Before watching your excellent video, I attempted to cut the clock frequency in half by selecting the 3.3V , 8 mHz option in the Arduino IDE before loading my program. The code I loaded was originally written to run at the standard 16 mHz, and it generates various tones separated by delays. I'm using a 5V, 16mHz Arduino Pro Mini, but (except when programming) I'm bringing in battery power on Vcc (bypassing the 5 Volt regulator), on the assumption that the 328P won't care about the voltage as long as I stay within the voltage limits of the 328P. To my surprise, this actually made the delays shorter and the tones were a higher pitch, as though the clock was running at a higher frequency than 16 mHz. Then, I added your two lines of code to Setup, to change the clock frequency to 8 mHz, and I reloaded the program using the 5V, 16mHz option in the IDE. Now the delays are twice as long and the tones are lower in pitch than they are at 16 mHz, which is what I expected. I'm still trying to figure out why using the IDE to change the clock frequency to 8 mHz seemed to cause the 328P to act as if the clock frequency had been doubled, instead of cut in half. What am I overlooking?
Thanks so much. Great idea for lower power consumption in less demanding applications without having to deal with waking up etc. Of course, I suppose you'd want to actually use sleep modes if you've gone that far, but definitely a good programming time/power savings ratio!
This was so helpful! Quick question I can't seem to find the answer two: How does the prescaler affect the baud rate for the serial log? Thanks again for this video, I have enjoyed watching the others as well
Firstly: Thanks for the vid. Secondly: Did you notice a current reduction when in deep sleep and reducing the clock? Because I was testing with an arduino nano, and without sleep mode the current goes down quite a bit with clock reduction (from about 15 mA at 16mHz to about 5.8 mA at 1 mHz). However, in deep sleep (using the low power library, ADC_OFF, BOD_OFF) the clock frecuency doesn't seem to have any influence (1.90 mA for ALL frecuencies). I suspect that those 1.9 mA are due to the USB port (arduino gets powered throught the 5V pin)
Like this video! Does the OLED have a built in memory or does it need constant input from the MCU? There are even more possibilities to reduce the consumption. You can deactivate unused components like the the ADC and some timers by setting the PRR register. The best results can be achieved using the watchdog timer if you have idle phases. I was able to go down to 5µA during idle using the ATiny. You can check the video Mini project: Bluetooth keys finder (aka "No worries keyfob") around minute 5:40, source links are available in the description. Another observation was that reduced voltage to minimum also reduced the leakage current. But since you are using a battery the regulation would eat up the difference.
The OLED has its own memory. In fact that scrolling is also done by the OLED chip. I did play with the ADC and PRR - shutting off the ADC helped a bit, the PRR did surprisingly little given its name.
Julian Ilett So it even could be possible running the display while the mcu is shut down... nice. The PRR is truly disappointing but easy to apply so worth to be mentioned.
Could you do the same for a battery powered arduino 7 segment based clock using a RTC module. So a button to reset, display the time and then go back to sleep?
If you turn off the ADC, Analog Comparator and Brown Out Detect circuits, you will get it down to 1uA. However changing the BOD level requires modifying FUSE bits.
Great video!! Have you tried waking up the mini with an interrupt using wd ? I am trying similar but having issues with configuration of wd with nano the nano just crashes
Julian, have you produced a video on how to lower the brown out detector to 1.8 volts? Can you use another arduino to reset the bod or do you have to use programmer? Thanks.
Hi Julian, thanks for the detailed info. I did the same as you described with the clock dividing. code is running well but serial monitor on pc always has only special chars and not the ones it should. Is there a trick to let the serial monitor work correct after clock speed changing?
Can the clock frequency be changed dynamically? Or can it only be changed at start up? Ie, can I use this to reduce power consumption during ‘down time’ but speed it up during other times ?
You can also use inline assembly for the sleep instruction. After setting up the registers just write the following line: __asm__ __volatile__("sleep"); Ok, unfortunately youtube makes the "asm" italic and removes two underscores. It's two underscores before and after asm. Also, if you wanna save power you should probably start with the simple stuff, like setting all your pins high or low, so they are not floating (as that causes the ATmega and many other microcontrollers to draw a little more power) and disabling peripherals like the ADC, UART and/or timers that aren't needed etc. (keep in mind that timer0 is used by millis, delay etc.) then if you wanna save even more power you can put the ATmega to sleep and get the current under 1 microamp if needed.
no but the sleep modes on a wemos d1 mini use about 1/10th of the power as this arduino. I use a wemos d1 mini in a weather station and can get over a week out of a 400mah lithium battery if i disconnect the solar panel.
for some reason, when I try to upload the code, nothing happens. Ive tried various versions of the code but nothing happens when I am monitoring power consumption. When I use the sleep modes, it works, but when I do the prescaler code, nothing happens. Can anyone let me know what I might be doing wrong?
How about dynamically adjusting the speed in the code? Instead of adjusting all delays, just give it the prescale command again when waking up so it can run at full speed until sleep.
In the deepest sleep modes, the external resonator/crystal is stopped altogether, so there's no need to prescale the clock in deep sleep. Dynamically adjusting the clock prescale in code is perfectly viable as a power saving measure.
I also disabled the ADC and Brown-Out and got the board down to 22uA. I had trouble with the onboard Linear Regulator. Mine was a SOT23 L05 (3 Legs), yours is a SOT23-5 LG50. (5 Legs) It was drawing 2.2mA at start, but current draw kept increasing as time went by. I tried powering the board from 12V initially, but then I powered everything from 5V and I got the same 2.2mA current draw. So I started removing components from the board (Crystal, Linear Regulator, Resistors, Capacitors) I started looking into 7805 datasheets and I saw a Typical Quiescent Current of 3.2mA. So be aware of cheap Chinese boards if your are looking into Low Power stuff.
@@fabirodriguez3677 the chinese Arduino consumes more. The voltage regulator was the problem, I removed it. But I also needed to find a replacement with low quiescent current and a fair price, preferably in a SOT23 package size.
Now I know why my project with an Attiny programmed with Arduino code runs much slower then on the 16 MHz Arduino. The Attiny only runs at 8 MHz / 4 MHz so any timers that runs on it will need dividing by 2 for the 8 MHz clock and 3 for the 4 MHz clock. I thought it was something to do with the system clock . Thanks !!
I always thought it was mad to include a deep sleep function that requires a physical interrupt to bring it back to life. Why not just deepsleep(1Hour)? Anyway, best way to do this is by removing the power completely to the Arduino with a TPL5110 power timer breakout chip - really easy to use. The only downside is it can only delay by 2 hours max. Data sheet for TPL5110 says 35nA when timing. In reality it is 20uA, which is pretty good!
You need to read datasheet section CLKPR register. This section explains that CLKPR register setup clock prescale system and it's has 7 bits for read/write register. So the is"CLKPCE" sub-register on CLKPR and allow you enable a change over clock system before you setup clock prescale value with the bit 3 to 0 on CLKPR. The BIT7 CLKPCE register is only "OK "setup, if you simultaneosly written to zero the bits6 to bits0 on register general CLKPR. So this is bit7[1]-bit6[0]-bit5[0]-bit4[0]-bit3[0]-bit2[0]-bit1[0]-bit0[0] = 1 0 0 0 0 0 0 0 and if you take bits in two group of 4 bits you obtain [1000]-[0000] that is [8]-[0] in hexadecimal value. So you can use CLKPR = 0b10000000 or CLKPR=0x80 when index "b" is binary value and "x" is binary value register setup.
Could anyone tell me why u can estimate the current consumption by measuring between the GND of the FTDI and the GND of the arduino mini ? Thanks a lot !
It's a little late, but here's my answer: You can place a current meter anywhere in series with the circuit. It does not matter if it is placed on VCC or GND. Take into account the internal shunt resistance of the meter.
Julian Ilett Thanks again for this video! I have just spent a few hours playing with this on my Uno and it works a treat. I did figure out that if you use OneWire or anything relying on precise timing you do have to set the speed to 16MHz again while it does it's stuff, then you can drop down again. Thankfully that's easy as pie with those two simple lines. I made two functions, clock_high() and clock_low() which I can call when needed in my sketches. Specifically I've been using a DS18B20 temperature probe which prompted me to find a solution otherwise it just didn't show a value when probed.
abyssunderground I don't think there's much inside the microcontroller which is unaffected by changes in the clock prescaler. The block diagrams in the data sheet will help identify those that are. Very pleased that you've found a use for it.
Julian, I am trying to follow along. The part a don't understand is "have to set bit 7 and all other to zero", I don't know how bit 7 equal CLKPR = 0x80. I do understand the second CLKPR = 0x01 for / by 2 , 0x02 is / by 4. Keep the videos comming
There is sleep library called Narcoleptic - code.google.com/p/narcoleptic/. I also did the same prescale experiments quite a time ago and the Narcoleptic library is on my HDD from back then. Unfortunately i don't remember the results using it. May be worth a try again. Cheers
abyssunderground ...and it's not strictly true that it isn't doing anything because it does have interrupt detecting routines still running in order to wake up.
You may be better off with the 3.3v Pro Mini. There's a graph in that 440 page datasheet which shows the maximum clock frequency recommended for different Vcc voltages. It's difficult to tell from the graph, but running at 16MHz at 3.7v or less looks a bit borderline. I think that's why the 3.3v Pro Mini has an 8MHz resonator.
Julian Ilett It's out of spec but what I've read it runs just fine. And I remember they sell those where you can change the voltage from 5v to 3.3v (16mhz) on ebay.
***** The problem with running out of spec isn't it stopping to work. I've run AVRs at around 30Mhz. The problem is, that it SOMETIMES and irreproduceable craps out. It could depend on temperature or on input voltage or the "weather of the day". Some things just get unreliable as soon as you move out of spec. The same AVR (speced for 16MHz max), that I could run at 30MHz with normal code, would corrupt the EEPROM for example when just running at 20MHz. There is a reason the manufacturers give you specs for a device.
Hi Julian, The sparkfun tutorial www.sparkfun.com/tutorials/309 "Adventures in Low Power Land" is a great resource regarding lowering the power consumption as musch as possible. -Bjorn Erik
Also look at the KickStarter project called µCurrent GOLD. Dave Jones has shed some light into the Low Current Land and limitations when measuring very low currents. Ref: www.kickstarter.com/projects/eevblog/current-gold-precision-multimeter-current-adapter I have ordered this adapter and I am looking forward to use it. From the video you can see the MCU current consumption duing startup from sleep using a DSO. Very informative :-)
if you buy the cheap nano withbuilt in usb they tend to ship with a 5v regulator, you can swap these to a 3 but then the poor thing is running at 3v/16mhz, which is essentially overclocking it, so you could use the clkpr command to bring it back down to 8mz, and i'm sure if you tell the arduino IDE software its an 8mhz device, all the timings for delay and serial comms etc will be fine. please note, if you change the 5v reg to a 3v, when you connect the usb cable then by the way these are designed you put 5v into the output of the 3v regulator , it does not like this!, so remove the diode on the usb 5v line and power the chip with the 5v from usb connected to vin/vraw with a diode or a jumper
Running atmega328p on 20mhz/5v, 16mhz/5v, 8mhz/5v, 8mhz/3.3v is okay, but running at 16mhz/3.3v is running above specification - maximum frequency depends on voltage.
"16 times slower" God please no, please stop using this idiotic incomputable phrasing of multiplying a relative modifier!! It makes NO sense, it is completely arbitrary, and people who actually want to learn about the stuff you're talking can't reliably tell whether it's now 1/16th the speed, 1/17th, 1/15, or anything else. "A sixteenth of a thousand" - See? Was that so hard? Perfectly logical, computable, and reliable.
To be honest dude the way he put it was alot easier to understand than "thousands" that's just more shit you gotta make your head compute, stop being a smartass you internet troll
Before watching your excellent video, I attempted to cut the clock frequency in half by selecting the 3.3V , 8 mHz option in the Arduino IDE before loading my program. The code I loaded was originally written to run at the standard 16 mHz, and it generates various tones separated by delays. I'm using a 5V, 16mHz Arduino Pro Mini, but (except when programming) I'm bringing in battery power on Vcc (bypassing the 5 Volt regulator), on the assumption that the 328P won't care about the voltage as long as I stay within the voltage limits of the 328P. To my surprise, this actually made the delays shorter and the tones were a higher pitch, as though the clock was running at a higher frequency than 16 mHz. Then, I added your two lines of code to Setup, to change the clock frequency to 8 mHz, and I reloaded the program using the 5V, 16mHz option in the IDE. Now the delays are twice as long and the tones are lower in pitch than they are at 16 mHz, which is what I expected. I'm still trying to figure out why using the IDE to change the clock frequency to 8 mHz seemed to cause the 328P to act as if the clock frequency had been doubled, instead of cut in half. What am I overlooking?
Great video, man...exactly what I've been searching for going on days now! Thanks!
You've inspired some more confidence in me as I have the same voltmeter!
You are one clever dude...how you can look at a data sheet and make sense of it is beyond me lol
This was super useful thank you! Need to read up on waking from sleep mode with interrupts next
Awesome sauce, I never knew there was a way to input assembly to Arduino - cheers mate!
Here is some info on that: www.nongnu.org/avr-libc/user-manual/inline_asm.html
Thanks so much. Great idea for lower power consumption in less demanding applications without having to deal with waking up etc.
Of course, I suppose you'd want to actually use sleep modes if you've gone that far, but definitely a good programming time/power savings ratio!
This was so helpful! Quick question I can't seem to find the answer two: How does the prescaler affect the baud rate for the serial log? Thanks again for this video, I have enjoyed watching the others as well
Firstly: Thanks for the vid.
Secondly: Did you notice a current reduction when in deep sleep and reducing the clock? Because I was testing with an arduino nano, and without sleep mode the current goes down quite a bit with clock reduction (from about 15 mA at 16mHz to about 5.8 mA at 1 mHz). However, in deep sleep (using the low power library, ADC_OFF, BOD_OFF) the clock frecuency doesn't seem to have any influence (1.90 mA for ALL frecuencies). I suspect that those 1.9 mA are due to the USB port (arduino gets powered throught the 5V pin)
Like this video! Does the OLED have a built in memory or does it need constant input from the MCU?
There are even more possibilities to reduce the consumption. You can deactivate unused components like the the ADC and some timers by setting the PRR register.
The best results can be achieved using the watchdog timer if you have idle phases. I was able to go down to 5µA during idle using the ATiny. You can check the video
Mini project: Bluetooth keys finder (aka "No worries keyfob")
around minute 5:40, source links are available in the description.
Another observation was that reduced voltage to minimum also reduced the leakage current. But since you are using a battery the regulation would eat up the difference.
The OLED has its own memory. In fact that scrolling is also done by the OLED chip.
I did play with the ADC and PRR - shutting off the ADC helped a bit, the PRR did surprisingly little given its name.
Julian Ilett So it even could be possible running the display while the mcu is shut down... nice. The PRR is truly disappointing but easy to apply so worth to be mentioned.
Could you do the same for a battery powered arduino 7 segment based clock using a RTC module. So a button to reset, display the time and then go back to sleep?
Great video. Really helpful.
thank you for this it's really useful for me!
Great video - is there a way to sleep for a duration (eg. if you wanted to wake up and do something every 5 minutes)?
If you turn off the ADC, Analog Comparator and Brown Out Detect circuits, you will get it down to 1uA. However changing the BOD level requires modifying FUSE bits.
Great video!! Have you tried waking up the mini with an interrupt using wd ? I am trying similar but having issues with configuration of wd with nano the nano just crashes
Does this work for the arduino Nano the same way it did for the mini?
Julian, have you produced a video on how to lower the brown out detector to 1.8 volts? Can you use another arduino to reset the bod or do you have to use programmer?
Thanks.
Now that's a vedio worth subscribing
What's the voltage at 5 ma and 8mhz ??????
hello Julian i have tried this many times and one problem i have is that when in sleep mode i can not get below 2.06 ma no matter what i do.
Your are converting 5v pro mini into 3.3 volt 8mhz that is already available in market . Two versions 5v and 3.3 volt pro mini !!????
Hi Julian, thanks for the detailed info. I did the same as you described with the clock dividing. code is running well but serial monitor on pc always has only special chars and not the ones it should. Is there a trick to let the serial monitor work correct after clock speed changing?
You have to multiply the baud speed in your sketch by the amount you divided the clock speed by.
Does it makes any sense to put a (NodeMCU-8266 + MAX7219) project into Sleep mode if the display time is only updating every 1 minute?
Can the clock frequency be changed dynamically? Or can it only be changed at start up?
Ie, can I use this to reduce power consumption during ‘down time’ but speed it up during other times ?
Could you also disable the ADC to lower the power requirement?
4:24 Oh my a pro like you using an old maplin knock off ? Mind you I have the same one - it must be around 15 years old now and is still working
You can also use inline assembly for the sleep instruction. After setting up the registers just write the following line:
__asm__ __volatile__("sleep");
Ok, unfortunately youtube makes the "asm" italic and removes two underscores. It's two underscores before and after asm.
Also, if you wanna save power you should probably start with the simple stuff, like setting all your pins high or low, so they are not floating (as that causes the ATmega and many other microcontrollers to draw a little more power) and disabling peripherals like the ADC, UART and/or timers that aren't needed etc. (keep in mind that timer0 is used by millis, delay etc.) then if you wanna save even more power you can put the ATmega to sleep and get the current under 1 microamp if needed.
Great video ;)
Could you clockdown the processor on a wemos d1 mini pro although it has not an atmel chip?
no but the sleep modes on a wemos d1 mini use about 1/10th of the power as this arduino. I use a wemos d1 mini in a weather station and can get over a week out of a 400mah lithium battery if i disconnect the solar panel.
for some reason, when I try to upload the code, nothing happens. Ive tried various versions of the code but nothing happens when I am monitoring power consumption. When I use the sleep modes, it works, but when I do the prescaler code, nothing happens. Can anyone let me know what I might be doing wrong?
goodjob very useful thank u sir
How about dynamically adjusting the speed in the code?
Instead of adjusting all delays, just give it the prescale command again when waking up so it can run at full speed until sleep.
In the deepest sleep modes, the external resonator/crystal is stopped altogether, so there's no need to prescale the clock in deep sleep. Dynamically adjusting the clock prescale in code is perfectly viable as a power saving measure.
I also disabled the ADC and Brown-Out and got the board down to 22uA.
I had trouble with the onboard Linear Regulator. Mine was a SOT23 L05 (3 Legs), yours is a SOT23-5 LG50. (5 Legs)
It was drawing 2.2mA at start, but current draw kept increasing as time went by.
I tried powering the board from 12V initially, but then I powered everything from 5V and I got the same 2.2mA current draw.
So I started removing components from the board (Crystal, Linear Regulator, Resistors, Capacitors)
I started looking into 7805 datasheets and I saw a Typical Quiescent Current of 3.2mA.
So be aware of cheap Chinese boards if your are looking into Low Power stuff.
Hello, I did not understand if it is Chinese the arduino consumes more? can't remove the regulator?
@@fabirodriguez3677 the chinese Arduino consumes more.
The voltage regulator was the problem, I removed it.
But I also needed to find a replacement with low quiescent current and a fair price, preferably in a SOT23 package size.
Now I know why my project with an Attiny programmed with Arduino code runs much slower then on the 16 MHz Arduino. The Attiny only runs at 8 MHz / 4 MHz so any timers that runs on it will need dividing by 2 for the 8 MHz clock and 3 for the 4 MHz clock. I thought it was something to do with the system clock . Thanks !!
I always thought it was mad to include a deep sleep function that requires a physical interrupt to bring it back to life. Why not just deepsleep(1Hour)? Anyway, best way to do this is by removing the power completely to the Arduino with a TPL5110 power timer breakout chip - really easy to use. The only downside is it can only delay by 2 hours max. Data sheet for TPL5110 says 35nA when timing. In reality it is 20uA, which is pretty good!
can u explain to me..why we used the coding
CLKPR = 0 x 80;.....the 80 is stand for what?
0x80 is hexidecimal for the decimal value of 128.
Hope this helps!
You need to read datasheet section CLKPR register. This section explains that CLKPR register setup clock prescale system and it's has 7 bits for read/write register. So the is"CLKPCE" sub-register on CLKPR and allow you enable a change over clock system before you setup clock prescale value with the bit 3 to 0 on CLKPR. The BIT7 CLKPCE register is only "OK "setup, if you simultaneosly written to zero the bits6 to bits0 on register general CLKPR. So this is bit7[1]-bit6[0]-bit5[0]-bit4[0]-bit3[0]-bit2[0]-bit1[0]-bit0[0] = 1 0 0 0 0 0 0 0 and if you take bits in two group of 4 bits you obtain [1000]-[0000] that is [8]-[0] in hexadecimal value. So you can use CLKPR = 0b10000000 or CLKPR=0x80 when index "b" is binary value and "x" is binary value register setup.
I suggest getting a good understanding of binary and hexadecimal notations before jumping on to projects.
Nice !
Could anyone tell me why u can estimate the current consumption by measuring between the GND of the FTDI and the GND of the arduino mini ? Thanks a lot !
It's a little late, but here's my answer: You can place a current meter anywhere in series with the circuit. It does not matter if it is placed on VCC or GND. Take into account the internal shunt resistance of the meter.
I didn't know about the pre-scalar, I'll have to look into this one for my current project. Thanks =) Also, do you have a link to that massive PDF?
www.atmel.com/Images/doc8161.pdf
Julian Ilett Thank you very much =)
Julian Ilett Thanks again for this video! I have just spent a few hours playing with this on my Uno and it works a treat. I did figure out that if you use OneWire or anything relying on precise timing you do have to set the speed to 16MHz again while it does it's stuff, then you can drop down again. Thankfully that's easy as pie with those two simple lines. I made two functions, clock_high() and clock_low() which I can call when needed in my sketches. Specifically I've been using a DS18B20 temperature probe which prompted me to find a solution otherwise it just didn't show a value when probed.
abyssunderground I don't think there's much inside the microcontroller which is unaffected by changes in the clock prescaler. The block diagrams in the data sheet will help identify those that are. Very pleased that you've found a use for it.
you are setting CLKPR to two different values. Why?
That is what the datasheet instructs you to do in order to follow the correct procedure.
Watch here @4:30
IQ +10 Thank you.
Julian, I am trying to follow along. The part a don't understand is "have to set bit 7 and all other to zero", I don't know how bit 7 equal CLKPR = 0x80. I do understand the second CLKPR = 0x01 for / by 2 , 0x02 is / by 4. Keep the videos comming
0x80 (hexadecimal 80) is 10000000 in binary - so bit7 is set (leftmost bit) and all the others are reset.
I understand 1000000 is 0x80 but you were looking at the 7 bit. I am still missing something
bee Hive 10000000 sets the 8th bit, but since the bits are numbered from 0 that is the 7th bit
Thanks
+Julian Ilett it might be worth next time writing stuff as 0b1000000 rather than 0x80, I know hex is so much shorter, but might help newcomers.
clever
try OBS screen capture or uploading you source code
There is sleep library called Narcoleptic - code.google.com/p/narcoleptic/. I also did the same prescale experiments quite a time ago and the Narcoleptic library is on my HDD from back then. Unfortunately i don't remember the results using it. May be worth a try again. Cheers
Not for nothing but you do not show the power source hookup for us newbies so I give you one star
Given that it's doing absolutely nothing 230 micro amps is quite a lot ;-) But nevertheless an interesting video.
Most batteries have a self discharge higher than this so it is pretty low compared to that =)
abyssunderground ...and it's not strictly true that it isn't doing anything because it does have interrupt detecting routines still running in order to wake up.
Thats not how interrupts work. By thier very nature they are not polled
I've read you could remove regulator from this board and connect it straight to li-po battery. Voltage should stay in safe limits for this chip.
You may be better off with the 3.3v Pro Mini. There's a graph in that 440 page datasheet which shows the maximum clock frequency recommended for different Vcc voltages. It's difficult to tell from the graph, but running at 16MHz at 3.7v or less looks a bit borderline. I think that's why the 3.3v Pro Mini has an 8MHz resonator.
Julian Ilett It's out of spec but what I've read it runs just fine. And I remember they sell those where you can change the voltage from 5v to 3.3v (16mhz) on ebay.
*****
The problem with running out of spec isn't it stopping to work. I've run AVRs at around 30Mhz. The problem is, that it SOMETIMES and irreproduceable craps out. It could depend on temperature or on input voltage or the "weather of the day".
Some things just get unreliable as soon as you move out of spec. The same AVR (speced for 16MHz max), that I could run at 30MHz with normal code, would corrupt the EEPROM for example when just running at 20MHz.
There is a reason the manufacturers give you specs for a device.
Hi Julian,
The sparkfun tutorial www.sparkfun.com/tutorials/309 "Adventures in Low Power Land" is a great resource regarding lowering the power consumption as musch as possible.
-Bjorn Erik
Also look at the KickStarter project called µCurrent GOLD. Dave Jones has shed some light into the Low Current Land and limitations when measuring very low currents. Ref: www.kickstarter.com/projects/eevblog/current-gold-precision-multimeter-current-adapter
I have ordered this adapter and I am looking forward to use it.
From the video you can see the MCU current consumption duing startup from sleep using a DSO. Very informative :-)
That is a good tutorial - thanks Bjorn
if you buy the cheap nano withbuilt in usb they tend to ship with a 5v regulator, you can swap these to a 3 but then the poor thing is running at 3v/16mhz, which is essentially overclocking it, so you could use the clkpr command to bring it back down to 8mz, and i'm sure if you tell the arduino IDE software its an 8mhz device, all the timings for delay and serial comms etc will be fine.
please note, if you change the 5v reg to a 3v, when you connect the usb cable then by the way these are designed you put 5v into the output of the 3v regulator , it does not like this!, so remove the diode on the usb 5v line and power the chip with the 5v from usb connected to vin/vraw with a diode or a jumper
+lez briddon Nanos run on a atmega328p, they are rated for 20mhz, so no. 16mhz is not overclocking it.
Running atmega328p on 20mhz/5v, 16mhz/5v, 8mhz/5v, 8mhz/3.3v is okay, but running at 16mhz/3.3v is running above specification - maximum frequency depends on voltage.
"16 times slower" God please no, please stop using this idiotic incomputable phrasing of multiplying a relative modifier!! It makes NO sense, it is completely arbitrary, and people who actually want to learn about the stuff you're talking can't reliably tell whether it's now 1/16th the speed, 1/17th, 1/15, or anything else. "A sixteenth of a thousand" - See? Was that so hard? Perfectly logical, computable, and reliable.
To be honest dude the way he put it was alot easier to understand than "thousands" that's just more shit you gotta make your head compute, stop being a smartass you internet troll