SN65HVD72: Extra components.

HI Max

Thank you for your reply

My device will be communicate to our other device over a twister pair cable. The cable will simply will be at least Cat5 RJ45 cables.  SO i think i will simply add the footprint for the TVS, as an option for the future incase there is EMC issues.

Would Cat5 cables be considered as high cable loss?

Regards

Dharmesh

Hi Max

Is there away to determine or calculate my max cable length? I am aiming to have them about 10-15 meters maximum, in worst case.

I have read that i can have multiple device connected to the RS-485 line, so if i am correct i would either need to give each device an ID or power them individually using a load switch to insure on message conflicts?

Hi Dharmesh,

Usually the simplest way to determine the maximum cable length for a given data rate would be to test it out. You could see at what distance communication breaks down and add some safe margin on top of that. Or, for a given cable you could take some measurements like the ones shown in the app note I linked to see if you're comfortable with the signal quality. If you didn't have hardware to test, you could compute the impact of the cabling on the RS-485 signals if you knew the insertion loss (versus frequency) or impulse response of a certain length of the cabling. Sometimes a cable vendor can provide this (particularly if the cable is marketed for higher-speed data communication), but in many cases the data is unfortunately not available.

For a general idea, you might want to reference Section 10.2.1.1 of the SN65HVD72 datasheet, which gives a length versus data rate characteristic (Figure 21) that comes from the RS-485 specification.

You are right that it is OK to have multiple devices connected to a common bus, but you just need to be careful that you only have one driver enabled (typically via the "DE" control line on the transceiver) at a given time so there are no message conflicts. There are several schemes for ensuring this. If one node has a "master" function then it could poll individual slave nodes (each of which would need to be enumerated with a unique identifier). Or, in some cases users implement a "time division multiplexing" approach in which each node would have a dedicated period of time it is allowed to transmit. Or, a "token" may be passed between each node in sequence, with each node only transmitting when they have the token (which would be passed to another node once transmission completed).

I hope this helps - let me know if you have more questions.

Regards,
Max

Hi Max

Thanks.

I will purchase RJ45 cables at various length to try at what stage it fails. I am presuming with my low data rate i could get fairly large distance. From the it seems l could get several 1000 feet as the data rate will be 250kbs.

Sorry but are stub lengths? My transceiver will be placed next to the RJ45 connector.

Yes one will be the master and will also be controlling the enable line on each devices PCB LDO. 

Dharmesh,

"Stub length" refers to any wiring branch that is not terminated. Ideally you would have a totally linear bus with termination at the two distal ends, and then any other nodes could tap into it at any point through a short cable. Sometimes, though, other nodes cannot directly connect to the main bus except through a longer length of wire. In this case, the branch from the main bus to that node would be considered a stub.

Stubs can be problematic if the propagation down them in significant with respect to the timing of the signal being transmitted. This is because the unterminated end of a stub would generate a reflection that would then interact with the intended signal on the bus. The solution to this is typically to decrease the length of the stub, to use a transceiver with slower output rising and falling edges, or to reduce the communication speed.

I'm not sure if gating the power to each transceiver in order to disable it is the best approach for you. If your LDO is disabled, then the node will not be able to transmit data but also will not be able to receive. It is generally simpler to maintain a constant power supply and to use the enable controls of the transceiver IC instead. The "DE" (driver enable) pin can be set high for the transmitting node and low for all other nodes that need to receive data. (The "/RE" or receiver enable pin can be tied to DE directly or can be controlled independently; the device will only receive data when /RE is low.)

Regards,
Max

Hi Max

Thanks

In regards to Stubs, all my cables will have Rj45 connectors on both ends. If more nodes need to be added, i will be making another PCB which acts like a hub, where multiple connectors connecting to the one  main base pcb  Would this be considered a stub? would you have an image or diagram which shows an example to avoid any confusion.

In regards to gating the LDO this was control the power to whole PCB and not just the transceiver , my aim was to insure a device to device communication while the other devices PCB are simply disabled. But now i am thinking that on my hub board  could simply have simple pull up or pull downs for each additional RJ45 connector acting as ID indicator of some sort and then have a  message protocol to determine which device is in communication. Then this way i can keep the power always on for all devices.

In regards to the DE adn RE pins, would it be possible that the master decvice(main PCB) controls theses pins for all additional devices? So when its TX it forces all other devices to be in RX mode and then wait in RX mode while any device can TX a message back? This would mean RX and TX pin for all devices would be also connected in cable. 

Would the baud rate of my UART have any effect on my max cable length i could use?

Dharmesh,

If your "hub" PCB does not re-drive the signals, it sounds like it would introduce stubs. To get a better idea what stubs are, you could reference Figure 3 in this application note:

www.ti.com/.../snla049b.pdf

In case you wanted to design an active hub board capable of redriving signals, you could reference this example design:

www.ti.com/.../TIDA-01365

I understand your goal on the power gating now. My concern would just be how to "wake up" the nodes when it is their turn to take part in communication. It seems like you would need to send a control line across your cabling for each node, which could be difficult. It would be simpler to configure an ID in hardware (via pull-up/down settings on unused RJ45 pins or DIP switch settings, for example) and then develop a protocol to request communication only with certain IDs selected at a given time.

Similarly, I think it would become cumbersome to send DE and RE controls over the cable. These would need to be sent individually to each slave node, so we are talking a large number of wires. These are single-ended logic signals as well and may be difficult to transmit over longer cabling. (If that were possible, you could have just sent the TX/D and RX/R signals between each node without converting to RS-485 at all.)

A common approach is to have a master node and a number of slave nodes with unique hardware IDs. The slaves operate in receive-only mode (DE = 0 V) until they receive a message from the master with their ID. If this message is a request for a response, then the master will disable its transmitter and the slave can then enable its transmitter (DE = VCC) in order to send the response back. After that completes, then the master can retake control of the bus.

Yes, the UART baud rate is directly tied to the signaling rate on the bus (the transceiver only transforms the logic-level signaling used by most UARTs into differential signaling suitable for longer-distance transmission). So, increasing UART baud rates will correspond to decreasing allowable cable length.

Regards,
Max

Hi Max

 Thank You for all your help. 

At the moment the HUB is not designed, but by judging figure 3 as mentioned, it would have followed bus configuration ‘A’, which I now believe is wrong. I was simply going to connect all the ‘A’ and ‘B’ together and then only activate one transceiver at time. 

So would I be correct in saying if I was to adopt the principle of a daisy chain, each of my devices would then need two transceiver, one for itself and the other to pass on the message? 

I will make the change to my design, so that I can adopt a hardware ids for each device when connected to the hub so that I can communicate without the additional GPIO’s. The extra pins that I now have from RJ45 cable can be used for address setting, so each device will now its ID once it get plugged into a socket. 

In regards to UART I will try to use the lowest baud rate(9600) as possible as my data requirement is very low as well.

  

Best Regards

 

Dharmesh Joshi

Thanks

I will get to designing by boards.

If I was to use cat6 RJ45 cable ,should i expect improved results at low baud rates?