THVD1400: Failsafe Biasing using 3.3V Supply

Part Number: THVD1400

Hi,

I am looking to use the THVD1400 RS485 Transceiver in a design where the RS485 bus will be connected to third party equipment and I will not have any infomation about any biasing arrangements of this.

I therefore plan to add (switchable) biasing resistors as well as termination - 560R biasing and 120R termination. I am planning to operate the THVD1400 from a 3.3V supply (interfacing to a 3.3V microcontroller) so should the positive biasing resistor go to +3.3V to keep the bus around mid-rail of my transceiver or to +5V which may be what other devices on the bus are operating at? Or does it not matter as the differential voltage will be correct with either? Would I be better to run it from a 5V supply or perhaps there is a different transceiver which would work better? I have both 3.3V and 5V supplies available.

Many thanks for any advice.

 

  • Hi Neil,

    So the THVD1400 itself does not need idle fail-safe bias resistors - because the device is idle/open/short fail-safe by design. Also there is little to no reason to add a common mode bias to the bus - a differential may be wanted but at least from a device level adding a common mode bias just adds unnecessary loading to the transceivers. 

    However if you don't know what the rest of the bus will look like and you want to have a fail-safe bias just in case there are a few things to keep note of:

    1. The differential bias across the termination should be >= x; where x = 200mV + noise_margin and you select your noise margin. By RS-485 standard >= 200mV is valid and <= -200mV is valid - between -200mV and 200mV is invalid - while devices with integrated fail-safe mostly have the thresholds between -200mV and 0V - if you don't know what the bus is connecting to beyond "RS-485" you need to assume 200mV is the minimum. If you want noise protection on that you also need to add some noise margin - typically 30mV to 50mV is pretty standard. This means that with bias voltage set to 3.3V then 560 ohms will not be >= 200mV. While this may work if every device on bus is idle fail-safe (then you just need idle voltage >= 0V and realistically it just adds noise margin then) - I don't know if you can guarantee that to be the case. 

    2. These bias resistors are also going to be loading the bus. So you need to account for other loading as well - RS-485 standard allows up to 32 unit loads on 1 bus - the actual impedance of 1UL can be variable - but in general RS-485 applications will just use 1UL = 12k ; 1/2 UL = 24k .... 1/8 UL = 96k - this specific device and most modern transceivers are 1/8th UL - or 96k input impedance.  This means that the common mode impedance should be >= 375 ohms (32 unit loads in parallel) 

    3. Worst case differential impedance for the bus should be 54 ohms - but in standard busses it will be 60 (2, 120 ohm terminations in parallel) 

    With those facts in find we can look at the sizing equation:

    RFS = (VS/VAB +1)  * 27.8 - this assumes standard 120 ohm resistor for the side of the bus without the passive fail-safe - so assumes ~60 ohms total termination. 

    Where VS is the source voltage for the bias

    VAB is the set point for the differential bias 

    We can rework it to find the bias voltage for a given resistance

    VAB = VS/(RFS/27.8 - 1)

    So if VS = 3.3V and RFS = 560 then VAB ~ 172mV ; which is lower than 200mV required for a known level in RS-485.  If VS = 5V then VAB = 260mV - which is proper. 

    You also need to look at bus loading and will need to resize one of the terminations:

    The bus cannot exceed 32UL where the amount of loading budget you have which is going to be given by:

    Loading (in UL) = 1UL/RINEQ

    RINEQ = 1/(1/375 - 1/RFS); 1UL = 12k

    RINEQ(RFS = 560) = 1.135k

    Loading (in UL) = 10.57 of total 32UL budget remains - which means 10 1UL devices can be on bus, 20 1/2UL devices, 40 1/4UL devices, or 80 1/8UL devices (THVD1400 is 1/8UL - but 1UL devices can exist on RS-485 busses) 

    Each end of bus should be terminated with 120 ohms - since the fail safe resistor pair is only placed on 1 node on the bus (it should only be placed on one node) that one node cannot be terminated with standard 120 ohms because the input impedance is no longer 120 ohms. 

    New termination is :  1/RT2 = 1/RT1 - 1/(2*RFS) --> RT2 = 1/(1/RT - 1/(2*RFS)); RT2 (RT = 120, RFS = 560) = 134.4 ohms - if you keep standard 120 you will be adding reflections to the bus. 

    ---------------------------------------------------------------

    With all of the above information - now we can look at how you should handle your system:

    1. If you don't have any idea what type of RS-485 devices that will be on the bus and/or you don't have any idea how many nodes will be added then the correct design is going to be something like this:

    VAB = 250mV; RT1 = 120; VS = 3.3V

    RFS = 394.76 ohms

    UL budget Left: 1.6 UL - so 1 1UL device, 3 1/2UL device, 6 1/4 UL devices, or 12 1/8UL devices

    RT2 = 141.5 ohms

    If you increase VS (which is just the bus bias) to 5V with the same reqs

    then RFS = 583.8 ohms

    UL budget Left: 11.45UL - so 11 1UL devices, 22 1/2UL devices, 45 1/4UL devices, 90 1/8UL devices. 

    RT = 133.7 ohms

    So I would advise that you use a VS of 5V and use similar resistors to what you are using for the best performance. 

    If you know anything about the devices you will be connecting to you may be able to relax some of these requirements - but if what you know is "RS-485" then the best bet is going to be VS = 5V  and RFS ~580 ohms (even if IC VCC = 3.3V - it doesn't matter as RS-485 devices are all required to work within spec for common mode biases of -7V to 12V - so 5V or 3.3V won't hurt the ICs - but 5V will make it easier to size proper idle fail-safes).

    Your current plan will only work if every device on the bus is idle fail-safe - which if that is the case I wouldn't be adding that much loading because you only would need a differential of like 50mV instead of > 150mV that it currently would sit at. 

    Please let me know if you have any other questions

     

    Best,

    Parker Dodson

  • Thank you for that informative reply, Parker.

    I will use a 5V supply - looks like the THVD1400 will interface ok to 3V3 input signals - and use the resistors you suggested. The other devices on the bus (probably only 1, maybe 3 max) are completely unknown to me at the moment so I just wanted to be safe. I don't think the bus will be a very long transmission line either so it should not be very demanding.

    Thanks again.

    Kind regards,

    Neil

  • Hi Neil,

    The input range should be okay with 3.3V signals (<= 0.8V is low and >= 2.0V is high for digital inputs) - the output signal on "R" pin would be my concern as that will output close to 5V with 5V supply - so generally either a level translator or a controller with 5V tolerant inputs for safety. 

    Since the bias resistors are on the bus side you could power THVD1400 with 3.3V supply and just pull up bus to 5V - that is assuming you have both a 3.3V and 5V supply available - if it is one or another then you will have to deal with the 5V on the "R" pin if you are interfacing with a 3.3V controller (either level translator or controller with 5V tolerant inputs). 

    But beyond that - yeah it sounds like you have a good plan moving forward if you have any other questions please let me know!

    Best,

    Parker Dodson

  • Hi Parker,

    Yes, I have already added a 74LVC1G125 to translate 5 -> 3V and still allows me to tristate the RxD, while keeping /RE on the THVD1400 low. I do have both supplies available so sounds like a good solution.

    Thank you for your help, much appreciated.

    Kind regards,

    Neil

  • Hi Neil,

    No problem - glad I was able to provide some help here!

    Please don't hesitate to reach back out if your have any further questions. 

    Best,

    Parker Dodson