Voltage Regulation... seems simple

CharlieII

I have asked related questions over the course of the past couple years now trying to accomplish the same pathetic task in a multitude of ways, working off and on on this project, and I still have nothing to show for my time, and I still can't do something as simple as regulate voltage, or even understand how to do such a thing.

I literally have automotive nominal voltage of "12vdc" to work with. So figure between 8 and 14 volts depending on conditions.

I want to run servos that can handle 6 to 8vdc.

Seems simple enough right? Well I apparently cannot accomplish that task....


I can snag a simple linear reg and promptly turn it into smoke as soon as any real load is applied to the servo though. I'm good at that.



I want to know how to build a voltage regulator/converter that can produce various voltages (lower than input) whenever I please, at whatever voltage I please and with some USABLE amperage. Not weak 1A or 2A nonsense.

I'm talking 10, 20, 30+ amps if I wish.

How is this done?!?!


I need most common/simplest as I'm sure with anything semiconductor there are roughly 20 billion ways, types and methods of doing the same thing.



I realize this sounds impatient, but I honestly think that I might start smacking myself in the head with a 2x4 every half hour or so until I figure this out because I'm truly finding my efforts pathetic at this point.

I can't get from 12v to say 6v with any decent amperage capacity to save my life. It's pathetic IMO.


It has COMPLETELY stopped my project. In fact, this is the ONLY thing between my project and the finish line, just as it was about 2 years ago when I first ran up against this.


Any help is appreciated. I might lose it if I can't figure this out....

Ridiculous if you ask me.

Thanks.

CharlieII

Actually, I searched my own username and came up with my last thread...

http://forums.parallax.com/forums/default.aspx?f=15&m=373644



My first post there summarizes my needs in a much more patient and graceful way.

I never managed to put together a switching reg because you can't just buy the thing and stick it on a board. Inductors, diodes, yada, yada, yada....

And there's no list of components to say what's needed to complete one that I've found.

I did make some good smoke though. Getting better and better at that as I push harder and harder to get some simple voltage regulation.


Thanks.

PJAllen

10, 20, 30 Amps is not "simple."· That's simply not linear, step-down territory.

Leon

Look at the Nat Semi Simple Switchers. Designing with them is quite easy but you need to know what you are doing, and you must design the PCB properly.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Leon Heller
Amateur radio callsign: G1HSM

Erik Friesen

A good 5-6v reg with the biggest heatsink you can fit will be the simplest way to get what you want. Or, you can do a series of stepdowns, like 14-10-6, thereby spreading out the dissipated heat. 10-20-30 amps? Probably not. Enough to drive your servos? Probably if you keep it under 1-2 amps.

Do the math. 14-6=8 . 8*10 = 80watts. For 6 volts through a linear regulator at 10 amps you have to dissipate 80 watts of power. You probably won't find a heatsink that can handle that.

Peter Jakacki

Linear Technology are always the people to check for the really good stuff. They have triple output switches using external mosfets that go up to 60A. Here is a pic of a circuit that you can set for whatever voltage and up to 25A at high efficiency, no smoke.

There are many devices to choose from, here's one link:
www.linear.com/pc/productDetail.jsp?navId=H0,C1,C1032,P87747

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
*Peter*

Franklin

www.google.com/search?q=buck+regulator+wiki&rls=com.microsoft:en-us&ie=UTF-8&oe=UTF-8&startIndex=&startPage=1&rlz=

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Stephen

davejames

Charlie,

Providing the amount of current you've mentioned can be a formidable task, as you've learned by becoming proficient in smoking devices.

Reducing 14VDC to 6V (say) at 20A calls for a power dissapation of (14 - 6) * 20 => 8 * 20 = 160W!!!

That ain't trivial my friend.

The main thing you have to deal with is that power dissapation, and the only way to do so (yes, if working with linear regulators not switchers) is to provide enough heat sinking material in which to dump all that power.· And *that* translates to one big chunk 'o metal (aka "heat sink").

There are formulas that assist in deciding how large a heat sink is required.

A book that has been invaluable to me is the National Semiconductor Volatge Regulator Handbook.· And the last time I checked, it's in download form on the NatSemi website.· Get it.

As you've found, single chip linear regulators are not sized for the amount of current you wish.· That's when the design calls for a big, beefy pass element.· I've included·a pic of the simple concept using the venerable 2n3055 transistor; I'm sure there are better devices available, but you'll get the idea.

The simple voltage regulator provides the reference voltage for the 3055, and the 3055 handles the current - in very simple terms.· Granted, the Vout will be a diode junction less in voltage but there are ways around that.· Also there are a number of protection schemes that can be done including the lowly fuse.

Anyway - download the book, it'll help immensely.

Regards,

DJ

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔

Peter Jakacki

Charlie,
I'm not quite sure what the big problem with yada yada yada is. You just get the components just like the recipe for a cake. Yes there are recommendations but of course you can substitute. Never never consider a linear regulator for this job yada yada yada. Your biggest problem is that you keep wanting to do it the "easy" way but end up causing yourself grief and frustration over these years and never doing it at all. Go to a little more trouble for a lot less trouble in the long run, you will wonder why you procrastinated in the first place.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
*Peter*

CharlieII

Peter, you're right. Seeing that I have 12vdc and I only need 6vdc I did see this as a rather simple task.

When I think about the components on my truck consuming and handling on the order of 100, 200, 400, 500, 600+ amps it makes it all the more pathetic in my mind that I can't even drive a couple servos pulling 3A each on a bad day.

Voltage regulation in my mind seems akin to pressure regulation with most any other medium. So why is it so brutally easy to regulate air pressure, oil pressure or water pressure with a simple regulator and still have plenty of flow (amperage) yet with electrons it becomes such a PITA?

It's not like I'm trying to produce more than I started with. I just don't want all of what I already have.

I feel like I have a 20 dollar bill in my hand, yet I can't even buy a pack of gum because I don't know what to do with all the change I'll get back.


It's incredibly frustrating. But I realize the frustration is in the fact that I don't understand this, which is frustrating to anyone really.


I decided last night before I made my initial post that I was going to do nothing else until I had built a 6 volt regulator that could sustain 10amps output INDEFINITELY when run from a 13 volt source. I figure if I can manage that, then I'll have this thing licked.

I appreciate the hell out of the help. I'm going to look over those suggested schematics, probably build one or more of them verbatim, and I'm going to print off that handbook that was referenced and stick it in my bathroom..... where I know it will get read...

Now back to the part that makes things like this take soooo long, and kills motivation...

Waiting for parts to arrive just to realize in 3 seconds why the package won't work, so on and so forth.

CharlieII

Also.... there are about 14 or so external components on the example that Peter gave for instance. While it may seem trivial to you all, I can't make it clear enough how many times I've spent weeks just trying to assemble that stuff for one of these little "switchers" and come up empty because there seem to be a zillion different types of each one of those 14 things to choose from. Different ways of attaching them. Often times precluding the use of a breadboard to mock up the circuit as some of the components are simply not offered in through-hole packages....

So on and so forth.


I know this may be asking a lot here, and if so, I understand, but can anyone just give me a part number list that I can just type in to newark, or digikey or similar and then when it comes in I can just breadboard it?

Cause I've got a whole drawer full of useless brand new inductors, diodes, capacitors, so on and so forth that were wrong for one reason or another, even though the spec listed on the little schematic was correct.

For instance, when you look up capacitor.... well, what kind? How many of those things are there? I can't recall right now, but electrolytic.... some other kind, axial, yada, yada, yada.

Yet every one of those thousand combinations might ALL have the exact same uF specification.... which is all the little schematic tells you about.



If that sounds pathetic to you, then we agree. Which defines my level of frustration here. Having no formal background in this area I find mundane tasks rather complex. And some complex tasks come very easy.


Anybody list off a part number "cake recipe" for me? Cause that's something I could put together. I think if I could just see one work I would be good to go.

Thanks.

Peter Jakacki

Ok, main parts are:
search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=LTC3878EGN%23PBF-ND
search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=RJK0305DPB-00%23J0CT-ND
search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=513-1357-1-ND
search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=478-3345-1-ND
search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=565-3190-1-ND
search.digikey.com/scripts/DkSearch/dksus.dll?vendor=0&keywords=CMDSH-3
search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=495-2240-1-ND

The other Rs and Cs are non-critical. This search and compare specs only took around 15mins or so.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
*Peter*

TonyWaite

@CharlieII

When I've reached the 'banging head against the wall' stage and have exhausted time and money, I tend to
reach for an off-the-shelf module instead: this is always cheaper for prototyping, and is one more problem
out of the way so that you can progress your project.

This is especially important with power supplies, so that you can get to the starting-block!

Regards,

T o n y

Peter Jakacki

If you want OTS solution then try this tiny module, 6V @20A 91% efficient for around $25
parts.digikey.com/1/parts/1607158-convert-dc-dc-0-59-6v-20a-sip-nsr020a0x43z.html
au.mouser.com/ProductDetail/Lineage-Power/NSR020A0X43Z/?qs=DWPgn%2fG9gkxYgP6PIW5L%2fA%3d%3d

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
*Peter*

CharlieII

I think I'll take Tony's advice and use the little IC that Peter linked for 20 some odd bucks...

That's PERFECT!


Plug.... and play....


Thank you man. I'll mess with building specific power supplied later I guess. That will get this project DONE!


Did I mention thanks???





On Edit:

And I'm also copying that parts list as well. Bare in mind that while it only took you 15 minutes (which I appreciate tremendously) it would take me hours because I DON'T KNOW what the hell this stuff is. Little mundane things like package type, or any other of the usual list of 10 to 15 some odd parameters to choose from when filtering that I haven't the foggiest idea how to choose from destroys my ability to weed through the list. And often times choosing "13v" instead of for instance "8 - 14v" can make 99% of the options disappear and I don't catch it. It's linguistics to be blunt. I don't have a functional vocabulary.

For instance, I can go use google to search for baseball bats. Maybe search for "32 oz 33 inch easton bat". And that's what you can easily do. And something as simple as a baseball bat we take it for granted. However, imagine if when I searched in google, my knowledge base only afforded me a search string more akin to something like.....

"32 oz 33 inch Metal object, conical in shape used to strike spherical object"

If you don't know the terminology you can't search for something worth a damn. And the same search now takes forever as you have to just happen across the terminology as you go.

Post Edited (CharlieII) : 3/18/2010 3:51:06 PM GMT

kwinn

Here is an idea to reduce power dissipation in a simple linear regulator that has to cover a wide voltage range or output high current.

The center tap transformer/bridge rectifier/diode splits the voltage so Q2 has only half the total voltage on the collector.

Q1 has the full voltage on it's collector.

When Vr is below half the total Q1 puts a small drive current to the base of Q2, and Q2 provides the output current.

Once Vr rises above the half way point the Q2 base emitter junction acts as a diode and Q1 provides the total current.

Since neither transistor ever has to drop more than half the total voltage at full current power dissipation is reduced significantly.

*** This is not a complete power supply circuit, only a starting point.

Peter Jakacki

This assumes that transformer supply maintains it's voltage but this is not the case and a half-wave circuit will have even less regulation. Note too that the transformer would be a real transformer, not an ideal circuit and complicated somewhat by the imbalance plus half-wave as well. However, I do not believe a transformer is involved as we are talking about automotive voltages and very high currents so it sounds like it is running from a lead-acid battery.

Nonetheless, linear regulators are not a logical or good choice for high-current or efficiency. The switching module just works, without needing a steel shed to heatsink it and drawing just over half the current from the input supply (12V @ ~11A -> 6V @20A)

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
*Peter*

kwinn

@Peter Jakacki

Regarding linear regulators not being a logical or good choice.... couldn't agree more. Only reason I posted this was because he was looking for something simple. If he is using a battery this circuit is not much use to him.

There is no half wave rectification involved in this circuit. Because the outer connections go through a bridge rectifier and the negative output is common (or ground, which I forgot to show) the center tap will always be positive going with respect to the negative from the bridge. The purpose of the diode is to block the capacitor from discharging back through the transformer.

It definitely is not an ideal circuit. Even with a quad op amp, feedback and circuitry to make it as linear as possible it was not very linear at the switchover point. It did however make it possible for us to build a relatively simple power supply that could put out 5A from 0 to 48V using 2N3055's.

Peter KG6LSE

IMPO SMPS is the only way you could make this work ..
i have seen some MicroATX PSUs for car computers that might work
Most have 11-17V in with 5 / 12 /3.3 ect out .
and IF you tweak the setpoint devider you could use the 5V rail at 6V .
Most of thease things handle up to 120W so they might work..

Peter KG6LSE

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"Carpe Ducktum" "seize the tape!!"
peterthethinker.com/tesla/Venom/Venom.html
Never underestimate the bandwidth of a station wagon full of tapes hurtling down the highway. —Tanenbaum, Andrew S.
LOL