MCU, PT100 and multiple relays problem

[screen_cleaner]

MCU, PT100 and multiple relays problem [SOLVED]

Hello.
I have a problem with my project. It is a simple board with 4 PT100 sensors, 12 relays, 1 Character LCD and 1 4x4 keypad. The relays and PT100 sensors are on different boards. To simplify the schematic I've removed the LCD and the keypad and left just one relay and PT100 sensor.

The sensor circuit consist of constant current source(MCP1700 with a 2.5k resistor), 1 op-amp subtracting the wire resistance and 1 filtering and amplifying op-amp.
The relays are marked as 5V 200mW.

At first glance everything worked fine. The PT100 sensors are reading the correct temperature, the relays are swithing, the lcd and keypad are working as expected.

The problem is when all the relays switch on, the PT100 sensors get confused and display wrong temperature. The PT100 sensors get confused even with one relay switched on but to a lesser degree.

What I've done so far:
1. I've installed a sepperate voltage regulator for the sensors, relays and the MCU (not at the same time) with no success.
2.Installed capacitors everywhere with hope to filter the rippels of the power supply with no success.
3. Replaced the power supply with a laboratory one, with no success.
4. Bypassed the voltage regulator for the relays and hooked them directly to the unregulated power rail with no success.
5. Replaced the L78S05 voltage regulator with a more precision one BA05CC0T.
6. I've dug out the internet with no success.
7. Measured the constant current source of the sensors, but it was very stable.
8. Measured a voltage drop of 120mV on the unregulated power line.
9. There was no voltage drop on the outpud pin of L78S05 regulator.
10. Measured a voltage drop of 115mV on the MCU Vdd pin.

Since I'm just a hobbyist, not a professional in electronics, I've run out of ideas. I'll be much grateful if you give me some directions where and what to start to investigate further.

 

[ericgibbs]

hi SC,
Can you post a sketch showing the 0V/Ground line connects to the different modules.??

Poor 0V line to PSU layout can cause the problems you are reporting.

E,

 

[screen_cleaner]

The modules are connected with ribbon cables and IDC connectors. The length of the cable to the PSU is 1meter.

I hooked the oscilloscope to the L78S05 regulator's ground and made two measurements:
1. All relays off:
Relays board ground showed 21mV.
Sensor's board ground showed 6mV.

2. Eight relays on:
Relays board ground showed 198mV.
Sensor's board ground showed 25mV.

 

[ericgibbs]

The modules are connected with ribbon cables and IDC connectors. The length of the cable to the PSU is 1meter.

I hooked the oscilloscope to the L78S05 regulator's ground and made two measurements:
1. All relays off:
Relays board ground showed 21mV.
Sensor's board ground showed 6mV.

2. Eight relays on:
Relays board ground showed 198mV.
Sensor's board ground showed 25mV.

hi sc,
Looking at your images and the measured values, I would say that it is the 0V paths to the PSU which are causing your problem.

Try a 'star' connection for each module direct to the PSU 0V.

 

[screen_cleaner]

Thank you.

Very fast and accurate answer. That was exactly the problem.

Is there a better way for connecting separate boards?

 

[ericgibbs]

Thank you.

Very fast and accurate answer. That was exactly the problem.

Is there a better way for connecting separate boards?

When mixing analog and digital circuits I try to use separate connectors for the power connections to the modules, also I use heavier gauge wire if possible.

Molex type connectors are cheap and come in many pin counts.

Keep the resistance of the power wires as low as possible and connect the individual modules via their own connectors directly to the PSU output.

E.

 

[screen_cleaner]

Once again, thank you.

Here are some usefull materials for the future readers:
https://www.elmac.co.uk/pdfs/Lord_of_the_board.pdf
https://www.analog.com/library/analogDialogue/archives/39-09/layout.html