search

Tympan / Tympan_Rev_E_Hardware

Repository for Tympan Rev E Electronics and case designs
MIT License
12 stars 2 forks source link

Increase capacity of the 3.3V power supply #2

Closed chipaudette closed 4 years ago

chipaudette commented 4 years ago

In my experience running 4 channels at high speed with SD writing, the system becomes unstable. I believe that the power draw is too large for the existing regulator, given the voltage of the LiPo battery.

Recent current draw measurements suggest that we can pull 200-300 mA, which is nearing the max of the current AP2210 regulator. Specifically, if we were to pull 300 mA, the voltage drop would be 250-500 mV, which means that our battery would need to have a voltage of 3.55 to 3.85 V. Given that the nominal voltage is 3.7V (though higher after immediate charge), that's a recipe for brown-out and for much reduced battery life.

Moving to the Teensy 4, where we can move to higher CPU speeds, we will likely increase this power consumption even more. I think that we'll be regularly in the 300 mA range, with excursions higher. I think that we should consider bigger power supplies.

Looking at digikey, the nearest bigger regulator is the AP2213, 500 mA. What I see, however, is that the total capacity may be higher, but that the voltage drop is the same for a given current draw. So, if our problem is that our battery voltage just isn't high enough for the regulator to deliver the required current, switching to the 2213 doesn't appear to solve the problem.

@biomurph, @eyuan-creare : does anyone know of a low-noise regulator (similar to the 2210) that has less voltage drop when in the 200-400 mA range?

biomurph commented 4 years ago

I just did a quick look at Adafruit WiFi module (HUZZAH ESP32) as those things are usually super power hungry and Adafruit is good at power supply design. They use the AP2112, which has a high output current (600mA) and what seems like darn good drop out voltage. Not enough time right now to do a head-to-head comparison, but worth a look.

eyuan-creare commented 4 years ago

Good thought Joel. Never thought of looking at what other products use. Here's a nice LDO, but it's expensive ($6). Two package sizes: ADP7156ACPZ-3.3-R7 ADP7156ARDZ-3.3-R7

Specs (click to enlarge photos)

eyuan-creare commented 4 years ago

Noting the PSRR for the existing AP2210 regulator:

eyuan-creare commented 4 years ago

If we are happy with 0.29V dropout, then there are some for $0.50 with 700mA output and low PSSR. NCP167 series

eyuan-creare commented 4 years ago

AP2112 has a 0.25V dropout, which is nice. PSSR is around 65dB:

And AP2213 only defines PSSR for 10mA, but it is around 60-70dB

It has this note, which suggests that for good PSRR, Vin should be 1V higher than Vout. Likely true for other LDOs as well.
image

eyuan-creare commented 4 years ago

Nice reference on choosing LDO regulator specs from Analog

eyuan-creare commented 4 years ago

Alternatively, we could spec a buck boost converter for providing 5V to the BT module. That would keep it from sagging the 3.3V bus when it sporadically draws current.

chipaudette commented 4 years ago

Nice work, team!

For the AP2112, we have a working schematic from Adafruit and we have Adafruit's seal-of-approval. That seals the deal for me. Let's go with that.

chipaudette commented 4 years ago

Adding a Buck-Boost option (as suggested earlier), for completeness. Shouldn't be needed with a LiPo to support by 3.3V, but there is a certain appeal:

TPS6303x: https://www.ti.com/lit/ds/symlink/tps63031.pdf

Let's stick with the AP2112 because of the adafruit documentation.

Chip

biomurph commented 4 years ago

Moving to the Teensy 4.1 and reviving this topic. The schematic for the 4.1 is HERE Paul is using the TLV75733 Vreg. The output current limit seems to be adequate. @eyuan-creare is the PSRR and drop out OK?

erk1313 commented 4 years ago

@chipaudette : @biomurph suggested the TLV75733, which is part of the Teensy 4.1 design. It has a high output current (1A), a low dropout, and a decent PSSR. The PSSR is less than our existing Rev-D regulator, but that's probably not important as we are not feeding it from a high-switching supply, nor have components that create high frequency noise. I would say it's fine.

IC Max Current Dropout @500mA PSR @ 10kHz; 500mA PSSR Plot Link
AP2112 600mA 250mV 67dB datasheet
TLV75733 1000mA 175mV 45dB datasheet
chipaudette commented 4 years ago

Sounds good...let's go with the TLV75733.