A place users can post their projects. If you have a small project and would like your own dedicated place to post and have others chat about it then this is your spot.

User avatar
By davydnorris
#92388
jankop wrote:
davydnorris wrote:I have also switched all my sensors over to 2.8 or 1.8 volt versions, and run my ESP at 2.8V with a high efficiency buck converter instead of an LDO

I have experienced that the ESP8266 does not work reliably at voltages lower than 3V.


Not my experience at all - I only use the Espressif manufactured modules and have found 2.8V to be rock solid as long as you have enough current to start them.
User avatar
By davydnorris
#92389
Pablo2048 wrote:Do you have any efficiency and EMI measurements of your buck converter in whole range of loads? I didn't see any type of converter with efficiency > 70% overall in range of 100uA-500mA. They are way worse actually and produce large EMI. LDO in such case was better (at least according to my measurements...).


In my case I have never needed out to 500mA - the most I have ever seen is 210mA briefly on start up and have managed to tame this with a large capacitor (this also tames any ripple). By far the biggest issue I have found with LDOs is that they are terrible when the battery voltage is much higher than the supplied voltage - in this case they pull the full current required by your circuit at the full voltage drop across the battery and chew power like crazy until the voltage gets close to the circuit plus drop out.

I built some modules for testing using dry cells and loaded them up with 3 x D cell batteries in series. Those with a 3.3V LDO were dead in a week, whereas the buck converter lasted months. You could feel the heat coming from the LDO as all that power wasted into the environment.

For the tests above I actually used a buck/boost converter - the ISL9110, outputting at 3.3V. This converter has efficiency > 90% for the whole range of interest. It was the performance of the boost side at low battery voltage that made me realise it wasn't worth using boost - batteries were on their way out by the time the boost kicked in and then the current draw to make up the 3.3V just killed them in hours.

Lowering the operating voltage right across the circuit and switching to a high efficiency buck converter has worked a treat for me, both with dry batteries and with solar
User avatar
By Pablo2048
#92391
davydnorris wrote:I built some modules for testing using dry cells and loaded them up with 3 x D cell batteries in series. Those with a 3.3V LDO were dead in a week, whereas the buck converter lasted months. You could feel the heat coming from the LDO as all that power wasted into the environment.

Of course, but this is merely hypothetical test with near constant huge current. Here we are speaking about consumption over whole device range - say 100uA to 500mA peaks. This device would spent 99% of time sleeping, consuming just few uAmps and this is exactly the range where SMPS would fail.
davydnorris wrote:For the tests above I actually used a buck/boost converter - the ISL9110, outputting at 3.3V. This converter has efficiency > 90% for the whole range of interest. It was the performance of the boost side at low battery voltage that made me realise it wasn't worth using boost - batteries were on their way out by the time the boost kicked in and then the current draw to make up the 3.3V just killed them in hours.

Well the ISL9110 has quiescent current of 35uA LDO XC6203P332PR has quiescent current 8uA. If we rely on 100uA sleep current, the power consumption of ISL9110 would waste nearly 30% of energy. Also the diagrams of ISL9110 efficiency starts at 10mA, but we are talking of 100uA so we are unable to determine, how efficient the conversion would be. The ISL is constructed for shutdown and run only when necessary, but this approach complicates the whole schematics because we need external timer and/or cpu to start the converter. So my opinion is that in this usecase the LDO wins...
User avatar
By amadeuspzs
#92392 Thanks for sharing more details. I tried a TPS62203 which has a simpler footprint for hand soldering, but had stability issues with the high frequency feedback circuit.

I did some comparisons of the LDO and Buck with no load and a 22ohm load (~150mA):

Image

Image

Do you have any data on performance?

Note I only measured the outputs once as it was quickly apparent the LDO was more stable. It does draw more power at higher voltage, at load, although I don't believe heat dissipation would be measured by the ammeter (I used a https://lowpowerlab.com/guide/currentranger/) - waste by heat would be much harder to quantify?

TL;DR the LDO is working well and I get 9+ months from a 1200mAh single cell.