Propeller Mini in Low current mode pulling 15.9ma
JohnR2010
Posts: 431
To continue from an earlier post where I was asking if it was Okay to source the mini from 5v I set it up and did some testing today. As discussed everything worked as anticipated I tested sourcing the mini with 5v hooked up to its 5v pin 4 on J2 and all worked extreamly well. This will allow me to source high current 5v devices since I will be driving everything from an external 5v power supply. The next thing I want to do is use it in a low current configuration (from batteries) sleep mode. So I hooked up an external 3.3v regulated power supply to the 3.3v p4 on J1 of the mini. Everything ran fine my code executed just as expected but when i put it to sleep (used waitpeq on pin 9, all other cogs stopped) the propeller was still drawing 15.97ma?? I was expecting sub Milli amp readings. I'm measuring the current feeding into the 3.3v pin. What is using the current at this point I'm not using the two on-board voltage regulators??
Comments
I think that Parallax's data sheet on the Propeller shows the relationship of power consumption to clock rate. I am not sure how the PLL comes into play. But for really low power, you need to go slower. That might mean changing the actual crystal. In the extreme, a 32Khz watch crystal might be chosen.
Try the charts on page 26 and 27 of the Propeller's Datasheet PDF.
If you can change the clock to change the clock to RCSLOW when it goes asleep, that would save the most power.
And when it wakes up, change back to 80 Mhz.
Something to peruse:
http://forums.parallax.com/showthread.php/142124-Low-power-guidelines
However... I think your problem is that the LM2937 regulator is not designed to take a voltage on its output from an external source. Is that in fact the regulator you see on your 'mini? And the 'mini schematic shows an LM2940-5 for the 5V supply. I do know that if you take an LM2940-5 and apply 5 volts to its output, and leave the input open, unplugged, it will draw about 10mA of reverse current. Furthermore, if that 5V is applied to the output and the input is shorted or connected to a standard DC power transformer (DC short through the xfmr), it will draw an unsafe and possibly destructive level of reverse current.
Regulators that are designed to withstand a reverse voltage applied to their output usually advertise that fact prominently in their data sheet.
Tracy, your dead on with what I think is happening. I have re-run my test and with all 8 cogs stopped I'm still pulling 15.03mA so the problem is not in the clock speed. I fear you hit the nail on the head in that I shouldn't apply voltage to the outputs of either one of the regulators contrary to an earlier post from David Carrier http://forums.parallax.com/showthread.php/157425-Is-it-possible-to-run-the-Propeller-Mini-with-5v
I think the takeaway from my original question (Is it Okay to drive the mini from its output?) is: Yes it will work but the long term affects are not known and the chip was not designed to do that! As my end goal is to have a product I can sell It doesn't look like the Mini is going to be the solution. Its a shame I have the circuit done and laid-out for the mini I desperately need the small footprint! Back to the old 40pin through-hole beast (I do love this chip just need a smaller through-hole version!). I wish we had the capability to do SMT and work with QFP version!! Arrrgggg...
I doubt that there would be any "long-term" effect, just that side effect that you observed.
However, why wish you could do SMT? Everyone can, it is so so easy that I can get somebody who has never even soldered a lug, to paste and assemble fine pitch parts. Forget about micro-tip irons, you just want a dedicated el-cheapo toaster oven ($20-$30) with a top and bottom element, some solder paste, and some aluminium foil. You can run the solder paste across the fine pitch QFP leads rather than individually and plonk you components down, preheat the oven to max, place your assembled board onto the tray with the folded foil in-between to prevent hot spots, onto the middle shelf. Leave it in the oven for 4 minutes max (no profile), carefully remove without jarring and let it cool on a cooling tray for a couple of minutes. I find that it takes 3 minutes before the solder even starts to reflow but 4 minutes ensures that all parts have been reflowed otherwise you could have some bridges.
If your solder paste is not that fresh you might have some bridges but these are usually cleaned up with a touch of flux and a medium sized iron tip (the tip's mass grabs the excess solder).
However, that doesn't necessarily mean no current will flow.
Of course, now you are applying 3.3 volts to the output of the LM2937, a regulator with different specifications.
Have you looked closely at the Propeller Mini board to see if it's possible to cut the traces to (or maybe remove) the regulators so that your power supply is isolated from everything but the Propeller circuit?
"Regard the original question of whether you can power the Propeller Mini's Vin pin with 5 VDC, it should work, but it will leave the 5-volt regulator in full dropout, so the power consumption will be a littler higher than normal. You can also connect the 5-volt supply to the 5V pin to power the Propeller Mini. It may leak some power into the Vin pin, so I recommend leaving the Vin pin unconnected."
He points out that excess current will be drawn due to the regulator being in full dropout. I'd say that the voltage at the output is turning on all of the internal bias circuits in the '2940. That in itself won't damage the regulator. If your project can afford the extra current, okay, but not good for your purposes. David also points out that the connection may leak power back to the input, so leave Vin unconnected. I would state that MUCH more emphatically. I just tried it again with a current limited power supply, and with a short circuit at the input (or with a wall-wort transformer plugged in, unpowered). The 5V applied at the output quite happily delivered 0.5A back to the input through some path inside the regulator. By NO mistake do that!
It was pointed out in that other thread that the LM2940 is an automotive device, and it is rated to withstand a high level of voltage both forward and reverse. That applies to the Vin side only, not Vout. It won't be destroyed or pass a bad voltage to your circuit even if you hook up the car battery backwards or some big + or - kick comes from the starter motor. All kinds of load dumps and situations occur in the automotive environment, conditions that hit the Vin side.
The regulator on the mini looks to be an SOT-223 package. For comparison, look at the Linear Tech LT1129. On page 12 of its data sheet under applications you can see what I mean about full disclosure of possible combinations of input and output when backup power supplies are involved.
The champions of low quiescent current are regulators that have pmosfet pass transistors, such as the MCP17xx series. However, be careful with those when applying a voltage to their output. The mosfet substrate diode passes the voltage and current directly back to the input. But those could be a good substitute regulator for your project.
Peter, I have seen videos of this and I find it very interesting. I think I need to get a kit or something and give it a try. When I have boards made I usually get a 100 boards or more made and it ends up costing me $450 plus I don't have the guts to order a 100 boards and go this route without some first hand experience. How do you place the smaller SMT components like the capacitors and resisters? They look so small and I know the paste will hold them in place but I thought it would be so hard to keep everything lined up as you place it in the oven.
I like this suggestion as long as I can cleanly do it and it doesn't take a lot of time I'm open for it. At first glance it looks like LM2937 output trace is on the top of the board and I don't see any vias taking it to other layers so it looks doable. The trace is so small it is going to be hard to cut it and not nick the neighbors. I will take a look at the schematic and see if there is another place I can get at it.
You should know from the videos that exact lining up of components is not really necessary, as long as you don't have them off their pads what will happen is that they will be pulled back when the solder reflows. But this is something you have to try yourself, even on a spare bit of board and just try pasting, use long tweezers to pick up parts and place them. Give them a quick check and tidy up before placing in the oven. I find that I end up "smudging" the QFP around, to make sure all the pins are fully seated and nudge it into place. The tweezers should be clean and preferably non-magnetic. I take some components off the reels and place them into those little pill trays suitably labelled so that each compartment has it's own lid. If need be I tip one type out onto a business card with the long edge folded to stop parts falling off that edge and then pick them up from this card, afterwards I tip the remaining ones back into the tray using the fold as the chute. Using the card is also good for fiddly parts like SOT-23s etc as a lot of them are upside down but I just pick the ones that are right side up or tap the card to bounce them around.
A piece of cake, and sometimes I have had to remove some minor bridges but if the solder paste is good it will come out clean. A steady hand helps although I use a manual dispenser gun rather than a syringe as that gets too tiring when you do quite a few boards. I have done up some videos before but I need to do another one when I can get someone to hold the camera and zoom in at the right angles so I might call in young Jack to help me make and edit a video. All those little tips I've mentioned from the foil to the card really make a difference but it's seeing it all work that makes the biggest difference. When I get my new IoT5500 boards back then that might be a good time to document this.
Yeah, there is a lot of advice out there for DIY SMD but really and truly this is the simplest cleanest way, forget the frying pans, the hot air guns, the micro-fine tips, just paste and place, and always toast on full for 4 minutes with the foil, not less, works every time, week after week, year after year, design after design.
To be sure you must drive all pins into an inactive state, never leave any floating as stray voltages can cause input gates to be biased both on and off which will leak current of course.
BTW, a lot of those LDO regulators are totally unsuitable for micropower operation.
With a bit of luck, one might find one that can swap out whatever is on the Propeller Mini. At the same time the 5V regulator could be simply removed. Then you would pretty much have a micro-power Propeller Mini.
Do EEPROMs have a quiescent current draw? I have never tried to find out, but the PDF should mention such a spec. I peeked at a 64K I2C EEPROM. Operating current 0.1 to 3.0ma (highest when writing), standby current 0.01 ot 0.05ua... check for yourself.
ww1.microchip.com/downloads/en/.../21189f.pdf
Reran the test with all pins set to output and forced low didn't make any difference. To be clear I have a very simple app it blinks a LED 5 times and then ends the Pub method causing all cogs and execution to stop on the propeller. At this point the Propeller Mini is still pulling 3.95ma. So no cogs are processing anything. I have even pulled my LED off the board and the current didn't change. Something is still pulling stray currents. I just noticed another discrepancy in the propeller mini schematic. R1 in the Bill of Materials is listed as a 4.7k but in the schematic drawing it is drawn as a 10k ohm resister. If in realty it is a 1K ohm resistor and somehow it was leaking to ground through the eeprom or the propeller's P29 pin it would be pulling about 3.3mA and that would explain what is going on.
Does the 3.95mA include the current of some external voltage regulator, or only the current to the Prop 3.3V supply pins?
You mentioned above that you are considering the LM267N-3.3 simple switcher. Be aware that those are efficient at their high design currents, but they are not efficient at all at micropower currents. If you consider swapping out the LM2937, one possibility is the MCP1703, which operates at a meager 2 microamps of quiescent current. Compare that to 10 microamps minimum keep-alive current of the Prop itself (one acheived!). Other possibilities are the micropower PNP regulators like the LT1121 that is used on all of the multi-slot BASIC Stamps, quiescent current (zero load) = 20µA.
The 3.95mA is the current being delivered to the Prop mini's 3.3v p4 on J1. My current meter's (Fluke 28) negative lead is directly on P4 of the mini and the red lead is on the output of an external 3.3v power supply. So there are no other components in the path. When the Propeller is running with one Cog turning an LED On and Off the current varies from 4.82mA to 9.95mA. After the LED blinks 5 times my program runs to the end of the method and stops. At that point I see the 3.95mA
Thanks Loopy for checking that was on my list to do. I was thinking it was right in line with what you found. Why should I check it if someone else will??? So kidding I actually enjoy reading spec sheets just hadn't got to it yet. I should have looked before I even asked. Thanks again for looking it up!!
So the question still remains what is drawing my phantom 3.95mA?
It seems this question has attracted some high horse power resources and I wanted to let everyone know I'm sold on the mini and say thanks for the suggestions!!
My point about "inactive" states means either high or low depending upon the active state. For the I2C that means high, for a grounded LED, that means low. For an input that means it's either pulled-up/down already or ensure it is either high or low, never floating. Also double check the way that you are measuring, if for instance you measure 4ma then confirm the source of that by switching the meter into the supply paths etc. Be methodical before jumping to any conclusions. You always need at least one cog running (which has conditioned the I/O pins) even at a WAITCNT with RCSLOW to confirm low-power operation. You are doing something wrong, making some assumptions, not being methodical enough, I know it.
Re: Peter's comment about tapping surface mount components held on a business card in order to flip them right-side-up. I do that too. It is kind of a game, to see how many resistors etc. can flip at once. If all of them end up right side up; you win:-) I have a pile of little ceramic wasabi or teabag holders I found at an outlet store for that purpose instead of a card, but the tap trick still works. I can put different parts in the plates and write on the edge of the plate with a sharpie. When done, the sharpie mark comes off with alcohol.
The Circuit:
My test circuit has a LED and a 220 ohm resistor tied to P8 of the mini and a switch to ground on P9 (Yes I know I should have a pull-up on that switch). On my bench I have three power supplies 12v, 5v, and 3.3v that I will use to power the mini for the various tests. My test meter is a Fluke 289 multimeter set to measure milliamps.
The Code:
The code turns my LED on and off 10 times ever second, then it goes into a wait state for pin 9 to go to ground. Once pin 9 has been grounded it turns on the LED for 5 more seconds and then ends the method causing the propeller to shut down. At this point the LED is off and all Cogs are stopped and from my understanding all pins go to their input state. So clock speed, having the pins set high or low have no meaning at this point. The propeller is shutdown. Right??
The Test Results:
I ran 4 test and took 4 current measurements. For each test I measure the current going into the propeller with my current meter in-line (no current clamp). My meter is between the bench power source and the corresponding test pin on the mini. The pin I connected the meter lead to for each test is noted in the header row of the test grid. I took 4 readings during each test run. The first reading "1 Cog" was with the program running and the LED off. The second reading "1 Cog + LED" was when the LED was on. The third reading "Pin Wait" was when the propeller was waiting for the P9 to go low. The fourth reading "All Stop" was after the program ended and the propeller was shut down. I repeated these 4 reading for each of the 4 test. The first three test i simply move the power from the Vin pin to the 5v pin and then the 3v pin. For the last test I cut the output trace of the LM2937 and applied power to the 3v pin.
Tracy, I was just speculating that this could be a possible explanation to the phantom power drain. I have no idea what the real value of this resistor is as the schematic has had several errors I was just spit balling. I will see if I can find it and some how get my leads on it.
I agree we should find this!! I'm willing to keep at it until we do. I just may learn something!! Thanks again for the input.
We do dare! Though we have the same trouble the Fonz has in saying we were wrrrr...wrrrrrh.....wrrrrh.....oh forget it.
BTW, you can't stop all the cogs as all the I/O will be in input states and unless you have them pulled up/down you will have problems, CMOS logic inputs on any chip should never ever be left floating unless of course that is your intention, but the consequence is increased current drain at least when the ultra-high input resistance picks up every stray induced voltage etc. YOU MUST at least run with one cog with I/O set to INACTIVE and try RCSLOW at least, with an optional WAITCNT if you like. For your circuit just turn the LED (and button) off (low) and make all the other inputs including SCL high, simple. After that you can do a Homer, Doh!
Thanks for sticking with it.
BTW, I noticed that you have a button connected to the Prop but where is the resistor to pull it up or are you assuming something here that isn't so....
And I did learn something!! Thanks Tracy and Peter sorry I was so thick i didn't understand the importance of setting the unused pins to output and making them high as well as the clock speed. For some reason I had it in my head that the clock speed would have no affect during an WAITPNE don't know where I came up with that but sorry to make you guys repeat yourself so many times.
Here is the updated code:
Here are the updated test results with the new code:
Thanks for the help!!!