NNTP-Posting-Date: Tue, 19 Nov 2002 09:36:46 -0600
From: firstname.lastname@example.org (John Fields)
Subject: Re: Need help measuring pulses from Polaroid 6500
Date: Tue, 19 Nov 2002 14:56:37 GMT
Organization: Austin Instruments, Inc.
References: <email@example.com> <firstname.lastname@example.org> <3DD85DE9.email@example.com>
X-Newsreader: Forte Agent 1.5/32.451
X-Abuse-and-DMCA-Info: Please be sure to forward a copy of ALL headers
X-Abuse-and-DMCA-Info: Otherwise we will be unable to process your complaint properly
X-Received-Date: Tue, 19 Nov 2002 15:36:49 GMT (newsfep2-win.server.ntli.net)
On Mon, 18 Nov 2002 03:26:56 GMT, Fred Bloggs
>The logic works right but there are two problems of practical
>interest. Because you start count immediately and trigger the logic off
>both clock edges, your circuit has a mean range bias error of 0.75 range
>cell, whereas a single polarity edge would cause INIT to be late and
>ECHO to be early for a zero mean. Also, if there is no ECHO, your
>display remains frozen at the last valid return. Other than that, it's
>whizz-bang all the way Jose...
Thanks, but there seems to be something wrong with your reasoning.
With the circuit in its quiescent state (INIT and ECHO low) the
counter will be held in reset and disabled (ENT low) by INIT low and
partially enabled (ENP high) by the complement of ECHO\ appearing at
Looking at the transmitting side, once INIT goes high, ENT will be
pulled high, and the next high-going edge of CLK will cause the
counter to increment. Such being the case, the counter will not
start counting immediately unless INIT precedes the leading edge of
CLK by just enough to meet the 160's setup and hold requirements.
When this happens the counter will increment just as the acoustic
pulse leaves the transducer and will introduce the maximum error
generated on the transmit end, one CLK period. The minimum error
will be introduced when INIT goes high immediately _after_ the
leading edge of CLK. So, if the occurrence of the leading edge of
INIT with respect to CLK is random, which it is, I would suspect
that the mean transmitting error would always be negative (the
acoustic pulse would not have gotten as far as the counter
indicated) and would be one-half of one CLK period.
Now, looking at the receiving side,
CLK__| |____| |____|
CLK__| |____| |____|
In the first example, ECHO has gone high after CLK/ but before CLK,
so the counter will accumulate one extra count. Since the echo came
back before CLK went high and the counter went ahead and counted
anyway, this is an error which indicates that range was greater than
it really was.
In the second example, ECHO went high after CLK went high, so range
was actually greater than indicated by the counter.
Since the sense of the error is opposite depending on which side of
the leading edge of CLK ECHO happened to fall on and since the
relationship between CLK and ECHO is random, I would expect the mean
error here to be zero, leaving us with a mean error for the system
of +1/2 CLK period.
I didn't bother with the no-echo problem because the OP was only
looking for a gross indication that the machine was working, but an
easy fix would be to hit a retriggerable one-shot with the pulse
from U8-3 and then use the one-shot's output to activate the 4511s'
LT inputs should it time out.
Of course, the right way to do the whole thing would be with a small