1-wire middle line connections

[S. Petrukhin]

Hello, friends!
Which option of connecting sensors via the 1-wire bus do you think is more suitable for an average length of up to 100 meters?

[spostma]

I would use your left -side circuit, with GND/1-wire running over a wire pair.

Have a look at https://www.eevblog.com/forum/projects/ds18b20-1-wire-on-noisy-lines-over-cat5e-cables/
for some good suggestions for pull-up and termination...

[S. Petrukhin]

I would use your left -side circuit, with GND/1-wire running over a wire pair.

Have a look at https://www.eevblog.com/forum/projects/ds18b20-1-wire-on-noisy-lines-over-cat5e-cables/
for some good suggestions for pull-up and termination...

Yes, I also like the left circuit-it has anti-interference protection. But 1-wire is a very gentle protocol not only under the influence of interference, but also from the line capacity, from the returned reflected signal.

In the mentioned discussion, something different is discussed, and the scheme there is unsuccessful. I use this scheme:

[spostma]

I trust that you wil have determined the 910R and 1NF to be optimal for your 100m UTP cable;
the 910R is a bit low-ohmic for my feeling, and the 1NF a bit big for a reasonably-temerminated one-wire bus.

You could try to change the single 910R pullup resistor into two 1K8 resistors on both sides of R74;
that reduces the imbalance in low-signal level in the transmitting and receiving direction.

Maybe a 22R to 82R resistor in series with the 1NF capacitor could improve signals,
or splitting the 1NF into 2x 470pF ones, and one of the two capacitors using a series resistor...

Replacing the pullups by a current source (or two) should always give better (linear) flanges.

You could use schmitt-trigger inputs on your microcontoller side.

By the way, do you use some termination on the far end?

[S. Petrukhin]

I trust that you wil have determined the 910R and 1NF to be optimal for your 100m UTP cable;
the 910R is a bit low-ohmic for my feeling, and the 1NF a bit big for a reasonably-temerminated one-wire bus.

You could try to change the single 910R pullup resistor into two 1K8 resistors on both sides of R74;
that reduces the imbalance in low-signal level in the transmitting and receiving direction.

Maybe a 22R to 82R resistor in series with the 1NF capacitor could improve signals,
or splitting the 1NF into 2x 470pF ones, and one of the two capacitors using a series resistor...

Replacing the pullups by a current source (or two) should always give better (linear) flanges.

You could use schmitt-trigger inputs on your microcontoller side.

By the way, do you use some termination on the far end?

This is a competent scheme that corresponds to the research and specification of TI, it is proposed by TI, it is not my invention. I simplified it a bit by removing the active pull-up. I studied the signal, it is very good for my conditions, the equipment has passed industrial testing for more than a year. I tested this on a 200m line by deliberately laying an extra cable along noisy power lines, where power is supplied to powerful motors from frequency converters and soft starts.

If you add up 22 and 82 Оhms, you get 104 Оhms, which is very close to the UTP wave impedance. It was possible to achieve exactly 100%, but then we fall out of E24.

I have some doubts, I would be interested to know the experience of colleagues.

The terminator on the far end is not required, we have matching impedance on the MCU side.

[spostma]

I just found an excellent 1-wire design guide for long cables,
much more thorough than any Maxim/Dallas document I ever saw before:

https://sobek.su/Docs/1-Wire-Design_Guide_v1.0.pdf

I also has useful transmission data on common wiring types.

An important clue that this document gives is to limit the DV/dt slope rate of the falling edges,
especially at the master side to minimize reflections on the other cable end.

[S. Petrukhin]

I just found an excellent 1-wire design guide for long cables,
much more thorough than any Maxim/Dallas document I ever saw before:

https://sobek.su/Docs/1-Wire-Design_Guide_v1.0.pdf

I also has useful transmission data on common wiring types.

An important clue that this document gives is to limit the DV/dt slope rate of the falling edges,
especially at the master side to minimize reflections on the other cable end.

Thank you, spostma!

A really detailed and interesting document with good research. I, for example, did not know what is the difference between the capacity of one pair and between pairs.

[spostma]

I am glad that I could help you, S. Petrukhin!

It is the authors of this document that deserve credit for doing years of excellent research, not me 

By the way, your English is excellent!

With greetings to you, spostma