Hallo Nagarajan,

some modern RS-485 transcievers have many advantages: They are slew rate limited, have a driver output protection (current limit and thermal limit) and they provide symmetrical signal routing, which is the best methode for suppressing noise.

If you try to avoid them, a different solution would not be better at same cost. A solution with equal performance would need more parts and would cost more. So, if you try to use 74HC240 keep in mind, that same performance is hardly to be achieved.

Clock frequency is not the only important parameter, when thinking in terms of EMI (radiation), slew rate of signal is also decisive! Slew rate determines how high the content of harmonics is in the range, where cable shows resonances and where harmonics can be extremely amplified.

An example:
We choose symmetrical square wave for clock signal, having a frequency of 150kHz. When transition time of edge is zero, then Fourier spectrum shows a decay of harmonics of 20dB per decade. Means, amplitude of 1.5MHz harmonic is ten times smaller than that of 150kHz. Amplitude of 15MHz harmonic is 100 times smaller. This sounds to be quite acceptable. But 10m long cable shows first resonance at about:

fres = c / lambda = 3 x 10^8 / 20 Hz = 15MHz

with lambda / 2 = 10m.
So, resonance at 15MHz results in a drastical increase of amplitude of 15MHz harmonic again and non neglectable radiation might be caused.

Slew rate of classical 74HCMOS is about 7nsec for a capacitive load of 50pF. Finite rise time introduces a second pole in Fourier spectrum: fg = 1 / pi / Tr, where Tr means 'rise time'. Above fg amplitude spectrum decays with additional 20dB per decade. So, in our example, fg is 45MHz, which results in a spectrum, which decays from 150kHz to 45MHz with 20dB per decade and above 45MHz with 40dB per decade.
From this calculation we can see, that use of 74HCMOS will not help us, to prevent radiation at 15MHz. Even if we assume, that transition time of 74HCMOS is a bit bigger with the cable.

When we now use a driver with limited slew rate, like MAX487, we get a transistion time of about 1µsec. Then, additional pole is at fg = 320kHz. As consequence we get additional suppression of about 33dB at 15MHz, finally resulting in 73dB suppression. By other words, amplitude of 15MHz harmonic is now decreased to about 1/5000 of amplitude of 150kHz! Even if we do not know much about mechanism of radiation (common mode noise etc.), it's evident, that use of slew rate limited driver helps enormously to suppress radiation!

Slew rate limited driver also helps us to struggle with transmissionline effects: If we apply a fast voltage change to a long cable, new voltage cannot be present at any point of this cable at the same moment. Instead of this, the fast voltage change, the edge, is travelling with velocity of light from driver to reciever. Velocity of light in a media like cable is some smaller than in vacuum, about one half. Remember the term SQRT(epsilon r).... So, for 10m long cable it takes about 70nsec for reaching the reciever. If cable is not terminated at both ends correctly (termination impedance must equal characteristic impedance of cable), the travelling edge will be reflected at the reciever's end of cable, means, a copy of edge is travelling back to driver. Here, a further reflection can occur, and so on.
Those reflections are becoming problematically, when electronic at reciever's end is waiting for an edge, being more precise, waiting for ONE edge. Reflections can only be prevented when impedance termination is used. RS485 transmission uses it, and can handle it! It's not so easy to drive rather small termination impedance (about 120R).
But slew rate limiting does much more: We saw, that edge needs about 70nsec for reaching reciever. But what, when there is no edge, but a slowly rising ramp of 1µsec rise time? Then, even with non well matched termination there are no shatter echoes to be expected. The different reflections cannot be distinguished any more. For understandiing this, assume that slow ramp consists of 1µsec / 70nsec = 14 small steps. Then, the only, that can be reflected at ends of cable is also small steps! If you now increase number of small steps resulting in continuously rising ramp, reflections become smaller and smaller and finally disappear.

When using 74HC240 solution you must at least provide series termination at drivers side. Then, half portion edge (don't forget voltage dividing factor: termination impedance / (termination impedance + transmission line impedance) = 1/2) travells along cable, reaches reciever side, is totally reflected (which means a doubling of signal at reciever!), travells back to driver side, and looses all it's energy in termination impedance (no further reflection). So far so good. But you still have the radiation problem..

Cross coupling:
There's another point to be considered. You have to transmit three signals. And if cables are sitting next to each other, fast edge travelling in one cable can couple into other cables via stray capacitance between cables. Here, slew rate limiting also helps! As impedance of stray capacitance (Cstray) is inverse proportional to f, according to 1 / 2 / pi / f / Cstray, cross coupling via stray capacitance of tenth harmonic is increased by ten times!
If drastically decreased cross coupling (means, use of slew rate limited drivers) is combined with symmetrical signal routing, where cross coupling of signals of both polarities cancels each other to wide extent (cross coupling of +U AND -U occures, and because both signals have same value, but different polarity effect of cross coupling vanishes!), it's possible to route several, different twisted pair cable next to another, without getting any problems. Even screening of cable can be omitted in many situations!
That's why RS485 transmission performance is so superior.

If you would like to use 74HC240, you must take care for cross coupling. And, because you do NOT have the benefit of symmetrical signal routing, use of screened cable with both ends of shield connected to ground might be essential.

You can, of course, insert some filter at output of 74HC240 for limiting slew rate and thereby reducing cross coupling. Something like 1k resistor from driver output to cable line and 1nF from cable line to ground. But result is questionable. One disadvantage of this methode is increased driver impedance, which worsens suppression of interference for frequencies below 1 / 2 / p i / 1kOhm / 1nF = 160kHz, means hum. And you need some Schmitt-trigger input at reciever side, of course.
Having an output filter is better than nothing, but still worse than using RS485 technique. Remember, you still don't have symmetrical signal routing.

The benefit of symmetrical signal routing for long cables, carrying digital signals cannot be overemphazised!

Concluding, 74HC240 solutiion CAN work, but additional measures are neccessary. Having your application running in noisy environment I would only recommend a methode comprising symmetrical signal routing (twisted pair), means the choose of RS485.

Finally one important fact to all those people who think: 'Nah, I won't use slew rate limiting, I want to have the really fast stuff!'. If cable becomes longer and longer, something rather strange happens: Transition time of edges increases automatically, even if cable is driven by fastest possible signal! The mechanism behind is 'dispersion': A fast edge travelling along a cable consists, according to Monsieur Fourier, of a whole spectrum of waves with different wavelengths, like rectangle signal consists of several harmonics. But if an electromagnetic wave is travelling in a media, velocity of travelling depends on wavelength (dispersion). So, after some meters different velocities are no longer neglectable, and dispersion can be observed. When cable is about 1000m long, rise time has increased to about 2µsec!!! Signal does not look like pure rectangle any longer. If you compare the performance of two drivers, one with and the other without slew rate limiting, you will not see any difference of signals at reciever side!

Bye,
Kai