AVR-Chat

Peter Jakacki wrote:
> Russell Shaw wrote:
> 
>>Non-termination of such long lengths and of uncontrolled impedance
>>characteristics is asking for trouble. I'd only do that if i verified
>>the received waveform with a cro.
> 
> cro??? Of course I verified it with a scope and though I didn't "need" 
> termination I still used it.

Well if you verified it, that's all that's needed.

>>>>At my previous job, rs232 was run from the lab to various measuring equipment
>>>>outside (10-20m).
>>>>
>>>>I know of several radio hams that run rs232 from their radio room to things
>>>>outside and up radio towers (10-30m).
>>>>
>>>>RS232 was designed for unterminated lines which keeps things simple,
>>>>especially for multi-drop star networks. RS485 requires a single backbone
>>>>with terminator.
>>>
>>>RS-232 is not multidrop and at the low speeds it operates at would not 
>>>need termination.
>>
>>There are tristate rs232 chips for doing multidrop.
> 
> Please, I am fairly sure the RS-232 standard does not say multi-drop. I 
> have implemented multi-drop RS-232 in the distant past but I have better 
> things now. Read what Maxim themselves say (they must know I guess).

The maxim data sheet for max220-max249 say multidrop rs232.
http://www.datasheetcatalog.com/datasheets_pdf/M/A/X/2/MAX232.shtml

> "Typical RS-232 transmissions seldom exceed 100 feet for two reasons. 
> Firstly, the difference between transmitted levels (±5V) and receive 
> levels (±3V) allows only 2V of common-mode rejection. Secondly, the 
> distributed capacitance of a longer cable can degrade slew rates by 
> exceeding the maximum specified load (2500pF). Because the RS-232 was 
> designed as a point-to-point rather than multidrop interface" 
> (http://www.maxim-ic.com/appnotes.cfm/appnote_number/723)

A *long* cable might have 5-10nF of capacitance. A max3322 can
do 5V/us into 5nF, so it does a 10V swing in 2us. Assuming no
overshoot and ringing problems, a 12us cycle should be doable,
which is around 76.8kbaud (i'd usually use less than that).

http://www.maxim-ic.com/quick_view2.cfm/qv_pk/3612

>>>>RS232 can be fed thru radio coax and separated out with a filter, unlike
>>>>rs485 which requires a balanced line.
>>>
>>>Cool, I like that, but is it really real RS-232 anymore? (I think not)
>>
>>Using RF chokes and isolating caps, the coax looks like normal wire.
> 
> I have done similar things myself using inductors and caps to isolate 
> signals but I like the way amateurs improvise according to knowledge 
> without trying to rigidly adhere to standards for the sake of standards 
> (no longer RS-232).
> 
> 
>>This is the root cause of unreliable comms.
>>
>>                  long wire-pair
>> Box_A <======================================> Box_B
>>   |                                              |
>>   | Each box 3kohms common-mode Z to ground      |
>>   |                                              |
>>  --- case-ground       Several square           ---
>>  --- capacitance       metres loop area         ---
>>   |                                              |
>>   |  conductive ground/soil/floor/wiring         |
>> --+----------------<-Ignd->----------------------+--
>>                      Ignd: ground-current
>>
>>Your receivers are highly vulnerable to having their common-mode
>>voltage range exceeded by voltage drop caused by ground-current,
>>and loop voltage induced by fluctuating magnetic field within
>>the loop area. Higher-frequency noise will have more effect
>>because the case capacitance impedes less and the loop voltage
>>is proportional to magnetic field rate of change. It is highly
>>susceptable to long-distance lightning interference.
>>
>>Ground wire *must* be used for reliability.
> 
> No, you are assuming that current will flow, I do design professional 
> equipment for a living, not just for my garage. If my equipment is not 
> line-powered then I like to opto-isolate the 485. How can current flow 
> in these two situations??

I assumed you were saying the ground wire was not required even in
non opto-isolated setups. You can still have galvanic isolation, but
excessive capacitive coupling to a common ground can still be a problem,
such as capacitance between windings in mains transformers, and MOVs
that connect PCB ground to mains ground etc. With real optos, the
capacitance is small enough.

>>>Anyway, I like the fact that you are not strictly abiding by the 
>>>standard and doing what engineers do best .... tinkering ;)
>>
>>I always work with known and bounded parameters. Never "tinker".
> 
> Call it what you will, this is like the chicken and egg thing. How do 
> you know and bound parameters unless you "tinker" in the first place????

You always imagine how everything is to work, then design/simulate and build it.
I consider cut-n-try poking at hardware "tinkering", but i do calculations or
simulations to avoid that. All the tinkering should only be during debugging,
finding problems about invalid assumptions and mistakes etc. Simulation is
really tinkering in software, but it has to be done for things not easily
calculable, such as monte-carlo tolerance runs.