Rene,

I apologize for taking so long to respond.  I've got too many projects to
work on at the moment. 

I do not like the local loop simulators supplied by TAS or PTT.  In my
opinion they are a poor representation of real world local loops. They are
sourced at 600 ohms and should be sourced at 900 ohms.  I don't use them
unless a client insists they must be used.  In my opinion TSB-37A testing
makes poor and good modems look good. 

My performance figures come from testing with real local loops.  I have
been designing modems since 1969.  My first modem was a discrete Bell 202. 
I was employee #13 at Vadic (we later sold it to Racal).  While there I
was in the development engineering department for 10 years and advanced
development department for 5 years.  During those years I was privileged
to work for and with John Bingham (the inventor of 1200 bps full duplex
modems).  During that time I learned a lot about full duplex modem from
FDX1200 bps to 9600 bps.  During that period I developed my knowledge of
DAA's, return loss, 2W to 4W hybrids, the effect real local loops have on
modem performance and how to squeeze all the modem performance possible
out of a modem. 

When I stated I've been asked to diagnose poor performing modems and come
up with a way to fix them I meant just that.  In almost all these cases
the original DAA was changed so the modem would pass approval in a
specific country.  This was done sans ever testing the modem performance
after the changes were made.  The changes to the DAA were made for one
reason and one reason only, to insure the modem would pass approvals.  In
my opinion it was done sans any regard or thought on how the changes might
effect modem performance. 

What me and Joe were trying to say was a return loss of 6 dB is adequate
for a modem DAA.  6 dB is not adequate for terminal equipment which is
used for voice conversations.  For obvious reasons. 

We were talking about two different things.  One is the requirement for
stringent return loss on a modem DAA .  And second I was saying the
addition of circuitry within the DAA to meet stringent return loss
requirements can effect the performance of the modem.  This is very true
if the return loss circuitry added to the DAA over compensates for return
loss and causes an inordinate transmit signal of loss within the DAA.  To
overcome the loss the transmit signal must be increased compensate for the
loss.  This will generally cause poor performance. 

I design my own external 2W to 4W hybrid that takes into account the
variable length of the local loop.  This hybrid will generally have a THL
of at least 20 dB on about 96% of all local loops and as much as 30 dB or
more is not uncommon.  It does not work well when the reflected impedance
is 600 ohms.  I have to provide an option strap for that.  I won't try to
describe how it functions.  This is not the forum, time or place to do so. 

Who is the manufacturer of the modem with Rockwell chips that has good
performance?  I worked for Rockwell 7 years and designed several hundred
DAA's for customers.  The one's I designed are not the passive hybrid
DAA's used in the Rockwell reference designs.  The Rockwell passive
designs were designed by someone else.  Mine always have an active 2W to
4W hybrid. 

When Joe Randolph and I were speaking about return loss we were referring
to PTT return loss speed to used a complex impedance as a reference.   I 
n the U.S. Part 68 does not have a return loss requirement.  The only U.S.
return loss specification I know is a recommendation in RS-496.  Which 
states:

"It is DESIRABLE the return loss conform to the tables below:

Frequency               Minimum Return Loss

260 - 500 Hz               7.0 dB   (SRL Lo)
560 - 1960 Hz             11.0 dB   (ERL)
2200 - 3400 HZ            14.0 dB  (SRL Hi)

The reference impedance is 600 ohms in series with a 2.16 uF capacitor.

In my DAA 's designed to source 600 ohms the return loss is generally
between 30 and 45 dB.  I am unable to get that much if the DAA is required
to used a complex impedance as a reference.  If I do the modem 
performance suffers. 

Amities,

Duane

Reply via email to