I designed and built an analogue processing circuit for measuring power quality. I believe I made some mistakes when setting up the system's power supplies and i'm looking for some advice or general standard on what to do in this situation.

The system is running off batteries (4xAA), so over time, the supply voltage drops below the minimum needed voltage. So, I placed a centralized buck-boost to provide a stable system voltage and hooked the rest of the system supplies to the buck-boost. Below is the system supplies: enter image description here

As the circuit is doing percise measurement; percision LDO's where used for ADC and offset voltages (drawing a couple of mA). What I found is the Buck-Boost switching noise was leaking into the LDO's outputs (be it Microvolts - but this damaged system sensitivity). Mistakabily, I hooked up an STM32 MCU (using about 50-100mA) to the buck as well causing system strain and leaked MCU noise into thte system.

My question - with the given 'Vout' voltages required and the supply varying from 6-4V, what would be the best practice - separating precision and the (relatively) high power supplies? I was originally thinking if I supplied the MCU separately, I would be attenuating 6V to 3.3V, burning lots of power and requiring a beefier MCU supply to handle this. Advice?

  • \$\begingroup\$ If you want really low noise then you should add a shunt regulator to the output to power your low noise circuitry such as op-amps and ADC. Research the TLV431: ti.com/lit/ml/slyc147/slyc147.pdf and ti.com/lit/ds/symlink/tlv431.pdf \$\endgroup\$ – Jack Creasey Jun 24 '18 at 18:17
  • \$\begingroup\$ I see this all too often. People design by trial and error then try to figure out the requirements and solve problems later. Start with MUST HAVE specs 1st e.g. speed, BW, resolution, accuracy , range , CPU speed , power drain Wh etc, etc. , then we can discuss best practice with current technology. But a sufficient understanding of EME,EMI,RFI or EMC which cover similar topics is mandatory. THis is mainly impedance of inductive loop noise and capacitive field noise combined with shielding, filtering, grounding and conductance or ESR ratios. ( a priori for a good design) \$\endgroup\$ – Sunnyskyguy EE75 Jun 24 '18 at 19:08
  • \$\begingroup\$ Geometry of interference must be included in spec. Source geometry and signal, and interference geometry and noise E,B field or V,I loop. Without this, we can only talk in generalities. \$\endgroup\$ – Sunnyskyguy EE75 Jun 24 '18 at 19:12
  • \$\begingroup\$ Noise specs include tolerance on supplies, noise and spectrum \$\endgroup\$ – Sunnyskyguy EE75 Jun 24 '18 at 19:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy