LowPowerLab Forum

Hardware support => Low Power Techniques => Topic started by: SadE54 on January 08, 2017, 06:45:48 AM

Title: LDO Choice / transient response
Post by: SadE54 on January 08, 2017, 06:45:48 AM
Hi,

I'm looking for a lower quiescent current LDO for a RFM69 board than the mcp1703 but it's not se easy :)
Low Iq LDOs have also low PSRR and very bad transient response. When you look at the MCP1703 when you're going from 100uA to 90mA of load current, there's a drop of 750mV during 300us (with 1uF output capacitor). and I guess it's almost the same case when the RFM69 is starting to send some frames at +20db.
Maybe it's a problem, maybe not . Some of you already had some sort of problem with bad electronic behavior with the mcp1703 ?
I found NCP170 / NCV8170 from On semi . They seem to have a better dynamic load response and 0.5uA of Iq
Title: Re: LDO Choice / transient response
Post by: ChemE on January 08, 2017, 08:34:34 AM
The input voltage range is so small it is hard to envision what it would be for.  Maybe stepping down from 4 AA NiMH batteries but it you were doing that, it would be better to just use 2 LiFeS2 batteries and skip the LDO altogether saving a further 500nA.  I do agree it is hard to beat the MCP 1702/1703; I've looked and never found better alternatives.
Title: Re: LDO Choice / transient response
Post by: captcha on January 08, 2017, 05:37:05 PM
I've been using the NDS352AP SOT23 P-Ch FET for various projects to successfully turn on/off a resistive voltage divider for periodic battery sampling of my motes. However, I've tried using these to switch on/off the RFM radios and it's constantly resetting the system as soon as I do the init(). I've had better results with adding extra capacitors around it but the voltage drop may explain why I'm seeing these brownouts with the 1703.

Quote from: ChemE
The input voltage range is so small it is hard to envision what it would be for.

For a wireless sensor project I'm using a 1S 150mAh LiPo feeding the MCP1703. I'm recharging using small 6cmx6cm 6V solar panels with a MCP73831T charge controller. The narrow input voltage of the NCP170 would be fine for this project as feeding the LDO with ~ 3.6V doesn't leave much headroom.
Title: Re: LDO Choice / transient response
Post by: TomWS on January 08, 2017, 08:18:09 PM
...<snip>However, I've tried using these to switch on/off the RFM radios and it's constantly resetting the system as soon as I do the init(). I've had better results with adding extra capacitors around it but the voltage drop may explain why I'm seeing these brownouts with the 1703.<snip>
I tend to go with bulk caps on the supply side when I'm using a load switch to switch in a load like RFM69 that can reach a load current of 125 mA in a hurry.  Even when switching in a low power load, but one with its own bulk caps, the transient response is too much for any voltage regulator to handle instantly.

Tom
Title: Re: LDO Choice / transient response
Post by: SadE54 on January 09, 2017, 03:23:25 AM
The input voltage range is so small it is hard to envision what it would be for.  Maybe stepping down from 4 AA NiMH batteries but it you were doing that, it would be better to just use 2 LiFeS2 batteries and skip the LDO altogether saving a further 500nA.  I do agree it is hard to beat the MCP 1702/1703; I've looked and never found better alternatives.

I don't need extra voltage for input , the max should be a 3.7v Lion or 5V from an phone external batt. The boards will have a way to shunt the regulator . And I want a regulator to avoid any problem with some components on the board , some are limited to 3.6V, even if i think a direct 3.7v won't hurt them  :P

For the transient thin , here's a nice little article :
http://www.digikey.com/en/articles/techzone/2014/jan/selecting-the-right-ultra-low-quiescent-current-ldo-regulator
Title: Re: LDO Choice / transient response
Post by: captcha on June 12, 2017, 08:07:38 AM
In one of my earlier posts I mentioned that I got brownouts and resets on the mcu whenever I powered up the RFM when using the MCP1703 LDO. I think I now have a solution that may be worthy of mentioning here in this thread. It's not so much about which LDO to pick to improve transient response, but (for me, in this particular project) a way around it.

My project uses a small, 100mAh (or 150mAh) LiPo battery because I want to keep the overall size small and this is about as big a battery that will fit in my enclosure. I always preferred running with 1S as it just simplifies re-charging. Since I was getting quite frustrated with the poor transient performance of the 1703 and didn't want to add big capacitors on an already small pcb, I investigated the option of running everything directly off the battery. My critical components are: RTC, ATmega328p, RFM (12B or 69) and a DS18B20 temperature sensor. Except for the radio module, all other components can run on as high as 5.5V. The RFM datasheets say that the maximum Vcc voltage is 3.6V for the RFM69's and 3.8V for the RFM12B's. They would be out of spec if I were to run them on 4.2V (a fully charged LiPo).

This is where I was stuck for a while until I tested with putting a diode in series after the fet that switches the power to the radio. A simple 1N4001 has a voltage drop of about 0.7V and if my LiPo is fully charged, the diode will drop it down to a safe 3.5V for the RFM. I noticed that the voltage drop changes somewhat when there's more or less current, but just 1x diode is enough for this application.

Losing the LDO made several improvements for this project:
- less current drain on the battery as the LDO is not taking up any power anymore
- more space on the circuit board as the LDO and its surrounding caps are gone (although I do now have to add the diode, DO-214 footprint)
- made the design somewhat simpler and cheaper
- improves power stability as the LDO no longer causes a dip whenever a big(ish) load is connected
- longer battery life as the LDO needs to overcome the dropout voltage (according to the datasheet a 3.3V 1703 is out of spec when fed with less than 4V on the input and supplying the maximum 250mA output). Not that I want to, but I can run my LiPo now down to 3.3V and still get the radio to work.
- no longer a restriction of 250mA current draw
 
I have this now in working operation with the RFM12B's and expect to upgrade to RFM69CW once I have redesigned the pcb's.



Title: Re: LDO Choice / transient response
Post by: TomWS on June 12, 2017, 09:25:00 AM
I can see the transient response benefit but wouldn't expect much battery life improvement as the load currents of the components powered directly from the battery are going to increase proportionally to the increase in voltage - probably pretty close to a wash given the low quiescent current of the MCP1703. 

Tom
Title: Re: LDO Choice / transient response
Post by: captcha on June 12, 2017, 07:12:41 PM
Well spotted.

True, the current draw may be more, now that the voltage can go up to 4.2V rather than everything running at 3.3V.

I was comparing apples and pears: the radio-always-on power consumption at 3.3V vs the switchable arrangement at direct battery voltage levels.

When I wasn't able to get the switch to work properly, power consumption was around 450uA, whereas now with the switched in place I get a sleep mode current draw of around 3uA. Night and day difference.
Title: Re: LDO Choice / transient response
Post by: TomWS on June 12, 2017, 08:56:05 PM
Well spotted.

True, the current draw may be more, now that the voltage can go up to 4.2V rather than everything running at 3.3V.

I was comparing apples and pears: the radio-always-on power consumption at 3.3V vs the switchable arrangement at direct battery voltage levels.
I'm not sure it would be measurably more.  It just won't be less., that was my only point.
Quote
When I wasn't able to get the switch to work properly, power consumption was around 450uA, whereas now with the switched in place I get a sleep mode current draw of around 3uA. Night and day difference.
Depending on your 'wake up' model, switching off the radio can save power, however, you should be able to get the radio to a single digit uA consumption.  In fact, unless you're using something heretical like a nA external timer of some sort,  shutting down the processor and using the radio as your timer has the best power profile.

Tom
Title: Re: LDO Choice / transient response
Post by: captcha on June 12, 2017, 11:42:55 PM
Quote from: TomWS
shutting down the processor and using the radio as your timer has the best power profile.

The application is to (some day) build a wireless sensor mesh network. I'm using the RTC as an accurate (multi-minute interval) wake-up timer so that all nodes wake up 1 second before data is exchanged across the network. Time-synchronization will be part of the protocol so that nodes can update their local clock and adjust the slew rate if their local clock is consistently out by too much. I haven't looked at using the radio's timers to trigger wakeups but I have read on this forum that people have been successfully using it.

Indeed, very interesting posts from those chasing sub-uA power consumption. Amazing..

For the moment I'm (very) happy with my 3uA! :)