I'm building a circuit where the boost-convert is not enabled unless you make contact with a touch sensor. then the boost-converter starts, which applies power to the arduino who holds the EN pin high on the boost converter for some time. After that the EN goes low and the system shuts off completely.
I also have a USB charger which should only charge the battery (shouldn't affect how power is managed after the boost-converter).
My problem is that whenever I apply +5V to the USB input bus it seems to immediately results in a failure of the boost-converter IC. Prior to applying power to the USB I get a steady 5v boost from just the LiPo (and all subsystems appear to be operating). If USB power is applied then removed I measure the boost output voltage and see it oscillating/flickering at levels near the battery's voltage. At this point the IC must be replaced.
While USB power is applied the battery does appear to be charging and the USB's +5v does provide power to the PCB and everything works, its just that at this point the boost convert is toast..
I was wondering if someone could look over my (simplified) schematic and possibly spot any error(s) that could be causing this behavior.
For reference I was looking at this schematic from Adafruit.com but I didn't see much that was different, which leads me to wonder if there is an error in the way I have connected the touch sensor?
Also as a note, the chip I'm having trouble with is a 16qfn and is a bit frustrating when required to keep re-working it after making changes that don't improve anything, which is why I'm requesting some help
Boardburner2:
How are you regulating the charge current to the battery ?
If you are referring to the charge rate setting of the MCP73871 LiPo battery charger IC then the charge current can be selected using resistors connected to the Prog1 input.
Boardburner2:
That chip shows 2 4.7 uF caps, where are they ?
They should be located right next to the charging chip to avoid instability.
C3 (10uf) is connected to the output of the charger (its not as close to the IC as it could be).
C12 (10uf) is across the battery and gnd (located very close to the IC).
C8 (4.7uf) C5(10uf) are connected to the input (not as close as they could/should be).
Could capacitors/placement be the only issue? I use clean power from a regulated P.S.
And it seems to destroy the boost-converter immediately.
The inductors and capacitors for a DC-DC converter chip need to be very tightly laid-out, as
per the manufacturer's datasheet, otherwise the stray inductance will cause any number of issues.
Also the types of inductor and capacitor are usually crucial - use what the chip manufacturer
recommends (or better).
megaBlocks:
I use clean power from a regulated P.S.
And it seems to destroy the boost-converter immediately.
?
Are you sure the earthing arrangement is correct ?
Is the gnd of the psu isolated from mains ground ?
Good psu will have a removable earth strap on the front.
MarkT:
The inductors and capacitors for a DC-DC converter chip need to be very tightly laid-out, as
per the manufacturer's datasheet, otherwise the stray inductance will cause any number of issues.
Also the types of inductor and capacitor are usually crucial - use what the chip manufacturer
recommends (or better).
Of course!!
I was definitely examining the example in the datasheet to try to get things close to the specified layout parameters. Everything I used was SMT and most of it is on the top layer with plenty of ground pour nearby. All power input and regulation is more or less on one side of the board while logic and highspeed stuff is on the other.
That's a polyfuse? The series resistance of that (even if not triggered) might mean the input voltage
to the boost converter is rather high, and if the input capacitor is the wrong sort or more than a few mm
away the converter's current ripple will be transformed into input voltage ripple superimposed on the already high input voltage, causing over-voltage pulses.
Its a 600kHz converter so the dI/dt values are severe - the layout guidance in section 12.1 of the datasheet
is mandatory.
That chip is expressly designed to work from a single LiPo under load, I doubt they expected it
to be driven direct from a charger circuit. If your polyfuse triggers you pretty much guarantee failure
of the boost converter as the raw charger voltage goes direct to the converter, even with good input cap
choice and layout.
Which component are you referring to? The fuses I used on my pcb do not reset.
MarkT:
That chip is expressly designed to work from a single LiPo under load, I doubt they expected it
to be driven direct from a charger circuit.
I tried to replicate the Adafruit Powerboost 1000c and has same components I used. powerboost charges from usb and I tested that it operates under no load.
Boardburner2:
When you connect your psu what current limit do you have set ?
Being cautious, I connected an Adafruit powerboost 1000c and watched the current draw (~0.14 amp). At that point I set the limit to just above (or about 0.2amp), then I connected my PCB with a USB breakout cable to the P.S. and never saw the CC light or any different behavior compared to the Adafruit board.
It is quite possible that when plugging in the USB, you are creating an input transient to the 5V boost converter that is damaging it. One way to tell is to run the converter exclusively from the battery, and have the charger only charge the battery. If the problem goes away, then this is the cause. You might be able to get around this also by temporarily disabling the boost converter prior to powering via USB to avoid the transient.
Initially, the boost converter will be running at a certain duty cycle. When the USB is plugged in, the amount of current through L1 will change, and the duty cycle will eventually reduce to bring the currents back in balance. If the transient causes too much current through L1, it will saturate and allow a damaging surge current. Alternatively, the higher voltage on the input will cause a higher voltage on the output that is beyond the Absolute Max Voltage rating for the Vout (6V) before the duty cycle has a chance to adjust.
