TIDA-00527: Distance vs power supply curent...?

Hello Federico,

For this reference design, the power delivery is somewhat independent of the RS-485 signaling.  The factors that you pointed out (the current rating of the coupling inductor and the DC resistance of the cabling used) are the primary factors limiting current delivery, while the voltage rating of the AC coupling capacitances used would limit the voltage achievable on the line.  Note that this DC voltage would typically not be seen by the transceiver since it would be blocked by the series capacitors, and so a high-voltage-tolerant transceiver would not necessarily need to be used.

One exception to this last point would be cases where the DC supply voltage is ramped up very quickly (or "hot plugged").  In this case the large change in voltage could make it through the capacitor and then remain until discharged via the input leakage current of the transceiver (or any other leakage paths such as bleeder or biasing resistors).  If that were something expected in your application, then I would recommend using a transceiver capable of withstanding 36 V such as THVD2450.

One thing to be aware of with this reference design is that it operates much better at high data rates (such as 1 Mbps) and with data that is encoded or scrambled such that there are no long periods of constant voltage levels.  This is because of the use of an AC coupling capacitance.  The capacitance forms a high pass filter with the input resistance of the transceiver (or termination resistor, if one is used), and the duration of each string of bits transmitted should be much less than the time constant of this filter in order to avoid distortion of the RS-485 waveforms received at the far end.  Because of this, it may be difficult to use this approach with slower data rates such as 9.6 kbps or 19.2 kbps.

Regards,
Max

Dear Max,

in first, thank You for your fast reply!

About current, my doubt is if the high current (max 500mA) can disturb / distort the comunication? I think it needs to choose the good inductor value...

I agree with You for the question about the high voltage tolerant transceiver... no hot swap events in my project.

Instead about the data rates on communication, yes I understand the question of high-pass filter, but if I increase a lot the data rates, the problem is the short cable lenght. In my case, the maximum cable lenght is about 200m and so I don't know if it's possible to arrive up to 1Mbps... I prefer to arrive up to 115200 bps. I think it need the best choice between L and C. Which is your opinion?

Thank You and Best Regards

Federico

Hi Federico,

I understand your point about the current disturbing communications.  A DC current should not have this effect (since the higher-frequency signal currents should be able to superimpose on top of this on the line).  However, if there is something like an abrupt change in the load current then it could result in differential-mode voltage noise due to the effect of the inductances.  If this noise were large enough it could appear to the RS-485 transceiver as a bit of the incorrect value.  Use of a large decoupling capacitance on the system power supplies (after the power has passed from the cabling through the inductors) will help with this.

It is a good point about the trade-off between data rates and cable distances.  Note that depending on the cable properties even 200-m lengths can support fairly high rates, though - you can reference this application note for some example measurements:

https://www.ti.com/lit/an/slla431/slla431.pdf

If you knew what to expect from the cabling you could make the best judgement which L/C values are right to optimize the trade-off between L/C size and high-pass cut-off frequency given the cabling's existing "low pass" effect.  Note that if the application can work without termination resistances it will allow for smaller capacitance values to be used, so I would recommend evaluating this.

While in general using as high of a data rate as possible is preferred, it still wouldn't solve the problem of many RS-485 applications sending frames that are relatively short compared to the time with the bus spent in an “idle” state. So, one other idea I wanted to propose is to use some resistances at the transceiver pins to bias them into a known state.

For example, you could use large-valued pull-up and pull-down resistances on the “A” and “B” pins to bias the A line to be (for example) 0.5 V above B. This would ensure the receiver interprets the bus state as high when no data is being transmitted (as is typical in UART communication). Then, if the word began with a low “start” bit it could still likely drive A enough below B to be considered a logic low. The main constraint would be making sure that the total time spent in the low state is not so long that the voltage has enough time to decay back to the high-level bias. You could determine this by considering the worst-case density of “low” bits (which, if there is no encoding, may simply be based on the number of bits in each frame and the space between frames). I hope I explained that well – please let me know if the idea is not clear.

Regards,
Max