[sci.electronics] Hardware Correlators

dcthomso@cs.strath.ac.uk (Duncan C Thomson IE87) (12/05/89)

I am doing a final year project which will require correlation of an
(ultransonic) signal with a reference signal.  I am currently trying to decide
whether to do this in software or in hardware.

Does anyone know of any hardware correlators?

How fast are they?

How much do they cost?

Who manufactures them?

Who supplies them (preferably in the UK?)

Thanks in advance,

	-Duncan

commgrp@silver.bacs.indiana.edu (12/05/89)

dcthomso@cs.strath.ac.uk (Duncan C Thomson IE87) writes:

>I am doing a final year project which will require correlation of an 
>(ultransonic) signal with a reference signal.  I am currently trying 
>to decide whether to do this in software or in hardware.

>Does anyone know of any hardware correlators?
>How fast are they?
>How much do they cost?
>Who manufactures them?
>Who supplies them (preferably in the UK?)
>Thanks in advance,
> -Duncan


One type of correlator can be made as follows:

       __     __     __         __     |\
in >--|__|---|__|---|__|- ... -|__|----|  \
           |      |____________________|   >----- out
           |___________________________|  / 
                                       |/

The input is an AC-coupled detected pulse train (e.g., a series of 
ultrasonic pulses converted to dc by a receiver).  Each box is an 
analog delay-line IC (Radio Sh** used to sell two-channel analog-
delay chips made by EG&G). The delay lines are clocked at n times the 
input pulse repetition frequency (prf), where n is the number of 
stages per delay line (typically 256).  The outputs of all the delays 
are summed in an op amp.

Repetitive signals marching through the delay line at the right prf 
are added; other signals and noise average to zero.  I don't remember 
the equations; I think signal/noise ratio improves as the square root 
of the number of delay-line taps (summer inputs).

The effect is that of a very narrow-band filter. The output is dc 
voltage.  Response time is slow, as it takes a while to fill the whole 
delay train with signal.

This interesting circuit was used in an inductive audio-frequency 
mine-rescue beacon receiver manufactured by General Instruments.  
Faster versions can be built with other types of delay line; some old 
radar receivers used acoustic delay-lines consisting of tubes filled 
with mercury.

--

Frank Reid   W9MKV    reid@gold.bacs.indiana.edu

gd@milkfs.itstd.sri.com (Greg DesBrisay) (12/06/89)

SAW devices can be used to make good correlators.  I'm not sure of the
cost these days.

If you simply want to correlate a digital data stream, then several
chips manufactured by TRW might do the job for you.  Some TRW
correlators are the TMC2023, TDC1023, TMC2220, and the TMC2221.
(ref. TRW LSI Products Data Book, 1988)

TRW Application Note TP-17 entitled "Correlation--A Powerful Technique
for Digital Signal Processing" by Dr. J. Eldon is listed on the last
page of the TRW catalog.  That might be a source of good info.

TRW's LSI Products Sales Office in England is:
	Unit #28 Frederick Sanger Road
	Surrey Research Park
	Guildford, Surrey GU2 5YD
	England
	Phone: (0483) 302364

Good Luck! 


Greg

michael@fe2o3.UUCP (Michael Katzmann) (12/06/89)

In article <1048@baird.cs.strath.ac.uk> dcthomso@cs.strath.ac.uk (Duncan C Thomson IE87) writes:
   >I am doing a final year project which will require correlation of an
   >(ultransonic) signal with a reference signal.  I am currently trying to decide
   >whether to do this in software or in hardware.
   >
   >Does anyone know of any hardware correlators?
   >
I know TRW make a few.

   >How fast are they?
   >
Very (10's of Megs I think, though its a few years sice I looked at the docco)
   >How much do they cost?
   >
If you have to ask, you probably can't afford them.

The CSIRO in Australia (division of Radio physics) built a custom correlator
for the AT (Australian Telescope), which is cascadeable and looked very nice.
I don't know whether there was ever thought about licencing it's manufacture
commercially.

A DSP approach may be possible (if a bit slow), I think INMOS has an app note
on one of their DSP chips for this sort of work. (See SGS-THOMPSON-RCA-INMOS..)


---------------------------------------------------------------------
email to 
UUCP:       uunet!mimsy!{arinc,fe203}!vk2bea!michael
						  _ _ _                    _
 	Amateur	|    VK2BEA	(Australia)      ' ) ) )      /           //
 	Radio	|    G4NYV	(United Kingdom)  / / / o _. /_  __.  _  //
	Stations|    NV3Z	(United States)	 / ' (_<_(__/ /_(_/|_</_</_

Michael Katzmann
Broadcast Sports Technology.
2135 Espey Ct. #4
Crofton Md. 21114 USA

Ph: +1 301 721 5151

sampson@attctc.Dallas.TX.US (Steve Sampson) (12/07/89)

TRW makes some correlators on a chip.  The TMC2023 for instance does
64 bits at 30 MHz.  There's a couple other models.  See the Data Converters -
DSP products book  P.O. Box 2472, La Jolla, CA, 92038.

tgg@otter.hpl.hp.com (Tom Gardner) (12/08/89)

About a decade ago Reticon Inc made correlators based around CCD
bucket-brigade analogue shift registers.

No, I don't have any more up-to-date info.....


tom gardner