Simple and efficient 12V->5V convertor?

by **RoganDawes** » Tue Jan 15, 2013 10:15 am

Hi folks,

I am trying to connect my Texecom alarm panel to the local network (and Internet eventually), so that I can monitor it remotely.

I have a USB-Com device (basically an FTDI 232R USB-serial convertor) that I plan to connect to a TPLink WR703N router running OpenWRT and tcpser to make the port available over the network.

My problem is with sourcing power to run the bits. The panel will provide 12V@200mA to power a "ComIP" device, and also uses this to provide power to the USB-Com (as opposed to taking power from the USB). This should be adequate for the WR703N, which is reputed to take around 180mA at boot (5V), and around 80-90mA in operation. My initial thought was that the USB-Com was converting 12V to 5V for the FTDI convertor, and I could simply use that to power the WR703N. Unfortunately, the convertor used is limited to 100mA, and the router simply power cycles endlessly.

My current thought is that I should find an alternative convertor to provide 5V at a slightly higher current rating. I've been looking at a bunch of TI chips, like the LM22674. Unfortunately, these seem to require quite a lot of surrounding components, which makes it more difficult to use. Surely there must be simple and efficient 12->5V convertor chips that do not need a lot of surrounding infrastructure to configure them?

Any suggestions?

Rogan

by **matseng** » Tue Jan 15, 2013 10:31 am

The easiest and also cheapest way would be to just get a dc-dc converter from ebay.

You can get a nice small module that eats voltages between 7 and 35 volts and can be adjusted for 5 volts up to 3 amps output for $1.50 including shipping. For instance from China http://r.ebay.com/8Dfqom

If you're located in US and want an US seller you have to pay a bit more than that of course ... about $9-$10 inclusive shipping. But it still seems like a decent price. http://r.ebay.com/y3nQOW

Just search for BUCK CONVERTER pick one...

by **hlipka** » Tue Jan 15, 2013 10:34 am

You can look e.g. at TIs LMZ12001 which is a complete SMPS on a chip and needs just some resistors and capacitors. But even simpler: why not use a simple 7805? The FT232R draws about 15mA, so the losses in the regulator are negligible...

by **matseng** » Tue Jan 15, 2013 10:41 am

It's the WR703N that will be powered, not the FT232....

by **systemstech** » Tue Jan 15, 2013 12:18 pm

@RoganDawes
Please be careful as modifying any fire or intrusion panel may void your insurance policy. I work with commercial panels like, Simplex, Fire lite and Notifier if any modifications are carried out they must comply with the UL listing for your unit. In other words What ever you attach to the panel must be made for and listed by UL for your specific unit. This is especially true if you have a monitoring service. Also the work done on any monitored panel must be done by certified personal.

by **RoganDawes** » Tue Jan 15, 2013 2:11 pm

Thanks all for the suggestions and warnings. Much appreciated!

by **baoshi** » Wed Jan 23, 2013 8:59 am

Hi @RoganDawes,

I suggest you stay away from LM2596 basesd eBay modules. Most probably these modules are from China, using counterfeit LM2596 chip. I bought some before, their ripple voltage can be as high as 200mV under load, and you can tell from the poor quality of the markings on the chip.
You may try search "KIM055L" or "KIM3R35" on eBay. Those are fixed 5V module salvaged from some Japanese scrap. It is based on TI TPS40057, which is far more advanced and most importantly harder to counterfeit.

by **matseng** » Wed Jan 23, 2013 9:11 am

Hmm? So you are telling us that the Chinese do have chip factories making counterfit ic's that actually work as the original but with slightly worse specs? Counterfit and badly remarked chips usually don't work at all.

I'd rather guess that they use low quality high-esr capacitors. And possibly an underdimensioned coil...

by **baoshi** » Wed Jan 23, 2013 11:33 am

matseng wrote:Hmm? So you are telling us that the Chinese do have chip factories making counterfit ic's that actually work as the original but with slightly worse specs? Counterfit and badly remarked chips usually don't work at all.

I'd rather guess that they use low quality high-esr capacitors. And possibly an underdimensioned coil...

There are no longer technological barriers to duplicate chips, especially "jellybean" chips like regulators or opamps. I had encountered many XX2596, XX2577, XX1117 so called "compatible chips". Better ones are sold normally in domestic market. But some sub-standard ones are remarked with reputable brands and dumped into gray market.
There also exist chip "refurbish" lines, where they salvage chips from electronic wastes, do silicon level repackaging and sold as "new samples" or even sold by reels.
The situation for passive components are even worse.
So I always avoid buying lower end components from Chinese vendors. Large scale ICs like MCU, FPGA are still ok.

by **matseng** » Wed Jan 23, 2013 12:47 pm

baoshi wrote:There also exist chip "refurbish" lines, where they salvage chips from electronic wastes, do silicon level repackaging and sold as "new samples" or even sold by reels.
The situation for passive components are even worse.

I've seen these claims from a lot of people on forums, but I've never seen any real news reports/articles in magazines with any kind of proof that this actually happens.

Sure, there are a lot of crap being sold, but that should be on a design/board level rather than component level. By using parts with lower specs they can shave off a few cents here and there and raise their profits.

I can't see factory rejects of cheap active parts be a huge problem either since, from what I've read, most testing and rejecting is done even before the individual units even are cut from the wafer. Cutting, placing, bonding and sealing a chip before testing is a waste of time & resources

I have a hard time believing that even with the (rather) low salaries in China it can be profitable to desolder, clean up, do silicon level repackaging and bonding, marking and re-selling a chip valued at $0.25 or so.

And making a profit of desoldeing an passive part with cut and soldered/bent legs from a junk pcb and then make it look brand new when a new part is worth a few cents, or less, is even more unbelievable.

It's a completely another ballgame to take a factory new $0.10 chip and grind off the laser markings on top and then laser engrave another more expensive part number on it. This is relatively easy to do and can bring in some good profits for a while until the buyers starts to complain.

by **RoganDawes** » Wed Jan 23, 2013 2:29 pm

Thanks for the suggestions.

by **arakis** » Thu Jan 24, 2013 4:02 am

perhaps an heatsinked linear LM317 is the simplest solution.. at 200mA and a 7v drop it will dissipate 1.4W at the most.. even the clip on heatsink should be able to handle that...

by **RoganDawes** » Thu Jan 24, 2013 2:47 pm

I'm trying to avoid as much wasted energy as possible, since the source can only provide a limited current. Hence some sort of switched supply seems best. Thanks for the input, though.

by **fcobcn** » Sun Feb 10, 2013 1:05 pm

I'm using a TL2575 from ti, wich has a max efficiency of 88%, 1A, works quite well, previously were using a 7805 wich needed heat sinking.
There's apparently no heat coming out from the TL2575 circuit.

by **SA_Penguin** » Mon Feb 11, 2013 10:25 pm

For those wondering- that could be a KTT (TO-263) package.
If it has a hole in the top for bolting on a heatsink (hard to tell for sure) it's a KV (TO-220) package.
Details are at: http://www.ti.com/product/tl2575-05

If you go by the PDF data sheet, the coil is 330uH, the input cap is 100uF, and the output cap is 330uF.
There's a schottky diode between output and ground - and that's the whole circuit. Pretty neat!