5v MCU powered from 5-24v

[mobbarley]

I want to run a 5V MCU with an input from 5-25V, whats the best way of doing this?

I only need ~10mA so I was thinking of a series resistor with shunt a regulator but if my calculations are correct the series resistor will be at almost 2W dissipation at 25V - a little high.

[Zero999]

I think you've got the decimal point wrong, I make it 200mW.

Another option is to use a low drop-out regulator and make sure that the circuit will still work when the input voltage is 5V and the regulator is operating in drop-out.

The LM2936Z-5 has a low quiescent current current of 15µA and the drop-out voltage will be just 150mV when the load current is 10mA so with 5V in, the output voltage will be 4.85V which is within the power supply voltage specification of all 5V logic components.

[saturation]

Build a regulator around LM7805 IC, this will accommodate the wide range in Vdcin and give you a Vo= 5Vdc within 4%.

[TheDirty]

7805 has a dropout of 2V.  That wouldn't work at 5V in.  You need something with a much lower dropout, like Hero999 suggested, or some other way to avoid the dropout as Vin gets close to 5v.  I'm not the best at analog and power.

[migsantiago]

A step-down switching supply would be enough. I was going to recommend the TL2575 but it only accepts an input from 10V to 40V and you have 5V to 24V.

The easiest choice is a 7805 as the other guys say.   Forgot the dropout voltage. 

[saturation]

you're right, at least 7Vindc.  I stand corrected.

7805 has a dropout of 2V.  That wouldn't work at 5V in.  You need something with a much lower dropout, like Hero999 suggested, or some other way to avoid the dropout as Vin gets close to 5v.  I'm not the best at analog and power.

[scrat]

If your spec is strict, and you have to obtain (in the worst case) 5V from a 5V- source, then you will need a switching regulator that can even step-up voltage (buck-boost or similar).
Otherwise, if an output voltage of little less than 5V is allowed, what about bypassing an LM7805 or similar with a Schottky diode? The IC regulator will work for high input voltages, while at low values (near to 5V) the diode will provide something above 4.5V, which could be enough.

[Mechatrommer]

LM317 input 8V-40V for Vo = 5V

[Strube09]

Is the input anywhere from 5-24V?

Or would it be used at certain levels from certain sources? Like 5V wall wart, 9V wall wart, 12 or 24V Automotive?

If so you could set up a comparator circuit that would just say anything less than 5.5V would just pass the supply to the controller and anything greater than 5.5 would pass the voltage through a regulator.

What kind of micro are you using. Microchip has a few high volt chips that have a internal shunt regulator that you could look at if you are using a PIC.

[scrat]

If so you could set up a comparator circuit that would just say anything less than 5.5V would just pass the supply to the controller and anything greater than 5.5 would pass the voltage through a regulator.

It ends up that a simple reference(zener?)+opamp+MOSFET could be a good choice. For little currents, series resistance of the MOSFET can have quite low dropout (even a 10 Ohm MOSFET will give 100mV @10mA).

[Zero999]

I wouldn't rule out either the LM7805 or LM317 just yes, it depends on the minimum voltage requirements of the PIC circuit. The drop-out voltage depends on the current, for example with an output current of 10mA a typical LM317 will drop-out at 1.5V, not 3V so if the PIC circuit is still fully operational down to 3.5V, it will be fine, otherwise you need a better regulator.

[RayJones]

The obvious answer would be a SEPIC buck/boost switching regulator design.
The not so obvious is the design of these, but SEPIC into Google will get you started.

The efficiency of a SEPIC is not as good as either buck or boost regulator versions, but still is a damn sight better than a shunt regulator @ 24V. ie heat dissipation.

Another possibility depends on how tight your 5V needs to be.
If you use a Low Voltage capable PIC, you could maybe simply use a 7805.
You will probably find the 5V output regulation stops when you get to 7Vish, but the output voltage will still follow the input albeit with a 2V or so difference.
This however definitely falls into ugly design territory.

[migsantiago]

The obvious answer would be a SEPIC buck/boost switching regulator design.
The not so obvious is the design of these, but SEPIC into Google will get you started.

The efficiency of a SEPIC is not as good as either buck or boost regulator versions, but still is a damn sight better than a shunt regulator @ 24V. ie heat dissipation.

I didn't know that kind of supply. Thanks for mentioning it.

[slburris]

So let me put in a plug for the design software on the National Semiconductor
site.  You can start a design right on the front page:

http://www.national.com

So I put in the input voltage 5-24 volts, output voltage 5 volts, current 50ma,
and out popped a nice little flyback switcher design using the LM2585, complete
with schematic and a parts list.  See attached PDF for details.

Scott

[RayJones]

So let me put in a plug for the design software on the National Semiconductor
site.  You can start a design right on the front page:

http://www.national.com

So I put in the input voltage 5-24 volts, output voltage 5 volts, current 50ma,
and out popped a nice little flyback switcher design using the LM2585, complete
with schematic and a parts list.  See attached PDF for details.

Scott

Gotta love that Design.pdf file.
It even lists the components, especially the transformer.
I admit the magnetics are always my weak point with switcher designs - choice of core windings etc.

[scrat]

So let me put in a plug for the design software on the National Semiconductor
site.  You can start a design right on the front page:

http://www.national.com

So I put in the input voltage 5-24 volts, output voltage 5 volts, current 50ma,
and out popped a nice little flyback switcher design using the LM2585, complete
with schematic and a parts list.  See attached PDF for details.

Scott

Good shot! I didn't even know how powerful it could be a tool like this.

[Zero999]

The obvious answer would be a SEPIC buck/boost switching regulator design.
The not so obvious is the design of these, but SEPIC into Google will get you started.

The efficiency of a SEPIC is not as good as either buck or boost regulator versions, but still is a damn sight better than a shunt regulator @ 24V. ie heat dissipation.

Don't you think that's a little overkill for 5V at 10mA from a 24V power source?

Presumably it doesn't need to be low power so I don't see any real benefit of using an SMPs.

Just for educational purposes: an alternative to a SEPIC is to use two switching transistors so it can work both as a buck and as a boost.

http://www.eetimes.com/design/automotive-design/4006655/Linear-regulators-vs-switchers-for-automotive-applications

Another possibility depends on how tight your 5V needs to be.
If you use a Low Voltage capable PIC, you could maybe simply use a 7805.

That sounds like a good enough idea to me.

You will probably find the 5V output regulation stops when you get to 7Vish, but the output voltage will still follow the input albeit with a 2V or so difference.
This however definitely falls into ugly design territory.

The dropout voltage will be closer to 1.5V with an output current of only 10mA (see the datasheet http://www.national.com/ds/LM/LM340.pdf) so it should be better than that, if the PIC will work down to 2V, it should be fine down to 3.5V which should be more than enough contingency for a circuit running off a 5V supply. Just make sure the output is well decoupled so ensure it's still stable when regulation is lost.

[RayJones]

Yeah a SMPS for 10mA is a bit of overkill. I must admit I skimmed the current requirement.
If you reckon the voltage regulator is the go then there is really no difference to the OP's original desire to use a shunt regulator and a series dropping resistor.

Either way 200mW needs to dissipated, it either going to be the series resistor or the series voltage regulator.
The series voltage regulator will however not regulate once we reach whatever the dropout voltage is.

The shunt regulator will always give a more stable "5V" than a series pass device.
There are some nice TO-92 shunt regulators. I encountered some in a, dare I say it, SMPS I recently repaired. http://www.fairchildsemi.com/ds/KA%2FKA431.pdf is the devices I came across.

A simple 4V7 zener would maybe all that is needed and an appropriate sized resistor value, but the varying zener current would be somewhat of an issue for a stable Vout over the specified range.
The KA431 would eat it for breakfast.

[Zero999]

The advantage of a series pass device is that it uses less current when the output current falls, a shunt just uses the same amount of power regardless. In this case though, the LM7805 has a quiescent current of 6mA and the requirement is 10mA so I agree, use the (LM/TL/KA)431.

[TheDirty]

Another thread that can't actually give any final answers because the OP never returns to answer any questions regarding their actual design rules.

[Mechatrommer]

so we just talking to ourself.

[Zero999]

I think we should be more patient. We've only waited just over a day for the original poster, give it a few days, then we can start complaining. It annoys me when people ask questions on forums then whine when no one responds within a day - let's not be like that.

[TheDirty]

I think we should be more patient. We've only waited just over a day for the original poster, give it a few days, then we can start complaining.

Thought the post date was earlier than that, but I was looking at the join date.

[Mechatrommer]

I think we should be more patient. We've only waited just over a day for the original poster, give it a few days, then we can start complaining. It annoys me when people ask questions on forums then whine when no one responds within a day - let's not be like that.

ok ok! ok "pa" i'm following you!

[mobbarley]

i'm here i'm here 

A lot of suggestions - I should have started out by mentioning that I had looked into a sepic / etc but I am really aiming for a tiny pcb that works first time - efficiency is not really a concern as it will be wall-powered from the device it is plugged into.

I thought the drop out of even an LDO would be around a volt so I started looking at a shunt regulator etc. However from the look of some data sheets I should be able to get around 4.8v out from a 5v source - this is close enough to not need any bypass mechanism.

All that and everything handled by an IC + small capacitor which is great for my small footprint of ~2 sq.inches.