[GreenKeys] Range and Bias

[Jeffrey D Angus]

On 7/21/2013 12:11 PM, George B. Hutchison wrote:
> Excellent explanation of the operation of any selector.
>
> The only comment I have is that the additional length of the stop
> pulse was not for selector "settlling", but for the fact that early
> on machines were in locations where the power could have been DC, 25
> cycles, 50 cycles or 60 cycles.

Well actually, yes, that's what I meant.
I'd written:
> The reason for the extra 1/2 width stop pulse it to give the machine
> time to settle in case it's not quite running at the correct speed. 
That 11 mS of "extra stop pulse time" was to allow the receiving machine on
the other end to get caught up if it was running a bit slow.

The entire operation is asynchronous. Sort of the classic "hurry up and 
wait"
mode of operation.

Depending on the position of the range finder, in the middle it's already
1/2 way past the 5th data bit when the machine has to finally "do something"
If the range finder is retarded closer to the end of a data bit, that 
leaves even
less time for the machine to "do something" and return to the "waiting for a
start pulse."

Coincidentally, that 1/2 bit extra on the stop pulse corresponds to the 1/2
bit delay of the sampling via the range finder.

Side thoughts:
On a mechanical system, timing is derived by number of degrees on a shaft.
7.5 "bits" would be equal to 48 degrees for each bit and 24 degrees for that
1/2 bit of time. Physically rotating the selector assembly via the range 
finder
either advances or retards it from the "ideal" point. For example, moving it
+/- 15 degrees would mean moving it +/- 6.875 mS from the ideal position.

On a digital system, most do not have a "range" adjustment. They have a
fixed count of 33 mS, sample, then 22 mS, sample, 22 more mS, and so forth.

To implement a ranger finder, you would have to first count off 26 mS, then
add the "range adjustment" that would vary between 0 and 14 mS, then
count the subsequent 22 mS times.

By digital, I mean how a "no moving parts" i.e, glass TTY would look at the
raw recovered mark/space data and determine what is when.

The 5th sample would be at 121 mS. If the baud rate was off roughly 11%
the digital sampling would occur outside of the 5th data bit resulting in
garbled print. This is why, with the range finder you should be able to
copy the TWX 66 WPM machines. But you can't do it with a computer
unless you change the baud rate from 45.45 to 50. Adding mark and space
distortion makes it even rougher.

Now I did mention HOW you could implement digital range finding. I don't
believe however, that anyone actually did that.

Jeff-1.0
wa6fwi








  you would have three sets of counters, perhaps running
at 1 KHz for example (And a stepped variation of 1 mS).
An initial "I have a start pulse" that counts down