Power 3.3 volt circuit from 24 volt battery, efficiently

Thread Starter

ybarigou

Joined Jul 14, 2021
9
Hi All (I am new to this forum),

I am trying to power a 3.3v circuit using a 24v battery efficiently. The circuit draws up to 100mA @ 3.3v when awake, and about 30uA @ 3.3v when in sleep mode. The ratio of sleep time to awake time is about even (1:1).

The challenge is to drop from 24v to 3.3v as efficiently as possible. Therefore, linear regulators (i.e. variable resistors) are not an option since in this case the input current is the same as the output current, and therefore 24v - 3.3v = 20.7v is waisted as heat, especially when the circuit is awake drawing 100mA.

The other option I was considering is to use a switcher (output current < input current). I tried multiple switchers but unfortunately, they all have a quiescent current of at least a few hundred uA (250uA minimum), which is not efficient enough since this current is drawn @ 24v, so not convenient for when the circuit is in sleep mode. These switchers do have an Enable Pin but unfortunately this pin switches the whole circuit off which is not an option.

Any suggestions on how to solve this problem for the sleep mode part ?
 

BobaMosfet

Joined Jul 1, 2009
2,110
@ybarigou I'm sure others will have good suggestions. My initial thought is to consider using a zener/transistor arrangement to drop the voltage from 24 to 5 before going into a 3.3V BUCK regulator. Or maybe investigate creative use of OpAmp with other parts to achieve the same goal of dropping the 24V to something lower to help drop the quiescent overall.
 

BobTPH

Joined Jun 5, 2013
8,812
100 ma @ 3.3V is 330mW.

250uA at 24V is 6mW.

Why are you so worried about the quiescent current when it is 0.2% of the power used?

Bob
 

Thread Starter

ybarigou

Joined Jul 14, 2021
9
@ybarigou I'm sure others will have good suggestions. My initial thought is to consider using a zener/transistor arrangement to drop the voltage from 24 to 5 before going into a 3.3V BUCK regulator. Or maybe investigate creative use of OpAmp with other parts to achieve the same goal of dropping the 24V to something lower to help drop the quiescent overall.
If I am able to drop to 5v from 24v efficiently then I could skip using a buck (switcher) converter and go directly with an HT7333 regulator which should suffice to go from 5v to 3.3v efficiently in both sleep and awake mode..
The problem is to go from 24v to that low range (3.3v ~ 5v) efficiently
 

Thread Starter

ybarigou

Joined Jul 14, 2021
9
100 ma @ 3.3V is 330mW.

250uA at 24V is 6mW.

Why are you so worried about the quiescent current when it is 0.2% of the power used?

Bob
You are correct. But actually, I mentioned the wrong use case (my bad!), as I am planning to use the same setup but with a ratio of sleep time to wake time of 1000:1.

Any suggestions ?
 
Last edited:

BobaMosfet

Joined Jul 1, 2009
2,110
If I am able to drop to 5v from 24v efficiently then I could skip using a buck (switcher) converter and go directly with an HT7333 regulator which should suffice to go from 5v to 3.3v efficiently in both sleep and awake mode..
The problem is to go from 24v to that low range (3.3v ~ 5v) efficiently
Okay, so.... have you looked at zener/bjt and/or opamp options.... Voltage and current don't work independently- use one to control the other.
 

Thread Starter

ybarigou

Joined Jul 14, 2021
9
Okay, so.... have you looked at zener/bjt and/or opamp options.... Voltage and current don't work independently- use one to control the other.
I looked but I'm unable to figure out how to go from 24v to 5v~3.3v using zener/bjt and/or opamp to minimize quiescent current (I'm new to this stuff). Any tips would be helpful
 

Ian0

Joined Aug 7, 2020
9,668
You are correct. But actually, I mentioned the wrong use case (my bad!), as I am planning to use the same setup but with a ratio of sleep time to wake time of 1000:1.

Any suggestions ?
Then use the linear regulator, because although the losses will be large when it is awake, it is awake so little of the time that it doesn't matter.
Same argument as @BobTPH in post #3, but reversed.
As an improvement, use a linear regulator to supply the circuit is sleep mode, but switch in a buck when it is awake.
 

BobaMosfet

Joined Jul 1, 2009
2,110
I looked but I'm unable to figure out how to go from 24v to 5v~3.3v using zener/bjt and/or opamp to minimize quiescent current (I'm new to this stuff). Any tips would be helpful
@ybarigou OpAmps satisfy your problem because they are signal devices on the input- low current (extremely). And provide whatever current the load needs on the output (within capabilities of the OpAmp).

Here is an _ideal_ simulation giving you an example of converting 24V to 3.3V into a 33-Ohm (100mA drawing) load, and it only draws 26uA from the 24V source (input side) - OpAmp will draw 100mA (whatever load demands on the output side if there is a load):

1626302635642.png

This gives you a starting point. Resistor divider establishes the 3.3V reference on the positive input, and the inverting feedback drives the voltage on output to match the reference. It rounded to 3.28, but you can fiddle with values and then use multiple resistors to get an exact value for each side of the divider.
 
Last edited:

Thread Starter

ybarigou

Joined Jul 14, 2021
9
@ybarigou OpAmps satisfy your problem because they are signal devices on the input- low current (extremely). And provide whatever current the load needs on the output (within capabilities of the OpAmp).

Here is an _ideal_ simulation giving you an example of converting 24V to 3.3V into a 33-Ohm (100mA drawing) load, and it only draws 26uA from the 24V source (input side) - OpAmp will draw 100mA (whatever load demands on the output side if there is a load):

View attachment 243550

This gives you a starting point. Resistor divider establishes the 3.3V reference on the positive input, and the inverting feedback drives the voltage on output to match the reference. It rounded to 3.28, but you can fiddle with values and then use multiple resistors to get an exact value for each side of the divider.
Thank you so much, this is incredibly helpful!
Last (quick) question: do you have any recommendation for an OpAmp (would the lm358 for example do the job) ?
 
Last edited:

BobTPH

Joined Jun 5, 2013
8,812
Using an opamp amounts to designing your own linear regulator. If you think you can beat the best commercially avaiable ones in quiescent current, go ahead and try.

The quiescent current if an LM358 is 500uA, plus, you need a reference and other circuitry which will take more.

Bob
 

tindel

Joined Sep 16, 2012
936
All these other methods are just linear regulators - all with poor efficiency. Your best bet is a buck converter. With that said... Here is an app note that uses the best of a switching regulator and a linear regulator topology to create a low noise regulated output with good efficiency. It's possible that you can use more modern parts to have the best of both worlds.
 

Thread Starter

ybarigou

Joined Jul 14, 2021
9
All these other methods are just linear regulators - all with poor efficiency. Your best bet is a buck converter. With that said... Here is an app note that uses the best of a switching regulator and a linear regulator topology to create a low noise regulated output with good efficiency. It's possible that you can use more modern parts to have the best of both worlds.
After searching plenty of buck converters online, it turns out after reading their respective datasheets that many of these buck converters advertised as "low current" are actually about the current when the buck is completely switched off through Enable PIN and not the actual quiescent current. This would not work for me since I have to keep the circuit powered on even in sleep mode.

Do you by any chance have any recommendations for a buck converter that would work for my scenario and has low quiescent current (< 250uA) ? The buck converter that was suggested earlier in this thread (MAX5096/MAX5097) seems to have a high quiescent current of 680uA (unless I'm reading the datasheet wrong)
 

BobTPH

Joined Jun 5, 2013
8,812
I don’t think you will find a buck converter that is efficient at 250uA. You could tap your battery stack to get 3V, but that would put more drain on that one cell. Perhaps have two in parallel as the fist cell?

Bob
 
Top