CAN bit timing, Seg2 =1Tq

classic Classic list List threaded Threaded
9 messages Options
Reply | Threaded
Open this post in threaded view
|

CAN bit timing, Seg2 =1Tq

Amitai Broyde
Hi,
I am using 1Mbps CAN network based on Cores running at 8MHz. I would like to extend the bus length by setting sampling point 125ns before end of bit time. Current bit sampling time is 250ns before end of bit time. 
Will it work?

Regards,
ab
Reply | Threaded
Open this post in threaded view
|

RE: CAN bit timing, Seg2 =1Tq

John Dammeyer

That puts the sample point at the 87.5% point and is really on the edge.  You haven’t described how long a bus you want to create or what type of cable you want to use.  All that will affect how long a bus you can make at that bit rate.

John Dammeyer

 

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Amitai Broyde
Sent: July-15-16 12:30 AM
To: [hidden email]
Subject: [CANLIST] CAN bit timing, Seg2 =1Tq

 

Hi,

I am using 1Mbps CAN network based on Cores running at 8MHz. I would like to extend the bus length by setting sampling point 125ns before end of bit time. Current bit sampling time is 250ns before end of bit time. 

Will it work?

 

Regards,

ab

Reply | Threaded
Open this post in threaded view
|

Re: CAN bit timing, Seg2 =1Tq

Amitai Broyde
Hi,
The cable is a controlled 120 OHM cable (5nS/m) by Axon Cables . Don't recall the exact p/n. I would like to extend the bus to about 30m.
The dominant to recessive loop worst case delay is 400nS, there are 16 nodes nodes. All nodes clocks' use XOs - several ppm accuracy.
What the risk of edgy sampling? it is still 125nS in.

Regards,
ab 

‫בתאריך יום ו׳, 15 ביולי 2016 ב-10:52 מאת ‪John Dammeyer‬‏ <‪[hidden email]‬‏>:‬

That puts the sample point at the 87.5% point and is really on the edge.  You haven’t described how long a bus you want to create or what type of cable you want to use.  All that will affect how long a bus you can make at that bit rate.

John Dammeyer

 

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Amitai Broyde
Sent: July-15-16 12:30 AM
To: [hidden email]
Subject: [CANLIST] CAN bit timing, Seg2 =1Tq

 

Hi,

I am using 1Mbps CAN network based on Cores running at 8MHz. I would like to extend the bus length by setting sampling point 125ns before end of bit time. Current bit sampling time is 250ns before end of bit time. 

Will it work?

 

Regards,

ab

Reply | Threaded
Open this post in threaded view
|

RE: CAN bit timing, Seg2 =1Tq

John Dammeyer
It's been many years since I've worked on the edge like that.  I found that with 35m Beldon Thick DeviceNet and 1MBps that I had to use at least 16 Tq per bit to sample at the 87.5% point.  

As I recall you need to take into account that a node at the far end may not detect the edge of the start bit until the end of its 125n/S clock window where the transmitter sends the dominant on the edge of its 8Mhz clock.  So you have a potential skew of almost 125nS.  That can put the point where the node tests as far out as 124nS which is too late.  

Some CAN devices have other kinds of Start Bit Edge Detection logic that my synchronize differently but if you think of a simple state machine that every 125nS looks at the signal, it may not detect the edge until 124nS after it was sent.  So you have almost a 1 bit skew on the signal not including the propagation time.

Having said that you can try it.  We found for the 35m segments we had to back down our network to 500kbps.  Although it required reprogramming and reinstalling 50 nodes, it was easy to do because the TKE Bridge we used could have each 50 node segment running at separate bit rates.  We fed it with 500kbps, it distributed the messages out the other three ports at 500kbps or 1Mbps so we never overran the bridge.

John Dammeyer


From: [hidden email] [mailto:[hidden email]] On Behalf Of Amitai Broyde
Sent: July-15-16 3:45 AM
To: [hidden email]; [hidden email]
Subject: Re: [CANLIST] CAN bit timing, Seg2 =1Tq

Hi,
The cable is a controlled 120 OHM cable (5nS/m) by Axon Cables . Don't recall the exact p/n. I would like to extend the bus to about 30m.
The dominant to recessive loop worst case delay is 400nS, there are 16 nodes nodes. All nodes clocks' use XOs - several ppm accuracy.
What the risk of edgy sampling? it is still 125nS in.

Regards,
ab


--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/
Report any problems to <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

Re: CAN bit timing, Seg2 =1Tq

Amitai Broyde
Thanks for the detailed explanations. 
I am not sure but my understanding is that you are concerned mostly from the signal distortion due to the length of the line, distortion that may cause more delay in the sampling time.
Does 20 m line would be acceptable for the above mentioned system?
I found that 100 OHM cable save about 20% in delay time. The Dominant level of CAN drivers driving  50 OHM load instead of 60 OHM almost the same. What is your opinion of using 100 OHM line? 

Regards,
ab

‫בתאריך יום ו׳, 15 ביולי 2016 ב-19:07 מאת ‪John Dammeyer‬‏ <‪[hidden email]‬‏>:‬
It's been many years since I've worked on the edge like that.  I found that with 35m Beldon Thick DeviceNet and 1MBps that I had to use at least 16 Tq per bit to sample at the 87.5% point.

As I recall you need to take into account that a node at the far end may not detect the edge of the start bit until the end of its 125n/S clock window where the transmitter sends the dominant on the edge of its 8Mhz clock.  So you have a potential skew of almost 125nS.  That can put the point where the node tests as far out as 124nS which is too late.

Some CAN devices have other kinds of Start Bit Edge Detection logic that my synchronize differently but if you think of a simple state machine that every 125nS looks at the signal, it may not detect the edge until 124nS after it was sent.  So you have almost a 1 bit skew on the signal not including the propagation time.

Having said that you can try it.  We found for the 35m segments we had to back down our network to 500kbps.  Although it required reprogramming and reinstalling 50 nodes, it was easy to do because the TKE Bridge we used could have each 50 node segment running at separate bit rates.  We fed it with 500kbps, it distributed the messages out the other three ports at 500kbps or 1Mbps so we never overran the bridge.

John Dammeyer


From: [hidden email] [mailto:[hidden email]] On Behalf Of Amitai Broyde
Sent: July-15-16 3:45 AM
To: [hidden email]; [hidden email]
Subject: Re: [CANLIST] CAN bit timing, Seg2 =1Tq

Hi,
The cable is a controlled 120 OHM cable (5nS/m) by Axon Cables . Don't recall the exact p/n. I would like to extend the bus to about 30m.
The dominant to recessive loop worst case delay is 400nS, there are 16 nodes nodes. All nodes clocks' use XOs - several ppm accuracy.
What the risk of edgy sampling? it is still 125nS in.

Regards,
ab


--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/
Report any problems to <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

RE: CAN bit timing, Seg2 =1Tq

Bram Kerkhof
In reply to this post by Amitai Broyde

The first thing I usually do in these kinds of situation is rethink whether I am:

A)      Using the right technology

B)      Using the technology right

 

The big advantages of CAN for me are that it is cheap, simple and reliable as long as you keep away from the bleeding edge (it’s called ‘bleeding’ for a reason ;-).

When pushing the limits like this, you’ll end up with a system that’s either cheap and simple or reliable, which is probably not what you want.

 

Keep in mind that impedance is everything at this point, and that you don’t only have to consider the backbone cable, but also all the connections in between. What works for 35m between two nodes, may not work over the same distance with multiple nodes on the bus.

 

Do you really need 1 mbits? Chances are you can simply drop down to 500kbits without too much work.

If throughput is an issue, consider splitting the network with a gateway. If timing determinability is an issue, consider a time-sensitive protocol implementation (like TTCAN) and a lower bandwidth.

 

Or use something else entirely.

 

Cheers,

Bram

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Amitai Broyde
Sent: vrijdag 15 juli 2016 9:30
To: [hidden email]
Subject: [CANLIST] CAN bit timing, Seg2 =1Tq

 

Hi,

I am using 1Mbps CAN network based on Cores running at 8MHz. I would like to extend the bus length by setting sampling point 125ns before end of bit time. Current bit sampling time is 250ns before end of bit time. 

Will it work?

 

Regards,

ab

Reply | Threaded
Open this post in threaded view
|

Re: CAN bit timing, Seg2 =1Tq

Amitai Broyde

Got it,
Thanks,

ab

‫בתאריך שבת, 16 ביולי 2016 ב-15:05 מאת ‪Bram Kerkhof‬‏ <‪[hidden email]‬‏>:‬

The first thing I usually do in these kinds of situation is rethink whether I am:

A)      Using the right technology

B)      Using the technology right

 

The big advantages of CAN for me are that it is cheap, simple and reliable as long as you keep away from the bleeding edge (it’s called ‘bleeding’ for a reason ;-).

When pushing the limits like this, you’ll end up with a system that’s either cheap and simple or reliable, which is probably not what you want.

 

Keep in mind that impedance is everything at this point, and that you don’t only have to consider the backbone cable, but also all the connections in between. What works for 35m between two nodes, may not work over the same distance with multiple nodes on the bus.

 

Do you really need 1 mbits? Chances are you can simply drop down to 500kbits without too much work.

If throughput is an issue, consider splitting the network with a gateway. If timing determinability is an issue, consider a time-sensitive protocol implementation (like TTCAN) and a lower bandwidth.

 

Or use something else entirely.

 

Cheers,

Bram

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Amitai Broyde
Sent: vrijdag 15 juli 2016 9:30
To: [hidden email]
Subject: [CANLIST] CAN bit timing, Seg2 =1Tq

 

Hi,

I am using 1Mbps CAN network based on Cores running at 8MHz. I would like to extend the bus length by setting sampling point 125ns before end of bit time. Current bit sampling time is 250ns before end of bit time. 

Will it work?

 

Regards,

ab

Reply | Threaded
Open this post in threaded view
|

Re: CAN bit timing, Seg2 =1Tq

Bertil Bäck-2
CANopen also specifies 800kbps as speed. If you don't want to drop down to 500kbps.

Depending on what you do you can also run other speeds.

Br,
Bertil


From: "Amitai Broyde" <[hidden email]>
To: [hidden email]
Sent: Saturday, July 16, 2016 5:01:57 PM
Subject: Re: [CANLIST] CAN bit timing, Seg2 =1Tq


Got it,
Thanks,

ab

‫בתאריך שבת, 16 ביולי 2016 ב-15:05 מאת ‪Bram Kerkhof‬‏ <‪[hidden email]‬‏>:‬

The first thing I usually do in these kinds of situation is rethink whether I am:

A)      Using the right technology

B)      Using the technology right

 

The big advantages of CAN for me are that it is cheap, simple and reliable as long as you keep away from the bleeding edge (it’s called ‘bleeding’ for a reason ;-).

When pushing the limits like this, you’ll end up with a system that’s either cheap and simple or reliable, which is probably not what you want.

 

Keep in mind that impedance is everything at this point, and that you don’t only have to consider the backbone cable, but also all the connections in between. What works for 35m between two nodes, may not work over the same distance with multiple nodes on the bus.

 

Do you really need 1 mbits? Chances are you can simply drop down to 500kbits without too much work.

If throughput is an issue, consider splitting the network with a gateway. If timing determinability is an issue, consider a time-sensitive protocol implementation (like TTCAN) and a lower bandwidth.

 

Or use something else entirely.

 

Cheers,

Bram

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Amitai Broyde
Sent: vrijdag 15 juli 2016 9:30
To: [hidden email]
Subject: [CANLIST] CAN bit timing, Seg2 =1Tq

 

Hi,

I am using 1Mbps CAN network based on Cores running at 8MHz. I would like to extend the bus length by setting sampling point 125ns before end of bit time. Current bit sampling time is 250ns before end of bit time. 

Will it work?

 

Regards,

ab


Reply | Threaded
Open this post in threaded view
|

Re: CAN bit timing, Seg2 =1Tq

heikki saha
Hello,

125ns from the end of the bit means 87.5%, which is far end of the standard SP location in 1Mbps CANopen. So, there should not exist any problems with that. But, please note that the earliest SP location is still 75%, meaning that there may exist other devices using such! So, you can only have 4TQ (=500ns) for signal propagation, from which transceivers take approximately half. There shall always exist at least 1TQ phase segment 1 enabling non-zero SJW. That means that after subtracting transceiver delays, you'll anyway have 250ns left for back-to-back propagation in the network cabling, resulting 25m maximum achievable network length (25m*2*5ns/m=250ns).

800kbps may be selected, but there is a practical drawback -- there are too many devices not supporting it. So, it is recommended to search 2-3 2nd source devices as a background. If such do not exist, one option may be the use of switch, which receives and re-transmits the frames and thus doubles the maximum achievable segment length.

Best regards,

-H

Heikki Saha, CTO
  M.Sc. Automation
  Dr.Tech. Electronics
TK Engineering Oy
Mail address:  P.O. box 810, FIN-65101 VAASA
Visit address: Yrittäjänkatu 17, FIN-65380 VAASA
+358 (0)50 588 6894
email: [hidden email]
skype: heikki.saha_tke
http://www.tke.fi/
http://www.canopen.fi/

----- Original Message -----
From: "Bertil Bäck" <[hidden email]>
To: [hidden email]
Sent: Thursday, September 1, 2016 3:06:12 PM
Subject: Re: [CANLIST] CAN bit timing, Seg2 =1Tq


CANopen also specifies 800kbps as speed. If you don't want to drop down to 500kbps.

Depending on what you do you can also run other speeds.

Br,
Bertil

----- Original Message -----

From: "Amitai Broyde" <[hidden email]>
To: [hidden email]
Sent: Saturday, July 16, 2016 5:01:57 PM
Subject: Re: [CANLIST] CAN bit timing, Seg2 =1Tq




Got it,

Thanks,



ab



‫בתאריך שבת, 16 ביולי 2016 ב-15:05 מאת ‪Bram Kerkhof‬‏ <‪ [hidden email] ‬‏>:‬






The first thing I usually do in these kinds of situation is rethink whether I am:

A) Using the right technology

B) Using the technology right



The big advantages of CAN for me are that it is cheap, simple and reliable as long as you keep away from the bleeding edge (it’s called ‘bleeding’ for a reason ;-).

When pushing the limits like this, you’ll end up with a system that’s either cheap and simple or reliable, which is probably not what you want.



Keep in mind that impedance is everything at this point, and that you don’t only have to consider the backbone cable, but also all the connections in between. What works for 35m between two nodes, may not work over the same distance with multiple nodes on the bus.



Do you really need 1 mbits? Chances are you can simply drop down to 500kbits without too much work.

If throughput is an issue, consider splitting the network with a gateway. If timing determinability is an issue, consider a time-sensitive protocol implementation (like TTCAN) and a lower bandwidth.



Or use something else entirely.



Cheers,

Bram



From: [hidden email] [mailto: [hidden email] ] On Behalf Of Amitai Broyde
Sent: vrijdag 15 juli 2016 9:30
To: [hidden email]
Subject: [CANLIST] CAN bit timing, Seg2 =1Tq







Hi,


I am using 1Mbps CAN network based on Cores running at 8MHz. I would like to extend the bus length by setting sampling point 125ns before end of bit time. Current bit sampling time is 250ns before end of bit time.


Will it work?





Regards,


ab
--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/
Report any problems to <[hidden email]>