• AVR Freaks

CAN BUS physical layer problems

Page: 12 > Showing page 1 of 2
Author
llima31
New Member
  • Total Posts : 23
  • Reward points : 0
  • Joined: 2010/08/24 09:46:48
  • Location: 0
  • Status: offline
2012/01/03 03:42:06 (permalink)
0

CAN BUS physical layer problems

Hello,
 
I have a project were several boards connect to each other using a CAN bus, however, I am having problems with the can physical layer.
If I add more than three nodes, they stop communicating.
 
I discovered that if I add terminator resistors in each node, lowering this way bus impedance, they resume communication but this should not be necessary. Also, this way the signal gets attenuated along the bus.
I also discovered that if I lower the bus speed from the original 250kbps to 125kbps, I can add two more nodes without having problems.
 
The hardware was not designed by me, but now I have to solve this problem, so I ask for your help
 
Below is an image of the circuit between the can transceiver and the can bus.
 
- Each node has two rj45 connections, to allow for the adding of more nodes
- We use UTP cat 5 cable for the connections, and we use one of the pairs for the CANL and CANH lines
- The power supply for each node is provided by the bus it self, using the other wires of the RJ45 connector
- Each two nodes are connected with a UTP cable from the "out" rj45 connector you can see in the diagram to the "in" rj45 of the next node
 
I am suspicious of the filtering and protection circuit that you see in the image, because we have an older version of these boards were the transceiver was connected directly to the bus(rj45 connectors) and we did not have this problem. I have been reading about CAN protection circuits and what I saw was similar to this one, but I also read somewhere that if these circuits are not very well dimensioned, they can cause problems on the bus, maybe that is the problem.

 
The tipical environment were we will use the system is in data centers, however it can be applied to factories or other harsh environments, and it was because of that that the designer implemented the protection circuit.

thanks in advance for your help

Attached Image(s)

#1

30 Replies Related Threads

    Artic
    Super Member
    • Total Posts : 2288
    • Reward points : 0
    • Joined: 2004/06/17 06:21:59
    • Location: Wales now in Sofia,BG
    • Status: offline
    Re:CAN BUS physical layer problems 2012/01/03 07:15:47 (permalink)
    0
    Hi,
    Looks like who ever added the CAN physical layer components did not have a clue - sorry (but you prob think this yourself). What I would do is to start fresh and remove all the components, etc. that have been added to the CAN H and CAN L lines.

    If it were me, I would do the following;

    1. Remove all the components attached to the CAN H and CAN L lines.

    2. Remove the Rs resistor value on the MCP2551 and pull the Rs to GND

    3. Identify the two end nodes on the CAN bus and fit an external termination resistor (120R) across CAN H and CAN L at each end of the CAN bus. Keep this termination external as it keeps things simple when you want to expand the CAN network, etc

    4. Check the PSU voltage signals for noise on the rails, etc

    5. Check the GND resistance path is nice and low - a high impedance GND will cause you alot of problems.

    6. Check the CAN Baud rate settings. Use a scope to double check the CAN baud rate and signal quality.

    Doing something like the above should give you a working CAN bus. If problems still exist then I would check the rest of the physical layer hardware and only then look into a software issue.

    Hope some of the above helps
    Artic
    #2
    Antipodean
    Super Member
    • Total Posts : 1746
    • Reward points : 0
    • Joined: 2008/12/09 10:19:08
    • Location: Didcot, United Kingdom
    • Status: offline
    Re:CAN BUS physical layer problems 2012/01/03 07:33:10 (permalink)
    0
    Artic

    Hi,
    Looks like who ever added the CAN physical layer components did not have a clue - sorry (but you prob think this yourself). What I would do is to start fresh and remove all the components, etc. that have been added to the CAN H and CAN L lines.

    If it were me, I would do the following;

    1. Remove all the components attached to the CAN H and CAN L lines.

    2. Remove the Rs resistor value on the MCP2551 and pull the Rs to GND

    3. Identify the two end nodes on the CAN bus and fit an external termination resistor (120R) across CAN H and CAN L at each end of the CAN bus. Keep this termination external as it keeps things simple when you want to expand the CAN network, etc

    He does have a switchable termination, and in some ways I would prefer this, as having a plug in terminator causes other problems when non-technical people start doing things (terminator, whats that, isn't he in a film?). The termination resistor could probably have its value changed, 82 ohms does look too low for CAT5 cable.
     
    Artic

    4. Check the PSU voltage signals for noise on the rails, etc

    5. Check the GND resistance path is nice and low - a high impedance GND will cause you alot of problems.

    I would add ground connections to pins 3 and 4 myself, to help keep the ground resistance low.
     
    Artic
    6. Check the CAN Baud rate settings. Use a scope to double check the CAN baud rate and signal quality.

    Doing something like the above should give you a working CAN bus. If problems still exist then I would check the rest of the physical layer hardware and only then look into a software issue.

    Hope some of the above helps
    Artic


    I would look at the manufacturers data sheets for the protection components, and see if they are intended for high speed bus protection - they well have two much capacitance etc for a high speed bus. Check out the components specified for USB protection, and another place to look is the book by the Würth book "Trilogy of Inductors" where the most recent editions have sections at the back on doing this sort of protection. http://www.wuerth-elektronik.de/web/en/wuerth_elektronik/start.php
     
    Personally I would be loathe to just remove all the protection, as having worked in industrial environments and seen the way noise can crash electronics I know that something is needed here. But you do have to choose your components carefully for the bus speed.

    Do not use my alias in your message body when replying, your message will disappear ...

    Alan
    #3
    llima31
    New Member
    • Total Posts : 23
    • Reward points : 0
    • Joined: 2010/08/24 09:46:48
    • Location: 0
    • Status: offline
    Re:CAN BUS physical layer problems 2012/01/03 08:49:38 (permalink)
    0
    Hello,

    Thank you both for your sugestions, I was already thinking on using Artic's approach, and that's what I will do just to make sure the problem is from the protection circuit.

    Regarding Artic's steps 4 and 5, these are the results:

    4. The power supply that travels on the UTP cable(9v) has a ripple of 6mv peak to peak when it reaches the nodes, after the rectification each node does to 5v, the ripple is 2mv peak to peak.

    5. I am not very sure on how to measure this, I have put an multimeter measuring resistance from the MCP2551 GND pin to the power supply GND pin with the system off and I got 0ohms

    I will now start removing the components.

    Regarding Alan's sugestion to check the components data-sheets, both ESD24VS2U and ZJYS81R5-2P25T mention that they are suitable for CAN bus, but they do not mention Hi-speed CAN bus.

    best regards
    #4
    MKC
    Super Member
    • Total Posts : 399
    • Reward points : 0
    • Joined: 2009/12/04 13:24:27
    • Location: Belgium
    • Status: offline
    Re:CAN BUS physical layer problems 2012/01/03 11:07:17 (permalink)
    0
    Hi,
     
    Just a few ideas,
    1/ use 120R at both end not 82R
    2/ RJ45 cable is ok for short distance for longer distance check the minimum wire section at the internet
    3/ remove the 22R resistors
    4/ The choke you are using is well suitable for CAN, but don't swap the direction between the two coils (input is one site output is the other side)
    5/ The protection diode used is specially designed for CAN so this seems to be correct
    6/ check the maximum speed for the cable length
     
    Marc
    #5
    llima31
    New Member
    • Total Posts : 23
    • Reward points : 0
    • Joined: 2010/08/24 09:46:48
    • Location: 0
    • Status: offline
    Re:CAN BUS physical layer problems 2012/01/04 03:15:19 (permalink)
    0
    Hello Marc, thank you for your sugestions, I have two questions though, and some comments
     
    1/ Done already, did not solve the propblem
    2/ Regading RJ45, what do you mean by short distances? We are plannig to have  maximum bus length of 250m
    3/ Done already, did not solve the propblem
    4/ I have checked the choke datasheet but there is no reference of input and output side! Then I doublechecked the choke reference and the actual part in the pcb says "ZJY2501", is this the same choke? All searches in the internet for this reference turn out empty!
    5/ Ok
    6/ Currently we are only using a short distance because we are still in our test bed, we are using 1m RJ45 cables to connect each node at a speed of 250kbps, tested also with 125kbps with no luck
    #6
    Antipodean
    Super Member
    • Total Posts : 1746
    • Reward points : 0
    • Joined: 2008/12/09 10:19:08
    • Location: Didcot, United Kingdom
    • Status: offline
    Re:CAN BUS physical layer problems 2012/01/04 04:15:23 (permalink)
    0
    llima31

    2/ Regading RJ45, what do you mean by short distances? We are plannig to have  maximum bus length of 250m
    6/ Currently we are only using a short distance because we are still in our test bed, we are using 1m RJ45 cables to connect each node at a speed of 250kbps, tested also with 125kbps with no luck


    I seriously doubt that you will get anything like 250kbps, or even 125 kbps with 250m of cable, especially in a noisey factory type environment.
      
     

    Do not use my alias in your message body when replying, your message will disappear ...

    Alan
    #7
    Artic
    Super Member
    • Total Posts : 2288
    • Reward points : 0
    • Joined: 2004/06/17 06:21:59
    • Location: Wales now in Sofia,BG
    • Status: offline
    Re:CAN BUS physical layer problems 2012/01/04 04:27:38 (permalink)
    0
    I agree. It was a shame the OP did not provide this information previously.
    Cheers
    Artic
    #8
    MKC
    Super Member
    • Total Posts : 399
    • Reward points : 0
    • Joined: 2009/12/04 13:24:27
    • Location: Belgium
    • Status: offline
    Re:CAN BUS physical layer problems 2012/01/04 12:56:06 (permalink)
    0
    Hi,
     
    CAN bus has its specification on cable (wire section and length), check the information on the internet.
    RJ45 cable section is not within the CAN specification. You need to use specified CAN cable to get the maximum distance and speed.
     
    For the choke:
    1) input side and output side can be swapped, buth BOTH lines (low and high) must be on the same side, this a common mode choke.
    2) choke type : you need to check the spec. For test you can remove the choke and bypass it.
     
    Marc
     
     
    #9
    llima31
    New Member
    • Total Posts : 23
    • Reward points : 0
    • Joined: 2010/08/24 09:46:48
    • Location: 0
    • Status: offline
    Re:CAN BUS physical layer problems 2012/01/05 10:21:38 (permalink)
    0
    Hello,

    I managed to contact the previous electronic designer, he told me that he uses this design with sucess in a network of 25 ndes, but at a speed of 25Kbps.
    Probably it can all be solved with a component values change.
    Anyway I am on the process of removing all the protection circuit to test the bus without it
    #10
    llima31
    New Member
    • Total Posts : 23
    • Reward points : 0
    • Joined: 2010/08/24 09:46:48
    • Location: 0
    • Status: offline
    Re:CAN BUS physical layer problems 2012/01/05 10:44:17 (permalink)
    0
    Hello,

    I managed to contact the previous electronic designer, he told me that he uses this design with sucess in a network of 25 ndes, but at a speed of 25Kbps.
    Probably this problem can be solved with a component values change, but I am already considering lowering the bus speed as you also sugested.
    Anyway I am now on the process of removing all the protection circuit to test the bus without it
     
    Thanks for your inputs
    #11
    bugbear
    New Member
    • Total Posts : 6
    • Reward points : 0
    • Joined: 2012/04/10 20:36:36
    • Location: 0
    • Status: offline
    Re:CAN BUS physical layer problems 2012/04/10 21:03:40 (permalink)
    0
    i am using mcp2515 and mcp2551 along with atmega128...i am usin a 3.578 mhz crystal and have set the speed at 10kbps by setting the baud rate prescaler at 01010.i can see the waveform on txd pin of 2551 on dso but nothin is seen oc canh and canl lines..my transmitter side is workin properly,it gives an error in sending msg.how do i receive properly..i ve no idea wats wrong with the 2551..i ve my ckt on general purpose board...if ne could help me...
    #12
    john.dammeyer
    Super Member
    • Total Posts : 416
    • Reward points : 0
    • Joined: 2004/01/02 20:07:06
    • Status: offline
    Re:CAN BUS physical layer problems 2012/04/10 21:50:20 (permalink)
    0
    There are a number of basic reasons people can't get CAN to work.

    1. The bus must be terminated with at least one 120 resistor and for best operation one at each end.  (120R between CANH and CANL)

    2. CAN requires a minimum of two 'functional' nodes to work.

    3. When developing a CAN module for the first time (or anytime for that matter) it helps to have a working CAN bus monitor or dongle that can display messages.   It's so easy to get confused with the filtering and masking bits, bit rates etc. 

    So first add the 120 ohm resistor.  Then, with no messages transmitting you can see that the voltage is 2.5V on both CANH and CANL.

    Try and send a message and you'll see CANH go to about 4V and CANL go to about 1V.  If not, check your wiring.  Remove the Tx line from the processor and force it high and low and look at the bus.  If the driver doesn't work nothing will.

    Now with only one node on the bus a message transmtted will continue forever.  At 10Kbps, you would need to have a 357.8 divisor to create a 100uS bit time.  RS232 is more forgiving than CAN for bit rate variations so change to a 4MHz crystal so you can evenly divide the clock into bit times.

    #13
    llima31
    New Member
    • Total Posts : 23
    • Reward points : 0
    • Joined: 2010/08/24 09:46:48
    • Location: 0
    • Status: offline
    Re:CAN BUS physical layer problems 2012/06/26 08:32:55 (permalink)
    0
    Hello,
    I would like to appologize to everyone that helped with this issue, because I was moved urgently to another project and never replied again to this thread.
    Now Hi have some time for this project again, and I am back trying to solve this issue.
     
    Currently this is the status of the assembly:
    - I have 7 nodes on the network
    - I have 1.5m of UTP cat5 cable connecting each node
    - I have one 120ohm resistor at each end of the bus
    - As suggested above, I removed Rs resistor on the MCP2551 and pull the Rs to GND
    - I have not removed the D1, L1 and C4 yet
    - I changed the CAN bit timing configuration, increasing the PropSeg and the SJW, see atached image.
     
    - The power supply that travels on the UTP cable(9v) has a ripple of 6mv peak to peak when it reaches the nodes, after the rectification each node does to 5v, the ripple is 2mv peak to peak
     
    The conclusions so far:
    - If I reduce the CAN speed to 50kbps the entire network works fine
    - If I add 120ohm terminator resistors to the 4 last nodes, the entire network works fine, at 250kbps, but has a bus impedance of 25ohm
    - I also followed the http://www.esd-electronics-usa.com/CAN-Bus-Troubleshooting-Guide.html, everything seems fine without transmitting anything.
    - With only the normal two terminators CANH and CANL voltages are at 2.52v and 2.63v, but when I try to transmit, they change to a state where the bus is continually changing from recessive to dominant and never leaves this state, see attached picture IMG_0691.JPG. If I add the other terminators I mentioned above, the voltages are the same, 2.43v. 

    - Attached is a picture IMG_0690.JPG of the wave forms while communicating at 250kbps, they seem to be very much distorted.

     
    Although I made a few progress since the las time, I am still far away from the 64 nodes goal, even If I reduce the speed to 50Kbps I am almost sure the network will not work with 64 nodes untill I find the problem source.
     
    Once again, any help is much appreciated
    post edited by llima31 - 2012/06/26 08:41:31

    Attached Image(s)

    #14
    llima31
    New Member
    • Total Posts : 23
    • Reward points : 0
    • Joined: 2010/08/24 09:46:48
    • Location: 0
    • Status: offline
    Re:CAN BUS physical layer problems 2012/06/27 09:26:19 (permalink)
    0
    Hello again,
     
    Today I started removing the remaining components attached to the CAN BUS, D1, L1 and C4.
    Removing D1 and L1 made no difference, but after removing C4, I managed to put the 7 nodes comunicating at 250kbps only with the normal two terminator resistors of 120ohm.
     
    So the capacitor was to blaim, but now, shoud I add the other components again? And will they work fine without the capacitor C4? These questions may seem dum for some of you, but please keep in mind that I did not design this circuit, nor am I an experienced in electronic circuit design :s.
     
    The waveform on the CAN bus is as follows, is the distortion that we can see normal?
     
     
    Thanks in advance
     
    Luis Lima

    Attached Image(s)

    #15
    tom_usenet
    Super Member
    • Total Posts : 502
    • Reward points : 0
    • Joined: 2011/12/07 21:32:51
    • Location: 0
    • Status: offline
    Re:CAN BUS physical layer problems 2012/06/27 17:05:26 (permalink)
    0
    You need an Electrical Engineer to look at this. Or anyone who has done at least one year of EE. Or a hobbyist. Basically anyone who knows Ohm's Law plus a few other basic bits.
     
    "picture IMG_0690.JPG" shows two different distortions.
     
    That big ripple at the start of the first bit looks like reflections from the transmission line. That's a symptom of a "long cable that isn't terminated properly" or that has branches or lumped loads in the middle.
     
    The more serious problem is the time it takes for the waveform to get back to zero. That is delaying when the receiver sees the transition. If you look at the waveforms between the controller and the transceiver you'll see the "one bits" and the "zero bits" would be a VERY different length. That distortion is why your comms doesn't work. Also why changing the sample point makes it slightly better.
     
    From your oscilloscope trace it looks like the RC time constant is about 5us. So "RC = 5us". That was with 7 nodes on the network and 4 terminators, together with 44 ohms in series with each "C4" capacitor (ignore these for now). So:
     
    "7 * 22 e-12 * 25 = .000385us. That's a LONG way from 5us (about 1000)! Solving for the capacitance:
     
    C = 1/7 5u/25 = 1/35u = 28nF

    So I'd guess your "C4"s are actually 22nF and not 22pF. Someone mis-ordered or mis-read the markings. Use the right capacitors and you should be able to go 1000 times faster or use 1000 times as many nodes :-).
     
    Tom
     
    #16
    llima31
    New Member
    • Total Posts : 23
    • Reward points : 0
    • Joined: 2010/08/24 09:46:48
    • Location: 0
    • Status: offline
    Re:CAN BUS physical layer problems 2012/06/28 03:04:52 (permalink)
    0
    Hello Tom,
     
    Thanks a lot for the detailed explanation, your guess is completelly correct, I measured the capacitors and indead they are 22nf instead of the expected 22pf. I will order the correct ones and solder back the rest of the components to check how it works.
     
    Regarding the reflections, in the last trace I posted, which has no electronic components attached to the CANL and CANH lines,  they still seem to appear, although smaller. I con think of some candidates that could becausing this, which do you think is the most likelly?:
    - I am using CAT5 cable, which normally has 100ohm impedance, with 120ohm terminators
    - I am using a daisy chain conection method with two RJ45 connectors instead of a bus with T connectors
    - In the pcb of each node there are two small 1cm connections comming from the CANL and CANH lines to two header pins. These were meant for possible expantion in the future
     
    Thanks again for your help, it was priceless
     
    Best regards
    #17
    tom_usenet
    Super Member
    • Total Posts : 502
    • Reward points : 0
    • Joined: 2011/12/07 21:32:51
    • Location: 0
    • Status: offline
    Re:CAN BUS physical layer problems 2012/06/29 08:33:52 (permalink)
    0
    > Thanks again for your help, it was priceless
     
    It's fun spotting that from another continent.
     
    > Regarding the reflections,
     
    They only matter if they corrupt the data. The Receivers are working from the difference between the high and low signals, and (from the data sheets) switch at about 800mV and 650mV. So as long as the reflections aren't severe enough to get near that level there aren't any problems. The Coply App Note (mentioned below) gives a good diagram of what to measure to check the signals are OK.
     
    The CAT5 cable is well enough terminated with 100 ohms. A daisy chain is a lot better than having stubs. A 1cm stub only matters at frequencies that approach 1cm, which at the speed of light is about 3GHz.
     
    Your next problem will be the power distribution. How much current does each unit take and what's the resistance of 250m of cable? At 0.2 ohm/meter for the loop that's 0.2*250/2 as you're using two pairs. That's 25 ohms. You'll lose 1V for 40mA worst case (power supply at one end, all the loads at the other end). You may want to have multiple power supplies on the cable, but you may need to take special care if you want to parallel power supplies, depending on their design.
     
    Watch for the plug/connectors. The daisy-chains mean you have 256 plug-socket connections IN SERIES for 64 nodes. One bad connector or plug crimp will take out your whole power bus as well as your data bus. So you'd be WAY more reliable if all units connect to a common bus using a short fly lead. I'd go so far as calling that design unreliable and undiagnosable (try finding which of the over 200 plugs is bad).
     
    Next problem. You may have the comms fail when the unit at one end of the bus transmits and the one at the other end causes an error when it sends its ACK pulse back. 250m of cable is 500m round trip, and with cable running at 0.7C that will take 500/(0.7 * 3.0E8) = 2.38us.
     
    That's 60% of your bit-time at 250kHz! You'll need optimum selection of the PROP segment and the other timing parameters for this. Also remember that with the CAN bus you can trade off clock accuracy and bus length. You can have a long bus with accurate clocks OR sloppy clocks on a short bus. Are all your CAN clocks crystal-perfect and exactly the same frequency? If not you have to derate the length. The following says 250M is possible, but I think that's pushing it:
     
    http://en.wikipedia.org/wiki/CAN_bus
     
    "Bit rates up to 1 Mbit/s are possible at network lengths below 40 m. Decreasing the bit rate allows longer network distances (e.g., 500 m at 125 kbit/s)."
     
    http://www.copleycontrols../motion/pdf/CAN-Bus.pdf
     
    "The CANopen DS 102 V2.0 Physical Layer for Industrial Applications, Appendix" says 250kbit/sec over 250m is ok. I wouldn't push it that hard.
     
    Tom
     
    #18
    llima31
    New Member
    • Total Posts : 23
    • Reward points : 0
    • Joined: 2010/08/24 09:46:48
    • Location: 0
    • Status: offline
    Re:CAN BUS physical layer problems 2012/06/29 08:56:13 (permalink)
    0
    These are very useful considerations, some of them are covered, others not...
     
    > Your next problem will be the power distribution...
    Each node will consume aproximatelly 70ma, 9V. Each node has a voltage regulator to convert 9v to the 5v operation voltage. Aditionally each node has a connection for an external power supply. This means that somewere along the bus we connect an external DC adapter, and it will break the power line comming from the bus, and inject the power from this adapter to feed the rest of the bus, untill the next node with an external adapter and so on.
     
    > Next problem. You may have the comms fail when the unit at one end of the bus transmits...
    We have been thinking of this problem too, that is why we are considering lowering the bus speed to 125kbps and changing the bit timming to something like this:

    This will be a better and more reliable solution, we can live with 125kbps anyway.
     
    I will take a look at the The Coply App Note.
     
    Best regards, and thanks
     
    Luis

    Attached Image(s)

    #19
    tom_usenet
    Super Member
    • Total Posts : 502
    • Reward points : 0
    • Joined: 2011/12/07 21:32:51
    • Location: 0
    • Status: offline
    Re:CAN BUS physical layer problems 2012/06/29 19:05:49 (permalink)
    0
    > 64 nodes
     
    If it isn't "top secret", how do you guarantee every node uses different Transmit IDs? Do they all have to be manually assigned a unique "Node Number" on the network or is it generated from a serial number? Or are you polling them so that no two devices ever try to transmit at once?
     
    > Each node will consume aproximatelly 70ma, 9V.
     
    I was about to estimate this when your email came in. I was going to guess 10mA for the MCP2551, between 30-90mA for something like a PIC24, then whatever else the board needs.
     
    If you have normal linear regulators, they need at least 2V across them to work. If the design has a diode to protect against reversed power supply from the bus, you need 7.7V minimum input voltage for 5V out. That allows for 1.3V of drop on the cable, which at 70mA and 0.1 ohm/meter is 185m per node, or 18.5m for 10 modes or 9m for 20 and so on. Adding design margins reduces that. If you've got "low dropout regulators" then you get double that distance.
     
    > Aditionally each node has a connection for an external power supply.
    > This means that somewere along the bus we connect an external
    > DC adapter, and it will break the power line comming from the
    > bus, and inject the power from this adapter to feed the rest of
    > the bus, untill the next node with an external adapter and so on.
     
    Looking at the schematic I see you have "9VIN" and "9VNP", so I see you can have different 9V supplies feeding separate segments. Is the power switching automatic or does someone need to flip a switch on the one with the local power? It would be very neat if it is automatic (and prone to misconfiguration if not).
     
    > considering lowering the bus speed to 125kbps and changing the bit
    > timming to something like this:
     
    You've got SJW=1. Are all of your CAN baud rate clocks matched close enough for that? Worst case difference must be better than about 0.5% (or +- 0.25%). All of your nodes should be OK if all using crystals, but if you have other devices (USB to CAN adapters) then they might have fractionally different clocking speeds to the nodes.
     
    Read the following for a good derivation of the timing parameters required for maximum bus length (min SJW) and maximum clock tolerance (max SJW and smaller distance):
     
    http://ww1.microchip.com/.../en/AppNotes/00754.pdf

    The above also reminded me that the propagation delay in the transceivers and controllers has to be added to the calculations. The Transceiver delays add an equivalent to 20m of cable.

    > Watch for the plug/connectors. The daisy-chains mean you have 256 plug-socket connections IN SERIES
     
    Most of my experience of these plugs and sockets is in helping a friend wire up his house for data. When he crimps plugs on cables or punches down into a socket, about HALF of them are bad. That's just him (and bad tools), but these things aren't 100% reliable.
     
    Tom
     
     
     
    post edited by tom_usenet - 2012/06/30 20:59:47
    #20
    Page: 12 > Showing page 1 of 2
    Jump to:
    © 2019 APG vNext Commercial Version 4.5