ESP32 Forum

I’ve been setting up a small automation system for my hair studio to control the mood lighting and the water temperature at the backwash stations. Everything works fine on the bench but once installed near the sinks the display starts glitching every time the electric pump kicks in. I thought my wiring was solid yet the interference from the heavy-duty motors seems to mess with the SPI communication. I'm not a pro with electromagnetic noise issues and I’ve tried adding a few capacitors without much success. Since I got most of my parts from garnour.shop, I’m wondering if it’s a specific shielding issue with the modules or just my setup. The environment is quite humid and crowded with cables so it might be a grounding problem I missed. Does anyone have a trick to shield the controller better in this kind of workspace?

How long are the SPI wires? Can you change to shielded cabling for them?

the interference from the heavy-duty motors seems to mess with the SPI communication. I'm not a pro with electromagnetic noise issues

Have you put a scope or logic analyzer on it to verify that it's _really_ EMI interference from the the motor(s)? i.e. does moving them away as far as you can without changing wiring help? Does your LA show corruption? Does your scope show wild and crazy 0.4's and 2.3's lasting a (potentially improper) fraction of a clock edge in crazy places instead of 0's and 1's that are all snuggled up to clock edges?

That you've mentioned adding caps means you've at least thought about the other possibility: power interference. Can you try powering the motors and the electronics from independent power supplies just as an experiment? You didn't list a numeric value, but the pull on a small motor moving fluid can be a substantial fraction of a second, so it's not like putting a .1µF across the power rails, near the electronics, is going to give you enough of a "flywheel" to coast through the resulting sag. You need to be thinking about chunky electrolytics. I'd probably START testing at 220 µF and flip to 470 before dismissing that idea.

Fluid physics are complicated, but motors can turn into generators when they stop, especially if there's a little weight of the fluid that spins them backward for a little amount of time while the fluids stabilize in the line. Look for back-EMF on your scope, too. Maybe your careful 5v supply is turning into an 8V supply for a fraction of a second. Maybe that's OK, depending on the electronics between the actual chip and whatever power source you're using. Maybe.

Sprite's tip is, of course, solid. The field combat guide to "change to shielded cabling" is "try wrapping them in aluminum foil with an alligator clip connected to ground." That won't tell you that it WON'T help, but it can make you feel better about ordering a dozen if it DOES help.

Unsolicited wisdom: Engineering is about measuring all you can and relying on guesses and intuition when you can't. Even a cheapo <$10€ LA can tell you a LOT about I2C/SPI/UART-class signals. (More UW: If you imagine using it more than once, spend more on clips than an analyzer of this class.) There are forums that will tell you a scope that costs less than a car is garbage - and it might be - but for problems like this, even a relatively humble (think "Fnirsi") scope can tell you way more about such things than even the best DVM can. You don't need pico- or even nano-anything to see the kind of interference being discussed here on either SPI or power.

Oh, and just to mess with your problem-solving skills, it might be both power AND interference.

P.S. I do think this is one of the best subject lines I've seen on a forum like this in a while. Great application! Came for the title. Stayed to (hopefully) help.

the interference from the heavy-duty motors seems to mess with the SPI communication. I'm not a pro with electromagnetic noise issues

Have you put a scope or logic analyzer on it to verify that it's _really_ EMI interference from the the motor(s) i.e. does moving them away as far as you can without changing wiring help? Does your LA show corruption? Does your scope show wild and crazy 0.4's and 2.3's lasting a (potentially improper) fraction of a clock edge in crazy places instead of 0's and 1's that are all snuggled up to clock edges?

That you've mentioned adding caps means you've at least thought about the other possibility: power interference. Can you try powering the motors and the electronics from independent power supplies just as an experiment? You didn't list a numeric value, but the pull on a small motor moving fluid can be a substantial fraction of a second, so it's not like putting a .1µF across the power rails, near the electronics, is going to give you enough of a "flywheel" to coast through the resulting sag. You need to be thinking about chunky electrolytics. I'd probably START testing at 220 µF and flip to 470 before dismissing that idea.

Fluid physics are complicated, but motors can turn into generators when they stop, especially if there's a little weight of the fluid that spins them backward for a little amount of time while the fluids stabilize in the line. Look for back-EMF on your scope, too. Maybe your careful 5v supply is turning into an 8V supply for a fraction of a second. Maybe that's OK, depending on the electronics between the actual chip and whatever power source you're using. Maybe.

Sprite's tip is, of course, solid. The field combat guide to "change to shielded cabling" is "try wrapping them in aluminum foil with an alligator clip connected to ground." That won't tell you that it WON'T help, but it can make you feel better about ordering a dozen if it DOES help.

Unsolicited wisdom: Engineering is about measuring all you can and relying on guesses and intuition when you can't. Even a cheapo <$10€ LA can tell you a LOT about I2C/SPI/UART-class signals. (More UW: If you imagine using it more than once, spend more on clips than an analyzer of this class.) There are forums that will tell you a scope that costs less than a car is garbage - and it might be - but for problems like this, even a relatively humble (think "Fnirsi") scope can tell you way more about such things than even the best DVM can. You don't need pico- or even nano-anything to see the kind of interference being discussed here on either SPI or power.

Oh, and just to mess with your problem-solving skills, it might be both power AND interference.

P.S. I do think this is one of the best subject lines I've seen on a forum like this in a while. Great application! Came for the title. Stayed to (hopefully) help.