Supply voltage 1.8V or 3V for microcontroller

Question

Hi. 
 Can you please tell me from what considerations the supply voltage for nRF52 is chosen? 
 1. In which cases is it better to supply it from 1.8V and in which cases from 3V? 
 2. Does the selected voltage affect the energy efficiency parameters? 
 3. Does the selected voltage affect the bluetooth performance parameters? 
 Thanks.

hmolesworth · Accepted Answer

I had to make design decisions for low-power devices using BLE, and my experiences might be useful; simply put there is no universal "better" case although item 3 is the easiest to determine. " Bluetooth performance parameters " are basically unaffected by supply voltage; the reason is that the internal 1.3 volt rail is used for both Radio and MCU and that is likely to be similar for either 1.8V or 3V supply voltage for the reasons below, in answer to items 1 and 2, and also where the external 1.8V or 3V would come from, and the use of the nRF52 output pins. 
 Minimum power consumption or optimum efficiency: The primary choices are to use battery and no external regulator, just the internal DCDC; battery and both external DCDC regulator and internal DCDC, or battery and external DCDC regulator and internal LDO. I started with the latter, an external 1.8V DCDC regulator since this would minimise the io pin current feeding other circuits such as an accelerometer or memory supply; either the internal LDO or DCDC could be used. The result was impressive in that disabling the external DCDC regulator and using just the internal DCDC regulator used less current from the battery than enabling the external DCDC with either the internal DCDC or LDO. The reason is the high efficiency of the internal DCDC and the Iq of the external DCDC, the result was a significant cost saving by chucking away the external DCDC.. 
 So my findings are that net consumption is lowest using a 3V supply, no external regulator and the internal DCDC enabled. 
 There are a couple of issues with this; the first is that the battery must not exceed 3.6 volts (nRF52832 or nRF52840 VDD) when new or fully charged or when charging; great for 3.4V CR2032 coin cells but no good for a 4.2V LiPO. Battery voltages above 3.6V require an external regulator (if not using VDDH), in which case make it a DCDC 1.8V and use the internal LDO and not the internal DCDC. Second issue is external peripheral devices, such as Flash memory, ADC stuff and the like. Some of those are fine on 1.8V and some not; some require (say) 2.8V. Those that require the higher voltage require either a buck or buck-boost regulator off the battery, or require the 1.8V being set higher to 2.8V. If the external regulator is now 2.8V instead of 1.8V, the internal DCDC has to be enabled. Note io voltage is as VDD input to nRF52; if using separate external 1.8V and 2.8V DCDC regulators the subsequent internal io at 1.8V consequently requires a level-shifter to receive signals from 2.8V external devices; though happily some external devices have both a 1.8V io supply as well as the 2.8V analogue supply, but sadly not Flash memory. Outputs are fine, but not inputs to the nRF52. 
 There are other reasons why the simplest power-efficiency of (say) coin cell direct to nRF52 with internal DCDC enabled may not be viable. The primary reason is that BLE transmissions require bursts of current which will pulse the battery voltage low, less of an issue with (say) LiPO. The resulting disturbance on external analogue signals can be minimised if a separate 2.8V rail and DCDC for the analogue circuits is used as described above since that isolates those analogue circuits from the battery voltage dip. Use of the internal DCDC generates more radiated noise compared with the internal LDO, potentially an issue for analogue circuits; this is why if using an external 1.8V DCDC regulator it is better to perhaps use the LDO instead of the internal DCDC to generate the internal 1.3V. Note also external DCDC regulator noise can be minimised by choosing shielded inductors. 
 (Note I don't have test results for using VDDH).

Supply voltage 1.8V or 3V for microcontroller

Top Replies