• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 1 answer
  • Subscribers 32 subscribers
  • Views 948 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

THVD1500: TVS capacitance on RS485 multipoint bus

NIKOLAS ST
Prodigy 60 points
Part Number: THVD1500
Other Parts Discussed in Thread: THVD1510

Hello all,

in order to protect the RS485 transceiver against ESD, EFT and surges we add a TVS device parallel to the bus lines.

Unfortunately, these devices add some capacitance on the bus lines, which, some how, affect the communication.

I feel that the bus speed will be restricted / decreased, as the total bus length too.

The number of nodes will also be decreased, as this number will actually be a multiplier of the effect.

My question is, how we can quantitate the effect of TVS capacitance on bus speed, bus length and nodes number,

in order to select the proper / best TVS device for our application.

An idea I have is to express the TVS capacitance to bus length equivalent capacitance

and then use the various graphs (length vs speed) available on many RS485 guides.

Assuming a cable capacitance for RS485 as 50pF/m, a TVS with 300pF would be equivalent to 6 meters of cable,

so each node equipped with such TVS would decrease the total bus length by 6 meters.

  • 0
    TI__Guru 62920 points
    Hi Nikolas,

    What you propose in terms of trading off increased capacitance with decreased cable length (of equivalent parasitic capacitance) would be a good general guideline, although this would assume that capacitance is the primary limitation restricting transmission distance. (For very long cables operating at low data rates, sometimes series resistance is the limiting factor instead.)

    It can be a little difficult to analyze this issue fully, and so in many cases the capacitance limitations are characterized via either hardware prototyping or system simulations (e.g., using the device's IBIS model). However, you can do some approximate calculations based on the information in the device datasheet.

    For example, based on the output voltage vs. current characteristic given in Figure 1 of the THVD1500 datasheet, we can compute that the equivalent output resistance of the driver is about 35 Ohms. This is obtained by choosing two points along the curve and calculating the slope. This output resistance would interact with the load resistance and capacitance (arising from cable parasitics, cable termination, leakage currents and capacitances of inactive nodes sharing the bus, etc.) to form a low-pass filter that would apply to the transmitted data waveform. Whether this filter has a significant impact on the output data stream could then be analyzed. For lower data rates it would likely just "roll off" some of the signal transitions, but for higher data rates it may introduce excessive jitter due to intersymbol interference or could attenuate the signal entirely.

    Regards,
    Max
  • 0 in reply to Miguel Robertson
    Prodigy 60 points

    Dear Max,

    thanks for the reply.

    Your approach seems to me as a complete electrical network analysis,

    but not easy to perform on a bus which is not known in detail.

    A lot of assumptions needed and in every installation all the elements of the network will be different.

    We are designing a  series of "slave" RS485 modules, the transceiver IC (THVD1500, THVD1510) has only 1/8 UL resistive load

    and although we could claim a capability of 256 nodes on the bus, the added TVS device/devices will introduce a capacitance on the lines which may lower the number of nodes we could claim.

    The higher bus speed we are interesting to is 115.2Kbs or 250Kbs (max).

    But the bus installation our modules are going to be used is every customer's decision, so it is not possible to know the actual bus speed (lower than the above upper limits),

    neither the nodes number / bus length / cable type.

    We would like to offer a guidance to our customers for the bus nodes number in relation to the bus speed and cable length, as recommended/safe operation conditions of our modules.

    This is a point where we would need some help.

    Regarding the driver equivalent output resistance (THVD1500), what you mean is to calculate the two slopes (of the high side and low side drivers) and add them?

    What Is a typical RS485 driver equivalent output resistance?

    Regards,

    Nikolas

  • 0 in reply to NIKOLAS ST
    TI__Guru 62920 points
    Hi Nikolas,

    What I meant for calculating the driver output resistance was to choose two points on the current/voltage curve and compute the slope:

    Output resistance = change in voltage / change in current.

    A typical output resistance value for devices like this would be between 20 and 50 Ohms.

    In lieu of a full network analysis (which I agree would be difficult), let's look at a few rough calculations:

    Generally for best signal integrity it is good to keep the signal rise/fall time to less than 1/3 of the duration of each bit. This allows for the signal to reach its full minimum/maximum amplitude for each bit, which helps to reduce jitter due to intersymbol interference. In this case, the minimum bit duration is 1 / (250 kbps) = 4 us. So, let's consider the worst-case allowable signal transition time to be (4 us) / 3 = 1333 ns.

    For a signal that is capacitance-limited, the 10%-to-90% transition time is equal to about 2.2 * Tau (the RC time constant). In our example this would mean that Tau would need to be less than (1333 ns) / 2.2 = 606 ns.

    Let's assume a worst-case output resistance of 50 Ohms. This would mean that the maximum allowable capacitance is (606 ns) / 50 Ohms = 12.1 nF.

    This maximum capacitance would be the total capacitance of all the nodes in parallel as well as the cable parasitic capacitance. Based on this, you could figure out how you could trade-off between node count (assuming the capacitance per node is known) and cable length (assuming the capacitance per unit length of cable is known).

    These are of course rough calculations, but they can give a useful approximation.

    Please let me know if this doesn't make sense to you.

    Regards,
    Max
  • 0 in reply to Miguel Robertson
    Prodigy 60 points
    Dear Max,
    many thanks for your analysis.
    Your approach is what was in my mind but I couldn't put it down in engineering terms.
    Everything you wrote makes perfect sense and I agree with your assumptions in general.

    I will do some work and I will come back with a model to share with you and the community.

    Best regard,
    Nikolas