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.

TXU0304: Suggestions on use for SPI over ~0.8m distance

Part Number: TXU0304
Other Parts Discussed in Thread: SN74HCS125, SN74LVC2G86, LM311, SN74HC86, SN74LV86A, SN74HCS86, LM393LV, LM339LV, TMP144

Hello,

We are trying to evaluate a low cost option for data transmission from single remote sensor to single master host MCU via SPI operating at ~500kHz clock over a CAT6 STP cable with signal & clock in different pairs paired with ground. The environement shall be near HVAC duct of fans with only electric motors that might generate any visible E/M noise. If required we can reduce the clock frequency further upto 200kHz, however using differential drivers is not possible due to cost constraints.

We want to understand before beginning these experiments, is TXU0304 a suitable device for driving a single SPI slave to transmit data at or below 500kHz over less than 1m distance?

Can we use existing SN74HCS125 to drive the SPI from master end?

Which one would be more suited given the configurations.

Kindly guide if any other better solution is available for this use case.

  • You already have twisted pairs. So consider using LVDS. Or creating your own differential protocol by using SN74LVC2G86 and comparators.

    If you really want to use single-ended signals, then your only choice is to add low-pass filters to remove high-frequency noise, and add error checking in software. The TXU and the HCS would be equally suited.

  • Hello ,

    Thank you for your feedback,any particular suggestions for selection of LPF cutoff? Being a digital signal, we select 8x cutoff. Also, do we/should we use a 100ohm termination resistor in this configuration?

    Can you share more details on the SN74LVC2G86 + comparator based own differential protocol suggestion? It sounds like a creative approach to experiment from experimentation & learning perspective, & if it works within the budget it'll be amazing!

    One of the key challenge I face is the cost of differential drivers, the use cases we focus on our extremely cost-sensitive. So ideally solutions that can enable single-ended transmission within 1m distance shall be ideal. I (& others in my team) we make sure we minimize the long distance communication without differential drivers (we use AM26C3x pair in our BoM).

    All guidelines from your end shall be as usual, highly appreciated.

  • On a side note, from an academic exercise perspective, I am thinking of doing this custom single-ended to differential conversion.

    Can you share your comments on the following component selections? 

    LM311 shall be a good choice for experimenting with a custom single-ended to differential conversion? It has a propogation delay of ~115ns (Typ)

    Also SN74HC86 instead of SN74LVC2G86?

  • About 8× is OK for digital signals.

    The termination resistor should match the characteristic impedance of the cable. CAT6 usually has 100 Ω.

    A differential transmitter outputs two signals that are inverted from each other. This can be done with two XOR gates (this minimizes propagation delay skew). You might need to add series resistors at the transmitter to limit the current to something reasonable (also useful for the LPF). The comparator receiver is connected to both ends of the termination resistor (differential reveivers essentially are comparators).

    You can use any XOR gate as driver. SN74HC86, SN74HCS86, and SN74LV86A have about the same cost.

    The LM311 eats lots of power and is not particularly cheap. I would suggest something like the LM339LV/LM393LV (they are fast enough for 500 kHz).

  • Hello Clemens,

    Can you confirms following:

    • SN74HCS-125 & 365 both are same type of Line Drivers & Buffers except for number of channels, right?
    • Can we drive capacitive loads of upto 200pF (the Datasheet mentions to aim for less than 50-pF)
    • Do we need to install these on both ends of the cable for uni-directional signals like in SPI upto 500kHz maximum?
      • What is a better use - as transmitter or as receiver of long (~2m) cable single ended signals?
    • Any other key consideration we should be aware of - we are using 50Ω Terminatiing resistors to dampen the edges
    • We are doign proper Impedance matching for CAT6 cables
  • Yes, '125 and '365 have the same electrical characteristics.

    Capacitive loads look like a short and might exceed the absolute maximum rating for the output current, i.e., you might need to add current-limiting resistors. (For up to about 70 pF, this is not a problem.)

    You need a buffer at the transmitter end if the device that actually generates the signal has worse electrical characteristics. You need a buffer at the receiver end if the device that actually receives the signal cannot handle the voltages or slow edges.

    Also see [FAQ] What happens when I connect a logic device's output to a 50 ohm transmission line?

  • Hello Clemens,

    When using SN74HCS125 - can we use multi channles in parallel to increase the effective capacitive drive strength?

    We have to interface TMP144 sensors to our MCU UART & we are planning to use the SN74HCS125 (already in inventory) for improved drive strength & to leverage buffer's schmitt trigger inputs to improve noise resiliance further. Maximum baud rate of this UART is 5kbps.

    • Using 3 channles in parallel will it effectively increase the buffer's drive strength to 210pF (assuming 70pF per channel?)
    • Can we use 3 channels to transmit at one end & single channel to receive at the other end?
    • We plan to use dampening resistor between 33Ohm to 50Ohm to match the characteristic impedance of the cables while accounting buffers Z
    • Suggest any analytical calculations (if available in documented Application Notes) that we can use for all future references of using these buffers
  • Yes, the outputs of CMOS devices can be used in parallel.

    But increasing the drive strength just to reduce it again with dampening resistors does not make sense. For 5 kbps, you do not need fast edges, so just use a single channel with a resistor that is large enough to limit the current (at 3.3 V, at least 68 Ω).