EEVblog
Вставка
- Опубліковано 2 сер 2024
- In this Fundamental Friday Dave will blow your mind as he shows you how you can power a microcontroller without a power or ground pin!
A potentially big trap for young and old players alike.
This is not something you'd normally get taught.
Can you guess how it's done before the secret is revealed?
Forum: www.eevblog.com/forum/blog/eev...
SPOILER:
A tutorial on how ESD protection diodes in chips work and how they can be used, mostly inadvertently to power a chip through the input and output pins without having the power or ground pin connected.
This example uses an MSP430 microcontroller, but is applicable to almost any complex or simple CMOS chip.
EEVblog Main Web Site: www.eevblog.com
The 2nd EEVblog Channel: / eevblog2
Support the EEVblog through Patreon!
/ eevblog
EEVblog Amazon Store (Dave gets a cut):
astore.amazon.com/eevblogstore-20
T-Shirts: teespring.com/stores/eevblog
Donations:
www.eevblog.com/donations/
Projects:
www.eevblog.com/projects/
Electronics Info Wiki:
www.eevblog.com/wiki/
💗 Likecoin - Coins for Likes: likecoin.pro/@eevblog/dil9/hcq3 - Наука та технологія
Diodes Inc. datasheet not telling you about the diodes.... YOU HAD ONE JOB!
+Fernrat LOL!
+Fernrat It could be that the worker who knew about diodes, was not in work the day the datasheet was being created.
+Fernrat lol!
The company for whom I work has a product that I found to have this error. I mentioned it in a engineering review for a new version of the board and was nearly thrown out of the room because "it works fine, you're wrong." Well, I wasn't wrong and I got a promotion out of the deal.
+Aaron Hammett Nice!
+EEVblog Still can't figure it out.... :( but cool thing... :) and interesting for learning...
Do not lie to us Dave. We know you have a batteriser plugged in behind that piece of paper
Oh my god. That double space. My OCD does not approve.
+TomSparkLabs My OCD goes into overdrive when I also see emails or text messages, which are all written in lower case with no punctuation at all! ARRRRGHHH!!!
+A. R. Jasso No it´s that "Free-Energy" thing, in where you can plug any extender plug in one of its outlets and all other units (refrigerators, microwaves, TV-sets and you electrical heater) as well and all will work simultaneously....
+A. R. Jasso still waiting for someone to send him a batteriser swagbag in the mail for mailbag monday.
+EgoShredder what do you mean
Another common case where this crops up... if your circuit has multiple power domains, like a battery-powered device where some chips get powered down and others remain powered up [soft-off], you have to worry about I/O between the domains. A logic high signal fed into an input on a powered-down chip will burn current through the protection diodes, especially when powering down a chip effectively ties its VCC to GND rather than floating it like your example.
lol the part where he says you can remove the ground pin too. Learned something interesting and new to me today by watching this video. +1.
+Eric Liang Glad you liked that bit!
Within the first 5 seconds, I said to myself "protection diode", and then I had to wait 8 minutes to confirm it. Dave, you missed a golden opportunity here to release a short video! :)
+Quentin Smith I didn't need the 5s. The title was enough to know what the video was about. My interest was only to know, why the video is so long. But I gave up. to boring.
+JackySixSeven It was instructional, and meant for people who haven't yet encountered this behavior. I only wish Dave had measured the voltage drop between Vcc and the input he was powering with, to give the viewers something to think about. "Hmmm, 0.6V drop, that's interesting! About a silicon diode drop..." He did this later after he had revealed the cause.
+Quentin Smith It was supposed to be that, but I ended up covering much more stuff.
+Quentin Smith it's relatively short, Dave's "quick" looks can easily get to an hour and even part 1, that's why we love it :)
and the signature "this was a lot longer then expected"
+EEVblog I was thinking it's too easy in the beginning too, but then I've learned something. And when I was a beginner, it would be really easy for me to understand. Great work!
3 years I'm wondering "why it is so" , when I am working with PIC uC. With this interesting tutorial, it clears very thing. Thanks Dave.
You did it Dave! Over unity! That's a bobby dazzler
+Kenneth Wilhelmsen but due to the lack of a housing we had to go without a "We're in like Flynn" today
+Big Whoop At least we "came a gutser" a few times though. Proud to be an Aussie.
Thank you so much, Dave!!! You've literally saved my project! I've been looking for the current leakage path for about a year, ant it turned out to be this effect! Unbelievable!
What I like about these videos is that there is no background music, giving you the freedom to add your own.
You could even power the microcontroller with only some Leds and shine some light on it. The LEDs can act as solar cells if you put a few in series.
"The Art of Electronics" mentions this in the paragraph that begins with the words, "CMOS can drive you absolutely crazy!"
Prof. Julius Sumner Miller and physics is his business. Respect.
This is why I watch this channel! Best video ever. Always great to learn from an experienced professional.
Really mind blowing facts!
I've witnessed some of these phenomena without being able to find suitable explanation for the situation.
Great video, thanks Dave.
I'm really glad you went deep into this explanation. I have been level translating 5v to 3.3 using mosfets for some of my applications, and now I will experiment with just using current limiting resistors! ((5v - 3.3v) - 0.6v) / 1mA! Easy!
Am I the only one that cringed at the 3.3V rail on the blue bus, and gnd to the red bus? ;)
+Wolf Tronix
It's at least the two of us. You really must not mess up the colors! It saves so much hassle (and parts destroyed by brain lag), if you enforce a color scheme. I go even further and use (for flying wires, not breadboard jumpers) fixed colors for most functions. Orange for RX or reset (they start with "r" and orange got one, too), yellow for TX (because it's right besides orange on the wire harness anyway), green for clock signals (the chinese say "gleen" it's starts like "clock" ;) ), blue for data (from german "blau" both have an "a"). So yeah, I cringe at random color usage (maybe also because I teach students in electronics at uni, where I demand proper color usage for the power rails, power it's a PITA to trouble shoot circuits, if you have to start by tracing supply voltages)
+superdau
Perhaps, it is because he is Australian... Everything is backwards down under. ;)
he answered this, saying he was taught that the top rail is positive and the bottom is gnd... he said old breadboards didn't have the colors...
The old way makes more sense, but I still go by colours personally.
+Wolf Tronix routinely work with 3.3V MPUs and 15V analog. I can't bear to look at that red bus carrying ground. It's just WEEERRRRAWNG!
excellent vid... just found you and based on this alone I'm subbing because you both know the material and have *such a passion for the material* you tend to ramble on which of course reminds me of my best professors back in avionics (& really in any field)... the kids in the group could get a seriously awesome free education from watching your vids alone, I'd gather. wow, 3.3 v and hct -- its advanced so much since I was taught the stuff I'll even have to look up the T.
My father taught electronics back in WWII and actually got a radiation classification from standing on hiroshima 9 months after the bomb went off. He instilled a great love of math and electronics and lots of fields in me. I look forward to more of your vids when I have the time.
I knew what this video was going to be about,before I even clicked it. I discovered this a while ago,with the MSP430 chips/launchpad board. I could disconnect everything but the serial lines,and it was still going. Or disconnect the serial lines,and leave the test/reset connected,and it would still run. The LED's were very dim,but they were still blinking/fading. These chips will run off of flea-farts,for power. All it takes is one LOW-ish pin,and one HIGH-ish pin,and they'll run. It was a head scratcher,for a minute.
Clearly it's a form of zero point energy Dave happened to of stumbled on, Dave is just amazing!
Figured this out the hard way, but never truly understood till this great video. Thanks!
Love your Fundamental Fridays!
5yrs later still MIND BLOWING!!!
Excellent Dave, keep them coming!
Thanks much! I don't build a lot of circuits, but I have built enough to find this video very educational. I learned something new today.
Dave,Merry Christmas and Happy New Year from Italy!!!!
Def worth a thumbs up. Excellent vid, good bit of fun for someone who thinks they know about micros and such already - didn't know about the ground trick
More like this Dave! Great fun.
Great Video Dave. Its a good feeling on a Saturday morning being able to guess most of whats happening. Not feeling quite as much of a young player more. But #StillMuchToLearn :)
Julius Sumner Miller - one of the greatest physics teachers to ever be filmed. Loved watching his shows when they were on air here in America. Your videos are also great, Dave and kepp 'em coming!
+drcobol2000 kepp? Really? Can't type to save my @$$.
Amazing co-incidence. My professor just discussed this today in class.
Thanks for the explanation. I actually stumbled onto this concept a few days ago while playing with some 74HC595 shift registers attached to an Arduino. I pulled the Vcc pin and was baffled to see my circuit still working. Couldn't find a proper explanation other than people calling is parasitic current from the other pins. Your explanation is much better and logical.
As always, very fascinating and entertaining. Your videos make me wish I'd gone a lot further with electronics when I was younger. Thanks and best wishes.
Love these quiz/debugging vids
I forgot to connect power and ground on a chip in a recent project..and it worked, but not as i expected like you said :(...now I know why and why it took me so long to find. Many thanks Dave, this was really interesting and really a trap :).
I built a reflow oven controller out of an MSP430G2 earlier this year, and ran into exactly this behavior. I was more than a little confused. Glad to finally know why it behaved this way.
You have to watch out if you have a project using something like an FT232R or some other serial to USB IC. If you don't have the power rail of your project and the power rail of the FT232R connected, one of them can power the other through an I/O pin if one of them is powered and the other is not.
Very interesting! Never seen this happen before, although I did know about the protection diodes.
This was very useful, thanks Dave!
Great practical video... thanks Dave.
Fascinating and pretty damn useful information. Thank you Dave.
I love the binary counter. I used to put a pair of them on my breadboard when I was a kid. Was my first piece of "test equipment".
Great video Dave.
Good to know, thanks Dave.
Very informative. Well done.
Had a similar problem trying to build an external monitoring panel for a PLC using 24v LEDs directly from PLC input/output terminals.Outputs were OK but inputs were a complete failure.
Now I know why.
Thanks Dave
Simple & good to describe!
ESD protection is only half of the story, the MOSFET transistors in the pin output drivers also have the intrinsic body diode associated with them to conduct. This effect is actually used in power electronics where the body diode is used to provide current return paths, for example in H-bridge motor drives.
I had this exact phenomena on my battery management system prototype board. It had a FTDI's FT230X RS232-USB converter and only it's RX and TX pins were mapped to my MCU, the 5V usb power only went to the FT230X. The MCU hanged if I first connected the USB cable and then the battery pack to the board (3V3 was regulated from the battery), but not if I first connected the battery pack and then the USB cable.
Thanks Dave!
This is like watching Dora the explorer. I'm shouting at the screen.
Oh boy, I did the same...
These power diodes were also mentioned in an interview (on youtube) of one of the engineers Sophie Wilson of the ARM instruction set back in the day. Something about measuring power usage of the processor and the device was powered by the protection diodes because of some mistake on the board .. Can't remember exactly. Go watch it. It's great. :D
Very interesting video. Thanks !
Ha! Professor Julius Sumner Miller! Used to love him on Hilarious House of Frightenstein. Never saw Why Is It So, will have to look some up! :)
Helpful. Thank you.
Thanks for pointing that out! Might be useful one day for trouble shooting.
Excellent Dave! Hope this video stops a few people finding out the hard way.... Point of caution is that the diodes redirect the voltage onto the supply rail. If the circuit is low power, the supply voltage can be exceeded when the input voltage is higher than the supply voltage, even if the esd diode is within its current rating. A zener diode across the supply will protect the circuit.
cheers
Andrew
In fact the ARM1 chip, on Acorn's test bed, during a power consumption test mishap, did not receive power from its power pins. However the chip is still working, powered by just the current from the bus and the decoupling caps. This proves to both Acorn and their customers the ARM is so power efficient, a trait that still runs in the series to this day.
This is actually related to something I caught in my design job. A feedback network was external on an previous level of silicon and moved internally on the next revision, too bad the supplier forgot that their ESD device was the first device internal to the IC! The maximum voltage rating of the pin was supposedly 42v but would blow once the ESD device conducted around 28v. Long story short, slap a series resistor in there that would limit the input current and not offset the divider drastically, not bad for a young guy out of college (first thing I thought of 😂).
Great video dave!
I just came across this same issue when I was fiddling with a cd4026be 7 segment decoder. I planned on making a simple counter with a 7 segment display and while I was setting up the clock generator from a 555, I figured I'd disconnect the 4026, so I pulled the wire on Vdd expecting that would do it. To my surprise/dismay the chip continued working the next time I connected power to test the clock. I was confused, but some friends on IRC cleared it up for me, and then I found this video. I remembered Julian Ilett's video on a similar thing happening to a 74hc126 decoder after he damaged it, so I feared I may have damaged this in some way, but apparently it's normal. :D
The protection diodes on the pins also can be used to measure bad solder joints on QFN or similar package pins or broken pcb trace to a connector when something is not working as you excepted! Just use your multimeter in diode measurement mode and connect the "positve probe" to the pcb ground and measure the connectors/traces with the other probe. Also you can use the vcc as the reference for this test just exchange the probes "polarity".
Perfect!
Thank you!
I remember those many years ago, I still catch where there a fault when other engineer can't figured out. nice one Dave.
Thanks Dave!
I've used this "trick" on my teacher back in high school. He was baffled. And at 7:49 you have shown the exact page of exactly the same file which I've have just opened to "see" those diodes :) Even with the same search keyword :D I still use these diodes as a crude form of logic level translation from 5V devices to 3.3V tolerant FPGA input pins (with 10k serial resistor). And after I wrote it, you have said it (commenting while still watching). Man, we think the same way.
lol "This is heavy"
nice Friday banter!
Thanks for the explanation! I experienced this with an Arduino and a TI Launchpad and couldn't figure out why.
Hi Dave
In the ATE world (Flying Probe and Bed of nails) we use these protection diodes to run open tests.
Drive from GND test point on your board through the device and onto the net of the I/O pin.
I never considered this could be used to power a device!
I have actually had a problem with my quadcopter not too long ago where a ground trace on my minumOSD had burned out unknowingly from accidentally applying 12V lipo power to the ground pin instead of the input to a voltage regulator. after replacing a few smoked wires I noticed that the minumOSD had stopped working on battery power but would function with USB power to just the main flight controller microcontroller! Stranger yet, the thing would also function if I connected a monitor composition input to the output, with the ground of the composition lead going to another, more accesable ground! That's all it took to make it work, took forever to try to fix it because it would work on my test bench, but not in flight!!! Grounding issues like ground loops can be a REAL trap, no kidding!
Stupid auto correct on mobile! composite, not composition
part of my failure was not using star connections for grounding. I was daisy chaining all my grounds together, which was very susceptible to problems. I have yet to fix that though!
that digilent powerbrick is super handy! i want one now!
There are some 4000-series devices that don't have input protection to both rails. The 4049UB hex inverter and 4050B hex buffer only have a protection network to the negative rail Vss.
There's another surprise these diodes can spring on you. The reason there's a maximum current given for the protection diodes is that if you push enough current through them you can trigger the parasitic SCR inherent in CMOS, which can cause the device to try to swallow the supply.
For those familiar with the standard two/three inverter RC CMOS oscillator circuit these diodes are why you have the third resistor in series with the first input - it's to isolate the RC timing component junction from the clamping effect of the diodes.
Hey Dave great video. I was working with my Arduino the other day and when I moved from one room to the other, the leds were dimmer. I quickly found my ground had come undone and when re-attached the leds came to full brightness. The arduino program still worked the whole time. I now know how the ground path was made though the leds, and most importantly... why.
Excellent!
I've actually encountered this thing with ATTINYs... I use that a lot... Thanks Dave :)
Yep .. happens to best of us..
I managed to to power pic16c84 from PC Serial port ... luckily had some resistors on those lines :-)
+DIYTAO You didnt opto-isolate the serial connection?, i see you like to live dangerously
I had this problem in one project, where AVR micro would measure battery voltage with ADC. ADC was connected straight to battery with small voltage divider, and AVR was powered all the time. It emptied battery even with main switching mode power supply turned off. After I realized the protection diodes were leaking the voltage in ADC pin to supply rail I changed voltage divider to really high impedance one with capacitor in ADC pin, and it solved the whole problem.
Actually I experienced this behaviour with an Atmel AVR by "accident" and I was "shocked" (exactly the way Dave explained: "magic for young players" hehe) how it is possible :) A simple LED blinker example still worked, though the LED was like dimmed. Surely, I figured out what was going on (well, with also utilizing the power of using google ...) but it's really a nice video to also answer questions like this, thanks!
As a kid I had a BBC Micro, and I made devices to plug into its "user port" which was an IDC header/socket connected to a bunch of I/O pins of the Beeb's 6522 VIA. In one project I accidentally left its power supply unconnected, so no chip had power to Vcc, but it still worked! It took me weeks to find out why. I overheated the Beeb's 6522 VIA, because its poor I/O pins were powering my project, and I damaged it. School fees I suppose, from the good old days.
good job Dave, thanks for making me more conscious for this bloody thing from now on.
- Does it also has to do with "not used pins on CMOS ICs", which should be grounded and not be left on the air??
awesome stuff
Dave did good!
My home-spun 6809-based computer showed this phenomenon and I never quite understood it. But I could power the whole computer as long as I kept the UART and Maxim chip powered. I always just attributed it to "leakage current" but I see now it was more complicated than that!
A number of manufacturing In-circuit Test platforms exploit this to verify that the pins of an IC are soldered down. It's quicker and less complex than powering the board and stimulating it with logic patterns. The company I work for makes a machine that uses two of these diodes to form a primitive BJT on the die; making it easier to find faults on bused nets.
Love it!
Great! I didn't know that you can get away without dedicated IC to convert the signal levels.
wOW! i've been working with MCUs for years and never knew this! thanks Dave. :)
This is very interesting! Thank you Dave!
One thing though, your videos are generally unnecessarily lengthy. Very informative, but lengthy. If they were shorter, people would be able to watch more of them!
I have a video of me (shot by my sister) literally sleeping to one of your teardown videos XD
Back in the day (early 1970s) we had a circuit that was all CMOS that read a serial clock signal off of an EEG recording and put the output into a video stream of a camera pointed at the pens and paper that the EEG was being drawn. One day the output was a bit flakey (occasionally dropping out but always having the correct values when it was displaying. Turns out that the power switch was off and the circuit was being powered from the composite video that it was modifying.
Interesting... Bit of trivia for you, back in the 80s, when Acorn of Cambridge were inventing the ARM processor, they had a curious bug which had them scratching their heads... Everything was working great, but would crash at the same point in the code... They couldn't understand it...
Then they realised they'd forgotten to connect the power lines to the ARM, it was powering itself from the data lines, and yes, you guessed it, the memory location that kept causing a crash had all the data lines at 0!
It was then that the low power consumption abilities of the ARM design were really noticed!
Sometimes even a pullup resistor for a button can be enough to boot the uC, has happened to me many times before.
I tought the diodes are there because they are parts of the push-pull MOSFETs of the GPIOs. But you showed they are also there in input only devices. I learnt something.
I just stumbled upon this the other day when I powered a memory extension for a Z80 microcomputer via the CPU bus.
Uooou! This video blew my mind. :)
I like how this video is in the playlist but is unlisted :)
Before I watched the video I thought it was referring to those "one wire" interfaces. Like the digital temp sensor "DS18B20". You only need to connect data and ground. You can leave the VCC line disconnected. it's using the positive data I/O to power the device.
mind blown!
You can power a pic to flash leds through it's IO pins as well.
Been there... Very annoying. Thanks for the video!
This reminds me of this thing I figured out on accident when I was a child. See, I had this vidjah game console peripheral that came with a universal adapter so it would function on all the consoles that were out at the time. I just so happened to own a playstation 2 and a gamecube and just to mess around, plugged in both the plug for the gamecube and for the playstation while the playstation was still on. Doing this caused the gamecube on light to glow and some of the bobs to twitch to life on the inside.
This happened to me with an LCD... The Vcc pin was connected to a transistor on the development/fun board. I didn't know that and everything worked very well - until I sent zeros to it....
ive seen this a bunch on little circuits i've played with... always wondered
+Carter Cole It ain't something they teach you at school!