Need advice regarding choice of SMPS topology in a robotics application

Thread Starter

steinar96

Joined Apr 18, 2009
239
I'm currently in a bit of trouble. I'm designing a SMPS for a robotics project which requires some amperage due to it's..special nature. It's basicly a mobile beer cooler on wheels :D. It's running on 2 lead acid batteries (12V, 7.2AH) so high effeciency is required to maximize the life-time of the batteries (which is still quite short).

It's a quite delicate project since the bot is supposed to be able to map the appartment where it is operated at. Armed with RF circuitry its features will include remote control so that once you (actually i;)) thirst for beer it should be able to home in on you anywhere in the appartment.
Among it's features is automatically docking with a recharching mechanism (circuit) to recharge it's batteries once they reach critical levels.

Now that you have gotten a feel for the project you propably realise that feeding a small fridge + motors + control circuitry is a minor headache. The fridge and control circuitry will be drawing pretty stable current but the motors will introduce fast changing loads.

Obviously since it draws it's power from the batteries a SMPS is the most effecient solution. I've been reading up for the last 4-5 weeks or so (smps is such complex topic :eek:) and found the flyback topology a likely candidate. However there is some conflicting information. The power output i need is likely around 60-90W (aprox atm and depending on a few factors).

Wiki and a book i have states that flyback topology is good for between 0-150W requirements. Yet a magazine article from a engineering company specalizing in SMPS states that flyback is only good for about 0-25W esp at higher currents which makes sense due to voltage spikes which supposedly limits it's use above ~25W.

Since i havent had real world experience with flyback smps i'm having trouble deciding which source to actally take seriously. Is anyone able to shed light on this matter. The question is whether i should continue with the flyback design at aprox 60-90W or move up to a half-bridge topology.

Any advice/feedback appreciated :)
 

THE_RB

Joined Feb 11, 2008
5,438
Aren't you overcomplicating things? The fridge section requires 12v so you just run from the 12v batteries. The motor drivers will most likely be PWM hbridge drivers which act like switchmode supplies (ie they only use the required energy to move the robot, especially if you have wheel feedback encoders). It should be pretty efficient from the start.

Maybe you could look at improving the efficiency of the fridge section, those little 6pack peltier style car coolers are ok and you could mess with its controller to reduce its power needs and add some extra insulation. I assume you are going to put cold beers in it, so it doesn't need to cool them it just has to keep them cool which is very different.

Sorry that doesn't directly address your question. :)
 

jpanhalt

Joined Jan 18, 2008
11,087
I agree with THE RB, but even if you pursue SMPS, you will need to be more specific about your voltage and power needs.

What voltage(s) and currents do you need from your 12-V batteries?

John
 

Thread Starter

steinar96

Joined Apr 18, 2009
239
Actually jpanhalt. I dont think the voltage levels matter to answer my question. The flyback topology is capable of step-up/step-down multiple outputs which is what matters. So does the half bridge-topology. The question is still whether i should go for the half-bridge topology or flyback. But if you are curious of what i had in mind regarding voltages it's just 5V (logic), 12V (fridge) and 24V. This might change depending on the requirement of the motors used to drive the whole thing (12 or 24V)

I've stated my power needs already, which are somewhere between 60-90W. I would prefer the flyback topology because it's less complicated and less components then the half-bridge. But if the flyback is not a viable option at power levels above 25W and at some amparage then i'm better off with the half bridge. So if someone with experience could shed some light on this then that would be great.

And THE_RB yeah i might be complicating things slightly but nevertheless i need multiple voltages for all kinds of different circuitry. The logic is 5V, fridge is propably 12V, the motors are either 12V or 24V. This project is supposed to be slightly challenging :)
 
Last edited:

AlexR

Joined Jan 16, 2008
732
I totally agree with THE_RB, you are over complicating things.
You have a 12 volt fridge, why not use 12 volt motors, then the only supply you would need is the 5 volt for the logic which can be done with a buck SMPS.
Don't forget that however efficient a SMPS may be it will always be quite a bit less than 100% efficient so why waste power converting 12V to 24V when you don't need to.
 

Thread Starter

steinar96

Joined Apr 18, 2009
239
You are most likely correct. Atm i am assuming however that the fridge voltage is 12V (most likely though). But if it is i'm propably best off using 12V motors also and aquire the 5V from a buck regulator. It is however tempting to build a 3 output smps:D but atm i find myself likely to take your advice if the fridge is running on 12V.

I still wonder though if someone is able to answer my original question.
 

jpanhalt

Joined Jan 18, 2008
11,087
5V logic can usually be supplied by a simple linear regulator. Much easier. As for 12V and 24 V, you have 2 batteries. You can use a center tap for the 12V. I would probably just go with 12 V motors and avoid that complexity. In fact, your 24V motors may work fine at 12V PWM for slow moving.

John
 

Thread Starter

steinar96

Joined Apr 18, 2009
239
It would be quite easy yes, but linearly regulating from 12V to 5V for alot of logic circuitry does introduce a bit of wasted wattage. It might not seem alot but it is a design goal to maximize effeciency.
 
Last edited:

dummhuman

Joined May 11, 2009
6
AlexR was right when he said that the wisest course of action would be to use a buck switched mode power supply for your logic, as this will provide you with well over 80% efficiency (probably well over 90%), which, compared to the linear regulator (<45%), is spectacular. You will have to take care to make sure the noise from your SMPS does not interfere with your logic circuits.
The one reason you might want to still use a SMPS for your fridge is that the fridge will likely have problems as the battery discharges, because most fridges are not designed for use with depleted batteries; for this I might suggest using an LM5022 to boost the battery output voltage to 12V, the benefit of this is that the LM5022 can operate down to 0% duty cycle, and has under voltage lock out, meaning that you can configure it to provide you with a very steady 12 volts until the battery is drained to the lowest safe level. In addition, this regulator is very efficient, when the battery is down to 75% of nominal voltage, it can still give you 95% efficiency. I am currently using this regulator in a project, and I have become a big fan.

Good Luck,
Nick
 

Thread Starter

steinar96

Joined Apr 18, 2009
239
Thank your for your post dummhuman, the original reason for a multi-output smps for the entire thing was partially to avoid proplems when the battery started draining. I expect them to go pretty fast since i expect the fridge to be quite power demanding.
I'll take a look into that LM5022, sounds like a good investment for this project.
Having seperate SMPS's for different circuitry in the project does involve some extra cost (3 pwm controllers) with motors connected directly into the batteries but i guess it does have its merits, if 1 fails the others are good.

My microprocessor is 3.3V, while other logic circuitry is 5V. The fridge and motors should be 12V. So i'm still tempted to build a flyback/half bridge multioutput smps but that will involve some extra design time compared to two simple bucks for 3.3V and 5V. Then LM5022 for the fridge.
 

THE_RB

Joined Feb 11, 2008
5,438
There is a little buck regulator here that will convert 12v to regulated 5v at around 80% efficiency (and uses just 2 transistors and an off the shelf inductor);
http://www.romanblack.com/smps/a00.htm

Now with the fridge, why would it need tons of power?? Those little chest fridges used for solar camping use about 2 amps when running and might only run 1/3 of the time. If you fill it with cold beers and it is well insulated it might use less than that.

If the goal is energy efficiency as you stated then the PSU losses are not the place to start... ;)
 

dummhuman

Joined May 11, 2009
6
For the sake of simplicity, it would likely be easiest to make a boost SMPS for the fridge and motors, and a buck SMPS for the +5V, then use an LDO linear regulator (powered by the +5V), for your microcontroller, as the power dissipated by the linear regulator will be minimal, and it will help get rid of the ripple in your uC supply. You will have to consider what effect ripple will have on your 5V logic supply as well, and it might actually be best to use linear regulators for both the 5V and 3.3V supplies, even though it will be slightly less efficient; because of the lower voltage ripple. Best of luck to you, and please keep us posted on your progress and results, I am quite interested to see how your project turns out.

Best Regards,
Nick
 
Top