2

Question TDC for TOF

My question is in regards to dynamic range of short time period TDCs, or alternative solutions to a time to digital application.

If further information is required, please ask.

TDC datasheet

Question.

TDC7200 datasheet says, Measurement Range: – Mode 1: 12 ns to 500 ns – Mode 2: 250 ns to 8 ms

I wish to measure 66.6ps as my fastest sample and 3.33ns as my slowest sample.

Both mode 1 and 2 are outside of my desired range, however I see the datasheet also says 55ps resolution. Further down it also says LSB = 55ps.

Does this mean that if I wait for 12ns then all samples thereafter are within 55ps accuracy?

In other words:

  • fastest sample = 12ns
  • next sample = 12.055ns
  • next 12.11ns
  • next +55ps
  • etc etc until the nBit register is full?
JRE
  • 67,678
  • 8
  • 104
  • 179
Lars Olof Lundh
  • 21
  • 3
  • I think there are some kinds of interferometers for such applications. –  Sep 25 '16 at 01:58
  • I was about to suggest something with a ring oscillator, until I read the TDC2700 note, which integrates it all! So, is your question 'can I do better than the TDC2700?' ? Because the datatsheet spells out the dynamic range. – Neil_UK Sep 25 '16 at 06:53
  • Thank you Gregory, I will study interferometers further. Neil_UK: I think I got myself confused. would you please confirm something for me? I have edited my post with the new question at the bottom. – Lars Olof Lundh Sep 26 '16 at 10:03

4 Answers4

2

Regarding

Does this mean that if I wait for 12ns then all samples thereafter are within 55ps accuracy. In other words. fastest sample = 12ns , next sample = 12.055ns

Actually the TDC7200 datashet specifies:

+-----------------+------------------------------+-----+-----+-----+------+  
|      Param      |         Description          | min | typ | max | unit |  
+-----------------+------------------------------+-----+-----+-----+------+  
| T1STARTSTOP_Min | t_min between Start and Stop |  12 |     |     | ns   |  
| T1STOPSTOP_Min  | t_min between 2 Stop Signals |  67 |     |     | ns   |  
+-----------------+------------------------------+-----+-----+-----+------+

table adapted from datasheet, formatted thanks to online tool

So if first sample is at 12ns then next sample cannot arrive before 12+67=79 ns.

matt__chv
  • 144
  • 7
1

Did you tried with ACAM chips? they are not in 66ps range. But they are pretty good and provide good support and kits.

Alternative way is to use ultrawideband communication. Please refer to decawave

Sanu - Open Maker
  • 137
  • 1
  • 12
  • Thank you Sanu! TDC-GP22 by ACAM looks great, I will research ACAM further! I think I understand how my application will work now. I just wish to confirm that LSB and resolution works in the manner that I have suggested at the very bottom of my question? thank you soo much for your suggestions! – Lars Olof Lundh Sep 27 '16 at 10:36
  • Please refer the datasheet. I didn't tried in that way. Also try to find a very stable tx/rx circuit for TOF. Happy to here that the answer is useful :) – Sanu - Open Maker Sep 28 '16 at 06:15
1

Take a look at this app note at TI: Interleaved Short Time Measurements using TDC7200. It explains a way to make measurements in the 0 to 12ns range as long as the measurement can be repeated. Assumption is that the measurement will not change significantly between repetitions.

Dave Tweed
  • 168,369
  • 17
  • 228
  • 393
filipmu
  • 61
  • 3
1

The crucial part here is:

  • what is the width of a pulse?
  • how many pulses are there in short succession?

The precision is not a problem. For example our company cronologic build the xTDC4 time to digital converter: https://www.cronologic.de/products/tdcs/xtdc4-pcie

It measures the arrival time of pulses at a precision of 13ps, which fulfills your precision requirement. However, pulses must be at least 200 picoseconds wide and consecutive pulses on the same input must be at least 5 nanoseconds apart. Many other TDCs are slower than that.

Using deserializers you can measure pulses as small as 30ps at rates of up to 15GHz with a precision of 30ps. We also habe a solution for that. However this can only measure a certain amount of pulses at once because the data rate is to high for continuous readout.

Kolja Sulimma
  • 11
  • 1