1
\$\begingroup\$

If you haven't read the original post, you can skip to the 2021/09/09 5P PST update

Shortly after starting to work with basic circuits (not too long ago), I was convinced that all electricity within a dc circuit had to originate and return to the same power source (with the exception of optocouplers). I now think that's not true thanks to the following:

  1. I have a cheap USB oscilloscope and it reads from one circuit while being powered by my desktop. The oscilloscope PCB doesn't appear to have anything that resembles an optocoupler. Also, digital multimeters.
  2. While tinkering with a Raspberry Pi Pico, I found this section in the datasheet (Section 4.5 https://datasheets.raspberrypi.org/pico/pico-datasheet.pdf) that discusses OR-ing a second power source with a Schottky diode.

First and foremost, what is the correct jargon for what I am trying to accomplish? Web searching could be a lot more fruitful if I knew. (While writing this up, I noticed the use of "back-powering" in the datasheet; I will begin searching this)

I have used the diode in my circuit and have had mixed results.

  • In the win column, I've connected a number of power MOSFETs' gate pins to the Pico's pins with their source/drain pins being powered by a 2A USB phone brick while the Pico itself is seemingly also powered by USB/CPU.

  • In the loss column, the above circuit but additionally connecting the SWD pins (which include a ground wire) to my CPU (a raspberry pi 3b), led to an overheating Pico linear regulator that ultimately cost it its life (I think the additional ground cable was responsible; I'm only 80% confident that I wasn't shorting something or accidently swapped anode cathode somewhere).

I want to avoid asking too broad of a question, but I'm in a pickle because I don't know what I don't know. Are there a set of DON'Ts/ best-practices that apply when using two power sources? My power MOSFETs where essentially just converting 3v3 logic to 5V logic for a self-powered BLDC driver board (another example where two power sources are operating in the same board) that presumably doesn't draw more current than what the Pico's pins can provide (50mV/pin, I believe). If I were instead driving a motor itself (so as to continue using the same MOSFET circuit, let's say single-phase), it's not clear to me that my circuit wouldn't again overwhelm the Pico's linear regulator.

I'm looking into the concept of impedance within dc circuits because I've heard it (high impedance) used when describing oscilloscope/multimeter. I have a suspicion this may be the answer to protecting my Pico (make it high impedance; not quite sure how one does that. MOSFET, since its gate is "high impedance"?). Am I on the right path?

2021/09/09 10A PST : In the past (on stack overflow), I've been pretty quick to resolve questions. I would like to think it's because I asked tightly scoped questions and not because I'm inpatient ; ) . However, in this case, I'm going to take a day or two to shore up some knowledge of what's happening between the pins of two independently powered MCUs. I think that encapsulates some of what I'm struggling with. I'll also take the time to compose some schematics to clarify any discussions. Special thanks to Jay for putting up with my obtuseness.

2021/09/09 5P PST: Hoping to clarify what it is I am searching for. Thinking about Jay's 2nd bullet point regarding our circular perception of a circuit and re-read some Wikipedia entries on transistors. Particularly helpful was this line from the Field-effect transistor entry, "Field-effect transistors have high gate-to-drain current resistance, on the order of 100 MΩ or more, providing a high degree of isolation between control and flow." This isolation is where I'm stumbling. To be as isolated as optocouplers, my circular perception demands four pins, not three. I decided to try this on my bench with a couple of iterations of a toy circuit.

Iteration 1

Iteration 1: No resistors. According to my circular perception, this shouldn't work since electricity won't flow from V1 into the gate without a return path. To my surprise, after the circuit is assembled but w/o power, providing power to V2 but not V1 (e.g. power supply off) the led would flicker unpredictably. I had read that MOSFETs gates can hold a slight charge and that MOSFETs can also have very slight leakage so I chalked this up to operating the MOSFET in an undefined/unpredictable parameter space and moved on.

Iteration 2

Iteration 2: Try to add resistors between gate and V1 kind of blindly following schematics such as the one here ( Question about mosfet gate resistor) . It more or less produced the same result, unpredictable flickering.

Final Working circuit

Iteration 3: Reasoning that I still wasn't closing the circle for the electricity flowing into the gate, I added a very high resistor between source and GND2 (The first/FINAL schematic). This works exactly as I wanted it to. With bench supply at 1A. At 0V, LED is off. At 3V, LED is on. At 4V+ LED flickers. I'm looking into this and whether it's a result of Vg > Vd (For example, learning about source-follower configuration and whether it applies for my transistor; It's not essential that I understand the ins and outs of this at this moment).

This toy example provides the necessary tool for me to join low voltage low current logic circuits to high voltage high current flow/load circuits. Unfortunately, I'm still at a loss as to why it works and under what parameters. Just beyond the source pin, it would seem that I am encountering electricity from both circuits. How do I dictate what electricity returns to GND1 via R5 and what electricity returns to GND2? Currently it doesn't seem to matter because I'm using low voltage low current on both sides. However, I need my understanding to increase so that in the high voltage high current flow/load circuit case, I don't overwhelm the GND1 return path or the resistor R5.

Many thanks for bearing with me.

\$\endgroup\$
4
  • \$\begingroup\$ You need to see it (the current) first ,before you can manage it. To see it, all that needs is a bit of intuition and insight, not a some amount of knowledge. Read this thread. It is a different material, but dealing with the same fundamentals. Please, let me know if you still don't get the answer. \$\endgroup\$
    – jay
    Sep 8 '21 at 16:08
  • \$\begingroup\$ Thanks @Jay. I've looked at the post but will take some time to let it ruminate. This answer electronics.stackexchange.com/questions/162257/… also resembles my question and provides the following intuition: control the amount flowing through the positive and the power can return through either (combined) negative. But the difference between conventional and electron flow trips me up. Is there really no possibility of all those pesky electrons creating too much heat in the Pico's puny components (like traces) or on mobo? \$\endgroup\$
    – artdv
    Sep 8 '21 at 16:39
  • \$\begingroup\$ In a follow up, hopefully helping me strengthen my intuition--if I connect grounds as recommended in Q 162257 above, if I add resistance sufficiently greater than the wall-wart's internal resistance between the two grounds, will the power-side's power (conventional flow) almost always prefer returning via the wart's beefy cables/components? How will the resistance impact the signal between the logic and power sides? \$\endgroup\$
    – artdv
    Sep 8 '21 at 16:49
  • 1
    \$\begingroup\$ "I don't know what I don't know" +1 for that important realization \$\endgroup\$
    – Rev
    Sep 8 '21 at 17:58
2
\$\begingroup\$

The oscilloscope PCB doesn't appear to have anything that resembles an optocoupler.

Most don't so you have to be careful that you don't connect the ground clips to anything that is at a voltage other than ground. Otherwise you will create a short-circuit, a high current might flow and damage result.

Also, digital multimeters.

Digital multimeters are fully isolated and have no reference to Earth. You can connect the two voltage probes to any points in a circuit.

... OR-ing a second power source with a Schottky diode.

This is fairly common.

schematic

simulate this circuit – Schematic created using CircuitLab

Figure 1. OR connected power supplies. Whichever supply has the higher voltage will power the lamp.

Usually "back-powering" or backfeeding would describe where, for example, V2 > V1 and tries to push current into V1. The diodes prevent this. (Current will only flow in the direction of the diode arrow.)

Are there a set of DON'Ts/ best-practices that apply when using two power sources?

In general, connect the grounds of the two supplies together so that voltages all reference the same common ground.

I have a suspicion this [impedance] may be the answer to protecting my Pico (make it high impedance; not quite sure how one does that. MOSFET, since its gate is "high impedance"?). Am I on the right path?

Ask another specific question on this topic and include a schematic.

\$\endgroup\$
1
  • \$\begingroup\$ Many thanks for the clarifications on back-powering and or-ing. I won't expand on the multimeter/oscilloscope example, it is not my primary concern. My primary concern is what to do when having a common ground would be very difficult (it's embed in the usb cable and comes from my motherboard; I suppose I could clip 5V/GND wires and see if the signal still reaches the MCU). What I want to avoid is current from the motor trying going to the GND of the larger power supply and not through the computer. \$\endgroup\$
    – artdv
    Sep 8 '21 at 23:19
0
\$\begingroup\$

Really confused myself over this one. Answering my own question. Still not sure about the direction of the flow in my dummy circuit but I've concluded that it shouldn't matter because I should never try to use this outside of experimentation like this. The unknown concept that I feared (and that fried my pico) is that of "Ground Loop" (pp 7 & 8 from presentation below were helpful).

The answer is, use isolation ICs. I knew of photo couplers, but my concern was twofold.

  1. What happens when you need to transmit an analog signal?
  2. Are they small enough, fast-enough, consume too much current on the logic side?

I've since discovered that there are a whole family of ICs built just for this purpose, including isolated ADCs, with varying current, speeds, etc. depending on need. TI documentation was particularly helpful.

Isolation Products Home: https://www.ti.com/isolation/overview.html

Presentation: https://training.ti.com/sites/default/files/docs/TI%20Webinar%20Isolated%20Gate%20drivers%20slides%20_0.pdf

\$\endgroup\$

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.