RS485 signal characteristics in the CONV-TTL-RS485 system
Note: contains some CERN specific information for General Timing Distribution applications
The CONV-TTL-RS485 board uses the Texas Instrument SN65HVD3088E
low-power RS-485 transceiver.
This IC is the fastest in the range and is suitable for applications
that requires signaling rates up tp 20Mbps.
Each RS-485 output of the CONV-TTL-RTM-RS485-DB9 is driven by its own
RS-485 transceiver chip.
RS-485 signal distribution
At CERN, the CONV-TTL-RS485 system can be used to distribute two types of signals:
- The GMT signal: These are data frames containing General Machine Timing (GMT) telegrams. The frames are Manchester encoded, making the 500Mbps signals into 1MHz pulses.
- Pulses: These are pulses that come at any time and with varying (low relative to GMT) frequencies.
Quick overview
For RS-485 signal distribution applications, the following is a summary of the main parameters to be taken into account. Further details on each item can be found further down this page.
Property | Value |
---|---|
Differential output signal strength without load | 7-8.6V |
Differential output signal strength with 50 Ohm load | ~3V |
Maximum cable length for GMT | 100m |
Maximum cable length for pulses - f<100kHz | 1200m |
Maximum number of nodes on RS-485 bus | up to 256 |
Data rate and bus length
As shown below on figure extracted from SN65HVD3088E datasheet, page 18, by increasing cable length, data rate capabilities are impacted, also introducing jitter on the RS-485 signal.
For the purposes of timing signal distribution at CERN, the GMT signal
is manchester encoded and is transmitted as a 1MHz signal.
For this data rate and from the figure above, maximum distances would
be about 100m. For longer distances reaching 300m, 5% jitter must be
expected. whether receiver boards such as the CTRV (CERN specific)
can handle this sort of jitter must be further looked at.
Early tests have confirmed the increase of signal jitter with length of
the copper cable. The test setup consisted of RS-485 signal distribution
through varying length of twisted pair cable, with termination on both
ends of the bus (input and output). The table below gives a summary of
the results (though the reason for the decrease at 1150m is unclear).
Cable length | Measured Jitter |
---|---|
150m | ~26 ns |
200m | ~26 ns |
350m | ~69 ns |
650m | ~101 ns |
1150m | ~88 ns |
The Timing Receiver board (CTRV) has been tested to receive long distance signals, and has been able to cope with 100ns jitter.
For single pulse distribution with Frequencies below 100kHz, the RS-485 system should be able to handle distribution for up to 1200m without signal quality being compromised.
Optional termination
An RS-485 bus consists of multiple transceivers connected in parallel
the the bus. The number of nodes specified by the RS-485 standard is 32,
bu the transceivers that have been used can support up to 256 nodes.
As shown in figure below (SN65HVD3088E
datasheet, page 17),
both ends of the cable must be terminated in order to eliminate line
reflections.
On the CONV-TTL-RS485, the RS-485 differential lines are routed to and from the RTM to the motherboard through the P2 VME connector. The tracks can be 100Ohm terminated, this is optional and a solid-state relay is used to activate this option.
Two dedicated dip switches can be used to activate, input or output
termination separately. Note however that if termination is enabled on
the input, then it will be enabled on all RS-485 input channel.
Similarly with output channels, if RS-485 termination is active, then
this will apply to all output channels.
RS-485 signal electrical characteristics
The following setup was used in order to measure RS485 signal characteristics coming out of a CONV-TTL-RTM-RS495-DB9 RTM.
Note that the output from a single output line on channel 1 is measured. The the RS485_p and RS485_n signals are both fed into a single input channel of the oscilloscope. For this specific setup, this was done with a DB9 to LEMO 0S 2-pin patch cable, then a LEMO 0S to LEMO 00 adapter was used to connect a LEMO 00 cable directly to the oscilloscope via a LEMO 00 to BNC adapter. The 50Ohm load is then connected to a LEMO Y splitter, in order to show waveforms with the 50Ohm load:
And without the 50Ohm load:
By feeding the differential pair RS485_p and RS485_n to two separate channels of the oscilloscope and performing a subtraction on the two waveforms, the no load differential signals looks like this:
To summarise depending on the measurement method, a no load measurement
of the differential pair on any of the output of the
CONV-TTL-RTM-RS485-DB9 RTM should lie between 7 and 8.6 Volts.
By loading the pair with a 50Ohm load, the differential voltage is
~3Volts.