Smallest available power supplies?

Status
Not open for further replies.
It is no problem to get phone chargers or other "wall warts" that have specs like 700mA/5V, which is of course enough for Teensy. It is oversized by one order of magnitude for most projects, though, as a Teensy would probably be fine with 70mA most of the time.

Now I understand that we are not talking about huge waste of energy, but rather just a tiny bit. Nevertheless losing about 1W in the power supply while doing the actual work with 50mA seems rather wasteful. Using batteries is no solution in many scenarios either (and charging batteries has even worse efficiency, probably). Projects with tens of independent Teensies are no problem. Having tens of wall warts idling at 1W is still a bit of waste.

Is there a 5V 100mA or 200mA power supply that can be bought? What do 0.5W/230V LED bulbs use as a power supply? That was the commercial product with the smallest wattage rating I found.
 
just because a power supply is rated at 700 mA @ 5 V (== 3.5 W) doesn't mean it consumes all this power from the wall continuously. Most name-brand 'wall warts' are reasonably efficient. I recently measured the Apple iPhone one. At no load it consumed 12 mW (this would cost $0.01/year if you just left it pluged in with no load); at 5 W it was 74 % efficient (very good). At light loads I would expect it to be good also. It's small and safe.
 
Most newer "wall wart" power supplies are designed for some form of the "Energy Star" efficiency specs. I'm not very familiar with those specs, but here's a page with lots of info.

http://www.energystar.gov/index.cfm?c=archives.power_supplies

I do know that before the Energy Star program published specs for small external power adaptors, the vast majority were the old style with just a transformer, bridge rectifier and capacitor. You can easily tell that type in 2 ways: it's heavy and after being plugged in for hours, it gets noticeably warm, even if the thing it's powering is turned off. The newer ones don't heat up at all when the device they're powering is off or using very little power.

If you care about saving power, use one of those newer ones. If you're buying a new one, there may not be any choice. The old inefficient ones seem to have mostly disappeared from the market.
 
Also, you can recognize the old ones because they are not universal -- only 110-120 V and 60 Hz in the US. The 'good' ones are smaller, lighter and typically 100-240 V, 50/60 Hz or so.
 
Switch-mode power supplies certainly have their benefits - lighter, more efficient, lower standby losses, more compact, etc. but there is at least one downside - more electrical noise. In most applications that can be designed around. Just be careful which power supply you source - pay a couple more bucks to get one from a reputable dealer with all the usual approvals (UL, CE, etc.)

FWIW, my power meter uses a switch-mode power supply from CUI to produce up to 1.2A at 5V. An on-board power voltage regulator then takes it down to 3.3V for the Teensy and the other on-board components. I hope this approach will allow the best of both worlds.
 
Hi Guys,

for this topic I may have some input.
I needed a very small power supply. For peripheral chips like LED driver and RS485 I used a Traco tsr2433, for the Teensy 3.2 I used a TSR2350. The USBVin was cut away, as described here
https://forum.pjrc.com/threads/2597...ower-Teensy-3-0?highlight=switch+power+supply
Pretty nice small power switching units!
I followed the reference design for suggested filter to comply with EN55022 Class A limits, plus used some SMD 4,7µF 0603 caps at the output/GND pins to iron out some ripples- some smaller ripples still exist.

Traco technical sheet: http://www.tracopower.com/products/tsr1.pdf

I also have a question: Today the teensy died. It run over 2 Weeks with no problems. Today It got pretty hot and now the on board led only dims at blink.ino. After a retry it wasn't able programm via USB.
The GND is all the same.

Now when I power on the CPU gets hot again but no LED is dimming. I tried 3,3V on the 3,3V_in (beside VBat) input and nothing happens.

Here is my question:

Did the voltage regulator die because of the switching transients?

here is my schematic. the 100µF are low ESR 7mm cabs, the 4,7µF small standard 0603

PSR.png
 
Last edited:
1. Class A is industrial; so this is relatively noisy.
2. Not clear if there are minimum load requirements per the data sheet.
3. All SMPS have significant PARD, at least when compared to linear regulators.
4. Has line/load regulation been measured for 50% step loads? Not uncommon for some SMPS to do bad stuff at end of load step.
 
1. Class A is industrial; so this is relatively noisy.
2. Not clear if there are minimum load requirements per the data sheet.
3. All SMPS have significant PARD, at least when compared to linear regulators.
4. Has line/load regulation been measured for 50% step loads? Not uncommon for some SMPS to do bad stuff at end of load step.

for 2) The data sheet says "Minimum load -> not required"
for 3) PARD is usually expressed in mV peak-to-peak or rms, at a specified bandwidth, typically 20 MHz.
The data sheet says: 50 mV max
for 4) the data sheet says: dynamic load response 50% load change (upper half) -> 150 mV max. peak variation and 250 μS max. response time

Does the Teensy 3.2 voltage regulator die from those ratings?

For a real measurement I need an example to measure the load variation. The TSR delivers up to 1.0 A, so I don't know what passive part can be used.
Do you have any ideas?
 
I was also looking for the smallest possible power supply, and this little bugger was the best I could find. Not the best efficiency, but a very small PCB footprint and rather exotic dimensions that can make it fit in between other components, increasing component density.

I did not pursue this route however, because to make it EMC proof and improve output stability you still need a lot of external capacitors and inductors. I stayed with the Vigortronix VTX power supplies. Also not the best efficiency, but still very small footprint and has all components integrated.
 
Some of these extremely small DC-DC converters skimp on capacitors. Years ago, at least one person suffered a lot of frustration which was ultimately solved by adding an input capacitor. The datasheet implied this extra capacitor wouldn't be necessary for lower input voltages. It was.

https://forum.pjrc.com/threads/23647-Powering-Teensy-3/page2

Hi Paul,

thank you for the hint. On the schematics I postet you can see, that after the lowpass there is placed 100µF before each Traco TSR. I'll try it with 10µF.
But do you have any idea, why the Teensy 3.2 could die? After two weeks of running serveral hours a day the CPU gets pretty hot now and the LED at Pin13 lights up for blink example veeery low. I can't programm the unit anymore. Is that a sign for a dead voltage regulator? The V_in pin was cut trough, btw.
 
If you suspect the teensy's voltage regulator you can test it by loading the 3v3 pin whilst powering the board from the Vin pin (or resolder the vin-vusb trace and use usb). It should be happy to source 250mA (assuming enough power is commming in), connect a 33ohm (or similar) resistor which should draw 100mA. This is well within the specs - if you see that the voltage is too low or that the regulator is running hot it may be faulty.

Unless the regulator has failed as a short, in which case you should read an over voltage on the 3.3 pin, for the MCU itself to get hot the problem probably lies within the MCU, or what it's trying to do.

That's not really enough information (in my opinion) to trouble shoot your project. Running something for two weeks can turn a bit of bad practice into a big problem, and that can be anywhere from your hardware to your software. You might want to consider starting a new thread (as this is pretty of topic to the orignal post) with a lot more information about your project, including what it was doing and how (both hard and soft ware).
 
If you suspect the teensy's voltage regulator you can test it by loading the 3v3 pin whilst powering the board from the Vin pin (or resolder the vin-vusb trace and use usb). It should be happy to source 250mA (assuming enough power is commming in), connect a 33ohm (or similar) resistor which should draw 100mA. This is well within the specs - if you see that the voltage is too low or that the regulator is running hot it may be faulty.

Unless the regulator has failed as a short, in which case you should read an over voltage on the 3.3 pin, for the MCU itself to get hot the problem probably lies within the MCU, or what it's trying to do.

That's not really enough information (in my opinion) to trouble shoot your project. Running something for two weeks can turn a bit of bad practice into a big problem, and that can be anywhere from your hardware to your software. You might want to consider starting a new thread (as this is pretty of topic to the orignal post) with a lot more information about your project, including what it was doing and how (both hard and soft ware).

You are right, thank you for the explanation.
 
Status
Not open for further replies.
Back
Top