Tektronix 1502 TDR degraded risetime

[tggzzz]

I'm in the process of resurrecting a Tek1502 TDR, simply because I like them. This one has a somewhat strange bandwidth/risetime issue that I haven't seen on other Tek 1502s. I'd like to find out if other people have seen the problem.

The two traces below both show the internal 50ohm track terminated in the special shorting BNC socket. You can see the TDR pulse as it whizzes past the CR1732 sampling diodes and then shortly thereafter the pulse reflected from socket J0421 cancels it out. The horizontal scale is nominally 0.1ft/division, i.e. ~4.5cm/div (don't you just love non-metric units!).

The difference is:

• the slow sweep trace has the expected rise/falltime, i.e. distance resolution (ignore the different trace brightnesses due to strobing between the camera's shutter and the sweep)
• the normal sweep trace has a noticably degraded rise/falltime and hence resolution

So, my question is what might cause the slow sweep speed to have a normal risetime but the normal sweep speed to have a poor risetime?

Background: internally it looks like someone has "been in there beforehand"...

There is substandard soldering on the PSU board, but it works and I don't see it causing this problem.

The diode CR1691 is in the "right area" to cause the problem and is mounted poorly; I suspect it has been replaced by a "1N914" diode probably with a slightly higher capacitance. However, I'm unsure that is the cause of the problem; I'd have expected the problem to be nearer the sampling diodes.

Your thoughts would be welcome.

EDIT: attached two different traces, sigh.

[tautech]

I'm in the process of resurrecting a Tek1502 TDR, simply because I like them. This one has a somewhat strange.............. 

The horizontal scale is nominally 0.1ft/division, i.e. ~4.5cm/div (don't you just love non-metric units!).

1ft =~ 305mm, therefore 1/10 =~3.05cm. 

[tggzzz]

I'm in the process of resurrecting a Tek1502 TDR, simply because I like them. This one has a somewhat strange.............. 

The horizontal scale is nominally 0.1ft/division, i.e. ~4.5cm/div (don't you just love non-metric units!).

1ft =~ 305mm, therefore 1/10 =~3.05cm. 

Annoyingly, you're right - and that's a better match for the distance inside the cavity. Doesn't change the risetime difference, though

[texaspyro]

Your two trace pics are the same.

BTW, the high voltage caps on the CRT power supply are known failure modes.  I've replaced dozens of them.

The thermal strip chart recorders use an un-obtainium chart paper with a perforated timing track along the edge.  If you can find it, it sells for like $50 a roll.   I made a little circuit board with an ATtiny13 processor and a pot that simulates the timing track pulses so I can use $2/roll EKG paper.  You disconnect the optical interrupter cables and connect the board to where the interrupter was.

The timing belt on the chart recorder is also known to turn to goo.   Stock Drive Products have replacements that work.

Blown 20mA tunnel diodes are a killer.  I have used some 20 mA GaAs diodes from Russia.  They have a higher output voltage than the originals so you have to modify the output amplifier to use them.

[tggzzz]

Your two trace pics are the same.

Sigh. This must be the most careless post I've done, and I can't think of a decent reason - other than senility.

Anyway, I've changed the first post.

BTW, the high voltage caps on the CRT power supply are known failure modes.  I've replaced dozens of them.

I've had to do that once.

The thermal strip chart recorders use an un-obtainium chart paper with a perforated timing track along the edge.  If you can find it, it sells for like $50 a roll.   I made a little circuit board with an ATtiny13 processor and a pot that simulates the timing track pulses so I can use $2/roll EKG paper.  You disconnect the optical interrupter cables and connect the board to where the interrupter was.

The timing belt on the chart recorder is also known to turn to goo.   Stock Drive Products have replacements that work.

Yes indeed.

Nice hack for renovating the printer to use available paper. Have you written it up anywhere?

I kludged together an XY plotter plugin for the 1502 that I'm keeping (because it has a m/div x-axis, for ft/div), and have used it with a conventional scope to get a plot. That's been replaced by a pukka XY plugin, so I could have a TDR that works underwater - in theory, since I doubt my ability to seal it up properly and I don't fancy 2kV 1cm from my fingers

Blown 20mA tunnel diodes are a killer.  I have used some 20 mA GaAs diodes from Russia.  They have a higher output voltage than the originals so you have to modify the output amplifier to use them.

I haven't found that, yet. Again you have a neat hack that might be worth documenting.

If necessary I would consider replacing the TD with a fast PECL comparator, something like the ADCMP582 with a 37ps risetime. The mechanics might be entertaining, though.

[texaspyro]

I dug through my TDR stuff and found a pile of the chart recorder pulse fakers.  They don't have the ATtiny13A's on them, but that is easy to fix.   I'll do a separate thread on them and offer them for sale.

They are not a perfect solution... there is no timing feedback so they have a pot for adjusting the timing.  Also, they can't detect end-of-paper so you have to watch the chart for the tell-tale red stripe that indicates you are near the end of the roll.

EKG paper comes in rolls with the sensitive side in or out.  You want sensitive side in.   The cheapest paper with the best matched grid to the Tek paper that I have found is the Philips 40457C.  Also paper for the Medtronic Lifepak 10.

I also found a receipt for the replacement belt from Stock Drive Products.  It is a 40DP, 44 tooth belt P/N "A 6B 6-044012"  They are still available for a whopping $2.75 each.  Actually I *think* that is the proper belt.  I had problems counting the rotted belt teeth and first ordered belts with tooth counts around 44 teeth and then ordered 10 of the one that fit...  the receipt was for 10 44-tooth belts (in 2007)

Another guy did the amplifier mod for using the Russian tunnel diodes... I'm not sure what he changed, but it should not be difficult to work out.  The Russian diodes had a small "puck" with flat leads coming out the sides.  You fold the leads over the puck, trim them, and stick it in the spring loaded clamp for the original Tek diode.