|
In that case the app you are using is doing framed packets, RTC/CTS stuff for example.
|
|
|
|
|
fwiw, I own both ends of the connection  hence the mystery......
Charlie Gilley
<italic>Stuck in a dysfunctional matrix from which I must escape...
"Where liberty dwells, there is my country." B. Franklin, 1783
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
|
|
|
|
|
Well your app/transport layer has to be using RTS/CTS DCD/DTR then. THat can be the only explanation.
|
|
|
|
|
People ask me why I still hang on to my breakout boxes, you just never know... I do agree with Munchies_Matt though, it's probably a hardware handshake issue.
"the debugger doesn't tell me anything because this code compiles just fine" - random QA comment
"Facebook is where you tell lies to your friends. Twitter is where you tell the truth to strangers." - chriselst
"I don't drink any more... then again, I don't drink any less." - Mike Mullikins uncle
|
|
|
|
|
charlieg wrote: Older machines, larger boxes may still have built in com ports exposed at the back, but most of us are living in the land of usb to serial.
surprisingly even the latest motherboards often have a serial header - pretty standard layout, it's the case manufacturers that don't provide the header. I believe there are headers out that that just replace one of the expansion card slot covers with a DB9/25 and plug via a short ribbon into the mobo header.
Another option I've seen used are boxes that connect serial via network (RJ11/45 one side, 9 pin t'other)
- really handy for connecting remote items (i.e. old manufacturing equipment) where there's already network cables/routers running nearby.
- usually have a web interface for setting the serial options as well as IP address etc (no need to fart about with dip switches)
- have a few different modes of operation... the easiest is it makes the device look like any other TCP/IP connection/stream (as a TCP/IP client)
- all the serial madness set/hidden in the web interface = (in above mentioned mode) no need to use for example the .Net Serial Comms API which [BTW, barely maintained] although not bad does have quirks - particularly if the device/cabling is a bit flaky or keeps dropping/closing the serial link.
Message Signature
(Click to edit ->)
modified 14-Feb-19 10:22am.
|
|
|
|
|
Hi Charlie,
AFAIK you have to connect RTS / CTS, this is a hardware-Handshake.
Have a Look to Wikipedia for an Explanation
Greetings,
Klaus
|
|
|
|
|
I made a bunch of various such cables back in the 90s, and three-wire doesn't always get the job done.
Not everything I had to connect used EIA 232 (RS-232). All of our terminal servers were from DEC so EIA 423 (DECconnect) was most common and there were also some others.
Eventually I wrote up a Word document with all the pinouts I knew and the types of cross-wirings I had to make. I still have the document, but some of the diagrams are missing because I hadn't inserted them properly.
The last time I had to make such a cable was to connect a DB9 RS-232 USB adapter to the MMJ DECconnect console of one of my AlphaServers and it didn't work reliably as three-wire. It doesn't help that the MMJ console port is flaky.
modified 14-Feb-19 12:28pm.
|
|
|
|
|
It depends on the pin-out of whatever you're connecting it to -- "standards" is often considered an optional term, in serial comms.
RS232 only uses three wires, but not every manufacturer always uses the same pins.
What I often do is cut the end off an RJ45 (Ethernet-style) cable, and solder it up to the pins of the DB9 or whatever as described in the manual for the serial device (or by using a meter to find which pins are used).
You can afford a bit of trial and error (mixing up Tx and Rx is quite easy, because a wire that's Tx at one end is Rx at the other, and the documentation for serial devices can be confusing, on that point), because it's only comms voltage, so you won't burn anything out.
I wanna be a eunuchs developer! Pass me a bread knife!
|
|
|
|
|
RS232 is actually an electrical/mechanical specification, within that you can use as many of the pins as you like, depending on the protocol used, such as SDLC requiring CTS/RTS framing.
|
|
|
|
|
What piece of software are you using on PC side?
In one of our legacy app we are using the WIN32 API to control the COM port. Since USB2Serial Adapter we made the often the expierience that we need additional to ...
dcb.fInX = 0;
dcb.fOutX = 0;
... where dcb which is explained here: _DCB | Microsoft Docs[^]
It does not solve my Problem, but it answers my question
modified 19-Jan-21 21:04pm.
|
|
|
|
|
I know that link well. On one side is a win32 bastard on WEC7 (as well as CE 5.0). The other side is a PC (Xp, Win7, Win10) - very generic. Since the same API is supported it's pretty much the same.
We're at very, very basic settings.
Charlie Gilley
<italic>Stuck in a dysfunctional matrix from which I must escape...
"Where liberty dwells, there is my country." B. Franklin, 1783
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
|
|
|
|
|
Ah yes, the DCB  When I inherited this code, no one thought to initialize that structure before use....
The s/w on the PC side is a simulation test tool (implements communications protocol for some ancient piece of equipment). I have full control over it. I'll see what's going on over there...
Charlie Gilley
<italic>Stuck in a dysfunctional matrix from which I must escape...
"Where liberty dwells, there is my country." B. Franklin, 1783
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
|
|
|
|
|
 good luck
It does not solve my Problem, but it answers my question
modified 19-Jan-21 21:04pm.
|
|
|
|
|
You still need to wire up a few more pins.
RS-232 and RS-422[^]
3/4's of the way down are diagrams for null modem / crossover cables. See also the diagram after the null model cable, on the software handshaking cable. Even if you don't use software handshaking, wiring it up that way is a good idea.
Latest Article - Slack-Chatting with you rPi
Learning to code with python is like learning to swim with those little arm floaties. It gives you undeserved confidence and will eventually drown you. - DangerBunny
Artificial intelligence is the only remedy for natural stupidity. - CDP1802
|
|
|
|
|
I have a date with my soldering iron on Saturday. It's okay. The wife of 40+ years will be working, and it's going to rain where I live. 
Charlie Gilley
<italic>Stuck in a dysfunctional matrix from which I must escape...
"Where liberty dwells, there is my country." B. Franklin, 1783
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
|
|
|
|
|
Yeah - I should know that list of alternate null modems by heart; in the 1990s I was teaching that stuff to engineering students. All the different alternatives for flow control.
Soldering it was a minor problem, compared to making the students understand the need for flow control - and the need for so many different mechanisms. With X-ON / X-OFF (software control) coming in as yet another alternative.
The survey you link to is a good description of the issues. Recommended reading for anyone who needs to use COM port communication anno 2019.
|
|
|
|
|
It's going to depend on how the software on the computer side sets up the port. If it sets it up such that it needs to see some particular signal line enabled before it will talk, then you will have to set up that line and enable it on your side.
Or, if it's doing that, it may also be enabling an outgoing one as well. If so, you can always just loop that one back on the one it wants to see high, allowing it basically to enable itself. So wire RTS to CTS and/or DTR to DSR. Most likely the former I would guess.
|
|
|
|
|
Years ago I bought 100m normal twin flex (2 wires only) and cut it into 2 pieces - a 75m and a 25m. Soldered them booth up to db25 connectors (pin 2 to 3 and pin 3 to 2 ; no pin 5) used both with laplink, the 25m at work (office to office) and the 75m between my flat on the 4th floor and a friend on the 2nd floor in the next building of flats. worked fine.
(well, lost a serial card, you tend to forget that it is connected. there was some lightning in the surrounds, and the computer shut down a second or two before I heard the thunderclap. Fortunately only the serial card was lost - The computer started up normally when the serial card was removed)
|
|
|
|
|
Two most common causes of mistakes I've faced:
1. Cable crossed or straight. That ties in neatly into male vs. female connectors. At the end, I've built myself an own cable tester to find out which is straight and which is crossed.
2. Hardware flow control lines. Some devices use them, some don't and if a device uses that, you better solder the lines in place.
|
|
|
|
|
If you can get at the comm port settings, set Handshaking to NONE. It should then ignore everything except the RX/TX.
|
|
|
|
|
Remember to cross your fingers as well, hoping for the receiver always being ready to pick up another byte when the sender insists on transmitting it.
Flow control was included for a reason. Sure, most of our modern CPUs are so fast that they can "always" keep up with a COM port, but if it in exceptional cases are busy with something else, it is still nice to be able to tell the sender "Could you hold it for a little while?"
|
|
|
|
|
If I have the option - and flow control is needed, I prefer to use software handshaking (XON/XOFF)
|
|
|
|
|
And it works over a 2 line connection, such as a POTS line.
BUT: It is not bit transparent. And it requires the sender to respond fast enough. The latency is generally a lot higher with software flow control.
I was teaching networks at college level in the 1990, when 622 Mbps B-ISDN and ATM was still an option, and optic fibers were becoming commonplace, running for 50 km or more without repeaters. (They could run for much greater distances even in the 1990s, but towns are usually no further apart.) I used to start first lecture on B-ISDN with a question: How big is a bit? How many centimeters long? Light in a fiber moves at roughly 200,000,000 m/s. If you sende 622,000,000 per second, that gives them about 30 cm each. Before the first bit has come to the end of a 50,000 m long fiber, you have poured in more than 150,000 bits. And before the STOP!!! signal has come back to you, you have sent another 150,000 bits down the line. Now, let us see how well our basic frame mechanimsm from lower speed lines works under those conditions...
Lots of the students first thought I had made an error in the calculation, missing by a factor of a thousand. After checking and counting zeroes, they shook their head with a "But... What shall we do then?". (The next layer-1 protocol we treated was DQDB - another now forgotten alternative, which is a pity; it did have something going, with its reservation scheme. It never made any success, but we should learn for failures, too - failure in the market doesn't prove that it was without merits. Around 1990 it taught the students that there is more than one way to skin a flow control cat!)
|
|
|
|
|
I've read through everything, and I'll suggest the one thing that noe one else has
You said you have a number of pins not connected right? WOuld they at all happen to all be grounds?
A 10ft cable that's NOT properly shielded will loose quite a lot to stray radio waves in the surrounding environment. I have a 10 meter one that I can drop down out of my Loft hatch and into my PC in the office below, fro when I need to reprogram the switch in my server rack.
Over the years, it's starting to get old, and I can tell straight away when the shielding is starting to come loose.
The shorter the cable is, the less shielding matters, the GPS plugged into my PC for example has a really thin unshielded cable on it, but it's only a meter in length, the switch programming cable is a nightmare, we have so much wireless kit in our house with Phones, Thermostats, WiFi based TV's and Wifi capable smoke alarms that it's often better for me to find the shorter cable and drag either the PC system unit or one of the older laptops with a 9 pin on up intot he loft and do the work up there.
Cisco flat cables with an RJ45 on that are longer than about 3 meters are terrible, they pick up all kinds of crap, so much so I have some old ferrite chokes lying around that I salvaged from old electronics, and wraping the cable through them a couple of times, one at each end near the plugs, usually helps a lot.
|
|
|
|
|
Gentlemen,
Excellent suggestions all with debugging the cable differences. I'm going to have to file this under face-palm. 
It is a true statement that simple serial communications only require the 3 wires, we do it all the time on various projects. Points go to Mr. DCB suggestion man - if one's handling of the data is fast enough, you really don't need Dtr and Rts control wires. But if your application enables them in the DCB, you're going to need to hook them up.
So, the mystery is solved, and there is a lot of good info in this discussion. Thanks to all of you.
Charlie Gilley
Most life and software issues can be solved by checking your inputs....
"Where liberty dwells, there is my country." B. Franklin, 1783
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
