Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me.
|
Forum Index : Microcontroller and PC projects : Cable shielding
Page 1 of 2 | |||||
Author | Message | ||||
OA47 Guru Joined: 11/04/2012 Location: AustraliaPosts: 899 |
I have had issues with a project that is getting radio errors when a large soft start electric pump is in operation. To overcome this I have decided to try and locate the radio unit on the antenna. I will be using a 10 meter length of 4 core shielded cable between the radio unit and the RS232-USB converter to supply the 3V3, GND, Tx & Rx. My question is should I ground the shield of the cable at both ends or just one? OA47 |
||||
PeterB Guru Joined: 05/02/2015 Location: AustraliaPosts: 639 |
G'Day OA47 Just to make a real mess of your day I would use a 6 core cable. +5V or so to be regulated to 3.3V up the top GROUND Balanced + Rx Balanced - Rx Balanced + Tx Balanced - Tx Ground the shield at the bottom......probably The reason for only grounding one end is that if both ends are grounded, current will flow in the shield and hence radiate noise.............I think. Some smart young bloke will now shoot me down Peter |
||||
BrianP Senior Member Joined: 30/03/2017 Location: AustraliaPosts: 292 |
+1 B |
||||
mikeb Senior Member Joined: 10/04/2016 Location: AustraliaPosts: 173 |
Not shoot you down but just clarify what is happening. It's easier to picture if you draw the circuit out once you ground both ends. You actually produce a 'shorted' turn. When you consider transformer action, and how magnetic fields propagate around a conductor, you can see that induced currents are imposed on your conductors within the shielded cable. High frequency noise can be rectified, by the protection diodes built into most uC inputs, and produce a standing DC offset to the A/D converter also. In my line of work I come across many Soft Starters, and Variable speed Drives, which cause havoc with signals. A common solution to most of my problems is to pass my signal cables through a ferrite ring. No 'hard and fast' rules on number turns. Just as many as will fit. Regards, Mike B. Edited 2020-03-19 15:29 by mikeb There are 10 kinds of people in the world. Those that understand binary and those that don't. |
||||
Paul_L Guru Joined: 03/03/2016 Location: United StatesPosts: 769 |
Yep! Grounding both ends of the shield might produce a circulating loop current which can develop a voltage drop along either the ground wire or the shield. This would offset the ground reference voltage for the balance point of the Tx and Rx twisted pairs on one end or the other which would likely be amplified by one of the active elements and produce a background noise added to the signal. The Tx and Rx pairs should be true balanced differential pairs, tightly twisted so that any electrostaticaly induced voltage will be identical on both wires and thus cancel out at the receive end. Paul in NY |
||||
Volhout Guru Joined: 05/03/2018 Location: NetherlandsPosts: 3496 |
Grounding the shield: It is very simple. 1/ disconnect the shielding of the serial cable, but leave all other wiring connected. 2/ Take a multimeter and measure ohms between the 2 devices (at the shield connections). 3/ Is it 0 ohm (or < 10 ohm) then there is another ground connection => option A. If > 10 ohm => option B Option A: Connect the shield only on one side (avoid ground loops), generally this is the controller side (computer). If it is a very dirty environment (i.e. strong RF signals) connect the other side through a capacitor (i.e. 10nF Y cap). Option B: Connect both sides of the shield. The shield will also act as a discharge wire to avoid devices become charged. Do this for every device in a RS485 or similar system. It is very common to see the shield only hard connected at the main computer/PLC, and capacitor coupled to every device. Good luck, Volhout P.S. a Y capacitor is typically used becuase of it's design. When the capacitor fails it becomes and OPEN, not ashort. Edited 2020-03-19 17:49 by Volhout PicomiteVGA PETSCII ROBOTS |
||||
Tinine Guru Joined: 30/03/2016 Location: United KingdomPosts: 1646 |
Yeah I would go straight for the 232—422 converter with termination resistors. |
||||
Turbo46 Guru Joined: 24/12/2017 Location: AustraliaPosts: 1590 |
RS485 is made for this situation. I would use an RS232/RS485 converter at each end with a 120 ohm termination resistor. Send 5 volts up the wire with a 3.3v regulator at the top. Earth the shield ONLY at the computer end to avoid earth currents inducing noise into the signal lines. Communications should use a CRC with Acknowledge so the message can be repeated if corrupted. Bill Keep safe. Live long and prosper. |
||||
Tinine Guru Joined: 30/03/2016 Location: United KingdomPosts: 1646 |
Pretty useful ebook: RS422-RS485-Application-Guide-eBook.pdf |
||||
PeterB Guru Joined: 05/02/2015 Location: AustraliaPosts: 639 |
Good morning All Since non of you younger blokes want to pick a fight with me I will pick (another) fight with Paul. Paul stated "tightly twisted cable" Somewhere in the back of my mind there is something about "not too tight" and looking at photos in the eBook that seems about right. Is it that a tightly twisted cable will have increased C & L and hence reduced bandwidth ? I await your responses. Peter |
||||
Tinine Guru Joined: 30/03/2016 Location: United KingdomPosts: 1646 |
No, the way I remember it is a twist every 50mm. IIRC, this is from Serial Port Complete by Jan Axelson but I don't have the book to hand. It appears that twisted pairs are more common than shields when it comes to differential transceivers. Also, 10m probably wouldn't require termination resistors as I suspect that your BAUD is relatively low. 120R can be power hungry which would be a concern for a battery powered device. |
||||
Paul_L Guru Joined: 03/03/2016 Location: United StatesPosts: 769 |
Good afternoon All /KERFUFFLE MODE ON Hey Peter, the question is how much to twist rather than to twist or not. My answer is that I do not really know how tight the twist should be. I do know that Amphenol specs their CAT5E cables at 1Gbps or 350 MHz, CAT6 at 10 Gps or 550 Mhz, and CAT6A at 10 Gps or 650 MHz. The twist distance decreases from CAT5 to CAT6 to CAT6A. <www.cablesondemand.com>. Within a twisted pair capacative and inductive coupling between the conductors would increase in proportion to the both the gauge and the length of the individual conductors. The length of the conductor would increase as the rate of twisting increases. However, resistance would decrease in proportion to the gauge of the conductors which would reduce the time constant (R*C). This would seem to indicate that the higher CAT numbers would load a transmitter more with increasing length. Between the conductors of a twisted pair and other cables external to the CATn cable jacket both the inductive and capacative coupling would decrease with both decreasing twist length and increasing conductor gauge due to the increasing thickness of the conductor insulation and the cable jacket. This would seem to indicate that the higher CAT numbers would pickup less common mode interference from external circuitry. It sounds like a toss up. I'd buy the highest CAT number you can afford. /KERFUFFLE MODE OFF Tinine -- I just disected a CAT6 cable here and found a twist about every 12 mm. Paul in NY |
||||
Tinine Guru Joined: 30/03/2016 Location: United KingdomPosts: 1646 |
Heck there is even a calculator for this stuff |
||||
PeterB Guru Joined: 05/02/2015 Location: AustraliaPosts: 639 |
Australian overhead telephone lines are/were twisted. About 1 twist every 5 poles. How does that fit in with your high level maths? Peter |
||||
Turbo46 Guru Joined: 24/12/2017 Location: AustraliaPosts: 1590 |
PeterB, I can recall telephone lines in the country on poles with 4-6 cross-arms with unpteen wires on them but I don't recall them being 'twisted'. Paul, Kerfuffle is a good word. All OA47 asked was about earthing the shield after all. He already may have the cable that he intends to use and it may or may not have twisted pairs and that would not be a deal breaker although twisted pairs would be better. I don't believe that TTL RS232 will be viable over 10 meters whether twisted pair cable or not. As I said RS485 is made for this situation and 10 metres is peanuts for it. It should be OK using an untwisted pair over that distance if necessary. Regarding the 120 ohm resistors, yes they consume power but ONLY while a device is transmitting, I would still use them to reduce the effects of noise. Bill Keep safe. Live long and prosper. |
||||
PeterB Guru Joined: 05/02/2015 Location: AustraliaPosts: 639 |
G'Day Bill If you looked closely at telegraph poles you would see 4 insulators every so often instead of the normal 2. This provided a way of crossing the wires over or twisting them and it was done to prevent cross talk to adjacent pairs. On the subject of RS232. It should be banned! Peter p.s. The 4 insulators also provided a better place to build a nest I think. P Edited 2020-03-22 16:19 by PeterB |
||||
Tinine Guru Joined: 30/03/2016 Location: United KingdomPosts: 1646 |
As mentioned in another thread, my latest machine controller is based on the E100 which is the master of the RS-485/422 (full duplex) motion control network. This is quite the noisy environment due to the high power servo drives driving brushless motors @20KHz PWM. The network is constantly running @230K BAUD, every message has a checksum. Never had a glitch and I'm only using unshielded phone cable. This machine has never required the "Microsoft Reset" but the megabucks Trumpf laser, sitting next to my machine, constantly freezes, requiring a power-cycle. The laser machine has an industrial Windows PC (Siemens) on the front-end, talking to a Sinumerik CNC controller... MM/E100 ROCKS! |
||||
lizby Guru Joined: 17/05/2016 Location: United StatesPosts: 3009 |
I've successfully run 3.3V TTL RS232 over 50 feet of Cat6 on several occasions. I'll admit that this has not been extensively tested, or tested at all in complicated environments. PicoMite, Armmite F4, SensorKits, MMBasic Hardware, Games, etc. on fruitoftheshed |
||||
PeterB Guru Joined: 05/02/2015 Location: AustraliaPosts: 639 |
I still think RS232 should be banned Peter |
||||
Paul_L Guru Joined: 03/03/2016 Location: United StatesPosts: 769 |
RS232 works find over short distances in interference free areas, like where Lizby is in his barn. If you have interference or noise or are running more than 10 meters you should use a balanced method to screen out common mode junk pickup. Shielding only helps with capacitive coupled noise, it doesn't attenuate inductive coupled noise much at all. Paul in NY. |
||||
Page 1 of 2 |
Print this page |