Go Down

Topic: Fist Chip for Time Domain Reflectometer TDR Project (Read 4568 times) previous topic - next topic

DROBNJAK

Sep 17, 2015, 01:28 pm Last Edit: Sep 17, 2015, 01:29 pm by DROBNJAK
I watched this great video on YT: #88: Cheap and simple TDR using an oscilloscope and 74AC14 Schmitt Trigger Inverter and thought following the suit.

Beside the chip that he suggest, the 74AC14 Schmitt trigger, is there any other chip with faster rise times?

I found this one LTC6993-1, but it still has rise time of 1.1us.

Since I am working with cables that are short 100-300mm, I need something even faster. Is there a chip with rise time in low pico second range, that doesn't cost much. Preferably in DIP format or at worst in SOP?

Peter_n

74AC14 : 2ns
LTC6993 : 1.1ns
ATmega328P : ? (I can't find it in the datasheet, it is 3.6ns for SPI)
PO74G14A : 0.8ns

I think that normal rise times are about 10ns. You need a chip for the GHz range. A Schmitt-trigger for GHz is hard to find. The 74AC14 is already very fast.
Do you have an oscilloscope that can measure into the GHz range ?

Riva

For such short cables your also going to need a fast oscilloscope.
Why do you need to fault check/measure such short cables?
You could maybe put a fixed length cable between test cable and the pulse generator to negate the need for such short rise times by delaying the reflection, this would need a good, repeatable quality connection on the cables to prevent the joint causing to much reflection.
Don't PM me for help as I will ignore it.

DROBNJAK

#3
Sep 17, 2015, 02:25 pm Last Edit: Sep 17, 2015, 03:17 pm by DROBNJAK
Yeah, I have a secondhand 1GHz oscilloscope. So I can go pretty fast.

I am working on laser rangefinder idea.

But as well for general learning. I want to learn about reflections in cables etc., like transmission line theory. But most of cables that I have are quite short.

PO74G14A is fast, but pity not available any more.

For future reference, I searched Farnell's catalog for "comparator" and found few interesting ones:

- ADCMP573BCPZ-R2  HIGH SPEED COMPARATOR with 150ps rise time, but it is in a $30 range and it is in that awkward LFCSP-16 package with connectors under the IC.

- ADCMP552BRQZ in QSOP package, with 750ps rise and cost of about $10.

- MAX9601EUP in TSSOP package, with 200 ps rise time, for about $16.0.

-

Paul__B

For cables that short, you are by definition using the wrong concept - you need not a Time Domain Reflectometer, but a Frequency Domain Reflectometer.

Basically, a spectrum analyser (or more correctly, an impedance analyser).

pegwatcher

#5
Sep 18, 2015, 05:30 am Last Edit: Sep 18, 2015, 05:56 am by pegwatcher
Years ago I did a couple TDR designs. One used an 8 bit Flash AD converter with output to fast RAM. No good for your short distances.

The other was a small handheld instrument that monitored the reflections with a threshold detector. You would slowly adjust a  potentiometer to lower the threshold until the strongest echo triggered it, which then stopped a counter that was driven by a high speed clock. You read out the distance (the count) on 7 segment displays, very simple. High impedance faults have a positive echo and low impedance have negative echos (I Think, this is from old memory :) ) I used a couple LEDs to indicate whether it was a short/ground or an open.

The second example might be possible for your needs (haven't put a lot of thought into this, maybe someone else will have better input on using this method for your very short distances.)
I'm not a complete idiot. Some parts are missing.


pegwatcher

#7
Sep 18, 2015, 07:42 am Last Edit: Sep 18, 2015, 07:50 am by pegwatcher
Quote
Maybe a PECL or LVPECL driver? Something like MAX9321B,
That is one fast chip and differential inputs makes it perfect; one for echo and one for threshold voltage. Trouble is, C is so darn fast. My back-of-a-napkin calculations show a best resolution of about 50 mm if you happen to get a chip with Differential Input-to Output Delay (min) of around 145 ps . ..Maybe...   Layout would be critical, of course.

My calculation could be off by an order of magnitude or so :smiley-confuse: because it's past bedtime here.

[edit] Whoops - forgot to add in propagation delay in stopping the counter.
I'm not a complete idiot. Some parts are missing.

Go Up