MarkT

The divider should limit the current to way less than needed to power up a T4 - say 47k : 15k divider will limit to 0.3mA from 15V

If you have a divider and the clamping diodes you should be ok - note that an opto coupler would be more robust especially against variations in ground potential - depends how fast a signal though.

No idea, it was and maybe still is a silicon bug, or it might have been fixed. The nice thing about the MCP23S17 is you can multiplex 8 on the same SPI bus with only one CS line, which makes it a no-brainer to chose SPI over I2C if its...

No, with I2C the A7 and B7 pins cannot be used reliably as inputs. This is a known silicon bug in the MCP23017, the SPI version (MCP23S17) is OK though - so its 7 encoders per 23017, 8 per 23S17.

Yes because they don't interfere with each other when the mux switches.

If you had an RC filter at the analog input of the Teensy you'd have to wait about 8 or 9 time constants after changing the multiplexer before taking a reading to allow it to settle to 12 bit accuracy. You can have individual RC filters on each...

Table 8 in the SGTL5000 datasheet explains the various combinations of sample rate and clock multiplier supported: https://www.nxp.com/docs/en/data-sheet/SGTL5000.pdf Audio DACs/ADCs typically claim to support only a set of fixed rates, but in...

I adapted it for T4.0 (change CS to pin 4), and it worked OK, clocks and data going out on pins 13 and 11, at 4MHz as expected.

Those are transformer-based couplers which can be pretty noisy for EMI, especially at high speeds or the ones with internal oscillators. Fast opto and capacitive couplers are also available.

Hmm, uncalibrated and no mention of apparent power v. real power, so that 4W could mean almost anything from 0W to 10W?

It should mean the transmitter and receiver side are galvanically isolated. Typically TX and RX both need supplies, often denoted Vcc1 and Vcc2.

Above about 30Mbit/s you start to require careful attention to track impedance and routing, for instance don't expect things to work reliably on a breadboard up at 80Mbit/s, even with careful layout and short wiring.

There are 3 main types of isolator, optical, capacitive and magnetic. Of the optical sort there are standard phototransistor types (all very slow), and faster ones using photo-diodes and amplification. Capacitive and magnetic ones are usually...

Piezo elements typically electrically look like a type II ceramic capacitor (which is kind of what they are) in parallel with a current source, and its best to add some protection around them. All type II ceramic caps are microphonic.

How did you measure the wall wart power consumption?

Why not use a suitable H-bridge driver chip - many advantages that way, like undervoltage lockout, programmable deadtime etc. Then standard PWM can be used. The HIP4081A has too short a deadtime, but I'm sure there's something else suitable out...

MCP23S17 interrupts? That's 16 (or maybe 8) interrupt pins unless you common them as open-drain? That would reduce the bandwidth needed on their SPI bus.

Yes, those are standard dupont-style female headers(*). Just make sure the plating matches, tin for tin, gold for gold. Mixing tin and gold is bad news over time. (*) Hundreds of manufacturers make them

In general if the processor has an FPU use it... As I understand it with dual-issue support the T4's support single float ops upto 1.2GFlops, and double floats upto 0.6GFlops. Assuming the compiler gets everything just right of course...

Note that a zener diode won't prevent damage if the Teensy is unpowered when the LEDs are powered - then the protection diodes do all the work. Its best to use Schottky diodes to back a CMOS chip's internal protection diodes, as Schottky diodes...

rotary encoder switches, or on motor shafts? The bandwidth requirements are very different... What polling frequency are you using with the buttons and encoders?

Cheap hook up wires are often AWG 26 or thinnner and CCA rather than copper. Basically they save money by using wire so thin its not compatible with the crimps used, so its not unknown for bad wires to occur (correctly crimped wires are...

And there's contact resistance to consider, should be milliohms but can be > 0.1 ohm in bad cases, and there are 4 contacts in the power/ground cable wiring all in series too...

Well its certainly fast enough - I guess the question is memory footprint?

That's not an air-cored transformer, its a toroidal transformer(*). Could it be a thermal-cutout switch operating? Or a damaged winding internally responding to thermal changes? (*) the core is ring shaped within the windings.

74HCTxx series will do the 3.3 -> 5.0V conversion reliably (74HCxx won't, according to the datasheet)

An RTOS might make sense, a large OS, not. Proper computers have 10^5 times more memory!

Conformal coating you mean? For a marine environment I'd recommend conformal coating _and_ hermetically sealed enclosures and connectors, salt water and salty air really destroys electronics and rapidly.

Maybe its scrolling the wrong direction? Maybe there's an interaction with any reflection settings?

ESD? induced transient from nearby high voltage wiring? - can happen - its best to protect low bandwidth inputs with capacitors to ground as that will protect against such transients. For a pot input 100nF is going to do nicely.

Also if in slave mode you need to use the hardware one.

Sounds like the pad is fried for that pin. Does it work for digitalRead or digitalWrite ?

Metal cans like crystals/oscillators absorb radiated IR heat less well than epoxy components, so I suspect the reflow profile wasn't quite right, leaving some un-flowed solder paste under them.

Years ago I did something to handle high datarates on another system using SDRAM (original 100MHz style, not DDR). Basically ran a datapath using SDRAMs, 16 bit D-Q latches (TI "widebus" series, 74AC16374), and generated the control signals for...

Oh yes, I'd never used that - I don't know the answer.

More power supply decoupling per strip, quite large values are needed, maybe 330uF to several mF, as the base frequencies are in the kHz range. Some strips have access to the power at both ends, try adding decoupling at both ends.

Or correctly? Can you check with a spectrum analyzer?

They often are individually free-running PWM so are already at random phases to each other, although you might need a high speed camera to check this.

You need to explain your circuit I think, there is not on/off input defined to my knowledge. From the T4 product page:

What kind of strips? Individually addressed? Or dumb LEDs with PWM on the power?

Triangle waves by definition are 50% duty cycle. Do you mean a sawtooth wave?

Heavy filtering? USB is notoriously noisy and its probably wise to only use USB cables with big ferrites at each end. It may also just be noise from the 5V power over USB which comes from the laptop - a powered hub might be worth trying...

Perhaps a DDS chip could do the synthesis? Many are set up through SPI.

Important datapoint on AI: https://www.nature.com/articles/s41598-025-24658-5 Its a tool in your tool bag - use it wisely.

And makes sure its 1.8V version too - I got caught out, I realized I'd ordered the A version, so I ordered R version but didn't check the voltage, so had to reorder again...

.step works with FreeCAD BTW.

Many discrete 12 bits ADC are fully monotonic and highly linear, but yes no ADC can remove noise or such, but careful PCB design (separate analog and digital ground planes) around the ADC will always be much better than any microcontroller with...

You mean you are not reading the EOC signal - which you need to!

Keep any inductor away from the crystal oscillator pins which can be sensitive to pickup. Normally with a BGA device the underside is peppered with lots of small decoupling caps so they can be close to the chip.

ADCs on microcontrollers are always a compromise as they are in a noisy environment and made with a less than ideal process. Go with a external 12 bit ADC I reckon, that's plenty of resolution for a rotary knob (5 minutes of arc!). There's a 16...