RS-232 Standard Voltage Levels

[msimunic]

Hi everyone,

I'm wondering why are chosen such voltage levels for RS-232, inverted, non symmetric forbidden band.
Standard defines signal levels as "0" = +4V .. +12V; "1" = -3V .. -12V.
Then there are control signals (RTS, CTS, DTR, DTE, DCE) which have inverted logic "OFF" = -3V .. -12V; "ON" = +4V .. +12V
I'm aware that there are different voltage levels for RS-232 which can be used (3V up to 25V) but I think that original standard defines +12V.

I'm curious for historical reason why 12V, why forbidden band from -3V to +4V and why inverted logic for control signals were chosen for the RS-232 standard?

[ebastler]

Where do you find this asymmetrical definition of the dead band?
All sources I see describe a symmetrical dead band (-3V to +3V).

Must be for "hysterical" reasons indeed. 

[rs20]

If there was no dead band, then a device could output +0.001V for 0 and output -0.001V for 1, and would be classifiable as a compliant RS-232 device.   Instead, the contract is that receiving devices don't need to be that sensitive, they only have to be able to understand any signal stronger than 3V, and the transmitter must provide a signal stronger than 3V -- that way, all RS 232 compliant devices will be compatible with each other.

Guesses for other questions:

Higher voltages are more immune to noise (shout louder to be heard over the noise.)

Inverted logic is either just an arbitrary, harmless decision, or a historical optimization to save a few transistors (a logical inverter has less transistors than a logical buffer - a buffer is most often implemented as 2 inverters.)

[ebastler]

If there was no dead band, then a device could output +0.001V for 0 and output -0.001V for 1, and would be classifiable as a compliant RS-232 device.   Instead, the contract is that receiving devices don't need to be that sensitive, they only have to be able to understand any signal stronger than 3V, and the transmitter must provide a signal stronger than 3V -- that way, all RS 232 compliant devices will be compatible with each other.

That's clear (to me, and I believe to the OP as well). His question was why the dead band is asymmetrical (-3V .. +4V). But these asymmetrical levels may just be incorrect information; I could not find them confirmed anywhere.

[Bruce Abbott]

RS232C  is +-5V to +-15V transmit and +-3V to +-25V receive.

The first RS232 standard was released in 1960 as a 'recommendation' for connecting teletypes to modems. Early teletype machines were totally electromechanical, using a relay to receive the signal. With a high signalling voltage they could communicate over quite long distances via telegraph lines. Negative line voltage was usually used because a positive earth doesn't corrode (same reason some motor vehicles used positive earth). 

The negative RS232 'mark' voltage was probably chosen chosen to match an existing teletype interface. Op amps typically operate on +-15V, so that may have been a common supply voltage in modems.
           
The RS232 Standard

[ebastler]

RS232C  is +-5V to +-15V transmit and +-3V to +-25V receive.
       
The RS232 Standard

Thanks, Bruce. From memory I would also have stated the 5..15 V transmit voltage range -- but the standard you quote does not seem to mention it?

[msimunic]

Thank you. Yeah, I understand reasons about dead band.
Regarding asymetrical dead band - I saw some handouts from class about communication networks. May be an old information, variant of RS-232 compatible standard or just an error.
Most certainly that voltage levels were agreed to maintain compatibility with some older systems. Communication would work with other voltage levels also.
This standard is quite old and now days almost nobody knows why exactly those levels were chosen, but perhaps, someone encounters information about it.

[rsjsouza]

If you are interested, check another interesting reference that talks a bit about the history and more specifically about the latest revision F:
http://www.ti.com/lit/an/slla037a/slla037a.pdf

[Bruce Abbott]

Thanks, Bruce. From memory I would also have stated the 5..15 V transmit voltage range -- but the standard you quote does not seem to mention it?

The problem is that the actual standard is not available online. That website was the best I could google up in a few minutes.

Here's an app note on EIA/TIA–232–E from Dallas Semiconductor, which mentions the +-15V spec. it says,

"The original RS–232 standard was defined in 1962. As this was before the days of TTL logic, it should not be surprising that the standard does not use 5 volt and ground logic levels. Instead, a high level for the driver output is defined as being +5 to +15 volts and a low level for the driver output is defined as being between –5 and –15 volts. The receiver logic levels were defined to provide a 2 volt noise margin. As such, a high level for the receiver is defined as +3 to +15 volts and a low level is –3 to –15 volts."

[woodchips]

I remember RS232 being used to wire VDUs around factories and offices, where the supposed length limit, 200' wasn't it, would be exceeded by many times, always worked.

Are you sure TTYs were for RS232, I used the 20mA current loop interface, another that seemed to be able to run for miles.

[SeanB]

RS232 length is also going to depend on the data rate, you can go a lot further if you go slow, and higher data rates mean a shorter maximum length. Going over 200 feet with 9600 baud was doable mostly because most drivers were capable of higher current than the spec, and most receivers were capable of better performance, the specs published were a must meet spec, all would be somewhat better than this.

Current loop was capable of kilometres at 50 BAUD, seeing as this was the Telex data transfer method and 50 BAUD was the rate.

Think I had till recently some ISA serial cards with the requisite optocoupler and drive transistor with 220R resistors to provide the drive current for a current loop.

[rsjsouza]

IIRC correctly, the 3270 terminals at the university and on my dad's work used simple telephone twisted pair - some links had more than 1km.

[Astrodev]

Somewhere in a box I have got a copy of the EIA spec as I used to work with these interfaces a lot back in the 80's and still do to an extent today, as I remember the NRZ approach was to give better long distance transmission without the cost overheads of using a balanced lines as all the signals could be referenced to a single ground.

Obviously due to the need to have a degree of hysteresis in the received signal there has to be a deadband around 0V but some receive circuits did work closer to this and seen to remember that some would work to as little as +/- 1.8V.

One thing that used to cause quite a few problems was that some of the equipment we used to work with was parasitically powered off the handshake signals and if the transmit voltage was low (below 15V) there would sometimes be issues with reliability due to the lack of sufficient power.

[dacman]

+/-12V is used a lot because personal computers have those voltages already, and if +/-12V will work, then +/-15V supplies do not need to be added.

[rx8pilot]

It is funny to see this topic in Vintage Computing.....

I just designed in 2 channels of RS232 for a brand new, super high tech project. I thought the customer was mistaken.....but RS232 is still very alive so many decades later.

[T3sl4co1l]

FWIW, receivers usually have an actual threshold around 1.5V.  This is so a dead line condition is still a well defined logic level, and so that, in a pinch, raw TTL (still inverted, mind) can be used.  Supposedly, a lot of PCs cheaped out on the interface, using TTL or LVCMOS outputs (3.3V supply, CMOS pin driver e.g. 74HC).  Similarly, the parallel port became LVCMOS, though this isn't far off from the original TTL so it's okay.

Tim

[Rerouter]

Also the moment your dealing with peripheral interfaces in automotive, its almost always rs232 rs485 or rs422.

J1939 canbus tends to only be for integration level stuff used by OEM's. Because its veiwed as risky to broadcast in to a canbus you dont have permission by the OEM for.

And so long as you really go heavy on your esd protection, they remain uneffected by everything from bad alternator connections to cb radio antenna cables in the same bundle.

[DeanCording]

I'm curious for historical reason why 12V, why forbidden band from -3V to +4V and why inverted logic for control signals were chosen for the RS-232 standard?

The inverted logic on the control signals may have something to do with current balancing.  A typical RS232 cable only has a single wire for signal ground, so the design may be intended to minimise current flow on that wire.  Mind you, current flow should be minimal anyway for just signalling, but some interfaces did use the signals to drive indicator LEDs.  Whilst the RS232 specification doesn't mention current, the commonly accepted value is about 45mA per line.

As for the forbidden band, it is intended to account for voltage drop from long cable runs which may drop the signal below the noise level. 

[2N3055]

From TIA/EIA-232-F Electrical Specification....

All 232 circuits carry voltage signals, with the voltage at the connector pins not to exceed ±25 V.
All pins must be able to withstand a short circuit to any other pin without sustaining permanent damage.
Each line should have a minimum load of 3 kΩ and a maximum load of 7 kΩ, which usually is part of the receiver circuit.
A logic 0 is represented by a driven voltage between 5 V and 15 V and a logic 1 of between –5 V and –15 V.

At the receiving end, a voltage between 3 V and 15 V represents a 0 and a voltage of between –3 V and –15 V represents a 1.
Voltages between ±3 V are undefined and lie in the transition region. This effectively gives a 2-V minimum noise margin at the receiver.

Hope that clears it up..

Regards,

[David Hess]

FWIW, receivers usually have an actual threshold around 1.5V.  This is so a dead line condition is still a well defined logic level, and so that, in a pinch, raw TTL (still inverted, mind) can be used.

If you had not posted this than I would have.

Offsetting the switching threshold away from 0 volts, around the TTL switching threshold is typical, allows the various auxiliary control signals to be left out without causing problems if you only need transmit, receive, and maybe a pair of flow control lines.

Supposedly, a lot of PCs cheaped out on the interface, using TTL or LVCMOS outputs (3.3V supply, CMOS pin driver e.g. 74HC).

Some designs did cheap out and use TTL or TTL threshold CMOS logic with some input protection instead of a true RS-232 receiver.  This is not quite as dumb as it seems at first glance because it is not significantly different from the input structure of a 1489 receiver.

The inverted logic on the control signals may have something to do with current balancing.  A typical RS232 cable only has a single wire for signal ground, so the design may be intended to minimise current flow on that wire.

This does not make any sense because current balancing could always break down; there are no forbidden states.

What really pisses me off is that the RS-232 standard assumes that the level shifters will be inverting (the 1488 has one inversion and the 1489 has three inversions in series) so all UARTs work with all level shifters but many "modern" products produce "RS-232 compliant formatting" (in the small print) with 0 to 5 or 0 to 3 volt levels but invert the signal from the UART so a standard level shifter will not work!  This mostly works with RS-232 receivers that have a TTL level input threshold but it does *not* work with an RS-232 standard level shifter so there is no easy way to use these with RS-232 devices which are expecting a minimum of +/-3 volts.

[guenthert]

It is funny to see this topic in Vintage Computing.....

I just designed in 2 channels of RS232 for a brand new, super high tech project. I thought the customer was mistaken.....but RS232 is still very alive so many decades later.

I have a hate/love relationship to RS232.  On the one hand there are a million (or maybe just 32) ways some RS232 devices might be actually wired, otoh terminal concentrators are ubiquitous in data centers (ok, last time I had a look, it's been a few years) and robust.  I presume RS232 lives on because it's "good enough".

[rx8pilot]

My last project required 2 channels of RS232. It was part of a very sophisticated digital image processing system just recently released for sale.
RS232 is still very common in the motion picture camera business - which is dominated with bleeding edge tech.