Teensy4.1 using an external 3.3V supply?

[W5ZZO]

I have used a Teensy 4.1 from a 5V regulated supply, and taken a small amount of power from the 3.3V pin (near the 5V in) for a voltage divider circuit (so that I had the same reference voltage that the ADC was using). I also know how/why to cut the little trace so as not to mix my USB 5V with my supply when using an external 5V supply.
I was thinking (very dangerous concept at my age) that I should cut the trace and run a project I am planning all from a small 3.3V regulated supply (everything else is also 3.3V).

So my question is, can I just power the Teensy 4.1 by connecting a 3.3V supply to the 3.3V pin next to pin23/A9? Or is that dangerous in some way I have not grasped?
I know I won't have any +5V for the USB host, and for this design I can sacrifice that capability in the interest of a more minimal project size/complexity.

- Wes, W5ZZO

 

[defragster]

From pjrc.com/store/teensy40.html

3.3V Power
Teensy 4.0 has a voltage regulator which reduces the 5V VUSB / VIN power to 3.3V for use by the main processor and most other parts. Additional circuitry may be powered from the 3.3V pin. The recommended maximum for external 3.3V usage is 250mA. Teensy 4.0 is not meant to receive power on its 3.3V pin, but this can be done with [URL="https://forum.pjrc.com/threads/64468-Teensy40-only-3-3V-power-supply?p=259469%26viewfull=1#post259469"]special modificaton[/URL].

That note is for T_4.0 with liked post by Paul : Teensy40-only-3-3V-power-supply
Perhaps that applies to T_4.1 as well as far as connect details as the same note appears here : pjrc.com/store/teensy41.html#power

 

[W5ZZO]

Thanks for pointing those links out. I searched for some info b4 I asked, but didn't see those... probably lousy search terms on my behalf.
I got the explanation I needed to determine that I was on the wrong track, my idea will NOT simplify my project.

Thanks again!

- Wes, W5ZZO

 

[sicco]

However, the 3V3 LDO regulator (U4, which was TLV75733P and now is NCV8186AMN330TAG) has an enable input. driven by the Teensy. When that pin is low, the LDO not only turns off, but also shorts the 3V3 rail to GND, thinking that it is discharging just capacitors out there. But what happens if you push 3V3 externally on the output of the LDO U4? Think a scenario where you had the Teensy4.1 powered externally with a 1+ Amp capable 3V3 source, and have a battery on VBAT, and have it in sleep mode with PMIC_ON_REQ low, and then plug in USB with 5V... Or think what happens if you push the Teensy into sleep mode by pressing its switch for x seconds... Note the new NCV8186AMN330TAG shorts its 3V3 output to GND when disabled, the NCV8186BMN330 does not, and the TLV75733P does it via an on-chip resistor. With NCV8186AMN330 the exposure is high and I think you may kill it eventually...
Just thinking aloud.
Fact or fiction?

Looks like the thermal shutdown in the NCV8186A is of little use if the power comes in via its OUT pin...

 

[Uija]

I had a raspberry pi Pico connected to the 3.3V of the Teensy, so it is powered up, when the Teensy has power.
Everytime I moved the USB cable from the teeny to the Pico, the Teensy booted up successfully (made it awesome to debug that way).
So I am not saying that it will always run, but it didn’t break anything.

 

[PaulStoffregen]

Or think what happens if you push the Teensy into sleep mode by pressing its switch for x seconds... Note the new NCV8186AMN330TAG shorts its 3V3 output to GND when disable

Indeed it is shown as a N-channel mosfet shorting to GND in Figure 2 of OnSemi's datasheet. But in Table 4 on page 3 the Output Discharge Resistance is specified as 34 ohms. It's not just a fraction of 1 ohm, as you might expect from a discrete N-channel mosfet. I doubt there's actually a 34 ohm resistor fabricated inside the chip. They probably just designed the discharge mosfet so its on resistance is approx 34 ohms. That 34 ohms probably varies with temperature and is likely much less accurate than you'd anticipate from experience with discrete resistors.

You can expect to need about 100 mA current to overcome the output discharge, but that's approximately the same current as Teensy 4.1 will draw when running at 600 MHz with USB active.

I had a raspberry pi Pico connected to the 3.3V of the Teensy, so it is powered up, when the Teensy has power.
Everytime I moved the USB cable from the teeny to the Pico, the Teensy booted up successfully (made it awesome to debug that way).
So I am not saying that it will always run, but it didn’t break anything.

You should be aware that startup with power applied only to 3.3V depends on the voltage rise time. It needs to be quick. Plugging a USB cable into an already-powered USB port is probably fast enough, assuming you haven't added large capacitors added to the 3.3V line.

But if you use a 3.3V power supply with slow "soft start" circuitry, where the voltage rises from 0 to 3.3V over several milliseconds, Teensy 4.1 can fail to start up. The reason is because VDD_SNVS_IN and VDD_SNVS_CAP need to be powered up first, or at the very least together with other power inputs. See the power up sequence documentation for details. You can see on the schematic VDD_SNVS_IN gets powered from 3 possible sources due to diodes D1 & D2. The intended startup is for either of the other 2 sources to have power first, then the 3.3V power takes over once the 3.3V regulator is enabled. When neither of the other 2 sources have power, during a slow rising 3.3V ramp, VDD_SNVS_IN is approx 0.5V "behind" the 3.3V startup. If the IMXRT chip sees 3.3V come up before VDD_SNVS_IN and VDD_SNVS_CAP, it can get stuck and fail to boot.

 

[Uija]

As I only had that setup, when I was working on the Pico firmware, the cable was plugged in into a powered USB Hub. Never did that scenario in other environments!

 

[sicco]

Indeed it is shown as a N-channel mosfet shorting to GND in Figure 2 of OnSemi's datasheet. But in Table 4 on page 3 the Output Discharge Resistance is specified as 34 ohms. It's not just a fraction of 1 ohm, as you might expect from a discrete N-channel mosfet. I doubt there's actually a 34 ohm resistor fabricated inside the chip. They probably just designed the discharge mosfet so its on resistance is approx 34 ohms. That 34 ohms probably varies with temperature and is likely much less accurate than you'd anticipate from experience with discrete resistors.

You can expect to need about 100 mA current to overcome the output discharge, but that's approximately the same current as Teensy 4.1 will draw when running at 600 MHz with USB active.
.

Maybe the LDO survives that. Maybe not.
An ignorant Teensy user/designer might think it's a good idea to power it for a portable/wireless application via 3V3 rails with say two penlites, 3VDC. And have a coin cell on VBAT. We now know that those penlites, let's say 2 Ah, will have drained in ~1 day while one would have expected the Teensy in sleep would drain microamps. Not 100 mA.
Maybe add some hardware integration guidance in https://www.pjrc.com/store/teensy41.html on to prevent such design errors?
Same for the reverse current issue https://forum.pjrc.com/threads/71253-5V-3V3-VUSB-power-supply-struggles-Teensy4-1-what-kills-my-NCV8186AMN330TAG-ics?p=314074#post314074 here.