# Line termination for single-ended to fully-differential conversion

I am trying to convert a single-ended signal from a sensor to a differential one by using THS4500 from TI. The output of the sensor goes through a buffer before being connected to the fully differential amplifier. There are two configurations that I have simulated and tested. or My question is, when is R5 (the termination resistor) necessary? From what I can tell, there is no difference in the output.

• What is the frequency range of your signals and how long is the path from U4 to U1? Dec 4, 2021 at 2:37

R3 is the input impedance to the virtual ground while R5 is in parallel, so logically from a unity gain voltage source with low impedance, R5 is redundant. I can't think of a good reason to use R5 on U4, as I assume you are choosing differential to send the signal over longer distance to improve CMRR.

This assumes you have good signal integrity between signal and all grounds.

My question is, when is R5 (the termination resistor) necessary?

• When the signal frequencies of interest are high AND
• The highest important signal frequency wavelength approaches one-tenth of the connection distance between sending buffer and receiving amplifier AND
• When R5 is the correct value for the termination resistor (which it probably isn't) AND
• You can't use a series termination resistor at the sending/transmit end (unlikely IMHO)

Then, if that criteria is met, you should consider what value the termination resistor should be and, in your example, 500 Ω is probably far too high to be effective at preventing reflections. Reflections disrupt and corrupt the signal received.

Most single ended termination resistors are in the range 30 Ω to 100 Ω for comparison.

From what I can tell, there is no difference in the output.

Well, that's good because it probably means you don't need a termination resistor but, because 500 Ω is highly likely to be far too high to be a proper termination resistor, then you might be fooling yourself because: "there is no difference in the output" doesn't mean that there isn't an improvement if you do terminate with the right resistance.

Then there is the question of positioning an appropriate termination resistor. Given that the input impedance looking into R3 towards the THS4500 is actually the value of R3 (due to the negative feedback via R2) it's possible that R3 can be lowered to the correct value of termination resistor. Or, you could add a series termination resistor at the output of U4 before it feeds signals to R3.

• Thank you for the explanation. This was very helpful. I should have mentioned that the signal, in this case, is an analog dc signal. Based on the answers above, it makes sense that the termination resistor isn't necessary.
– pmoh
Dec 6, 2021 at 17:24