nRF52 at 1.8V - DCDC efficiency?

[samsonc]

Hi, Is there any information on how much improvement the DCDC has over the LDO at 1.8V? I'm working on a heavily space-constrained application, and I'm trying to make a decision as to whether or not it's worthwhile to use the board space for the inductors.

Thanks!

[RK]

The product spec has run currents at 3V supply with LDO and DCDC, that should tell you pretty much everything you need to know. Roughly, the the nrf52836, the DCDC currents are around 50%, so even adding a margin of error, that's 50% or more extra battery life.

[samsonc]

Unless I'm misunderstanding, I'm specifically asking about the system with a 1.8V rail. My system has no 3V rail. As I understand it, most of the logic and RF on chip runs at ~1V internally, but still requires an internal LDO before the buck converter, so I don't know the actual percent loss with LDO vs. DC-DC at 1.8V, where the improvement is far more marginal. Given the min operating voltage is 1.7V, I'm suspecting the improvement with a switcher is rather marginal from 1.8V, especially accounting for typical efficiencies of decent buck converters (85%ish?).

[Hung Bui]

Hi Samsonc,

I'm afraid we don't have much to provide you. The gain of using DCDC largely depends on the application you have, which peripherals running (CPU or Radio or other peripheral) and how often. The supply voltage is just one of the parameters. You may want to give us some more description such as BLE connectivity parameters, how many bytes you sent, how long is the CPU processing time, we can give you back the effectiveness of using DCDC in your use case.

[samsonc]

Thanks. I was hoping this was a common question that perhaps someone at Nordic had quick answers to. I'm planning on doing the measurements using the PPK (kit in the mail), but an early answer would be useful, as I'm working on a fairly short deadline prototype at the moment, and any derisking would be helpful. This project uses a 1.8V and is far more power and size sensitive than all of the previous projects I've done with the nRF5x. The proposed connection parameters at the moment are BLE peripheral mode, 1s update interval, and about 10 byte payload (including L2CAP & ATT headers). The CPU is on for only an additional ~100us to read data from a custom ASIC.

[samsonc]

Actually, a potential recommendation: for me, anyway, it would have been useful to see the numbers in the power profiler for DC-DC on and off. With the availability of some genuinely miniscule modules (and applications) for the nRF52 (many which don't have a built-in inductor at smaller sizes), I think there will be others who might also be asking if they really want the extra inductor hogging board space.