Thread View: comp.sys.transputer
2 messages
2 total messages
Started by Lee Campbell
Fri, 31 Jul 1992 23:27
B415 tram: link speed vs cable length?
Author: Lee Campbell
Date: Fri, 31 Jul 1992 23:27
Date: Fri, 31 Jul 1992 23:27
33 lines
2278 bytes
2278 bytes
Howdy pardners! we're using T805's to control a 2-legged running robot and we're sending attitude and dynamic state data out over differential links to be saved for later analysis. We are using B415 trams and Belden 9804 shielded twisted pair cable (awg 28, capacitance 15.5 pF/foot) to connect the differential links. My question is: what is the specification for maximum length cable that can be used for each data rate? We are currently running at 20Megabits/ sec over a 50 foot (17 meter) cable. From what I can see in the manuals this is outside the spec, but the manuals are not all that clear. In the future we may want to run longer cables (40m or more) and we want to know what data rates we must use. Manual citations follow. In "THE TRANSPUTER DEVELOPMENT AND iQ SYSTEMS DATABOOK" 2nd edition 1991 INMOS doc # 72 TRN 219 01 in chapter 28 (IMS B415 Differential link buffer TRAM) on page 229 it says "With cable lengths up to 10m, 20Mbit/s link speed is possible. Longer cables up to 100m support lower link speeds." Unfortunately, it never says what lengths at what link speeds. Similarly, in "TRANSPUTER TECHNICAL NOTES" copyright 1989 INMOS doc # 72 TRN 200 in chapter 2.3.3 (Differential line drivers/receivers) page 35-36 it says "This system has been use by INMOS for reliable link transmission over 160 metres of twisted pair at 5Mbits/s. It should be noted that the RS 422 specification limits the maximum bit rate to 10Mbits/s at a maximum distance of 15 metres using 24 awg twisted pair. At 5Mbits/s the maximum length is about 25 metres. The RS 422 specifi- cation is, however, deliberately conservative." This seems to imply that *any* communication at 20Mbits/sec is outside the spec but maybe we can do it anyway? Thank you for your help. - Lee Campbell <elwin@ai.mit.edu>
Re: B415 tram: link speed vs cable length?
Author: Don Goodeve
Date: Tue, 04 Aug 1992 18:08
Date: Tue, 04 Aug 1992 18:08
61 lines
3284 bytes
3284 bytes
In <26118@life.ai.mit.edu> elwin@ai.mit.edu (Lee Campbell) writes: >Howdy pardners! we're using T805's to control a 2-legged running robot >and we're sending attitude and dynamic state data out over differential >links to be saved for later analysis. We are using B415 trams and Belden >9804 shielded twisted pair cable (awg 28, capacitance 15.5 pF/foot) to >connect the differential links. >My question is: what is the specification for maximum length cable that >can be used for each data rate? We are currently running at 20Megabits/ >sec over a 50 foot (17 meter) cable. From what I can see in the manuals >this is outside the spec, but the manuals are not all that clear. In >the future we may want to run longer cables (40m or more) and we want >to know what data rates we must use. In the tech note on links the only hard specification (apart from data levels - from transputer databook) is timing skew. I beleive that the maximum allowable edge timing skew on a 20MBit data stream is 5ns, that is a pulse (lo-hi-lo or hi-lo-hi - whatever way you want to look at it) should not vary in length by more than 5ns due to the effects of your transmission/reception kit. The only effects you have to watch therefore seem to be dispersion in your twisted pair. I think the 15pF per foot spec is misleading -- we are talking transmission line effects here. As long as your drivers can drive it, then consider it driven. If you are getting significant reflection effects in the cable, this too may cause problems - solved by well matched termination. The only other constraint I can think of wrt the twisted pair is the attenuation factor. Hopefully this should be fairly low over the distances you want. Final consideration - timing skew due to drivers/receivers. This may be where the RS422 spec falls down. If drivers are designed to this spec, then it could be that there timing parameters will not be as good as you require. The easiest way around such problems is probably to design your drivers/receivers using either known tight-spec MSI components or discretes. This is not a complete answer -- I am just attempting to can my experience into a short tome. If you have any queries on what I have expounded above - drop us some email and I will see if I can help further. Cheers Don ------------------------------------------------- |Don Goodeve email: don@uk.ac.york.ohm | |Adaptive Systems tel : (+44) 904 432381 | |Electronics Dept. fax : (+44) 904 432335 | |University of York ----------------------------- |UK | Give it to the Softboys.... ------------------- If only life were in an open D tuning... -- --------------------------------------------- | Don --- Well why not? Someone has got to be!| ---------------------------------------------
Thread Navigation
This is a paginated view of messages in the thread with full content displayed inline.
Messages are displayed in chronological order, with the original post highlighted in green.
Use pagination controls to navigate through all messages in large threads.
Back to All Threads