serial cable too long

[Dan]

I'm using a cat 5 cable for serial communication between a PC and an
embedded controller card. I can get chars back from the card (using a
terminal program) but I can't send. Moving the PC next to the card solves
the problem for now. I'm just wondering if there's any way to boost the
signal. The cable stretches about 50 or more feet.

Dan

 

[Peter Bennett]

I'm using a cat 5 cable for serial communication between a PC and an
embedded controller card. I can get chars back from the card (using a
terminal program) but I can't send. Moving the PC next to the card solves
the problem for now. I'm just wondering if there's any way to boost the
signal. The cable stretches about 50 or more feet.

Dan

What baud rate are you using? Are you using Real(tm) RS-232 line
drivers and receivers at both ends?

You should have no problems getting reliable RS-232 communications at
well over 50 ft at 9600 baud.

 

[zliminator]

What baud rate are you using? Are you using Real(tm) RS-232 line

drivers and receivers at both ends?

You should have no problems getting reliable RS-232 communications at
well over 50 ft at 9600 baud.

I'm just using a teminal emulator (tglmicro.com) and the card is from
technologic (embeddedx86.com). I've used this card (ts-5300) and the ts-2800
and I've noticed some flakieness with the serial port, but it DOS and easy
to work with and besides, I'm just now getting into using winsock on the PC
to talk to the card which has the Watcom driver and it also has a flash
card. So there's more than one way to skin a cat.

Dan

 

[Martin Riddle]

There is a maximum cable length speced out at 50', but thats at a fixed capacitance value.
The max speed for this combination is 300 baud. I forget the cap value, but the cat5 cable should be
well under it. TI has a nice paper on 232/485.
Either lower the baud rate or get a short haul modem.

Cheers

 

[The real Andy]

There is a maximum cable length speced out at 50', but thats at a fixed capacitance value.
The max speed for this combination is 300 baud. I forget the cap value, but the cat5 cable should be
well under it. TI has a nice paper on 232/485.
Either lower the baud rate or get a short haul modem.

Throw a loopback tester on the cable to check its not faulty. If it works,
implement a proper (non max232) driver IC on the far end. Use Jim Thompsons
ripper 232 drivers, that is if they are still available!!

 

[Robert Baer]

What baud rate are you using? Are you using Real(tm) RS-232 line
drivers and receivers at both ends?

You should have no problems getting reliable RS-232 communications at
well over 50 ft at 9600 baud.

--
Peter Bennett, VE7CEI
new newsgroup users info : http://vancouver-webpages.com/nnq
GPS and NMEA info: http://vancouver-webpages.com/peter
Vancouver Power Squadron: http://vancouver.powersquadron.ca

I understand that 2 miles of twisted pair worked at SLAC...

 

[Roger Lascelles]

I would never consider RS232 for a 50 foot distance, not for serious work.
The two devices could have ground potential differences also.

Why not put an RS232 to RS422 adapter at each end ? If its noisy, use opto
isolated RS422.

Roger

 

[John Popelish]

I'm using a cat 5 cable for serial communication between a PC and an
embedded controller card. I can get chars back from the card (using a
terminal program) but I can't send. Moving the PC next to the card solves
the problem for now. I'm just wondering if there's any way to boost the
signal. The cable stretches about 50 or more feet.

Dan

Actually, reducing the voltage swing may work better. RS232 ports
have a limited current output (300 ohms output impedance, usually),
but can swing anywhere between about 3 and 12 volts positive and
negative. That means that is the swing is near the upper limit, most
of the current capability of the driver is used to drive cable
capacitance, delaying the peak voltages, considerably, making the
delay dependent on data pattern. But since the spec allows the
voltage swing to be down to three volts, most receivers work with
voltage swings well below that.

I have made long serial lines work by clamping the voltage swing at
the receiving end (of transmit and receive) with a pair of green leds
across signal to ground, one facing each way. This limits the swing
to about 2 volts, and keeps the capacitive delay to a minimum. It
also provides a nice signal display. Parallel any spare conductors
with the signal ground.

 

[Tim Shoppa]

I'm using a cat 5 cable for serial communication between a PC and an
embedded controller card. I can get chars back from the card (using a
terminal program) but I can't send. Moving the PC next to the card solves
the problem for now. I'm just wondering if there's any way to boost the
signal. The cable stretches about 50 or more feet.

I would advise against "boosting" the signal, but do make sure the
drivers and receivers at both ends conform to the RS-232 standards.

In good conditions (small ground differential, good shielded cable,
no induced noise) you ought to be able to do 19200 baud for 50 feet
quite reliably.

What baud rate are you using? In almost all cases going to a lower baud
rate will get you more distance with slew-rate limited RS-232.

Tim.

 

[Jim Thompson]

I would advise against "boosting" the signal, but do make sure the
drivers and receivers at both ends conform to the RS-232 standards.

In good conditions (small ground differential, good shielded cable,
no induced noise) you ought to be able to do 19200 baud for 50 feet
quite reliably.

What baud rate are you using? In almost all cases going to a lower baud
rate will get you more distance with slew-rate limited RS-232.

Tim.

Most modern serial chips *DON'T* conform to RS-232, they use a much
smaller signal than the "standard" of ±10V.

If you can find some of my old original 1488's and 1489's it should
work over longer distances.

Or roll your own with comparators.

Or, maybe the best bet, use differential transceiver chip sets and
convert to RS-232 on each end.

...Jim Thompson

 

[Jim Backus]

Actually, reducing the voltage swing may work better. RS232 ports
have a limited current output (300 ohms output impedance, usually),
but can swing anywhere between about 3 and 12 volts positive and
negative. That means that is the swing is near the upper limit, most
of the current capability of the driver is used to drive cable
capacitance, delaying the peak voltages, considerably, making the
delay dependent on data pattern. But since the spec allows the
voltage swing to be down to three volts, most receivers work with
voltage swings well below that.

I haven't got the 232E spec in front of me, but I'm fairly sure that
the generator voltage is specd at +/- 5 volts minimum. Nominal
resistive load is 5 k and the capacitance is (IIRC) 2500 pF. Receivers
are supposed to detect +/- 3 volts.

I sometimes wonder if 232 is the most abused standard!

 

[George]

You should have no problems getting reliable RS-232
communications at well over 50 ft at 9600 baud.

Indeed. By the way, how do you suppose they do this:

http://www.ebusinesscables.com/ultimate_usb_extension.htm

If it really handles fast USB over those distances, how do
you think they do it? And it appears that the amplifiers
work off of the 5V on the USB.

 

[Tim Shoppa]

Most modern serial chips *DON'T* conform to RS-232, they use a much
smaller signal than the "standard" of ±10V.

MAX232's may not be "modern", but they do swing
to +/- 7 or 8 V with 5V Vcc. It's not too different than a 1488 driver
running from +/- 9V.

If you can find some of my old original 1488's and 1489's it should
work over longer distances.

Don't get me wrong, I love the 1488 and 1489. But for shortish cable runs
(<100 feet or so) at high speed and minimal ground differential, I believe
that the design decision of using a smaller swing at the driver is the
right choice. It lets you do higher speeds at the same slew rate, and that
seems to be what the majority of users want.

Or roll your own with comparators.

I do like the tunable threshold of 1489's, because I don't like the
"manufacturer suggested" bias. I grew up with bipolar relays as RS-232
receivers

Or, maybe the best bet, use differential transceiver chip sets and
convert to RS-232 on each end.

Or just put a differential card in the PC and forget all the RS-232 levels.
Again, nothing against RS-232, I've been using it since Model 33 teletypes,
but differential is a real win.

I think the original poster was attempting to do some ridiculously high
speed (115200 baud?) over a 50-ft serial cable, and that's beyond any
reasonable expectations. 38400 will often work at 50ft but it is very
touch-and-go.

Tim.

 

[budgie]

(SNIP)

I sometimes wonder if 232 is the most abused standard!

You can stop wondering. It is.

 

[John Popelish]

I haven't got the 232E spec in front of me, but I'm fairly sure that
the generator voltage is specd at +/- 5 volts minimum. Nominal
resistive load is 5 k and the capacitance is (IIRC) 2500 pF. Receivers
are supposed to detect +/- 3 volts.

I sometimes wonder if 232 is the most abused standard!

I appologize for my poor memory of RS232 specs, but your corrections
only reinforce the validity of my suggestion (in my humble opinion).

 

[Robert Baer]

I would never consider RS232 for a 50 foot distance, not for serious work.
The two devices could have ground potential differences also.

Why not put an RS232 to RS422 adapter at each end ? If its noisy, use opto
isolated RS422.

Roger

You are full of it.
1) RS-232 is speced to handle up to 25V difference.
2) Even parallel printers can be driven over cable longer than50 feet
with zero problems.
3) RS-232 serial works reliably over hundreds of feet of cable with zero
problems; i have done that so many times that i have lost count 30 years
ago.
And no special equipment like line drivers, line receivers or
repeaters.

 

[Robert Baer]

I haven't got the 232E spec in front of me, but I'm fairly sure that
the generator voltage is specd at +/- 5 volts minimum. Nominal
resistive load is 5 k and the capacitance is (IIRC) 2500 pF. Receivers
are supposed to detect +/- 3 volts.

I sometimes wonder if 232 is the most abused standard!

Hell, during the CP/M daze, almost nobody honoured the specs.
No regard as to which sex the connector should be, no regard as to
RX/TX pins should be, unbelievable variations for handshaking (if any),
etc, etc and etc (as Yul Brinner in the King and I used to say).

 

[Robert Baer]

I would advise against "boosting" the signal, but do make sure the
drivers and receivers at both ends conform to the RS-232 standards.

In good conditions (small ground differential, good shielded cable,
no induced noise) you ought to be able to do 19200 baud for 50 feet
quite reliably.

What baud rate are you using? In almost all cases going to a lower baud
rate will get you more distance with slew-rate limited RS-232.

Tim.

Slew rate limiting is an excellent way.
In the original Fairchild daze, they made a triple driver and triple
receiver; the driver used internal capacitance and switched current
source/sink to produce rate limited signals for the line driver; i think
the receiver did not need a negative swing, and certainly worked with as
little as 3V on the input.