Design for a custom board

[glire]

Hi flash folks

I'd like to make an adjustable voltage regulated buck driver.
It should accept 2 li-ions input and have an output between 3.7V to 4.2V, which is like 1 li-ion (say 4.0V).
I'd like constant 2A output, 3A for 10-15s burst, as efficient as possible, and of course small and light.

There are zillion of cheap drivers using LM2596 available everywhere: too big, too heavy.

I found 2 quite small boards that might do the job:
http://www.goodluckbuy.com/mini-siz...8-20v-output-step-down-voltage-regulator.html
http://dx.com/p/dc-4-75-24v-to-dc-0-92-15v-voltage-step-down-module-blue-151065

I didn't purchased them, so I don't know how they perform. But they are probably not as efficient as I would like.

Unfortunatly I'm nowhere about designing PCB, soldering smd, not even purchasing "good" components (the "right" capacitors, the "right" inductors, etc).
I just did some basic circuits using stripboards a long time ago.
So I need help here.

I found some circuits:
ADP2303
AP3503
AP6503
AP65200
AP6507
MP2307
BD9329
MP1584
NCP3155
G5642

Maybe someone could choose the "best" circuit from that list (or another), and desgin an PCB for me.

Many thanks

 

[glire]

Ok, if no one can help me, can I be advised any forum where to get started designing and making PCB, then soldering SMD ?
Thanks.

 

[Yagon]

Without being cheeky or horrible to you , you want :
Someone to choose the components , design the circuit , design the layout and foil pattern , fabricate or get fabricated a PCB .
You see if that was me doing all that work , it would be my design and fabrication and I would sell you a unit .

I would start by reading some electronics books , learning what the various components do , how buck converters work etc.
Then buy some electronics kits , and a good temperature controlled iron , and practice with these until proficient .
Then look at the sample circuits that come with the "interesting" chips , choose one that is closest to your application .
Modify the values of this circuit so the numbers are right , then you can build this up on stripboard and test , take measurements to see if the numbers are with the tolerances of the devices and your design spec .
Then see about drawing up a foil pattern , maybe a freeware cad program , fabricate or get fabricated the PCB .
Build up several of these boards and test again (soak test , overload test , enclosed thermal test etc) , if it is all you wanted great .
Now you can work further on the design adding more features , more power or functionality .

To do this you will need the knowlege , the skill set and the tools (they are far from free) .

I always find it best to climb a ladder one rung at a time .

 

[glire]

I need to get started somewhere, right ?
Everybody did
I already have some knowledge, skills, maybe common sense... from the 80's.
I also have an old analog Hameg.
SMD came up since.

To me, the most difficult part is the dedicated tools needed, and the (very small) size.
An amateur can't make vias, can't solder exposed pads below ICs, can't make 8 layers PCBs.
But I hope there are workarounds.

I can start with a recommended and existing layout.
Take this IC for example (ADP2303), the PDF sheet is really complete.
It even teachs me in two words the benefit of X5R dielectric over Y5V
Well, it doesn't tell though why ceramic and not tantalumcapacitors.
http://www.analog.com/static/imported-files/data_sheets/ADP2302_2303.pdf

That sheet contains it all I think, from pcb layout to recommended parts with vendor ref.
Just need to purchase components and try.

digikey, farnell, rs-components ? Any other ?

Anything more to really get started ? (no books, dedicated electronics forum should help better don't you think ?)

It will take weeks for me.
Maybe it would take an evening or two for an experienced guy. I'd prefer that route if possible

 

[dc38]

I need to get started somewhere, right ?
Everybody did
I already have some knowledge, skills, maybe common sense... from the 80's.
I also have an old analog Hameg.
SMD came up since.

To me, the most difficult part is the dedicated tools needed, and the (very small) size.
An amateur can't make vias, can't solder exposed pads below ICs, can't make 8 layers PCBs.
But I hope there are workarounds.

I can start with a recommended and existing layout.
Take this IC for example (ADP2303), the PDF sheet is really complete.
It even teachs me in two words the benefit of X5R dielectric over Y5V
Well, it doesn't tell though why ceramic and not tantalumcapacitors.
http://www.analog.com/static/imported-files/data_sheets/ADP2302_2303.pdf

That sheet contains it all I think, from pcb layout to recommended parts with vendor ref.
Just need to purchase components and try.

digikey, farnell, rs-components ? Any other ?

Anything more to really get started ? (no books, dedicated electronics forum should help better don't you think ?)

It will take weeks for me.
Maybe it would take an evening or two for an experienced guy. I'd prefer that route if possible

In all fairness, I'm sure that we would all love to help you out. You've even done a little research on parts and such, but I think many of us would prefer if others would fully educate themselves before helping them out...like Yagon mentioned, practice makes perfect. I'm not saying that you are doing the following, but it sounds like you don't want to do the extensive research :/. Good things take good time, patience, and proper education with experience. There are geniuses who can build a great circuit with no experience just by reading a book and doing some research. That does not apply to most of us. People can pick parts for you and inform you how to assemble a circuit. They cannot help you realize the gravity of determination and self motivation. Fully learning the process and theory behind something often provides insight and appreciation into whatever you're doing.

In short, not to be rude, but for safety and educational purposes, I believe that you should do some more research for yourself and bring the question to the table again when you can ask it in a more objective and less personally subjective way.

 

[calipsoii]

Given the information you've provided in your post, my honest suggestion would be for you to research existing drivers on the market and purchase one for several dollars.

 

[glire]

I expected a few links, design tools, vendor names (I provided some, just confirming or warning them already helps), I didn't even got that.
I assume I was too generic and you were afraid
So I'll narrow my request.

I started playing with Eagle CAD. Seems like a 20 years old program that didn't improve ergonomics since.
I found others, but google gives links, not user ratings. So I'd like your advice or experience on pcb design (free) tools.
Thanks.

 

[Chodes]

How many CPF members could give decent advice on driver design?
Most of the benefit of CPF, as I see it, is learning that there a a lot of cheap drivers available.
Lot's of reports on performance testing reveals shortcomings.
Then there are Taskled offerings, good quality, good performance, a lot more money.
Only CPF members I am aware of that have achieved similar quality to Taskled is Der Wichtel, prices are not cheap.
Apologies to anyone I have overlooked.
Good incan drivers / softstarts hyave been made by several members, prices never near the cheap current LED drivers though.

So it seems anyone with the ability, persistence,determination,skill and intelligence to accomplish a good driver design and build will expect monetary reward and not want to share too much and I can appreciate and respect that.

 

[calipsoii]

Chodes basically hit the nail on the head. Goodness knows I'm no expert but I've been mucking with PCB's for a couple years now. My latest driver design barely works, has taken 9 months of evenings/weekends and has cost several thousand dollars to date.

You're being rather flippant/dismissive of the community, glire, and I think that's why you're getting a negative reaction. Everyone here is always willing to help out a member who's undertaking the task of making something new. My gut feeling is that you don't understand the sheer volume of learning and work that is ahead of you (while a lot of others do) hence all the suggestions you're getting to just purchase an existing driver.

If you are still serious about your project, my suggestion to you would be to:
- draw an electrical diagram on paper
- purchase the individual components from DigiKey
- solder them onto some protoboard
- test, test, test, test, test that your design works

Those are the first steps, long before you go into Eagle and have PCB's professionally created. Those first steps will cost $100 dollars. The next steps rapidly get into the thousands and all of your free time.

Good luck!

 

[SemiMan]

Glire,

The simple fact is you are too far from a point where the amount of advice I would need to give you would be anything less than essentially doing all the work.

Best thing is to buy off the shelf. Other option is an evaluation board from a semi vendor but you will find these to be rather expensive.

Hobbies are fun, but remember that people spend their lives doing this as well.

Semiman

 

[hiuintahs]

Semiman pretty much sums it up. Designing switching power supplies isn't for the novice. Board layout is critical. Hand-wired switchers really won't work well if at all. So if you can find a board already made up is your best bet.

National Semiconductor's (now merged with TI) "simple switchers" (LM2596 & a host of others) work pretty good. With the cheap price of those on Ebay, I have to wonder if they are legitimate or fake though. Keep in mind if you're wanting 2 amps, then the inductor and switching FET, whether that is built into the IC or not will necessitate a certain amount of size for power dissipation.

A lot the new switching power supply controller IC's use a package that is getting harder and harder to solder onto a board without specialized equipment. TI makes a lot of switching regulators, especially since they and National Semiconductor are the same company. You could download TI's free "SwitherPro" software tool and it will design it for you. But then you still have to get it built and the layout is critical. Pick one that works and download the data sheet and you will see what I mean as far as info......there is more to them than meets the eye. Here is a few part numbers that might work.........the LM25xx simple switchers, TPS54331, TL4001. This is just a couple that come to mind. I'm mostly familiar with TI. There are a bunch of other companies. I'll be honest the simple switcher is going to be your best bet based on simplicity. So I know that you mentioned it was rather big, but maybe you buy one anyhow as your prototype just to see how it works.

 

[uk_caver]

I started playing with Eagle CAD. Seems like a 20 years old program that didn't improve ergonomics since.
I found others, but google gives links, not user ratings. So I'd like your advice or experience on pcb design (free) tools.
Thanks.

I've only seriously used Eagle.
I did previously try a different free package suggested/provided by an electronics distributor, but it was terrible - bad enough to put me off the idea of using PCBs and keep me making things on stripboard for a while longer.

Eagle isn't perfect, but there are a lot of people using it, and therefore quite a lot of helpful tips around on the internet.

 

[glire]

A zillion thanks to all.

OK, I think I begin to get it. But please, do not blame me if I'm still naive.
Yes, really, I thought it was easy to assemble a chip, an inductor and a few caps and resistors.
Isn't that all needed for a buck voltage regulator ? Probably, certainly.
But, I agree, it might be a bit complex if I want efficiency and reliability on the smallest size.

So, I started with 2 ready to use regulators, LM2596 based: huge, eliminated.

I received from ebay 2 more ready to use regulators, a lot smaller.
1) MP1584 based, 23x17mm, 1.96g
2) MP2307 based, 17x11mm, 1.36g
They are supposed to reach 3A peak, 1.8A continuous. I didn't test them yet.
I hope they're good enough for me to start understanding and play with.

And thank you for pointing me on Texas Instruments, I was missing them.
TPS62130 -> light
TPS54628 -> easy
TPS54821 -> a beast
TPS53319 -> a bigger beast
Integrated switch, LDO, light load efficiency (but maybe the TPS54821), I like those features
But I don't like QFN packages, not sure how an hobbyist could solder them without an oven.
Only the TPS54628 might be OK with a solder iron.

I was almost upset a week ago, now I'm getting fun, even if I spent hours reading and did nothing yet
Thanks again.

 

[hiuintahs]

Ya, you are right. A lot of these IC's have specialized pcb footprints and are hard to next to impossible to solder without the proper tools.

Thus for hobby stuff I limit myself to the SO-8 stuff and parts that I can solder myself. When it comes to a switching regulator, you can either get one with a mosfet built in or an external mosfet. If I'm doing high power then I go for the external mosfet. The MP1584 will work but its still just an SO-8 package and thus can't dissipate a lot of heat. I noticed that it has the solder pad on the underneath so it will do a little more power than a standard SO-8.

Note the 2nd to last page of the data sheet:
http://www.monolithicpower.com/Lists/Product_NonSynchronous Switcher_5727/DispForm.aspx?ID=77

In most cases, they will tell you a suggested layout so as to make it work properly. The problem with these switchers is that the copper traces add unwanted inductance to the circuit. Thus short runs and properly placed components on a pcb is mandatory. That is why I mentioned earlier that you really can't hand wire switchers. You could get it to work that way but it would be noisy.

I'm using a part which looks like the MP1584 with a heat sinking solder pad underneath it. That has to be connected in order to maximize the full capability of the part. I'm using a maximum 1 amp though and its rated at 3 amps. I don't like to use these parts at their maximum power rating, but like to operate below so as to guarantee myself reliability. I will test it with one amp for several hours and see how warm things get.

It sounds like your MP1584 is an already built up power supply in a small board. This is a good start. Test it with a load, to see how warm it gets and if it will work for your needs. You mentioned 2 amps. The MP1584 should work OK. You may need to adjust some things. I did a search on Ebay and didn't see what board you got.

 

[torukmakto4]

I'd like to make an adjustable voltage regulated buck driver.
It should accept 2 li-ions input and have an output between 3.7V to 4.2V, which is like 1 li-ion (say 4.0V).

Did you mean current regulated? Feeding a constant voltage to an LED is not going to work out, so I am assuming you either meant current regulated, or plan to take a common single cell LED driver and feed it the output of your hypothetical buck circuit.

1. Why not just parallel the cells? If this is 2 cells stacked in a flashlight body, I would sooner design a compact parallel holder/contact assembly than a custom DC/DC converter.

2. There are plenty of single-unit flashlight drivers already out there that will take 2 cell input voltage range and drive a typical emitter (3.something Vf) at around 3 amps. I have a cheap chinese budgetlight with such a driver. Check Fasttech and DX and the like.

 

[glire]

Here is a picture of, left MP1584, middle MP2307, right LM2596:

Look for "super mini DC-DC step down" on ebay and you should find them

This is not (yet) for driving LEDs directly (I'm not looking for a current regulator).
It's a challenge where I'd like to "simulate" one Li-Ion cell from 2 cells in series, in other words a ~4.0V buck powered by 2 cells.

 

[hiuintahs]

OK, I think that MP1584 will work. The inductor is kind of small but since its only a 4.7uH, the series resistance should be small. At least the schottky diode is of decent size. Looks like SMA package and is bare minimum for your current application. I wouldn't trust the design to run continuously at 3 amps though. There is a reason the LM2596 looks like it does.

Pin 4 of the MP1584 is the feedback (FB) pin. You'll need to adjust one of the feedback resistors so that the output regulates at 4.0v. Download the pdf datasheet. Page 10 has the equation for determining what value of resistors will give you your desired output voltage. Also comparing the datasheet schematic with what you have, you will be able to see the function of the components associated with this design.

Once the output is set as close to 4.0v as you want, then put some loads on it to see how it works. Will need to get some power resistors to do this.

The board is a voltage regulator. If wanting to convert a buck to a current regulator, you use a very small resistor in series with the LED on the ground side and feed that voltage back to the FB pin. But a better way to do it is to use a very small resistor and an op-amp that gains up the voltage to the necessary level for FB regulation.


Since you have the other boards, you might as well play with them too.

 

[glire]

Thanks.
I looked at the MP2307. It's supposed to be more efficient than the MP1584 which doesn't have a low side FET on the switching output.
It seems recent and more efficient circuits have both high and low side switching FETs, and no need for external diode.
Less external components means less weight and less space, one of the goals

So I tested it with a ~3.4 ohm resistive charge under 4V (about 1.15A) as is, no heatsink.
I left it powered about 30min.
I don't have precise instruments to measure current and efficiency. So I used the "finger" method
The resistive charge was hot and very unconfortable to touch (temp was ~83-84°C, it's actually 2x 6.8 ohm 10W in parallel).
The regulator was barely warm (temp on the small 7x7x3mm inductor was 60.4°C, surprisingly it did not appear hot at all to my finger).
(Temp measured with Turnigy Infrared Thermometer)
I estimate 4.9W input and 4.6W output... +/- 0.1W... or +/-0.2W. Anyway, efficiency probably above 90%.

About PCB makers, I found seeedstudio. Any feedback about them ?