I just purchased a A* 32U4 Mini SV and it is pretty nice board.
But there is a little detail unclear, the schematics does not give full clarity of the /SHDN input pin.
There is a NPN transistor Q1 that inverts the signal and drives the enable pin of the buck switcher.
Trouble is with the gate biasing resistors that has no values. I tried to follow the tracks on the PCB and to me it looks like those resistors do not exist at all!
Guide just says pull high this pin to shutdown the regulator but what is the actual voltage for high state? Is it 5V? What is a safe maximum input voltage?
In my opinion there really should be a current limiting resistor at the gate, like 100k. Or, are the resistors somehow built-in to the transistor Q1? Or could it actually be a MOSFET, I could not find what is the part number.
The voltage where the transistor turns on is not controlled enough to count on it for an undervoltage lockout circuit; you will probably need a higher-accuracy external circuit.
Very true, my intention was not to use the Q1 as is to create the lockout circuit.
The buck regulator has internal POR treshold of 2.4…2.95 Volts. My challenge is to create an external lockout circuit that is able to control the SHDN input properly at this voltage. My first attempt to do this with TLV431 failed, it started to oscillate at the threshold voltage. Trouble is when the buck is enabled it draws a lot of startup current and the input voltage drops, turning the lockout circuit off.
Buck has the PG output, it is unfortunate this is not used to drive the AVR reset or it would be accessible like the SHDN/STAT pins are. Now the AVR begins to run at about 4.4 volts at the BAT terminals. The AVR has power-on threshold 1.4…2.3 volts so I am not sure what causes the startup at 4.4 volts. I guess it must be the BOD detector.
Basically I’d like to keep the input current low before the voltage goes above the threshold voltage. Now the board consumes 1…8 milliamps when the BAT voltage varies around 2.2 to 4.4 volts.
Buck regulator should draw some microamps if the EN input is driven low. If I connect the SHDN to the same voltage at BAT terminal my DMM measures 0.12 … 0.14 mA current in the range 2.2 to 4.4 volts.
Yes, the brown-out reset is the reason for the 4.4 V threshold. For the regulator, are you just saying that in the 2.2V to 4.4V input range, it draws more power when enabled than when disabled? I am not clear on whether you are reporting some observations about our board, telling us about the performance of your circuit, or trying to confirm a problem/limitation with our design.
I was observing how the A-Star board is behaving. And it really does that, if SHDN is not connected it draws more power than SHDN connected to the BAT+ terminal (within 2.2 to 4.4 range).
I also measured the buck output voltage from the 5V terminal. If SHDN is not connected the voltage is roughly the same as the input voltage, in the range 0.5 volts up to 5 volts. If SHDN is tied with the BAT+ to the lab supply, the 5V output stays at 0.5 volts all the time I raise the voltage up from zero. The 0.5 volts is just a residual charge at board bypass capacitors.
This is of course not a huge deal. In normal use the board powers up just fine, the BOD prevents any misbehavior with insufficient input voltage.
I’m trying to do a batteryless system with some energy harvesting system charging a capacitor and when there is enough voltage the board should power up. Goal is to minimize the power draw while the board should be off. I could also use external fet to cut off the BAT+ supply completely but first I wanted to try the SHDN input.
I have been drafting an undervoltage lockout circuit with LMC555 timer IC that should be operational from 1.5 volts up and that can either control the SHDN input or a separate mosfet. With this I have good command of power-up and power-down thresholds and SR-flipflop prevents any oscillations at startup.
There is a tiny change that you could do if you ever plan to update the A-Star Mini design. Add a 100k resistor from REGEN net to ground. This together with 300k pull-up to VIN creates a voltage divider that enables the buck regulator when the input voltage is about 4.3 volts. And when voltage falls below about 3.5 volts or so the buck is turned off. This way the power draw is lowered significantly at these voltages. This would not change the overall behavior at all, the Q1 would still disable the buck regulator when powering from USB, just like the original use case is.
Resistor could be made even lower, like 82k. Then the buck is enabled about 5.2 volts when going up and switched off around 4.8 Volts. EN input at the buck can handle the max voltage the divider can generate, as it has same abs max 42V as the VIN.
I tested only 100k, the other value is only spice simulated.
Alternatively you could add a pin terminal to REGEN net so a user could manage the power on/off thresholds externally.
Now I noticed the ULV variant do have that resistor (R10) but it is not populated. This is a third alternative; that a user could populate this with a value that fits the use case.
Thank you for the suggestion. The regulator IC on the ULV version of the A* mini has a much more accurate under-voltage lockout threshold, which is why we included a spot for a pull down resistor on that board. As I mentioned previously, the threshold on the SV’s regulator is not accurate enough to be reliable. We will keep in mind making that pin more accessible in the future though.
Earlier discussion was about the Q1 transistor and related threshold, not the enable pin at the buck regulator.
The ULV version really needs a more accurate lockout threshold, because of really low input voltages. I don’t really see why the SV version should have that accuracy, as the voltage range is so much wider.
I soldered a 33k 1% resistor at the end of the 300k resistor and this gave me power-on threshold 12,3 volts and power-off 10 volts, this accuracy meets my needs. Well, at least based on testing with just one board those limits are pretty deterministic and I had no oscillation issues. The datasheet does give quite grim numbers for the EN threshold performance, I’m not sure what are the conditions where this variability occurs. Is it over temperature range or part to part variance, or something else.
So from my perspective this is about adding or taking away options to choose.
