Microcontroller controlled by pushbutton for LED flashlight

Yes, you're on the right track. I would use a small micro such as an ATtiny or a PIC10F200 for an application like this. You could do it with discrete logic but it would be physically quite a lot bigger once you include latch/unlatch and debouncing.

Why do you need two MOSFETs? Does the output need to be push-pull?

How many boards do you plan to make?

Will you hand-assemble or use reflow?

What components have you chosen so far?

Can you tell us more about the power source and the load that needs to be controlled?

 

Right. The PIC10F200 is the same size and a bit cheaper, and is also up to the task.

ATtiny4: http://www.digikey.com/product-detail/en/ATTINY4-TSHR/ATTINY4-TSHRCT-ND/2477315 USD [email protected]
PIC10F200: http://www.digikey.com/product-detail/en/PIC10F200T-E/OT/PIC10F200T-E/OTCT-ND/2178264 USD [email protected]

There are lots of suitable MOSFETs. Here's one: http://www.digikey.com/product-detail/en/DMN1019UFDE-7/DMN1019UFDE-7DICT-ND/3451602 USD [email protected]
That one is in a no-lead package so hand soldering would not be feasible; reflow would be required. It may not be the cheapest suitable part either. I found that part on Digi-Key using a filter for N-channel SMT MOSFETs with logic level gate and R_DSon less than 10 mΩ. Unfortunately, Digi-Key's selection information has only one R_DSon specification per device; mostly, it applies at V_GS=10V, not at lower V_GS voltages like 4.5V or less, where you will be operating the device. So to be sure, you really need to download each data sheet and look at the R_DSon maximum specification for the lowest V_GS voltage you can guarantee, given the battery voltage range.

Edit: Maximum R_DSon is the most important parameter because you need to limit power dissipation when the MOSFET is ON.

P = I² × R and I² = 25 so R = P / 25. Power dissipation for a little package like that should be limited but I'm not sure how much - I haven't worked with such small devices before. Their size implies lower dissipation but the board will provide some heatsinking. My gut feeling is that you should be fine at 0.25W which is why I suggested 10 mΩ R_DSON. If you do some research you may be able to get a real life number.

 

Give some thought to what will turn it off also.

 

Yes that may be OK. 17 mΩ at 5A will dissipate 425 mW. That seems pretty high for such a small package without an intimate bond to the PCB.

Have a look at these documents:
www.analog.com/static/imported-files/tutorials/MT-093.pdf
ww1.microchip.com/downloads/en/AppNotes/00792a.pdf
www.maximintegrated.com/en/app-notes/index.mvp/id/3930

I assume that pressing the button when it's latched ON will turn it OFF.

Edit: Other suggestions:
http://www.digikey.com/product-detail/en/DMN1019USN-13/DMN1019USN-13DICT-ND/4898885 R_DSon(max)=10mΩ@V_GS=4.5V SOT-23 USD [email protected]
http://www.digikey.com/product-detail/en/AO4402/785-1549-1-ND/3621495 R_DSon(max)=5.5mΩ@V_GS=4.5V SO-8 USD [email protected]

 

Yes, roughly, over the working range (0~5A in your case).

 

Hi TT, if you go the PIC route you would need a programmer like:
Pic Kit 3
This is a complete solution. There is a programming IDE and software to upload your code to the pic itself and it comes with the necessary cables. I don't think you will have room on the board for a ICSP so you would probably want to build a dev. board to program the chips. I would build one with a zif socket and a sot adapter so that you can pop the pic's out easily and quickly. There are other programmers that you could use with MPLab's software, but for the current reasonable price, why bother with potential conflicts.

 

Both are normally programmed in C or assembly language. Microchip's free programming environment for the PIC includes a C compiler and an assembler. I think Atmel's equivalent does too.

They are very resource-constrained devices and the C language implementations have a number of limitations but would be fine for what you want to do. Writing in assembly language requires more learning, and the PIC and AVR have different instruction sets.

Well, in my opinion the AVR is a much tidier architecture, but this is only really significant if you're programming in assembly language. If you're programming in C, the compilers take care of those details, so no. Both devices have similar built-in peripherals - reset on power-up and brown-out, simple timer, on-board clock oscillator, weak pullups.

PIC and AVR are both fairly widely used, but I think the PIC is more popular. I think they're a pretty poor architecture, so that's unfortunate, but they certainly are cheap!

Like chopnhack, I have used the PICkit 3 programming adapter, which plugs into a USB port and is supported by the MPLAB X development environment. It's a long time since I used the AVR, and back then I used a very simple programming interface using the PC's parallel port, but modern machines don't have parallel ports, and USB parallel ports don't work, and I don't know what Atmel's entry level programming system is now.

Yes, you're right. You'll need to program the important internal peripherals - set the brown-out voltage, enable the watchdog if you want to use it, configure the I/O directions and weak pullups, configure the timer, and structure your code so that it does the debouncing, timing and output control, and is tidy and reliable.

We have several members here, including me, who can advise you on how to implement your requirements, once you've made a start.

 

Right, that adapter isn't suitable - you would have to solder the device into the adapter, program it, desolder it, and solder it into your board. Micros are normally programmed in-circuit and Microchip have lots of documentation for how a programming header can be added as part of the design.

You don't have the space or money to waste putting an actual connector on the board, but you could make the programming connections using small pads in various places on the board, or directly to the pads that the micro is soldered onto, using a custom-made programming jig with five individually spring-loaded sharp points made from pins (or old gramophone needles, if you have any!). Like an upside-down "bed of nails" of the kind that's used for board testing.

You can probably find products ready-made to do this. I don't know what keywords to Google or what companies to try, sorry, but someone else here probably does!

 

Any socket solution is likely to be too large for your 14mm diameter board.

The kind of pins for making temporary connects that Kris referred to are called spring-loaded contacts.

http://www.digikey.com/product-sear...ee=0&rohs=0&quantity=&ptm=0&fid=0&pageSize=25

You would use these by making pads on your target board in a specific pattern, then soldering the spring-loaded contact to another board in the same pattern and connecting them to a header for the PICKIT. Then you just press the pins onto the contacts on your target board when programming.

I have not used these myself, but have heard of others using them successfully to solve this problem.

Bob

 

This is what I was referring to as well - using the socket on a developer board to make it easy to program all your chips. They could be popped out and then soldered into the final device. The sockets are pricey as you have found.

Another option would be to make contacts at the edge of the board, maybe you can fashion an appliance on a hinge that when closed will make contact with the edge pads on the board. Think like the ram slots of computers.

FYI - the pictured item is a counterfeit MPLab's device. Look at the logo, its a w (see pic below, its a M). See if you can cancel. The reason its close to home is a bit of a lie, its a bulk chinese distributor that states they have a "local" warehouse. This may or may not be true. Most of the Chinese sellers that offer e-packets for shipping have very good transit time.

 

Yeah, chopnhack is right that the one you ordered is a counterfeit. It's best to order direct from Microchip. It might cost a bit more, but it's a lot more likely to work, and if it doesn't, you have some kind of recourse.

I strongly recommend providing in-circuit programming using spring-loaded pins - either through pads scattered around the board, or by direct contact with the device's pins on its mounting pads - because there's a pretty good chance you'll be wanting to reprogram devices after they've been soldered into your board, at least during the early stages of the project.