Author Topic: What's the convention for labelling grounds with low side shunts?  (Read 758 times)

0 Members and 1 Guest are viewing this topic.

Offline e100

  • Frequent Contributor
  • **
  • Posts: 409
I have a daisy chain power distribution system for a number of microcontrollers and at various places I have low side milliohm shunts to monitor the current.
This means that effectively the only true ground point on the circuit diagram is at the power supply/battery terminals. Everywhere else is a few millivolts above true ground. Do I just have Gnd, Gnd A, Gnd B an so on, or should I not even call anything that is not a true ground Gnd?
 

Offline Feynman

  • Regular Contributor
  • *
  • Posts: 125
  • Country: ch
Re: What's the convention for labelling grounds with low side shunts?
« Reply #1 on: December 05, 2020, 10:19:51 am »
There is no "true" ground. Ground can be anything that is convenient. In your case I would simply call the negative battery terminal something like "Bat-" and give each sub-circuit its own "GND_X" with a proper ground symbol. These are the "grounds" for your sub-circuits after all.

But wouldn't it be easier to measure on the high-side with a chip like the INA270 and use a common ground for all sub-circuits?
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 18166
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: What's the convention for labelling grounds with low side shunts?
« Reply #2 on: December 05, 2020, 10:58:54 am »
I would strongly recommend that [high side sensing], so you can maintain a solid ground plane under everything, improving signal quality and reducing EMI.

As for mV here and there, no, it doesn't really matter, except where it obviously does (e.g. ADC or other analog I/O, RF, etc.).  In particular, digital logic (LVCMOS) doesn't care.


The details:

In principle, the admonition should really be that at least one supply remains tied to its plane.  You could float local ground, so long as it's all bypassed to VCC as appropriate, and the VCC plane in turn covers much of the board.  This is more acceptable when loads are largely VCC-oriented, i.e., active-high signals with loads sinking to GND.  However, both device capability, and tradition, go against this -- CMOS pins pull down stronger than they pull up, so it's preferable to use active-low signals (loads sourcing from VCC), and so local grounding is more critical.  This isn't to say you can't do otherwise, but you won't have as much capacity (total mA load?) or as good performance (e.g. saturation voltage, speed) doing it the other way, and that can make the difference in some applications.

Also, as it happens, most devices are specified with total equal supply current limits.  So, unless you need the lower voltage drop, or lower total power dissipation -- you'll only be able to use up to the total capacity in a given polarity, and that's it.  The consequence is, you have the most total capacity using a mix of both.  That is, if you want to light up as many LEDs as possible from a single MCU, use a mix of active-high and active-low loads!

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline asmi

  • Super Contributor
  • ***
  • Posts: 1991
  • Country: ca
Re: What's the convention for labelling grounds with low side shunts?
« Reply #3 on: December 07, 2020, 04:41:04 pm »
Yea, floating ground is a nightmare for anything high-speed as it kills signal integrity. So I double the advice to use high-side current measurement.


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf