2 Jun
2009
2 Jun
'09
1:45 a.m.
cctalk-request at classiccmp.org wrote:
Message: 21 Date: Mon, 01 Jun 2009 20:16:20 -0500
From: John Foust <jfoust at threedee.com> Subject: Re: 1964 Antique MODEM Live Demo
To: <cctalk at classiccmp.org> Message-ID: <6.2.3.4.2.20090601201242.03e96008 at
mail.threedee.com> Content-Type: text/plain; charset="us-ascii" At 03:24 PM
6/1/2009, Gordon JC Pearce wrote:
Huh? The RS-232-port isn't supposed to deliver any power at all. Some
devices admittedly abused the RS-232 by using something like DTR to
actually supply the power needed to drive the thing, but that is abuse.
But I'd be surprised if a circa-70s modem was ever designed to use the
power from the RS-232 port to drive the modem itself. I'd expect it to
have an external power supply.
The problem description is a bit vague. How do you know it is generating
incorrect tones? Have you used a frequency analyser to check the tone
frequencies? Or might it just be that what has been observed is that the
modem don't work when used with the USB-to-serial? If so, then we have a
number of possible problems, none of which neccesarily is a problem with
the modem. Or is it just a case of listening to the modem while plugging
in the cable, and not the same thing happening on the different serial
ports. Might the USB-to-serial not start driving signals until something
talks to it on the software side?
I'd start by checking that the USB-to-serial really delivers correct
voltages. Break out the oscilloscope.
Next, I'd check if the USB-to-serial is happy without all kind of signal
"right", and if the modem supplies all the "right" signals. Also check
cables.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt at softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
>Okay, let's try again. I have a
circa-70s 300-baud modem, that will
>generate incorrect tones when it's plugged into a USB-to-serial
>converter, but not when it's plugged into a real serial port.
>Why might this happen?
USB only guarantees to supply 100 milliamps. Up to 500 mA can be
negotiated with the controller, depending. Might your older device
depend on more amps than the USB adapter can supply? 
2 Jun
2 Jun
9:06 p.m.
It was written...
Huh? The RS-232-port isn't supposed to deliver any
power at all.
Not quite true, but a case can be made.... As per the spec, pins 9 and
10
are reserved specifically to provide test voltages. The current available is
minimal and designed mainly for testing purposes with breakout boxes, etc.
But it is there (from those that chose to implement it - which back in the
day was most folks). Not only that, but it was not uncommon - in fact very
common - to find lots of devices that got their power from DTR, etc.
These devices had a lot of trouble with RS423, where the voltage was much
less than RS232. I don't recall the exact numbers but RS232 was something
about 12V swings and RS423 was about 3.5V swings. Other than that, the two
were pretty much identical. People often connected RS423 to 232 devices with
no problem at all. But those devices that expected power from DTR or pins 9
or 10 (on a DB25) had a bit of difficulty when plugged into a rs423 port.
But I'd be surprised if a circa-70s modem was ever
designed to use the
power from the RS-232 port to drive the modem itself. I'd expect it to
have an external power supply.
I can't imagine a 70's era modem getting
power from the interface. Well, not
a modem designed to interface to analog POTS lines. A line driver or short
haul modem? Heck yeah, I've seen scads of them that got their power from the
RS232 interface.
Jay West
3 Jun
3 Jun
8:37 a.m.
On Jun 3, 2009, at 12:06 AM, Jay West wrote:
These devices had a lot of trouble with RS423, where
the voltage
was much less than RS232. I don't recall the exact numbers but
RS232 was something about 12V swings
+/- 15V
and RS423 was about 3.5V swings. Other than that, the
two were
pretty much identical. People often connected RS423 to 232 devices
with no problem at all. But those devices that expected power from
DTR or pins 9 or 10 (on a DB25) had a bit of difficulty when
plugged into a rs423 port.
Some Sun systems have jumpers to switch their serial ports between
RS232 and RS423 modes.
-Dave
--
Dave McGuire
Port Charlotte, FL
9:03 a.m.
On Wed, Jun 3, 2009 at 11:37 AM, Dave McGuire <mcguire at neurotica.com> wrote:
On Jun 3, 2009, at 12:06 AM, Jay West wrote:
Formally, yes, that's the max signalling voltage, but in practice,
there's a lot of 1488/1489 circuits powered by 7812s and 7912s out
there. I am pretty sure that the MAX232 can't get things better than
about +/-10V (but I could be mistaken about that).
-ethan
These devices had a lot of trouble with RS423, where the voltage was much
less than RS232. I don't recall the exact numbers but RS232 was something
about 12V swings
?+/- 15V
9:10 a.m.
On Jun 3, 2009, at 12:03 PM, Ethan Dicks wrote:
Formally, yes, that's the max signalling voltage, but in practice,
there's a lot of 1488/1489 circuits powered by 7812s and 7912s out
there.
Very true; I was talking about the formal spec. I *think* the
spec actually says +3V to +15V and -3V to -15V, but I don't recall
for sure and I don't have it in front of me.
These devices had a lot of trouble with RS423, where
the voltage
was much
less than RS232. I don't recall the exact numbers but RS232 was
something
about 12V swings
+/- 15V I am pretty sure that the MAX232 can't get
things better than
about +/-10V (but I could be mistaken about that).
The MAX232, if memory serves, is specced at 10V but usually tops
out at 8-9V in my experience. I sure do love those chips. :)
-Dave
--
Dave McGuire
Port Charlotte, FL
9:19 a.m.
On Wed, Jun 3, 2009 at 12:10 PM, Dave McGuire <mcguire at neurotica.com> wrote:
There you go... that makes sense. I've even been known to tap a few
mA off of one of the negative charge pumps for LCD bias voltage for a
serial-interfaced LCD panel (192x64 pixels).
?Very true; I was talking about the formal spec. ?I
*think* the spec
actually says +3V to +15V and -3V to -15V, but I don't recall for sure and I
don't have it in front of me.
That is my memory, too, but I don't have it in front of me either (but
I used to make datacomm hardware, so I'm reasonably certain of it).
?I am pretty
sure that the MAX232 can't get things better than
about +/-10V (but I could be mistaken about that).
?The MAX232, if memory serves, is specced at 10V but usually tops out at
8-9V in my experience. ?I sure do love those chips. :)
Yep. I love 'em too, especially when they can let me stick with a
+5V-only design. Not as beefy as 1488s and 1489s, but a snap to use.
-ethan
9:31 a.m.
On 3 Jun 2009 at 12:10, Dave McGuire wrote:
The MAX232, if memory serves, is specced at 10V but
usually tops
out at 8-9V in my experience. I sure do love those chips. :)
I get about +/-10 out of the MAX232Ns from TI, but I use 10 uF/25v
caps instead of 1uF on them (just have more of them in my hellbox).
There are also the cap-less models from Maxim (can't recall the part
numbers) that are pretty tempting.
Some modern equipment gives you +5/0 for "RS232" levels. My DTV set-
top-box with a DE9 on the rear panel labeld "RS232" is such an
example.
Although, I suppose we're going to have to start calling it "TIA-
232".
--Chuck
11:11 a.m.
On Wednesday 03 June 2009, Chuck Guzis wrote:
On 3 Jun 2009 at 12:10, Dave McGuire wrote:
I've used the MAX 3233E recently, and it seemed to do what I wanted
(allow me to hook a serial console up to a device with a 3.3V port),
and I've used the 5V version (3235) before as well.
It's nice having a 1-chip design for an RS-232 level converter.
Pat
--
Purdue University Research Computing --- http://www.rcac.purdue.edu/
The Computer Refuge --- http://computer-refuge.org
The MAX232, if memory serves, is specced at
10V but usually tops
out at 8-9V in my experience. I sure do love those chips. :)
I get about +/-10 out of the MAX232Ns from TI, but I use 10 uF/25v
caps instead of 1uF on them (just have more of them in my hellbox).
There are also the cap-less models from Maxim (can't recall the part
numbers) that are pretty tempting.
Some modern equipment gives you +5/0 for "RS232" levels. My DTV set-
top-box with a DE9 on the rear panel labeld "RS232" is such an
example.
Although, I suppose we're going to have to start calling it "TIA-
232". 
23 Jun
23 Jun
9:42 a.m.
----- Original Message -----
From: "Chuck Guzis" <cclist at sydex.com>
To: "General Discussion: On-Topic and Off-Topic Posts"
<cctalk at classiccmp.org>
Sent: Wednesday, June 03, 2009 11:31 AM
Subject: Re: cctalk Digest, Vol 70, Issue 3
On 3 Jun 2009 at 12:10, Dave McGuire wrote:
I was interested to see this comment because I have a DS232AS I need to
replace, and the board it is on uses 10uF caps as you mention. The Dallas
Semiconductor data sheet mentions 10uF as an option "to reduce ripple".
However, neither the Maxim nor the TI datasheets I've downloaded mention
anything higher than 1.0uF. I was concerned that if I substitute one of the
other vendor's parts I would need to change the caps as well, but I gather
this is not the case, so thanks for that comment. I was concerned that the
higher capacitance value might somehow load down the charge-pump circuitry
or cause it to malfunction in some manner with a TI or Maxim part.
Later,
Charlie Carothers
My email address is csquared3 at tx dot rr dot com
The MAX232, if memory serves, is specced at
10V but usually tops
out at 8-9V in my experience. I sure do love those chips. :)
I get about +/-10 out of the MAX232Ns from TI, but I use 10 uF/25v
caps instead of 1uF on them (just have more of them in my hellbox).
There are also the cap-less models from Maxim (can't recall the part
numbers) that are pretty tempting.
Some modern equipment gives you +5/0 for "RS232" levels. My DTV set-
top-box with a DE9 on the rear panel labeld "RS232" is such an
example.
Although, I suppose we're going to have to start calling it "TIA-
232".
--Chuck
3 Jun
3 Jun
11:41 a.m.
On Wed, Jun 3, 2009 at 11:37 AM, Dave McGuire <mcguire at neurotica.com> wrote=
:
Formally, yes, that's the max signalling voltage, but in practice,
I thoght that one state was a positive voltage between +3V and +25V wrt
signal ground and the other state was a negative voltage between -3V and
-25V wrt signal ground. Anything between -3V and +3V is illegal.
There's no requiremnt, AFAIK, for the 2 votlages to be of the same
magnitude. A signal which swings between +12V and -5V is perfectly valid.
This explains a little circuit that I've sene used in small RS232
peripherals that run off their own wall-wart. Namely to use a 7905 (-5V)
regulator to power the logic. The common +ve raill (wall wart +ve,
regulator common pin) is the +5V line, the regulator output is the logic
ground (so there's the normal 5V supply, right way round, for the logic).
The RS232 driver is a discrete transsitor circuit running between the +5V
line and the (unregulated) input to the regulator -- the -ve side of the
wall-wart. Provided the latter is large enough, this does meet the RS232
spec.
-tony
On Jun 3, 2009, at 12:06 AM, Jay West wrote:
>
> These devices had a lot of trouble with RS423, where the voltage was muc=
h
> less than RS232. I don't recall the exact
numbers but RS232 was somethin=
g
about 12V
swings
=A0+/- 15V
12:25 p.m.
On Wed, Jun 3, 2009 at 2:41 PM, Tony Duell <ard at p850ug1.demon.co.uk> wrote:
I thoght that one state was a positive voltage between
+3V and +25V wrt
signal ground and the other state was a negative voltage between -3V and
-25V wrt signal ground.
You might be on to something there... I checked and the 1488/1489 pair
is rated up to +/-30VDC.
Anything between -3V and +3V is illegal.
Illegal, yes, but as others have pointed out, 0V might work with some
modern equipment.
There's no requiremnt, AFAIK, for the 2 votlages
to be of the same
magnitude. A signal which swings between +12V and -5V is perfectly valid.
True, and I've seen +12V/-5V designs when the manufacturer didn't want
to spring for -12V but had -5V lying around. I couldn't tell you
where right now, but it was in something from the 1970s.
-ethan
3:02 p.m.
Ethan Dicks wrote:
On Wed, Jun 3, 2009 at 2:41 PM, Tony Duell <ard at p850ug1.demon.co.uk> wrote:
Well, just to wade in with another 'opinion', if I have it correct:
the transmitter must drive the lines within the range +/-5V to +/-15V,
the receiver must respond correctly to signals in the range +/-3V to +/-25V.
The hysteresis provides some noise immunity; the wider receiver specs allows
for some line drop, noise, and common-mode bumping. There is also something
like a 1K to 7K impedance spec on a receiver.
--
If there were one thing about RS232 and (smart) modems that I wish had been
differently, it would be that the hayes modem standard had permitted some
technique to switch between data mode and control mode without having those 2
second guard periods. That was (is) so annoying waiting for dial-up software to
get the modem into control mode.
I haven't thought it through completely in relation to the RS-232 protocol
reqs., but perhaps using the control sigs in some manner, such as dropping DTR
and then sending some special character sequence; or using some
otherwise-unlikely combination of control sig states.
I thoght that one state was a positive voltage
between +3V and +25V wrt
signal ground and the other state was a negative voltage between -3V and
-25V wrt signal ground.
You might be on to something there... I checked and the 1488/1489 pair
is rated up to +/-30VDC.
Anything between -3V and +3V is illegal.
Illegal, yes, but as others have pointed out, 0V might work with some
modern equipment.
4 Jun
4 Jun
10:32 a.m.
Well, just to wade in with another 'opinion',
if I have it correct:
the transmitter must drive the lines within the range +/-5V to +/-15V,
the receiver must respond correctly to signals in the range +/-3V to +/-25V.
I think that is correct.
If there were one thing about RS232 and (smart) modems
that I wish had been
differently, it would be that the hayes modem standard had permitted some
technique to switch between data mode and control mode without having those 2
second guard periods. That was (is) so annoying waiting for dial-up software to
get the modem into control mode.
The point weas that the Smartmodem would work using only the TxD and RxD
data lines (it didn't use the control lines, simply because you couldn't
rely on them even existing on some RS232 interfaces) _and_ it could not
be fooled if the special sequence of characters to go to command mode
occured e.g. in a file download. It was assumed (I think quite
reasonably) that you'd not get the characters _and the pauses_ in a download.
-tony
3:37 p.m.
Tony Duell wrote:
It's no mystery what the original point was.
If there were
one thing about RS232 and (smart) modems that I wish had been
differently, it would be that the hayes modem standard had permitted some
technique to switch between data mode and control mode without having those 2
second guard periods. That was (is) so annoying waiting for dial-up software to
get the modem into control mode.
The point weas that the Smartmodem would work using only the TxD and RxD
data lines (it didn't use the control lines, simply because you couldn't
rely on them even existing on some RS232 interfaces) _and_ it could not
be fooled if the special sequence of characters to go to command mode
occured e.g. in a file download. It was assumed (I think quite
reasonably) that you'd not get the characters _and the pauses_ in a download. 
7:44 p.m.
Tony Duell wrote:
The point weas that the Smartmodem would work using
only the TxD and RxD
data lines (it didn't use the control lines, simply because you couldn't
rely on them even existing on some RS232 interfaces) _and_ it could not
be fooled if the special sequence of characters to go to command mode
occured e.g. in a file download. It was assumed (I think quite
reasonably) that you'd not get the characters _and the pauses_ in a download.
It was a bear to properly emulate that sequence in tcpser, as many BBS
systems used it to hang up:
+++
ath0
in lieu of dropping DTR.
I'm not quite sure I have it *exactly* emulated, but it must be close
enough, as I never get bug reports on that aspect of it anymore.
Jim
10:27 a.m.
On Wed, Jun 3, 2009 at 2:41 PM, Tony Duell <ard at p850ug1.demon.co.uk> wrote:
Of course. Most devices didn't bother to detect an illegal state on an
input line and moan about it. They just had an input stage with a
threshold somewhere between -3V and +3V. Any valid RS232 signal would be
handled correctly,
IIRC the well-known 1489 RS232 receiver chip has a threshold a little
about 0V. You can drive the input (RS232 side) of such a chip with a TTL
level signal and expect it to work -- I've done it myself many times for
testing/quick hacks. If I know the RS232 device I am tryign to drive uses
a 1489, I might well feed TTL signals into it. But I wouldn't do that in
anything I ecpected others to use.
Incidentally, I am suprised that any older RS232 device required 15V
signals and wouldn't work on 12V signals. I've got some older HP RS232
interfaces (11205, 11206) that use 741 op-amps to drive the RS232
outputs. They run off a +/-9V supply The slightly later HP11284 uses 1488
drivers, again running off +/-9V. AFAIK that is totally within the RS232
spec.
I thoght that one state was a positive voltage
between +3V and +25V wrt
signal ground and the other state was a negative voltage between -3V and
-25V wrt signal ground.
You might be on to something there... I checked and the 1488/1489 pair
is rated up to +/-30VDC.
Anything between -3V and +3V is illegal.
Illegal, yes, but as others have pointed out, 0V might work with some
modern equipment. There's no requiremnt, AFAIK, for the 2
votlages to be of the same
magnitude. A signal which swings between +12V and -5V is perfectly valid.
True, and I've seen +12V/-5V designs when the manufacturer didn't want
to spring for -12V but had -5V lying around. I couldn't tell you
where right now, but it was in something from the 1970s. 
3 Jun
3 Jun
11:30 a.m.
It was written...
Was it? I can't think of many classic RS232 devices that have those pins
wired.
Huh? The RS-232-port isn't supposed to
deliver any power at all.
Not quite true, but a case can be made.... As per the
spec, pins 9 and 10
are reserved specifically to provide test voltages. The current available is
minimal and designed mainly for testing purposes with breakout boxes, etc.
But it is there (from those that chose to implement it - which back in the
day was most folks). Not only that, but it was not uncommon - in fact very But I'd be
surprised if a circa-70s modem was ever designed to use the
power from the RS-232 port to drive the modem itself. I'd expect it to
have an external power supply.
I can't imagine a 70's era modem getting
power from the interface. Well, not
a modem designed to interface to analog POTS lines. A line driver or short
haul modem? Heck yeah, I've seen scads of them that got their power from the
RS232 interface.
5660
days inactive
5681
days old
17 comments
11 participants
participants (11)
-
ard＠p850ug1.demon.co.uk -
bqt＠softjar.se -
brain＠jbrain.com -
cclist＠sydex.com -
csquared3＠tx.rr.com -
ethan.dicks＠gmail.com -
hilpert＠cs.ubc.ca -
jwest＠classiccmp.org -
lists＠databasics.us -
mcguire＠neurotica.com -
pat＠computer-refuge.org