SparkFun Forums 

Where electronics enthusiasts find answers.

Questions relating to designing PCBs
By Bee
#109312
Hi,

I need to design a power supply for an Xbee device. The device is powered from 24v and I must say I am a bit lost selecting the right components. My device is just a Xbee device with a RS232 connection. Because my application is industrial I want to use a fanless enclosure which is aluminium.
Here is where I am now:
First I tested using a linear power supply with the LM317. I worked, but the heat generated was excessive and I don't have room for a large heatsink (and I don't have a fan so the heat can't really go anywhere).
Then I wanted to use a switching power supply (based on the mc34063), but read in the Xbee documentation that when using a regulated power supply the switching frequency must be at leat 500kHz and the ripple must be no more than 250mV. However, the components that match these criteria are all tssop or smaller packages and are normally regarded to be rather advanced...maybe not the best choice for my first real-life project.
On the other hand, using a good switching power supply will probably generate the least amount of heat because they are very efficient.
My in-between solution is to use a simple mc34063-based switching power supply, step down the voltage to around 4.5v and use a LD1117V33 (a low dropout linear regulator) to get to a stable 3.3v to drive my circuit and xbee. So basically I would be mixing high and low efficiency and the resulting efficiency will probably be rather good, while using simple and cheap components.

My question: does anybody have experience designing such a power supply? Should I invest more time in the switching power supply or is my mixed solution ok? Maybe the mixed solution still generates too much heat (I use the XBee Pro module and my total circuit can use 500mA), but I haven't been able to test it yet.

I'm very interested in what you think. This is my first project, so please be gentle with me :)
By allanw
#109323
It's easiest to use a premade switched mode power supply module. Look under DC DC Non-Isolated - Point of Load (POL) on Digikey.
By waltr
#109325
The mixed solution does work well. Using a high efficiency switcher to drop the 24 to ~4.5 then a linear for the finial regulation gets you some of both worlds, good conversion efficiency and good low ripple regulation. Just be sure to add a filter after the switcher to kill the switcher frequency.

We are doing that on the latest project where I work. However we are using high frequency buck converters (~2MHz). These are all rather fine pitch devices from Linear Tech. The high frequency does keep the induction sizes smallish and the high switching frequency is easier to filter.
By 48X24X48X
#109327
Hi,

You can use LT3502A from LT. It is quite suitable for 24 V industrial application. It comes with a MSOP-10 package, it can still be soldered with hand without any problem. And it switches at 2.2 MHz, which means you can use smaller inductors and probably cheaper. I have been using LTC parts for years now and they have been my favorite for power supply section besides TI.
By Bee
#109392
Thanks for the replies.
I have looked at the premade switching power supply modules but it seems the reasonably priced ones aren't 500+KHz, but mainly 50KHz to 250KHz. The LT3502A from LT looks good but costs around 5-6 bucks. I was thinking I could make a switching 24V - 3.3V DC-DC converter for at most 6 dollars....maybe I am wrong...but I like a challenge.
I tested the 4.5V to 3.3V linear regulation some more and found my heatsink became too hot to touch after a few minutes. I can touch the heatsink for around three seconds before I have to let go. After half an hour the heatsink became so hot that I shut the whole thing off. Not very precise engineering but it tells me even the drop of 1.2V is too much :wink: . My heatsink has about the same size as the Sparkfun PRT-00121 and my enclosure doesn't have room for a larger heatsink.
Anyway, this got me rethinking (i.e. aborting) the mixed solution: my project must be ON 24/7 and it should run as cool as possible (within the budget) so I will spend some more time on the switching power supply. I bought a few LM2964 high frequence buck converters (1MHz, cost around 3 dollar) and I am calculating and selecting the inductor, the capacitors and the resistors (using the example in the datasheet :-) ).

Update: I used a LD1117 linear voltage regulator for my test. That component has a 1.2V dropout. Some additional research brought me to the TLV2217-33KC which has a dropout voltage of 0.5V and that might produce a reasonable amount of heat if I drop the switching regulator to 3.8V.
By thibault
#162991
Hi,

I know you did this project 3 years ago but do you remember which solution appears to be the best at the end of the day?
I am facing the same problem where I have to convert 24 to 3.3 for an Xbee module.
Regards,

Thibault
By skimask
#162997
thibault wrote:Hi,

I know you did this project 3 years ago but do you remember which solution appears to be the best at the end of the day?
I am facing the same problem where I have to convert 24 to 3.3 for an Xbee module.
Regards,

Thibault
What have you tried thus far?
What ideas do you have?
Do you want somebody to drop a circuit in your lap or are you going to build it?
By thibault
#163023
Hi,
First of all it is my first electronics project in a company. It would have been good if Xbee-RS485 converter was still on sale ( https://www.sparkfun.com/products/retired/9083 )...
I am developing a cheap interface for converting 24VDC to 3.3VDC in order to use the Xbee and I also need to use a RS485 converter such as the MAX485 or MAX481 or RS485SOIC. I still didn't try anything I am now checking several solutions and put a cost on each one of them in order to find out the most cost efficient solution.

I am wondering about which solution is the best, using a MicroPowerDirect DC/DC converter which will provide a good efficiency and so will not heat a lot. Because I don't obviously need isolated components so I found other one: MCP16301 it comes with a schematic of the surrounding components in the datasheet and it is cheaper.
I could also use a linear regulator. I found cheap ones: the UA78M33CK and LM2937IMP-33. Both solutions could be good for my project.

Obviously the smaller is the part the better it is for me.
I think I will choose the MCP16301. I will of course add some parts to protect the converter. The company Droids ( http://droids.it), which design the Xbee interface cards uses the LM2937IMP-33 part which is a LDO regulator. I could also use it according to the datasheet of this part and the graph I put in attachment
The schematic I will use is in attachment.

Did someone already use one of these parts, I don't need a schematic just to know that it is possible.
Thanks,
You do not have the required permissions to view the files attached to this post.
By UNTEngineer
#163082
First off, I wonder if you are even allowed to discuss your electronics project with us, since this is for a company project. Id check with your direct supervisor before you accidentally release proprietary info.

Second, I worked on a project some months ago that dealt with switching from 16VAC to 5VDC. In my case I used a full bridge rectifier (which u dont need), and a 16v to 5v switching regulator. Those are high efficiency, low heat and low footprint regulators. I'd suggest looking at Recom DC/DC converters.
By thibault
#163204
Thank you for your advices. I will have a look at RECOM DC/DC converters. Regards