n00blibrarian
Posts: 3
Joined: Wed Aug 14, 2019 9:14 pm

Four dead Pis in one class!?

Wed Aug 14, 2019 9:32 pm

I'm a relatively new raspberry pi educator (I was fortunate enough to get to go to a Picademy last year) and have been running a weekly robotics club for teens this summer where the kids have been using the 3B+ and an explorer hat to build a simple two wheel drive rover. This week we tried moving from 'bench testing' to actually letting the things loose and it has been one problem after another. The battery packs, which I bought from a hobby shop here in NYC at the last minute (after our original plan fell through) and which were specifically labeled as for the Raspberry Pi, ran the Pis but we got the 'low voltage' icon whenever we tried to activate the motors. Oh, well, the shop didn't have enough of them anyway, and the other type I picked up for the rest of the sets, which they assured me would also work fine, actually FRIED my personal Pi which I was using last night to run through the lesson and make sure everything worked, which meant the kids were going to have to share the ones we had anyway.

So we set those aside and started tinkering with the robots still connected to the AC power again. This is where things really went south. Over the course of a one hour session I had FOUR Pis die. Three of them were lighting up the red light only, and the third didn't light up at all. On all four I tried disconnecting everything from the Pi except the power to see if I could get a green light, and no dice. They all died at the same two stations, as when the first one went kaput I swapped in a spare with explorer hat still attached and left all the same peripherals and motors. In both cases, they worked okay for a while and then died again. Also during the session we had trouble at multiple stations where Pis were boot cycling - for those, I had the kids disconnect everything and start again from scratch, and that worked fine. But it was quite a mess, especially given how smooth things have been so far.

The really odd thing was that, because we didn't have enough batteries to start with, only one of these four devices had ever even been run connected to one of them, and all were on AC when they died. We've been running for over a month with no issues from any of the devices other than some problems with a few SD cards, and now this? Is there any easy way to do an 'autopsy' on a dead Pi to figure out what on earth went wrong with it? I'll need to buy at least two new ones to keep the club running and without knowing what went wrong I'm leery of spending more money. If I can't figure out what went wrong this club may be dead in the water, and I don't want to do that to my kids. Help!

ejolson
Posts: 3423
Joined: Tue Mar 18, 2014 11:47 am

Re: Four dead Pis in one class!?

Thu Aug 15, 2019 5:08 am

n00blibrarian wrote:
Wed Aug 14, 2019 9:32 pm
I'm a relatively new raspberry pi educator (I was fortunate enough to get to go to a Picademy last year) and have been running a weekly robotics club for teens this summer where the kids have been using the 3B+ and an explorer hat to build a simple two wheel drive rover. This week we tried moving from 'bench testing' to actually letting the things loose and it has been one problem after another. The battery packs, which I bought from a hobby shop here in NYC at the last minute (after our original plan fell through) and which were specifically labeled as for the Raspberry Pi, ran the Pis but we got the 'low voltage' icon whenever we tried to activate the motors. Oh, well, the shop didn't have enough of them anyway, and the other type I picked up for the rest of the sets, which they assured me would also work fine, actually FRIED my personal Pi which I was using last night to run through the lesson and make sure everything worked, which meant the kids were going to have to share the ones we had anyway.

So we set those aside and started tinkering with the robots still connected to the AC power again. This is where things really went south. Over the course of a one hour session I had FOUR Pis die. Three of them were lighting up the red light only, and the third didn't light up at all. On all four I tried disconnecting everything from the Pi except the power to see if I could get a green light, and no dice. They all died at the same two stations, as when the first one went kaput I swapped in a spare with explorer hat still attached and left all the same peripherals and motors. In both cases, they worked okay for a while and then died again. Also during the session we had trouble at multiple stations where Pis were boot cycling - for those, I had the kids disconnect everything and start again from scratch, and that worked fine. But it was quite a mess, especially given how smooth things have been so far.

The really odd thing was that, because we didn't have enough batteries to start with, only one of these four devices had ever even been run connected to one of them, and all were on AC when they died. We've been running for over a month with no issues from any of the devices other than some problems with a few SD cards, and now this? Is there any easy way to do an 'autopsy' on a dead Pi to figure out what on earth went wrong with it? I'll need to buy at least two new ones to keep the club running and without knowing what went wrong I'm leery of spending more money. If I can't figure out what went wrong this club may be dead in the water, and I don't want to do that to my kids. Help!
The time I built a Raspberry Pi controlled toy car, I used separate battery packs with a common ground to power the motors and the Pi. It worked fine, but I was likely a bit lucky as well as I am a not an electrical engineer but a mathematics teacher.

While the Pi is cheap, it is a real computer that does not have any of the built-in safeguards a toy might have to prevent damage when misused. As a result they're easy to break. When connecting wires power should be off and anti-static protection in place. Moreover, the wiring should be double checked both visually and electrically using a meter before the Pi is powered on.

There is obviously something wrong with the current rover. Until the problem is found, connecting another Pi computer to the same circuit will likely result in another broken computer. For practical help you will need to post schematics showing how you have connected things together as well as photos which show the actual work. People who have taken the trouble to post such details seem to receive good help on these boards.

Until you figure out what's wrong with the circuit, you may want to switch to a pure programming project, such as sorting algorithms, with the kids.
Last edited by ejolson on Thu Aug 15, 2019 2:49 pm, edited 1 time in total.

User avatar
davidcoton
Posts: 4033
Joined: Mon Sep 01, 2014 2:37 pm
Location: Cambridge, UK

Re: Four dead Pis in one class!?

Thu Aug 15, 2019 10:09 am

The Pi3B+ is particularly sensitive to 5V getting back to the 3V3 supply. This can happen if anything shorts pins 1 & 2 of the GPIO header, and is often fatal to the Pi. Symptoms are that the red LED stays on, green stays off, and the 3V3 supply (pin 1) is at 0V. If you test this, be careful not to short pins 1 & 2 while testing -- otherwise the Pi will be dead afterwards even if it wasn't before.

The good news is that the Pi4B is more resilient in this respect (testing that statement is not recommended, however :shock: ), but Pi4Bs required more power and different power and HDMI connectors.
Signature retired

alphanumeric
Posts: 2013
Joined: Tue Jan 19, 2016 2:17 pm
Location: Sydney, Nova Scotia, Canada

Re: Four dead Pis in one class!?

Thu Aug 15, 2019 11:24 am

I have two, two wheel rovers based on a Pi Zero W and explorer pHat. I used a powerboost 1000c and a 2200 MAH LIPO battery. No issues with frying my Pi's not so far anyway. They have been run for a hour or so a couple of times. I need to set them up to be controlled with a Bluetooth controller and just haven't gotten around to it.
I started with this kit, https://www.adafruit.com/product/2939
And used this as my electronic chasis. https://shop.pimoroni.com/products/mini ... 448026055
My powerboost is soldered to a proto zero board https://shop.pimoroni.com/products/protozero
I also have a shut down button, and power switch to turn off the powerboost on the proto board.
My build pictures are here if you want to have a look see. It's a link to my public One Drive folder.
https://1drv.ms/u/s!AjOYwiwlwDtpgrNiqCt ... ?e=hU7TgC

alphanumeric
Posts: 2013
Joined: Tue Jan 19, 2016 2:17 pm
Location: Sydney, Nova Scotia, Canada

Re: Four dead Pis in one class!?

Thu Aug 15, 2019 11:27 am

I meant to mention, I went with the Pi Zero W and explorer pHat to keep my weight down, maintain a small footprint, and lower my power requirement.

User avatar
mikronauts
Posts: 2717
Joined: Sat Jan 05, 2013 7:28 pm
Contact: Website

Re: Four dead Pis in one class!?

Thu Aug 15, 2019 12:37 pm

Sorry to hear about your problems, however you REALLY should be using different battery packs for powering the Pi and the motors (as others have mentioned)

I've built dozens of bots, and have not fried a Pi yet.
n00blibrarian wrote:
Wed Aug 14, 2019 9:32 pm
I'm a relatively new raspberry pi educator (I was fortunate enough to get to go to a Picademy last year) and have been running a weekly robotics club for teens this summer where the kids have been using the 3B+ and an explorer hat to build a simple two wheel drive rover. This week we tried moving from 'bench testing' to actually letting the things loose and it has been one problem after another. The battery packs, which I bought from a hobby shop here in NYC at the last minute (after our original plan fell through) and which were specifically labeled as for the Raspberry Pi, ran the Pis but we got the 'low voltage' icon whenever we tried to activate the motors. Oh, well, the shop didn't have enough of them anyway, and the other type I picked up for the rest of the sets, which they assured me would also work fine, actually FRIED my personal Pi which I was using last night to run through the lesson and make sure everything worked, which meant the kids were going to have to share the ones we had anyway.

So we set those aside and started tinkering with the robots still connected to the AC power again. This is where things really went south. Over the course of a one hour session I had FOUR Pis die. Three of them were lighting up the red light only, and the third didn't light up at all. On all four I tried disconnecting everything from the Pi except the power to see if I could get a green light, and no dice. They all died at the same two stations, as when the first one went kaput I swapped in a spare with explorer hat still attached and left all the same peripherals and motors. In both cases, they worked okay for a while and then died again. Also during the session we had trouble at multiple stations where Pis were boot cycling - for those, I had the kids disconnect everything and start again from scratch, and that worked fine. But it was quite a mess, especially given how smooth things have been so far.

The really odd thing was that, because we didn't have enough batteries to start with, only one of these four devices had ever even been run connected to one of them, and all were on AC when they died. We've been running for over a month with no issues from any of the devices other than some problems with a few SD cards, and now this? Is there any easy way to do an 'autopsy' on a dead Pi to figure out what on earth went wrong with it? I'll need to buy at least two new ones to keep the club running and without knowing what went wrong I'm leery of spending more money. If I can't figure out what went wrong this club may be dead in the water, and I don't want to do that to my kids. Help!
http://Mikronauts.com - home of EZasPi, RoboPi, Pi Rtc Dio and Pi Jumper @Mikronauts on Twitter
Advanced Robotics, I/O expansion and prototyping boards for the Raspberry Pi

n00blibrarian
Posts: 3
Joined: Wed Aug 14, 2019 9:14 pm

Re: Four dead Pis in one class!?

Thu Aug 15, 2019 3:02 pm

I did have one thought of something else that was different this week: we had the kids plugging and unplugging the peripherals much more frequently as they switched between tweaking their code and testing the rover's movement. Could that be the issue? We do have some chromebooks and I may be able to set them up to work headless to reduce that wear.

Thanks to everyone who's suggested two battery packs. The honest fact is I'm a little intimidated by that. I thought I was safer using one battery pack connected through the USB port rather than trying to connect power through the GPIO pins, because I didn't want to overpower or short the board. But of course we've apparently managed to do that anyway! I'm still trying to figure out what we're going to do on that front because we do have to have something workable by next week. (I unfortunately can't share any of the pictures I took during the club for workplace policy reasons.)

One person on the educator forum suggested I double check all the SD cards so I'm going to do that as soon as I get a chance to work with them. Then I'm going to go through all of their kits over the next couple of days and see if I can figure out what they could have used to short the board. And I guess I'm swinging by the hobby store in the city again this weekend, both to replace a couple of the dead boards and to pick up whatever batteries I need to get it all to run.

User avatar
jbeale
Posts: 3478
Joined: Tue Nov 22, 2011 11:51 pm
Contact: Website

Re: Four dead Pis in one class!?

Thu Aug 15, 2019 3:31 pm

It is true that using the Micro-USB power input is safer than trying to connect +5V directly to the GPIO pins, where there's one way to do it right and lots of ways to immediately smoke the Pi. However that is a completely separate issue from whether you are using two physically separate battery packs. It is safer to use two battery packs because suddenly turning motors on and off can cause large voltage changes, which is not good for the computer.

That said, I actually have one of these https://www.dexterindustries.com/gopigo3/ which is made with a motor-control HAT that drives two motors and the Pi board from just one pack of 8 AA batteries, and it has worked with no problems. I have not been physically reconfiguring it since the initial assembly though.

alphanumeric
Posts: 2013
Joined: Tue Jan 19, 2016 2:17 pm
Location: Sydney, Nova Scotia, Canada

Re: Four dead Pis in one class!?

Thu Aug 15, 2019 3:43 pm

The Explorer Hat, which is what the OP is using, does not support running the motors off of a separate supply. The H Bridge is integrated into the Hat and uses the 5V supply from the Pi's GPIO. That also limits you in the motors you can use. Two H-bridge motor drivers (up to 200mA per channel; soft PWM control). I'm using the smaller pHat version of that. I'm driving DC motors in micro servo shape. Geared motors.

This is what the OP is using. Based on comments in the first post.
https://shop.pimoroni.com/products/explorer-hat

ejolson
Posts: 3423
Joined: Tue Mar 18, 2014 11:47 am

Re: Four dead Pis in one class!?

Thu Aug 15, 2019 4:34 pm

alphanumeric wrote:
Thu Aug 15, 2019 3:43 pm
The Explorer Hat, which is what the OP is using, does not support running the motors off of a separate supply.
The royal road to robotics seems paved with good intentions. I guess there is a reason why it takes four years of higher education to train an electrical engineer. Fortunately, amateurs don't need a degree and become surprisingly knowledgeable after a few mishaps.

I'm not sure what to recommend. In some countries one could try suing for damages under the claim that the Explorer Hat was unfit for purpose and did not follow best engineering practices. Since such a course of action is unlikely to be productive, it might be better to buy or build an H-bridge that works with a second power supply.

jbudd
Posts: 960
Joined: Mon Dec 16, 2013 10:23 am

Re: Four dead Pis in one class!?

Thu Aug 15, 2019 4:43 pm

On the Pi Gpio the 3.3V and 5V pins are side by side 0.1" apart. If you make an electrical connection between them for a moment, the odds are that the Pi dies. And there are other ways to instantly kill the 3B+

I suggest you replace your dead pi's with model 3B not 3B+.
4B is newer but it needs a different power supply.

Maybe also invest in some of these Gpio labels?
https://rasp.io/portsplus.
Put them over the Gpio pins and hold them down with a jumper between the two 5V pins to provide some insulation
https://www.amazon.co.uk/SODIAL-2-54mm- ... B071J6NRYV

alphanumeric
Posts: 2013
Joined: Tue Jan 19, 2016 2:17 pm
Location: Sydney, Nova Scotia, Canada

Re: Four dead Pis in one class!?

Thu Aug 15, 2019 7:26 pm

ejolson wrote:
Thu Aug 15, 2019 4:34 pm
alphanumeric wrote:
Thu Aug 15, 2019 3:43 pm
The Explorer Hat, which is what the OP is using, does not support running the motors off of a separate supply.
The royal road to robotics seems paved with good intentions. I guess there is a reason why it takes four years of higher education to train an electrical engineer. Fortunately, amateurs don't need a degree and become surprisingly knowledgeable after a few mishaps.

I'm not sure what to recommend. In some countries one could try suing for damages under the claim that the Explorer Hat was unfit for purpose and did not follow best engineering practices. Since such a course of action is unlikely to be productive, it might be better to buy or build an H-bridge that works with a second power supply.
Why blame the Explorer Hat? I haven't seen anything that conclusively points to it as the culprit?
And what it does is clearly stated on its product page. It will drive two motors, providing you buy the correct motors for it.
My two Explorer pHats drive my two rovers around with no issues. They have the same H Bridge the full Explorer hat has. They just don't have all the bells and whistles the Explorer Hat has.

alphanumeric
Posts: 2013
Joined: Tue Jan 19, 2016 2:17 pm
Location: Sydney, Nova Scotia, Canada

Re: Four dead Pis in one class!?

Thu Aug 15, 2019 7:35 pm

@n00blibrarian You mentioned unplugging and plugging in peripherals. If your treating the Explorer Hat as a peripheral you do not want to unplug it or plug it in with the Pi powered up. You want to shut down and remove the power to the Pi before attaching or removing anything connected to the GPIO.

Also, plugging that Hat in one pin off left or right, or one row off up or down will most likely cause issues like your having.

User avatar
Gavinmc42
Posts: 3633
Joined: Wed Aug 28, 2013 3:31 am

Re: Four dead Pis in one class!?

Fri Aug 16, 2019 2:56 am

Use two packs.
Motors can make huge voltage spikes that can kill sensitive electronics.
Are there any suppression capacitors across the motors brushes?
Those motor brushes spark and are basically spark gap radio transmitters.

Motors, relays, solenoids, inductive loads all need suppression.
Best to have two separate power supplies.
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

n00blibrarian
Posts: 3
Joined: Wed Aug 14, 2019 9:14 pm

Re: Four dead Pis in one class!?

Tue Aug 20, 2019 12:45 am

Thanks for all the help, folks. An update. I've tried all four boards with newly imaged and tested SD cards and the standard power supplies we got from adafruit and they all do seem to be dead. I still haven't figured out what went wrong or why four boards died all in the same session but my best guess is that it might have been extra wear and tear as the kids were handling the boards more than usual to try out their code without cords in the way? The two stations they were at were my two youngest teams so that might be a factor.

So what I've done is started setting them up to run headless from the chromebooks I use for our code club. If they were indeed just managing to twist something the wrong way, hopefully this will solve it. I really liked having the kids run them like desktop computers with a monitor, keyboard and mouse, because I feel it helps drive home that all the magic is happening on that little board, but they've got several weeks' experience with that now and this will add a new dimension to their work that I think is pretty cool and make the later stages of the club easier to boot. I'm also going to replace everything else that I can from those two sets just in case (jump wires, keyboards, whatever I have extra of.) I'm also going to have a quick 'care and feeding' refresher this week and make sure the kids aren't 'adjusting' the HATs or anything like that.

The advice to use separate motors is well taken but not possible with my current setup because the explorer hat doesn't allow it (another lesson learned!) But I did buy these power banks to run them from based on suggestions. I've tested them out, they work great.

But it's been an exciting week!!! I've learned a lot and made lots of notes! Thank you all so much for all the advice, this is such a helpful community!

User avatar
Gavinmc42
Posts: 3633
Joined: Wed Aug 28, 2013 3:31 am

Re: Four dead Pis in one class!?

Tue Aug 20, 2019 3:13 am

That Adafruit setup with the Zero shows two batteries, one for the Pi, one for the servos.
Servos are generally better than just gearmotors, but those servo type gearmotors might not have the suppression caps in them.

Forwards then reverse quickly can cause big power spikes.
Bigger power banks are no good if they cannot handle the motor currents.
Too high a motor current will lower the 5V output and glitch the Pi3's.
Plugging and unplugging powered parts can do the same.

In the automotive electronics world, "Load Dump" is the term for big spikes that electronics need protection from.
I think your Pi's got killed by "load dumps" or short circuits
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

Return to “Troubleshooting”