Photodiode sensor data transmission via long cable

[0-8-15 User]

I have built a data logger that monitors the start-up star of my analog water meter and calculates the current water consumption. It works great, but now it's hard for me to decide how to get the data out of the basement where the water meter is mounted. Since there is no electricity in the basement, I have to lay a cable anyway.

Option a) Solar power it with a short cable and use wifi for the data transfer
Option b) Lay a 20 m long three- or four-core power and data cable

I prefer option b), but now I wonder if I will have problems if I simply extend the cables to the laser and the photodiode. It works in my test setup, but I don't quite trust it.
I pulse the laser diode at a rate of ~ 500 Hz, that means if I want to place the op-amp down in the cellar I would probably need at least a 4-core cable.

I basically see three options for the wired circuit:

• Simply extend the laser and photodiode wires (3-wires needed)
• Extend the op-amp output (4-wires needed)
• Add an additional microcontroller between the operational amplifier and the data logger for digital transmission of the signal. (3-wires needed)

Which route would you suggest and why?

PS: The data logger itself is currently an Arduino Nano with a DS3231 RTC, a microSD adapter and an OLED display. It logs every 1/6 revolution of the start-up star with millisecond resolution.

[tggzzz]

I pulse the laser diode at a rate of ~ 500 Hz, that means if I want to place the op-amp down in the cellar I would probably need at least a 4-core cable.

You need to consider EMI, crosstalk, and the effect of the cable capacitance on the transimpedance amplifier stability.

Consider reverse biassing the photodiode.

[Doctorandus_P]

Long cables are always a cause for concern. Long cables on sensors have been used extensively in industrial measurement and control systems but for the last 30 years or so the trend is ever more to processing sensitive signals locally.

This starts with many IC's which are available for directly digitizing data, of en with built in pre-amplifiers and signal conditioning circuitry.

My own approach would be to use a single UTP cable into the basement. One wire pair for power (24Vdc and GND) and the other for data in RS-485 format.
Then, in the basement you have some PCB with an SMPS circuit and a small microcontroller that handles both your whole circuit, analyzes it and sends data packets though RS485 to "elsewhere".

[Kleinstein]

The usual wires for home phone systems have 2 twisted pairs. These are also often used for a door bell and they are about the cheapest muti-strand cable to buy. So there is no real trouble using 4 wires.

If needed one would us 2 wires with power and data over the same pair. The pulsed operation can however make this tricky.

[0-8-15 User]

I chose the third option (with active signal processing in the basement), thanks for all the helpful tips.

[0-8-15 User]

Now that it works pretty well I would like to optimize it further and think about how to properly terminate the signal line. Right now I simply drive the output of the microcontroller in the basement high or low (depending on the pulse state) and another microcontroller at the other end of the ~ 35 meter long cable triggers on the rising edge of the signal. The max pulse frequency is around 25 Hz.

I played around a bit and came to the conclusion that both a series resistor of around 50 Ohms at the sending side or a 10 nF capacitor between the receiving pin and ground seem to work pretty well.

Should I worry about line termination at all? If so, which solution would you prefer and why?

[ledtester]

I would just get a pair of RS485 transceiver modules, e.g.:

https://www.amazon.com/s?k=rs485+transceiver

and follow the termination guidelines here:

https://e2e.ti.com/blogs_/b/analogwire/archive/2016/07/28/rs-485-basics-when-termination-is-necessary-and-how-to-do-it-properly

With RS485 you should be able to transmit/receive at whatever speed the serial port on your microcontroller is capable of.