit is probably in every
edition of the Radio Amateur's Handbook. The scope is expensive. If
you don't have at least a 1GHz digital sampling scope, don't bother!
--doug, WA2SAY, retired RF engineer
Hmm, with CPU clocks hitting 4 GHz, I wonder how expensive an ADC
converter to work at corresponding speeds would be? (Just an idle
question ;-)
If you have to ask, you can't afford it. Bring outrageous sums of money
This is typically done by every trick tek knows about fast analogue
circuitry, in the best units with long vertical deflection plates in a
custom made crt with teeny delay lines between the sections of the
plates so the signal is virtually traveling toward the screen at the
same speed as the electron beam is traveling. And its moving fast enough
at 22,000 volts, relativity can and does get in the way. Such scopes put
the plates so close to the beam that the beam is intercepted by striking
the plates at just a hair over 4 cm high, 2cm from the horizontal center
line. The only one of those I ever saw was in the early 1980's or so at
the NAB show in Vegas, and it was well into a 5 digit asking price then.

Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
r***@gmail.com
2018-12-24 03:30:01 UTC
Permalink
Post by Gene Heskett
Post by r***@gmail.com
Post by Doug
Theoretically, you can build your own with a fast pulse generator
and an oscilloscope. The trick is, you need a REALLY FAST
oscilloscope! The pulse generator is easy, just a couple of
transistors, maybe a diode. The circuit is probably in every
edition of the Radio Amateur's Handbook. The scope is expensive. If
you don't have at least a 1GHz digital sampling scope, don't bother!
--doug, WA2SAY, retired RF engineer
Hmm, with CPU clocks hitting 4 GHz, I wonder how expensive an ADC
converter to work at corresponding speeds would be? (Just an idle
question ;-)
If you have to ask, you can't afford it. Bring outrageous sums of money
This is typically done by every trick tek knows about fast analogue
circuitry, in the best units with long vertical deflection plates in a
custom made crt with teeny delay lines between the sections of the
plates so the signal is virtually traveling toward the screen at the
same speed as the electron beam is traveling. And its moving fast enough
at 22,000 volts, relativity can and does get in the way. Such scopes put
the plates so close to the beam that the beam is intercepted by striking
the plates at just a hair over 4 cm high, 2cm from the horizontal center
line. The only one of those I ever saw was in the early 1980's or so at
the NAB show in Vegas, and it was well into a 5 digit asking price then.
Yeah, but I wasn't thinking about actually using a CRT to display the signal
in real time, but, instead, collect samples (at maybe 4 GHz??), store them,
and then display them as a static display.

???
Gene Heskett
2018-12-24 04:20:01 UTC
Permalink
Post by r***@gmail.com
Post by Gene Heskett
Post by r***@gmail.com
Post by Doug
Theoretically, you can build your own with a fast pulse
generator and an oscilloscope. The trick is, you need a REALLY
FAST oscilloscope! The pulse generator is easy, just a couple
of transistors, maybe a diode. The circuit is probably in every
edition of the Radio Amateur's Handbook. The scope is expensive.
If you don't have at least a 1GHz digital sampling scope, don't
bother!
--doug, WA2SAY, retired RF engineer
Hmm, with CPU clocks hitting 4 GHz, I wonder how expensive an ADC
converter to work at corresponding speeds would be? (Just an idle
question ;-)
If you have to ask, you can't afford it. Bring outrageous sums of
money This is typically done by every trick tek knows about fast
analogue circuitry, in the best units with long vertical deflection
plates in a custom made crt with teeny delay lines between the
sections of the plates so the signal is virtually traveling toward
the screen at the same speed as the electron beam is traveling. And
its moving fast enough at 22,000 volts, relativity can and does get
in the way. Such scopes put the plates so close to the beam that the
beam is intercepted by striking the plates at just a hair over 4 cm
high, 2cm from the horizontal center line. The only one of those I
ever saw was in the early 1980's or so at the NAB show in Vegas, and
it was well into a 5 digit asking price then.
Yeah, but I wasn't thinking about actually using a CRT to display the
signal in real time, but, instead, collect samples (at maybe 4 GHz??),
store them, and then display them as a static display.
???
Thats essentially what our high bandwidth scopes do today, $400 gets you
a 1ghz sampler with an effective bandwidth of 200 mhz. I've got one, the
nice thing is that because it is digital, a once a second glitch in a 42
megabaud seriel data train stands out like a sore thumb because its not
limited to the screen writing speed of an analogue scope. So its just as
bright as the main signal that doesn't have the glitch. The operating
software is buggier than a 10 day old road kill in August though. I have
both. My analogue Hitachi is actually good to around 200 mhz.

Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
John Hasler
2018-12-24 05:00:01 UTC
Permalink
Post by r***@gmail.com
Yeah, but I wasn't thinking about actually using a CRT to display the
signal in real time, but, instead, collect samples (at maybe 4 GHz??),
store them, and then display them as a static display.
At that sort of frequency sampling scopes (including the old crt ones)
sample at far below the signal repitition rate.
--
John Hasler
***@newsguy.com
Elmwood, WI USA
Joe
2018-12-24 09:30:02 UTC
Permalink
On Sun, 23 Dec 2018 22:54:20 -0600
Post by John Hasler
Post by r***@gmail.com
Yeah, but I wasn't thinking about actually using a CRT to display
the signal in real time, but, instead, collect samples (at maybe 4
GHz??), store them, and then display them as a static display.
At that sort of frequency sampling scopes (including the old crt ones)
sample at far below the signal repitition rate.
Yes, they rely on the signal being repetitive. To see a single
transient, you want at least five samples in the relevant period,
preferably more.

There is normally a real Nyquist bandwidth quoted somewhere in the
small print. I used to repair and calibrate Hitachi scopes for a while,
until everyone stopped using them.
--
Joe
Gene Heskett
2018-12-24 13:10:02 UTC
Permalink
Post by Joe
On Sun, 23 Dec 2018 22:54:20 -0600
Post by John Hasler
Post by r***@gmail.com
Yeah, but I wasn't thinking about actually using a CRT to display
the signal in real time, but, instead, collect samples (at maybe 4
GHz??), store them, and then display them as a static display.
At that sort of frequency sampling scopes (including the old crt
ones) sample at far below the signal repitition rate.
Yes, they rely on the signal being repetitive. To see a single
transient, you want at least five samples in the relevant period,
preferably more.
There is normally a real Nyquist bandwidth quoted somewhere in the
small print. I used to repair and calibrate Hitachi scopes for a
while, until everyone stopped using them.
I wonder why, Joe. My V-1065 still works well enough to measure frequency
at a 1% accuracy, and its now pushing 35yo. In a pinch I've looked at
the output sample of an old analogue tv transmitter to adjust the
modulation depth when the monitor was bonkers, it was actually usable at
180 mhz.

Their absolute top of the line V-1085, 200 mhz had a totally duff
triggering circuit in it so I sent 2 of it back, but its baby brother, a
4 trace 100 mhz was an absolute doll, and the only analogue scope that
would let you do the final calibration after replacing the head wheel on
a Panasonic DVC-PRO broadcast vtr. At $1500-2000+ a copy, I've done that
about 100,000 dollars worth of that. That and replaceing the tiny
electrolytic caps by the 3 lb coffee can full. 3 of those IIRC before I
retired in 2002. Failure rate on those in the dvc-pro's was near 100% in
2 or 3 years. Had to buy 'em from Pan., because all the others were
physically too big. Glad those are now history and I've retired, bending
over to see thru a big magnifying light 6-7 hours a day for a week or
more at a time is half of whats wrong with my back today.

Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
Joe
2018-12-24 14:00:01 UTC
Permalink
On Mon, 24 Dec 2018 08:09:11 -0500
Post by Gene Heskett
Post by Joe
There is normally a real Nyquist bandwidth quoted somewhere in the
small print. I used to repair and calibrate Hitachi scopes for a
while, until everyone stopped using them.
I wonder why, Joe. My V-1065 still works well enough to measure
frequency at a 1% accuracy, and its now pushing 35yo. In a pinch I've
looked at the output sample of an old analogue tv transmitter to
adjust the modulation depth when the monitor was bonkers, it was
actually usable at 180 mhz.
I used to see about three or four of the 665/1065 series a week, mostly
ten years old with mains switcher problems. Mostly routine fixing, after
the initial pain of troubleshooting a mains switcher. Capacitors,
mostly. Even the high-ripple ones will only run so long before they dry
up.

Then over about six months, it dropped off to nothing, and we never
saw any more. I assume oscilloscope use had dropped off enough that
people could just put a dead one in a cupboard wand say 'we'll get it
fixed if we ever need it'. Fortunately, oscilloscopes were a minor
sideline, so I didn't run out of work. The company still has a 1085,
which I use occasionally. At home, I have a 35-yo Kikusui, and a
temperamental Tek 465B of uncertain age, but older than that.
--
Joe
Gene Heskett
2018-12-24 15:00:02 UTC
Permalink
Post by Joe
On Mon, 24 Dec 2018 08:09:11 -0500
Post by Gene Heskett
Post by Joe
There is normally a real Nyquist bandwidth quoted somewhere in the
small print. I used to repair and calibrate Hitachi scopes for a
while, until everyone stopped using them.
I wonder why, Joe. My V-1065 still works well enough to measure
frequency at a 1% accuracy, and its now pushing 35yo. In a pinch
I've looked at the output sample of an old analogue tv transmitter
to adjust the modulation depth when the monitor was bonkers, it was
actually usable at 180 mhz.
I used to see about three or four of the 665/1065 series a week,
mostly ten years old with mains switcher problems. Mostly routine
fixing, after the initial pain of troubleshooting a mains switcher.
Capacitors, mostly. Even the high-ripple ones will only run so long
before they dry up.
Then over about six months, it dropped off to nothing, and we never
saw any more. I assume oscilloscope use had dropped off enough that
people could just put a dead one in a cupboard wand say 'we'll get it
fixed if we ever need it'. Fortunately, oscilloscopes were a minor
sideline, so I didn't run out of work. The company still has a 1085,
which I use occasionally. At home, I have a 35-yo Kikusui, and a
temperamental Tek 465B of uncertain age, but older than that.
465B's were temperamental allright, and way the hell out of calibration
due to drifting R's on the custom made ceramic input attenuator, and tek
has had no service parts for that for nearly 40 years now, and wanted
nearly $400 for one then. Then they merged with GVG, and it all went to
hell. I needed the custom ceramic plate they made a video op-amp on in
about 1995 as we had bought KTLA's old 300-3A/B complete with a digital
effects unit. They had just one, $1500, as is, where is! I said nooooo,
I don't think so, and started looking in the chip books, finding a
single ended one from TI for just under $2. Should bought 2 sticks of
them, it was enough faster that it threw the color phase out more than
there was adjustments for. There was about 8 of them in each channel,
and 6 paths thru that switcher. I should have shotgunned them all, but
since the packaging was different, each one would have taken around an
hour to do neatly. And I was thinking of retireing by then so it never
got done. If I'd have done them all, the video bandwidth would have been
at least trippled. Way ahead of its time in what it could do, I think
that was close to GVG's last hurrah. But the controls were about shot
too. GVG had an accessory E-Disk so the tech directors could program
their own bag of tricks, gvg wanted 20 thou for it. But when we got the
first copy of that switcher from the Penny's production house in NYC, it
came with the edisk manual and the com protocol specs. I looked at that,
said I can do that on a trs-80 color computer, took an old coco2 and
added "the Forgitten Chip" to give it a hardware rs232, wrote the
utility in basic09, which ran on the after market *nix imitation, called
os9, and sold it and two disk drives to the tv station for $245. Later I
found out mine was 4x faster than theirs! Not to mention it gave the
operators english names for their bags of tricks. And because that gave
me a comm path into the switchers innards, I wrote some more utils that
could reach into its control logic and identify which logic chip had
died. Handier than bottled beer in the long haul.

Oh, and I lied about the year in a previous msg this morning, I was 16 so
that was 1950, and the scope was a Hickock 505.

Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
Gene Heskett
2018-12-24 12:20:01 UTC
Permalink
Post by John Hasler
Post by r***@gmail.com
Yeah, but I wasn't thinking about actually using a CRT to display
the signal in real time, but, instead, collect samples (at maybe 4
GHz??), store them, and then display them as a static display.
At that sort of frequency sampling scopes (including the old crt ones)
sample at far below the signal repitition rate.
And depend on being out of synch with a repetitive signal, so they in
effect build up an image that covers enough time samples to assemble the
waveform. So you do not see it in real time. But the glitch is still
caught, and you will see a full brightness dot or pulse thats out of
place. In a cnc machine running on stepper motors, the current
regulating of the drivers if the grounding is not truly single point,
can crosstalk at the regulating frequency, usually well above 20
kilohertz, at peak voltages well over what it takes to destroy an fpga
gate as the ringing in that event often peaks at over 100 MHz and 30
volts. Trivial to see on the samplers display, but turn off the room
lights and really study what you see on a 100+ megahertz analogue scope.
Probably also true of servo-motors today since they are often driven at
full power with pwm signals. Much more efficient to make the motors
inductance work for you instead of against you. The drivers will heat at
1% or less compared to an analogue drive with the same effective gain
and power. They heat only during the transition, with very little heat
when on, and no heat when off. So the faster you can make that
transition, the cooler it stays. I use such an amplifier thats not much
bigger than a pack of camel cigarettes, to run a 1hp at 90 volts at 9.7
amps rated spindle motor on a 127 volt, 20+ amp capable supply, so it
can do in a short 1/2 second surge, around 2 hp. I've never scanned the
heat sink with an IR thermometer and found it above 85F in a 75F room.
The only time I hear it is when this amplifier regulates the current,
which I've set at 17 amps, makeing the motor iron "chirp". I figure
thats enough considering that the gear train its driving is plastic.

Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
John Hasler
2018-12-24 13:30:01 UTC
Permalink
In a cnc machine running on stepper motors, the current regulating of
the drivers if the grounding is not truly single point, can crosstalk
at the regulating frequency, usually well above 20 kilohertz, at peak
voltages well over what it takes to destroy an fpga gate as the
ringing in that event often peaks at over 100 MHz and 30
volts. Trivial to see on the samplers display, but turn off the room
lights and really study what you see on a 100+ megahertz analogue scope.
For that sort of thing you want a storage scope. Old Nicolets are
readily available for prices that even I can almost justify.

I used Tektronix analog storage scopes when doing motor control design
in the seventies but the digital ones are orders of magnitude better.
Being able to trigger on things that already happened is cool.

Before that it was cameras.

As to TDRs, if you can get by without actually seeing pictures of all
the impedence bumps you can get by with a fast counter, a high risetime
pulse generator, and a couple of fast comparators. Maybe $20 at
Digikey.
--
John Hasler
***@newsguy.com
Elmwood, WI USA
Gene Heskett
2018-12-24 13:50:01 UTC
Permalink
Post by John Hasler
In a cnc machine running on stepper motors, the current regulating
of the drivers if the grounding is not truly single point, can
crosstalk at the regulating frequency, usually well above 20
kilohertz, at peak voltages well over what it takes to destroy an
fpga gate as the ringing in that event often peaks at over 100 MHz
and 30
volts. Trivial to see on the samplers display, but turn off the room
lights and really study what you see on a 100+ megahertz analogue scope.
For that sort of thing you want a storage scope. Old Nicolets are
readily available for prices that even I can almost justify.
I used Tektronix analog storage scopes when doing motor control design
in the seventies but the digital ones are orders of magnitude better.
Being able to trigger on things that already happened is cool.
Before that it was cameras.
As to TDRs, if you can get by without actually seeing pictures of all
the impedence bumps you can get by with a fast counter, a high
risetime pulse generator, and a couple of fast comparators. Maybe $20
at Digikey.
I dunno if I could tolerate losing the pix, this old f--- has had a scope
probe in one hand since 1951.


Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
Joe
2018-12-24 09:30:02 UTC
Permalink
On Sun, 23 Dec 2018 23:11:35 -0500
Post by Gene Heskett
Thats essentially what our high bandwidth scopes do today, $400 gets
you a 1ghz sampler with an effective bandwidth of 200 mhz. I've got
one, the nice thing is that because it is digital, a once a second
glitch in a 42 megabaud seriel data train stands out like a sore
thumb because its not limited to the screen writing speed of an
analogue scope. So its just as bright as the main signal that doesn't
have the glitch. The operating software is buggier than a 10 day old
road kill in August though.
Ah, Hantek (or one of its aliases). Yes. I have the bottom-end one,
mainly for DC and audio. The triggering is worthless.
--
Joe
Gene Heskett
2018-12-24 12:40:01 UTC
Permalink
Post by Joe
On Sun, 23 Dec 2018 23:11:35 -0500
Post by Gene Heskett
Thats essentially what our high bandwidth scopes do today, $400 gets
you a 1ghz sampler with an effective bandwidth of 200 mhz. I've got
one, the nice thing is that because it is digital, a once a second
glitch in a 42 megabaud seriel data train stands out like a sore
thumb because its not limited to the screen writing speed of an
analogue scope. So its just as bright as the main signal that
doesn't have the glitch. The operating software is buggier than a 10
day old road kill in August though.
Ah, Hantek (or one of its aliases). Yes. I have the bottom-end one,
mainly for DC and audio. The triggering is worthless.
I think mine is a Gratten, and the triggering is fine once you hit the
magic twanger, but the firmware that controls it is buggier than a 10
day old road kill in August.

I have a firmware update file kit for it, but it requires a dos box with
a usb port. When was the last time you saw one of them? Short answer is
never.

dfu on linux won't talk to it. And I've never seen a dos with usb
drivers, not even the last drdos. If anyone knows how to get past that,
I'd be a happy camper indeed. Or maybe they charge for the access key, I
see the unpacked firmware dir has an empty "key" directory. If anyone
knows how to get past that... The manual, ADS1000*****************.pdf
does very carefully NOT mention it.

Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
Carl Fink
2018-12-24 13:20:01 UTC
Permalink
Post by Gene Heskett
I have a firmware update file kit for it, but it requires a dos box with
a usb port. When was the last time you saw one of them? Short answer is
never.
Doesn't freeDOS support USB? I know they recommend booting off a USB
drive.

There's a commercial DOS driver, DOSUSB, but it's very expensive.
--
Carl Fink ***@finknetwork.com
Thinking and logic and stuff at Reasonably Literate
http://reasonablyliterate.com
Gene Heskett
2018-12-24 13:40:02 UTC
Permalink
Post by Carl Fink
Post by Gene Heskett
I have a firmware update file kit for it, but it requires a dos box
with a usb port. When was the last time you saw one of them? Short
answer is never.
Doesn't freeDOS support USB? I know they recommend booting off a USB
drive.
There's a commercial DOS driver, DOSUSB, but it's very expensive.
I'll have to check that FreeDOS out. Unforch this old Asus board won't
boot from usb.

Thanks Carl.

Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
r***@gmail.com
2018-12-24 17:40:02 UTC
Permalink
Post by Gene Heskett
Post by Carl Fink
Post by Gene Heskett
I have a firmware update file kit for it, but it requires a dos box
with a usb port. When was the last time you saw one of them? Short
answer is never.
Doesn't freeDOS support USB? I know they recommend booting off a USB
drive.
There's a commercial DOS driver, DOSUSB, but it's very expensive.
I'll have to check that FreeDOS out. Unforch this old Asus board won't
boot from usb.
There is (are?) dos emulator(s?) for Linux -- do any of those support a usb
port?
Gene Heskett
2018-12-24 18:30:02 UTC
Permalink
Post by r***@gmail.com
Post by Gene Heskett
Post by Carl Fink
Post by Gene Heskett
I have a firmware update file kit for it, but it requires a dos
box with a usb port. When was the last time you saw one of them?
Short answer is never.
Doesn't freeDOS support USB? I know they recommend booting off a
USB drive.
There's a commercial DOS driver, DOSUSB, but it's very expensive.
I'll have to check that FreeDOS out. Unforch this old Asus board
won't boot from usb.
There is (are?) dos emulator(s?) for Linux -- do any of those support
a usb port?
No clue, thats another round tuit I've misslaid someplace. And ATM. I
have 2 milling machines broken and am having to buy a third to fix the
1st one. But the ball screw I need to fix the 2nd one seems to be made
from pure unobtainium, an 8mm by 500mm. The best I can do is only 130mm
long. :(

Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
Ric Moore
2018-12-24 20:30:01 UTC
Permalink
Post by Gene Heskett
Post by r***@gmail.com
Post by Gene Heskett
Post by Carl Fink
Post by Gene Heskett
I have a firmware update file kit for it, but it requires a dos
box with a usb port. When was the last time you saw one of them?
Short answer is never.
Doesn't freeDOS support USB? I know they recommend booting off a
USB drive.
There's a commercial DOS driver, DOSUSB, but it's very expensive.
I'll have to check that FreeDOS out. Unforch this old Asus board
won't boot from usb.
There is (are?) dos emulator(s?) for Linux -- do any of those support
a usb port?
No clue, thats another round tuit I've misslaid someplace. And ATM. I
have 2 milling machines broken and am having to buy a third to fix the
1st one. But the ball screw I need to fix the 2nd one seems to be made
from pure unobtainium, an 8mm by 500mm. The best I can do is only 130mm
long. :(
Just so you know, they are republishing Carl and Jerry from Popular
Electronics. I just received the first year collection, sold over at
lulu.com :) Ric
Gene Heskett
2018-12-24 20:50:02 UTC
Permalink
Post by Ric Moore
Post by Gene Heskett
Post by r***@gmail.com
Post by Gene Heskett
Post by Carl Fink
Post by Gene Heskett
I have a firmware update file kit for it, but it requires a dos
box with a usb port. When was the last time you saw one of them?
Short answer is never.
Doesn't freeDOS support USB? I know they recommend booting off a
USB drive.
There's a commercial DOS driver, DOSUSB, but it's very expensive.
I'll have to check that FreeDOS out. Unforch this old Asus board
won't boot from usb.
There is (are?) dos emulator(s?) for Linux -- do any of those
support a usb port?
No clue, thats another round tuit I've misslaid someplace. And ATM.
I have 2 milling machines broken and am having to buy a third to fix
the 1st one. But the ball screw I need to fix the 2nd one seems to
be made from pure unobtainium, an 8mm by 500mm. The best I can do is
only 130mm long. :(
Just so you know, they are republishing Carl and Jerry from Popular
Electronics. I just received the first year collection, sold over at
lulu.com :) Ric
That was tongue in cheek at its finest.

Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
Dan Ritter
2018-12-24 16:00:02 UTC
Permalink
Post by Gene Heskett
Post by r***@gmail.com
Post by Doug
Theoretically, you can build your own with a fast pulse generator
and an oscilloscope. The trick is, you need a REALLY FAST
oscilloscope! The pulse generator is easy, just a couple of
transistors, maybe a diode. The circuit is probably in every
edition of the Radio Amateur's Handbook. The scope is expensive. If
you don't have at least a 1GHz digital sampling scope, don't bother!
--doug, WA2SAY, retired RF engineer
Hmm, with CPU clocks hitting 4 GHz, I wonder how expensive an ADC
converter to work at corresponding speeds would be? (Just an idle
question ;-)
If you have to ask, you can't afford it. Bring outrageous sums of money
This is typically done by every trick tek knows about fast analogue
circuitry, in the best units with long vertical deflection plates in a
custom made crt with teeny delay lines between the sections of the
plates so the signal is virtually traveling toward the screen at the
same speed as the electron beam is traveling. And its moving fast enough
at 22,000 volts, relativity can and does get in the way. Such scopes put
the plates so close to the beam that the beam is intercepted by striking
the plates at just a hair over 4 cm high, 2cm from the horizontal center
line. The only one of those I ever saw was in the early 1980's or so at
the NAB show in Vegas, and it was well into a 5 digit asking price then.
We're almost in 2019, which means:

https://www.mouser.com/ProductDetail/Analog-Devices/HMCAD1511?qs=sGAEpiMZZMvTvDTV69d2Qnrp4UY3MFmsfo%252bIC8KWhPQ%3d

Analog Devices HMCAD1511
1 GS/s, 8 bits, 1 channel: quantity 1 from Mouser: $64.76.

That's the cheapest GHz+ ADC they have; you can still go up to
much heftier prices for more channels, more bit depth, even
higher sampling rates.

-dsr-
r***@gmail.com
2018-12-24 19:20:01 UTC
Permalink
Hmm, intended to send this to the list, sent to John Hasler, only, who did
respond -- I hope he will copy his reply to the ist (or tell me it is ok to do
so).
Post by John Hasler
As to TDRs, if you can get by without actually seeing pictures of all
the impedence bumps
you can get by with a fast counter, a high risetime
pulse generator, and a couple of fast comparators. Maybe $20 at
Digikey.
Can you elaborate a little on how that would work?
John Hasler
2018-12-24 19:30:02 UTC
Permalink
Post by r***@gmail.com
Hmm, intended to send this to the list, sent to John Hasler, only, who
did respond -- I hope he will copy his reply to the ist (or tell me it
is ok to do so).
Go ahead.
--
John Hasler
***@newsguy.com
Elmwood, WI USA
Gene Heskett
2018-12-24 20:40:01 UTC
Permalink
Post by r***@gmail.com
Hmm, intended to send this to the list, sent to John Hasler, only, who
did respond -- I hope he will copy his reply to the ist (or tell me it
is ok to do so).
Post by John Hasler
As to TDRs, if you can get by without actually seeing pictures of
all the impedence bumps
you can get by with a fast counter, a high risetime
pulse generator, and a couple of fast comparators. Maybe $20 at
Digikey.
Can you elaborate a little on how that would work?
I can imagine the first fast comparator would enable the counter,
previously held in reset and the second comparator is then enabled, and
its threshhold adjusted for a stable stop of the counter, giving the
number of input cycles between the start and the stop. Decent calculator
math would then give you the distance to the major impedance disturbance
that caused the echo.

You need a GHz (at least) signal source to count, the higher, the more
accurate. The delay of course is 2 way, out to the fault and back, so
convert that to distance one way with a /2.000 after using the usual
hambooks 984/frequency derivation to get the wavelength in feet IIRC.
Ideally the frequency should be a wavelength short enough to give decent
accuracy because this method will only give you the number of cycles as
an integer. You send the pulse and wait for the echo to come back and
stop the counter. Multiply the wavelength by the counter to get how far
away the fault is, multiply that by the PV of the line being tested and
divide by 2 to get the one way distance.

Clear as mud, right? 2000 feet of 6.125" inch line can be fun due to
losses weakening the return signal. Specially a fast one like a tunnel
diode might make. And no bets at all for a broadcast antenna because the
pulse is not frequency shaped to match the antenna's operating
frequency, so from the fine matcher on into the element array, its a
broadband mess that splatters all over the TDR screen. Seeing that on a
real TDR with trained eyeballs seems to be the only way to tell that,
from a cracked and burned up elbow 50 feet below it at the tower top to
antenna feed connections. And that 50 feet might be a weeks work for a
tower crew with some antenna styles. And of course it always happens in
bad weather, making the high steel work dangerous because there may be a
6" thick layer of ice on everything 2000 feet up.

That broken elbow might be the starting point, but the fire then moves
down the line towards the power source and may burn up 600 feet of line
before the transmitters VSWR protection circuits can shut it down. Been
there, done that, several times at what was once NETV's KXNE transmitter
on UHF channel 19. Rosemounts rime ice detectors buried on the antenna
structure had an average lifetime of a year. They cost quite a bit, but
their failure mode was always safe mode, so you didn't know a thing
until the main beam power breaker opened in response to the rising VSWR.
By then you were out 10 to 50 grand and several days to get a crew
rounded up to work on it, to get it back on the air. More than once in
that decade we had to let them sit around, or go do another job while
the ice went away. One time litterally tons of it had formed a flag
about 8 feet long on the downwind side of that storm. Cleaning 600 feet
of that line and rebuilding it took time, rags by the pickup load of
bags, alcohol in several 5 gallon buckets, and teflon parts aren't free
either. I'm glad I'm retired now.

Now, back to the asinine arguments about boobs. Is this our regularly
scheduled programming for this week? :)

Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
r***@gmail.com
2018-12-24 22:50:01 UTC
Permalink
<resending to the list (with slight editing to show who wrote what),
originally from John Hasler>
Post by John Hasler
As to TDRs, if you can get by without actually seeing pictures of all
the impedence bumps
you can get by with a fast counter, a high risetime
pulse generator, and a couple of fast comparators. Maybe $20 at
Digikey.
Can you elaborate a little on how that would work?
Apply the pulse to the cable under test. The rising edge triggers the
counter and enables the comparators. One comparator triggers on the
first rising edge on a return pulse, latching the count and disabling
itself. The other comparator triggers on the first falling edge on a
return pulse, latching the count in a different register and also
disables itself.

The count latched by the positive comparator represents the distance to
first positive impedance bump encountered by the pulse. An open will
double the applied voltage. The other will represent the distance to
the first negative bump. A short will invert the voltage.

You'll want to make the thresholds adjustable, of course, and you will
also want to add range gating so that you can ignore all but a selected
section of the cable.

There are some time to distance converter chips that would be ideal for
this. You'll want to put the whole thing under the control of a micro,
of course. It could scan the range gates up and down the cable and
report impedance as a function of distance.
--
John Hasler
***@newsguy.com
Elmwood, WI USA
Continue reading on narkive:
Loading...