Hi all,
I've been (unsuccessfully) attempting to breadboard a small boost regulator. I would like to run my 3.3V Arduino off of a CR2032 coin cell and I heard good things about the MCP16251T boost regulator:
MCP16251T
I'm using the SOT-23 package that I have soldered to a little breadboard breakout adapter I purchased from Digikey.
I have carefully followed the datasheet for the 3.3V application. Since this is a breadboard layout and I needed through hole parts, I've found what I believe to be are "equivalent" to some of the suggested components in the datasheet.
Cout:
Cin
4.7uH Inductor
For the feedback resistor divider I used generic 1/8W resistors. I used the 330K and 560K. They are the closest values I could find at my local electronics store. They work with the equation on page 14 of the data sheet and are close to the values in example 5 on that same page for a 3.3V output.
With a fresh 3V CR2032 battery, the behavior I am seeing is that the output voltage starts somewhere around 2V and then slowly drops to about 0.5V after a minute and stabilizes.
I've checked and double checked and triple checked my circuit. It matches the one presented on the top of page 2 of the datasheet for the SOT-23 package.
Is there something glaringly obvious that I should be looking for? What does the behavior indicate is going on? I have limited experience with building boost power supplies. Any tips would be great!
Thank you
Two thoughts come to mind:
-
Perhaps your trying to get too much energy from the CR2032. Try testing with two AA alkalines and see if there is a difference.
-
Switching regulators are very sensitive to lead length, especially the grounds. Look at you physical layout; are the grounds of the input capacitor, chip and output capacitor connected to the physical same point?
Switching regulators require a PCB with a ground plane and a good layout. You don't stand any chance of getting things stable on a solderless bread board.
I used to manage professional engineers who specialised in switching regulators, and I never had one that could get a design going and stable in less that three iterations of a PCB, and they were experts.
JohnRob:
Two thoughts come to mind:
- Perhaps your trying to get too much energy from the CR2032. Try testing with two AA alkalines and see if there is a difference.
This switching regulator has been successfully used with CR2032 and arduino in applications such as the "whisper node." It should work...
JohnRob:
2) Switching regulators are very sensitive to lead length, especially the grounds. Look at you physical layout; are the grounds of the input capacitor, chip and output capacitor connected to the physical same point?
Currently they are connected to the same row on the breadboard with a short wire that runs back to the (-) terminal on the battery. I can probably make them even shorter if this will help.
Grumpy_Mike:
Switching regulators require a PCB with a ground plane and a good layout. You don't stand any chance of getting things stable on a solderless bread board.
I used to manage professional engineers who specialised in switching regulators, and I never had one that could get a design going and stable in less that three iterations of a PCB, and they were experts.
That doesn't sound hopeful lol. Maybe it was pure luck, but I recently got a LM5118MH boost/buck regulator and a LM22675MR buck regulator working on a breadboard. I used SMD breakout adapters the same way I'm trying now and I was able to achieve the desired voltage (at least that's what my dummy meter showed).
I would not consider any of the push type "breadboards" appropriate for a switching circuit. To make it work you need copper clad breadboard maybe something like this:
ebay clad board
It would not be the ideal case but should work as your currents are low.
Regarding the ability to operate on a CR2032, thanks for the input. I've not worked with a button cell yet so it is good to know.
Maybe it was pure luck, but I recently got a LM5118MH boost/buck regulator and a LM22675MR buck regulator working on a breadboard.
It is much more likely that you did not test the circuit properly and it does not work correctly. Did you check the traces with an oscilloscope? If not you could not possibly tell if there was oscillations on the waveform nor if the waveform was stable under all current draw values.
The datasheet gives a recommended PCB layout. Follow that exactly, and use the recommended components.
Otherwise you are unlikely to succeed.
G-M is right - these devices are tricky. The very fast current switching means that the impedance of any wiring ( about 1nH/ mm ) is significant. Breadboards are OUT.
Even looking at the signals requires care - the long 'tail' of the earth clip of an osciilloscope probe can introduce large errors.
Allan
doublec4:
This switching regulator has been successfully used with CR2032 and arduino in applications such as the "whisper node." It should work...
Can I suggest you try it.
Currently they are connected to the same row on the breadboard with a short wire that runs back to the (-) terminal on the battery
Can you post a picture of your project so we can see your component layout?
Thanks.. Tom...
Grumpy_Mike:
It is much more likely that you did not test the circuit properly and it does not work correctly. Did you check the traces with an oscilloscope? If not you could not possibly tell if there was oscillations on the waveform nor if the waveform was stable under all current draw values.
That is very likely. As I mentioned, I just tested the outputs with a multimeter and no scope. I had the voltage output I was looking for, and the components being powered from them are surviving but I do not own an oscilloscope so I am not sure how well it is actually performing.
Since the MCP16251T requires many less associated components, I (naively) assumed that it would be easy to get at least similar results for the purpose of a bread board prototype.
allanhurst:
The datasheet gives a recommended PCB layout. Follow that exactly, and use the recommended components.
Otherwise you are unlikely to succeed.
G-M is right - these devices are tricky. The very fast current switching means that the impedance of any wiring ( about 1nH/ mm ) is significant. Breadboards are OUT.
Even looking at the signals requires care - the long 'tail' of the earth clip of an osciilloscope probe can introduce large errors.
Allan
Thanks Allan, for the first revision of the PCB prototype I stuck to the recommended layout the best I possibly could. We'll see how that one turns out!
TomGeorge:
Can I suggest you try it.
Can you post a picture of your project so we can see your component layout?
Thanks.. Tom...
I will make a first revision PCB with the recommended layout in the datasheet and try it for myself of course I can post a picture when I get home tonight. Last night I started removing components from the breadboard and soldering them directly to the breakout board pins to try and make the lead lengths as short as possible. Didn't make a difference in terms of performance, but definitely made it look like a mess :o
Didn't make a difference in terms of performance, but definitely made it look like a mess
Reminds me of a story by supervisor told me at university. He used to do research into radar in the war. One time the Germans caricatured an air to air VHF transceiver and decided to copy it. However, they thought the actual wiring was a mess, typically British they thought, so they tided it up, typically German, you might think.
The copy did not work because, although it was neat, the Germans knew little about stray inductance and capacitance so it could not work if they made it neat.
So until you can measure the performance you can't say if it improved or not.
One time the Germans caricatured an air to air VHF transceiver
It sounds like they did - but surely you mean 'captured'?
Allan
I know switching converters need good layout - but why exactly? I thought it is because of EMI and possibly noise on the output. What features of solderless breadboard can make the switching regulator to fail utterly?
@OP: did you consider powering the Arduino directly from the cell unregulated? AVR based Arduinos should work with less voltage than 3V3 and most sensors also don't need such "high" voltage
As noted above the inductance of typical wire / pcb trace is 1nH/mm.
If you reckon the device is switching eg 1A in a few nS, and apply the standard
V = L di/dt
You'll see that even a few nH generates a large transient voltage, which screws everything up - particularly if there is common inductance shared between power and signal paths.
Hence the need for very careful pcb layout, and the complete unsuitability of solderless breadboard techniques.
It's also why SMD devices perform better - much lower lead inductance.
Allan