24VAC Power Supply options

Hi all.

I'm looking to replace a 24VAC power supply. Currently, the board takes 24VAC with a center tap. Piece of cake taking the + and - sides, rectifying them, sending them to 7812s, 7805s and 7912s, and away we go. I need to convert to one that does not use a center tap. I've eliminated the negative rail needs of the board, so now just need

+12/+5 VDC outputs. The current requirements are ~ 1.5A on the 12V line and 500mA on the +5.

I've used the national semi webbench stuff and it puts a nice little switching supply together for me. The problem is, it sticks a big ol' inductor on there. This thing is a beast (like 1.5" by 0.75" by

0.75"). I can find a horizontal mount one, but it takes up a ton of board real estate.

The linear regulator design gets to the Vin limits of a typical 7812. Max input voltage is 35VDC, and the full wave rectified voltage is about 36VDC (the actual transformer voltage varies between 19 and

28VAC.) Needless to say, that regulator gets too darn hot and eventually blows.

I'm just a lowly firmware guy, so any hardware types have any answers to a smaller component version of that power supply? I'd really like something that didn't take up so much room and have such a high profile.

Reply to
Dave
Loading thread data ...

I don't have any specific recommendations, but I think an LDO (the 7812) would be a problem. Dropping 20 volts at 1.5 amps gives you 30 Watts! That is what an older Intel CPU dissipates and they use a fan on the heat sink!

A switcher is what you want and they should not be hard to come by. There are lots of newer inductors around that are smaller and better. But the key to getting the inductor and capacitors smaller is to up the switching frequency. I bet the designs you came across are running below 200 kHz. Look for one that switches closer to 500 kHz or even 1 MHz. Try the TI site, I know they have some nice ones, but all the ones I have looked at only work up to 6 volts in. TI also bought PowerOne or someone similar who makes PS modules! No real design work, just pick the input and output voltages and current and you are done! *That* sounds like the way a firmware guy should go... ;)

--

Rick "rickman" Collins

rick.collins@XYarius.com
Ignore the reply address. To email me use the above address with the XY
removed.

Arius - A Signal Processing Solutions Company
Specializing in DSP and FPGA design      URL http://www.arius.com
4 King Ave                               301-682-7772 Voice
Frederick, MD 21701-3110                 301-682-7666 FAX
Reply to
rickman

Yikes! Typical linear regulators a terrible for big step downs in voltage, especially at any significant current. The problem is basic physics, you are stepping 36VDC down to 12V. That means the regulator is dropping 24volts at 1.5 amps, or about 36 Watts! Now go look at the watt rating on your soldering iron! :)

In a situation like this, you definitly need a switching regulator. I have seen some switcher supply bricks with most of the passive components built in. I think Lucent was one of the companies making these.

Try ST and Maxim (amongst many others), they have all kinds switching supply ICs. I have used the L5973D from ST with good results, it has settable output from 5 to 35 Volts and comes in an SO-8 package. The physical inductor size that we use is much smaller that what you are describing as well. Good luck

-J

Reply to
Mood

look at power trends. They have pincompatible switcher replacement for the old 78xx and 79xx linear regulators.

cheers

Reply to
Felix Bonjour

You didn't say explicitly: does the new one *have* to be a 24V one? Because those 24 Volts, or rather the rather large distance between that and the actual output voltages you need, are really the root of your problem.

OK, so that's 20 Watts DC output.

If you really think 0.85 cubic inches for a part that has to handle 20 Watts of electrical power qualifies as a "beast", maybe you'll have to re-check some physics. So let's conduct a little survey: go around your house and look --- is there *any* AC->DC power supply (wall wart or laptop power converter, e.g.) that comes anywhere near as small as

1 cubic inch in size? I doubt you'll find any.

Your best bet by far may be to get away from those 24V input voltage of that power converter. You'ld be a whole lot better off with 15VAC for input, e.g. Even investing in a small-ish transformer to make 12 VAC from your 24 would already simplify things a lot, and allow you to stick with your simplistic design using linear regs.

--
Hans-Bernhard Broeker (broeker@physik.rwth-aachen.de)
Even if all the snow were burnt, ashes would remain.
Reply to
Hans-Bernhard Broeker

Thanks Rick. Could the TI site by better for a firmware guy or what? Much nicer than National's. I did find another solution there that kept the size down, but added a bunch more components (guess that's probably the tradeoff.) But, their software gives you a complete BOM with part numbers and everything! Too bad they don't have a firmware site like that. Thanks for the pointer.

-- Dave

Reply to
Dave

That's the name, Power Trends. Aren't they the company that TI bought?

--

Rick "rickman" Collins

rick.collins@XYarius.com
Ignore the reply address. To email me use the above address with the XY
removed.

Arius - A Signal Processing Solutions Company
Specializing in DSP and FPGA design      URL http://www.arius.com
4 King Ave                               301-682-7772 Voice
Frederick, MD 21701-3110                 301-682-7666 FAX
Reply to
rickman

To use smaller inductors, you need a higher switching frequency. National has a newer line of their Simple Switcher chips, but when I last checked Webbench didn't support them. That was quite a while back, so it seems likely that they would support them now.

Lately I've been using Linear Tech switchers. They make a wide variety of high frequency ones, and they have free downloadable design software that's fairly good. The name is LTCad if I recall correctly. It will come up with designs for you, and includes a specially customized Spice simulator that you can use to tweak the designs if they aren't exactly what you want.

I was particularly impressed that they actually officially support running their software on Linux using Wine.

Reply to
Eric Smith

Most of the LT parts are pretty good, but most of their switchers are not synchronous and require the use of a fat Schottky diode. Many also require a diode for the bias voltage boost circuit. Adding a couple of diodes starts to eat up real estate that other chips don't need. Otherwise, LT makes very good opamps. :)

--

Rick "rickman" Collins

rick.collins@XYarius.com
Ignore the reply address. To email me use the above address with the XY
removed.

Arius - A Signal Processing Solutions Company
Specializing in DSP and FPGA design      URL http://www.arius.com
4 King Ave                               301-682-7772 Voice
Frederick, MD 21701-3110                 301-682-7666 FAX
Reply to
rickman

Thanks to everyone that responded. To summarize:

- Yup, I need 24VAC. That's what is present on the incoming power supplies I need to interface with.

- TI did buy power trends, and the software that TI puts up for power supply design is nice. Going with a high switching frequency makes the inductor smaller, but also puts many more parts on the board. Price-wise, it's actually better to go with the national 2576 switcher because it can actually support > 48VDC (which is a bonus for me).

- I checked on the LT parts also, but again, National's solution had fewer parts. Cost-wise, it works better.

- I still need to check through for those other higher frequency parts that aren't supported in WebBench.

You guys rock. I have the utmost respect for hardware folks.

-- Dave

Reply to
Dave

There are two things I don't understand. If you checked out the Power Trends stuff, how is that not the simplest solution? They are modules that need no external components, AFAIK. They are not chip level devices that need coils and caps.

The other is how are the higher freq parts more complicated? Are you talking about the compensation components? Counting components may have a value, but resistors and small caps don't cost much at all. A reel of

5000 resistors costs maybe $15. Caps are the same until you get above 0.1 uF or so. The input and output caps need to be larger and start costing some money (still under a buck), but the higher freqs allow these to be smaller and cheaper. Normally the chip is the most expensive part regardless.

Am I missing something? Are the slow, older converters just a lot cheaper than the new high freq parts?

--

Rick "rickman" Collins

rick.collins@XYarius.com
Ignore the reply address. To email me use the above address with the XY
removed.

Arius - A Signal Processing Solutions Company
Specializing in DSP and FPGA design      URL http://www.arius.com
4 King Ave                               301-682-7772 Voice
Frederick, MD 21701-3110                 301-682-7666 FAX
Reply to
rickman

Sorry - The power trends modules are also completely expensive (~30-60 bucks.) The entire board costs about 60 bucks before this part. A non-starter...

The National semi - low freq parts solution (~150kHz) was something like 4-5 components on the order of about 5 bucks total. Picking the same input/output parameters from the TI site put together about 25 components (my "complexity" argument.) I didn't even bother pricing them out.

I can handle the big inductor on the low freq part. It just looked weird on the board. Simple/cheap/but big. The board's all through hole anyhow (old equipment) so no biggie.

For other product, I'll probably use the TI solution (but will also look into National because someone mentioned some higher frequency parts that WebBench didn't know about.) These other products will need the smaller footprints because of more dimensional constraints.

I guess there is also the EMI problem you can create for yourself using the low frequency parts. Something else for me to think about.

Thanks again.

-- Dave

Reply to
Dave

Can you give me the part no. I thought that there were simple higher Vin regulators that would do what you want with the standard arrangement of components. 25 sounds very excessive unless this is a controller and you need external FETs and diodes, etc.

I am not a power supply engineer, but I have had to do a couple for myself. I don't really know much about EMI, but I would expect the higher freqs and smaller inductor would generate less EMI. But then again, the higher freq will be harder to deal with. So it may be a trade off.

--

Rick "rickman" Collins

rick.collins@XYarius.com
Ignore the reply address. To email me use the above address with the XY
removed.

Arius - A Signal Processing Solutions Company
Specializing in DSP and FPGA design      URL http://www.arius.com
4 King Ave                               301-682-7772 Voice
Frederick, MD 21701-3110                 301-682-7666 FAX
Reply to
rickman

I didn't save the design. I just downloaded the TI tool, input my parameters, and out popped a schematic and BOM. It did have external fets, diodes, etc. I'd do it again and report the part to you, but it's on my machine @ work.

Dunno. Guess I'll find out during UL and FCC testing. Gulp.

-- Dave

Reply to
Dave

the

Yes TI was hungry, PowerTrends was one of thouse that got eaten...... On the TI homepage I din't found those nice things again.

Reply to
News

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.