Rod, at 21:42 on Mar 19 1996, you wrote to Bill Grimsley...

BG> I've absolutely no idea how Rockwell interpret
BG> their Eye-Q Monitor reading, nor even what it means,

RS> Its based on the root mean square of the error vector.

Damn, I just KNEW I should have held onto my vector-scope!

RS> I assume you understand the basics of a trellis constellation pattern.

I only really understand the very basics, athough my bluff level is
considerably hgher than that  IOW, I get by on a good day...

RS> At a particular symbol rate and constellation pattern, you have an array
RS> of points which are the discrete states that the carrier can have, the
RS> constellation of valid points. Now each point is the ideal coordinates
RS> for that particular point, if the observed signal is close to one of
RS> those, its taken as that value. The difference between the ideal location
RS> of each point and the one actually seen is essentially due to the noise
RS> and jitter and group delay etc on the line. The root mean square is just
RS> a mathematical way of expressing how far away a particular point is from
RS> the correct location, the error.

OK, that's fairly easily understood.  I just pictured how I'd imagine it to
appear on a vector-scope and I was away...  :)

RS> You can go and take a bex and have a good lie down now |-)

Dunno, TCM is much easier to understand than say CTV camera theory, IMO.

BG> so all I can do is take notice of the manual when it tells me
BG> that values in excess of 10 are very poor (10 "what's", BTW?).

RS> Its just a scaling of the root mean square of the error vector
RS> which produces the EQM number. I think calling >10 'very poor' is
RS> mad. Its more usual to reduce the carrier rate if the EQM is >20.

Some modems actually report that the retrain level is set at an EQM of 098,
which looks next to useless up there.

Regards, Bill

--- Msgedsq/2 3.20