Multi-voltage power supply design

[Silverspur]

I need help/reviewing with the design a PCB power supply.

I'm designing a board (or a set of boards) to automate the management of a pool.

I have the following 'blocks' :

• An ESP32 SoC - 3.3V
• A pH & ORP probe "reader" board - low drift 5V
• Relays to control 230V pumps - independent 5V (connection to ESP32 outputs is made through optocouplers)
• A DS18B20 temperature probe - 5V

The 5V for the pH/ORP probes reading is done using one of two outputs of a LT6658 (low drift, high accuracy power supply). The other output will bes used to power the temperature probe.

Board(s) general power supply is AC power (220-230V).

I imagine the following combination:

• 220V - HLK-PM09 - 9V - LT6658 - 5V - pH/ORP probes (output 1) & DS18B20 probe (output 2)
• 220V - HLK-PM01 - 5V - relays
• ? - 3.3V - ESP32

The HLK-PMxx parts will be equiped with fuses and varistors for safety.

What do you think of this setting? Regarding the ESP32 power supply, is it best to use a 3.3V regulator (such as AMS1117) connected on the 9V supply, or else to use an additional dedicated HLK-PM03?

Thanks a lot for you help!

[PSR B1257]

What do you think of this setting?

Rather complicated.

Why do the pH and temperature-probes need independent power supplies?

HLK-PM09

Neat modules. I would only use one of them and derive every voltage from that.
Maybe a second one only for the relays, but than rather 12V/24V since that is a more common "industry-voltage" for relays and contactors. Speaking of relays, make sure, they are rated for motor-loads. A "10A relay" is only good for switching a 10A resistive load.

[Silverspur]

Thanks for your answer!

Why do the pH and temperature-probes need independent power supplies?

From what I undestand, the voltage supply needs to be very stable for the probes readings to be accurate (more stable than what a standard regulator can provide).

Neat modules. I would only use one of them and derive every voltage from that.
Maybe a second one only for the relays, but than rather 12V/24V

Perfect, this makes sense. I was planning to use 5V since I already have a 5V relay, but it is not clear that it is rated for motor-loads so I might have to check and adjust 

[Kleinstein]

PH probes are not that sensitive. There accuracy is anyway limited by temperaure effects. So one does not need a super stable reference. A normal 78L05 or 1117 type LDO should be good enough.
The PH probe itself is a high impedance voltage source and modern OP-amps for the amplification have a rather good PSRR. Battery operated meters may even use a 9 V block directly with no regulation. The tricky part with the PH probe may be getting good isolation and to avoid leakage in a high humidity environment.

Pool control also does not need lab grade accuracy.