Voltage Operating Range

Can the IMP operate in at min. voltage of 2.8V? 

If not, what is the min. permissible operating voltage and what happens if supply goes below this value?


The imp is specified to require a 3.3V supply, however in some circumstances it can run from a lower voltage. There will be implications for some of the functionality. You should contact Hugo (hugo@electricimp.com) to discuss your application and he’ll be able to advise on exactly what will be affected.

Fen Consultants, UK

Specifically, the wifi chip needs 3.3V to operate. The rest of the Imp will survive on anything between 1.8V and 3.3V, and there’s going to be a feature where the Imp itself can control an external 3.3V PSU, and only ramp up to 3.3V when it needs to use the wifi. However, that’s currently only partially implemented (it keeps the 3.3V up all the time except when asleep, because it keeps wifi up all the time except when asleep), so it’s not on the wiki yet.


I have a question about the opposite end of the voltage scale: will the Imp still operate at 3.6V or do I need a voltage regulator to bring it down to 3.3V? I was thinking of powering it from a Lithium Thionyl Chloride battery - those put out 3.6V max and have very high capacity. The problem is that the voltage regulator on the April board appears to not go into low power mode at that low a voltage and we'd like to take advantage of the deep sleep of the Imp.

So do I need to come up with my own board or can I use the April board or power the Imp directly?


3.6v is absolute max for the imp; in general, you can run down to 3.0v without an issue. As Peter says we can survive on much lower voltages (down to 2.0v, limited by the ATSHA chip) without wifi running.

For such a low drop, you probably should consider a low Iq LDO. Something like the TI TPS73533, which has 46uA Iq, would be fine though a little power hungry. One option would be to use a really low Iq LDO (TI do one with 500nA standby, the TPS78218) and use the “external enable” functionality described by Peter to enable a bigger, more meaty LDO when higher currents were needed.

This gives you the very low power standby, but also lets you supply the wifi when needed.

Is the April board regulating the voltage from the USB?

Yes, the April has a DCDC (~3.5v - 16v) between the power inputs and the imp.

Ok, I've just tried hooking up a Lithium Thionyl battery between the 3.3V and the GND pins of the April board. If I understand correctly (please correct me if I'm wrong), this bypasses the internal regulator and powers the Imp directly.

This appears to be working fine, I haven't left it running for more than a few minutes, but it does connect to WiFi and appear in the Planner.

Do I risk breaking anything? The voltage of the Lithium Thionyl battery is very stable under load, so it's 3.6V for most of the life of the battery, it sharply decreases at the end of its life. Will the Imp mind if I keep it powered at 3.6V (most of the time in sleep mode)? The battery I'm using has a capacity of 19000mAh, so it should keep it in sleep mode for about 361 years if I calculated it correctly. Or at a transmit power of 200mA, it should be able to transmit for about 4 days non stop.

Hooking to those pins will work, but there may be an unknown amount of leakage going into the supply.

However, as I said, this is not recommended. You need to stay below 3.6v and you’re right on the edge there - and from what I’ve read, the voltage on those cells doesn’t decline much during the discharge cycle.

A 50uA LDO would still give you 38 years of standby (thought there will also be some self discharge) and the Iq of the LDO is miniscule when the imp is awake.

Thanks Hugo! I’ve just measured the current consumption when powering the April board (well, the Imp) on the 3.3V and GND pins at 3.6V - it goes down to 25uA in deep sleep mode. Am I doing something wrong or am I just leaking some power into the voltage regulator and other components on-board the April? Not that I mind much, at 25uA it’s still going to last a long time, some 86 years :). But I’m wondering what I am doing wrong (if anything).

Is there any easy way to bypass the on-board regulator (other than unsolder it)? I still need the ID chip so I can’t just remove the card and power it separately, can I?

Not very easy to totally isolate without removing the IC. You can cut traces but that’s messy as it’ll break the path to the 3v3 solder pad too.

As has been discussed elsewhere, the DCDC on the april board takes ~20uA when operating BUT can go into a ~2mA mode when the voltage gets near 3.3v. You’re right in the “danger zone” of entering the 2mA mode at 3.6v, so using an LDO would give you much more consistent power draw.

I modified my April board by cutting the traces at the voltage regulator pins 6 & 7. I could then attach a power supply to the 3V3 solder pad and it all still works but allows measurement of the actual imp current (see my post “IMP Power Profile - Empirical Measurements”).

Cutting the traces at the IC will make re-attachment relatively easy as well (he says, not having done it yet…).

Would things be a bit safer if I added a Schottky diode to drop the voltage a bit? I’m not a hardware/electronics guy, but if I understand correctly, a Schottky diode will drop 0.15-0.45V and since the battery puts out some 3.65V that would put me well within the 3.0-3.6V range. Do I really need a voltage regulator in there or can I just use a (dirt cheap) Schottky diode?

I’ve run additional tests with the battery powering up the Imp directly and it appears to work fine, even at 3.65V, but I don’t want to risk frying it.

The diode drop will be near zero when the imp is taking microamps though, which isn’t ideal (though it’d be fine for when the imp was awake).

I soldered a 3V coin battery (measured as 3.3V) to P+ and P-. I could see the imp light is on, but it is dimmer orange, not green (if I switch to USB, it is green).

Could anyone recommend a 3v3 battery that can power up the April board?


Coin cells cannot supply the required current for the imp.

The April power supply accepts anything from about 3.5v to 12v (technically 16v but the reverse voltage FET was incorrectly specced so really you shouldn’t go higher than 12 without replacing it) - this is using the P+/P- battery pack pads.

So, 4xAA will work fine and are probably the easiest solution. You may also have luck with CR123 cells connected directly to the imp as these can supply the current required and are around the right voltage.

@bipingwu: you can use LiFePO4 batteries - http://en.wikipedia.org/wiki/Lithium_iron_phosphate_battery -, those are 3.3V and rated for very high currents. Something like this: http://www.a123systems.com/lithium-ion-cells-26650-cylindrical-cell.htm . You can easily find them in stores as batteries for RC cars or helicopters. The A123 Systems are supposedly the best (and most expensive).

Hugo and Drazvan:

Thanks a lot. I bought a CR123, which is 3.3V and 1550mAh, and it works fine. I bought another one with 800mAh, and it works too.

Because of high price of these battery, I am wondering: will a battery with 400mAh work for April board?

If anyone has tried, please share. We will try in the near future.


Yes, CR123’s will work. They have very high peak current handling capacity (at least 2x what we use).

Really though, those should be connected direct to the 3.3v on the april, not to the power input (unless you want to put 2 in series, in which case obviously feed it into the BAT pads).

Right now, unless you’re using server.sleepfor(), you will run these down kinda quickly. Low power wifi mode enable will come shortly which reduces general “on network” power by about 80%.

bigingwu I’ve been using SAFT-LS26500 3.6v with April 7.7Ah not cheap