Hi
Are there any known issues with the Pixy camera losing it’s signature data stored in flash?
We installed 4 Pixy cameras into 2 helicopters to track the cyclic and yaw pedal movement and they all worked fine for about 2 weeks.
They then stopped working and when I looked at them today it turned out that 3 of the 4 had lost their signature data. We only set a single signature (red coke cap).
Thanks
Had this issue again today with 1 of the 4 Pixy cameras. All of them are connected to Arduinos.
It was probably close to 0C at the time.
Hi Sean,
Hmm I haven’t heard of this happening yet. Did you re-program the Pixy’s by chance? Some of the firmware updates may lose your previous signature data if the data locations are re-mapped in memory. I’ll have to look at the temperature ratings of the processor to see what the normal operating range should be.
EDIT:
The only temperature info I could find for the LPC4300 said “Temperature range: –40 to +85 ºC”, which I assume would include the internal flash, so that probably wasn’t the problem. Under what conditions do you usually lose the signature data?
Thanks.
Scott
Hi
When we first received the Pixy cameras we updated all of them to version 0.1.44 and then set the signature on all of them to our red coke tops. We haven’t performed any further firmware updates since then.
However intermittently some of the Pixy cameras lose their signature data, and also protocol interface type, for example the ones setup for UART revert back to SPI.
It is an intermittent issue, but basically the helicopters fly during the day and then are put in a hangar overnight. And intermittently in the morning one or two of them, not always the same cameras lose their signature data.
Again this morning, 1 of the 4 lost their signature data. It was fairly cold again, -2C this morning when we switched them on.
Could some random noise on the SPI or UART connection from the Arduino trigger some code path on the Pixy to erase the flash?
Cheers
Hi
We lost the signature data again yesterday afternoon on one of the Pixy cameras when it was much warmer round 10C, so it may not be temperature related.
I have seen other requests for the ability to download and upload configuration data from/to the Pixy from a microcontroller. If that was implemented then we could use that as a backup, i.e. on startup our microcontroller could query the Pixy camera to check what signatures it has loaded and what protocol interface it’s using and if that data was been lost we could upload it from our microcontroller.
Cheers
Same problem on my Pixy: interface selection, I²C address and signatures are lost.
As I’m using the I²C bus with a specific address, this is really annoying, especially for an embedded system.
I have to reconfigure the Pixy through PixyMon and the USB cable.
Any hint? Thank you!
Hello world!
Im having the same issue. I have a bunch of robots that drive on a ground and look for things like purple dinosaurs.
After upgrading to a latest firmware, i noticed that once in a while my pixies (i have 3) would (randomly) loose color signatures
I dont think that i noticed this issue before upgrading to the new firmware, but again, i have not done much testing before upgrading to new firmware.
I am talking to pixy using i2c communication, default address. I have adjusted a lot of values in signature creation tab, upped the brightness a bit. I am running color codes. Temperature is not a factor in my setup. I am turning the power on and off to my Pixy rather frequently. I program my robot to do stuff and test it, and when the robot is not running, i turn the power off. The robot runs on a 3.7v battery which powers a 5v step up chip, thats where i get my 5v power to pixy. I think that 5v power output is well regulated, and the power is given to the power connector (± pins on that white connector).
Any ideas how we could fix this?
I was about to ask Rei Vilo which firmware version he is using to see if there is any correlation between firmware version and having this intermittent issue.
However I’m running firmware version 0.1.44, i.e. the released version before the latest beta firmware supporting color codes.
So based on your (Vadim) post this issue occurs with both the latest firmware version and the previous version (0.1.44).
In my case it’s also an ‘expensive pain’ since the helicopters cost on the order of $1,300/hr to operate and on flights when the Pixy camera goes awol we get no data.
I should also point out that Im doing all the pixy programming via PixyMon software. My pixy is set to recognize 2 color signatures and 3 color code signatures. When this issue happens, all of the signatures are gone, and so are my configurations. It looks like a factory reset, pretty much. Oh, and im using Arduino Fio to talk to pixy.
I’m using
@> version
Pixy firmware version 1.0.2
@
I guess either there is some sort of intermittent electrical issue which wipes out the flash memory storing the configuration data or there is a bug in the firmware being hit intermittently where maybe under certain conditions it fails to read the configuration data successfully from the flash and it then defaults to using the default configuration data. Although in this case it would also have to end up writing the default configuration data back to the flash since once the Pixy gets into this state it seems to stay in this default state.
So I guess we’ll need to review the firmware code that reads and writes the configuration data.
Hi everyone,
Thanks for all the info. If anyone has found a way to reliably reproduce this please pass it on to us. In the meantime we’ll see what we can find in terms of software bugs or hardware faults.
Thanks!
Scott
I just figured out how to erase the color signatures every time in my case.
Im currently using Arduino Mega, which talks to pixy via I2C bus. SDA and SCL pins on pixy are connected to SDA and SCL pins on mega (20,21)
Pixy gets its power from a 5V source. I am using a 3.7V lipo battery and a pololu step up voltage chip to get my 5V volts. These 5V power Pixy, a current sensor, and Arduino Mega (i run a jumper on VIn on mega).
Step 1) Disconnect 3.7V lipo battery from the source.
Step 2) Use a USB cable to teach Pixy some random colors via PixMon
Step 3) Disconnect USB cable from Pixy
Step 4) Plug in a USB Cable to program/power Arduino MEGA. As soon as this is done, pixy goes trough a complete reset
My 5v power rail is now getting voltage not from the voltage step up chip, but from the VIN pin on the arduino (and this is the only source). Actual voltage on the rail is 3.8V. There are 179mA going in to my circuit from the Vin pin (measured with Fluke multimeter), 165mA of which go in Pixy.
and this keeps on looping. Attempting to view raw or cooked video results in PixyMon crashing.
Intermediate fix to this problem is to make sure that Pixy is not connected to any other power source while its getting programmed via USB cable, so this is probably a hardware issue
Additional info:
I was trying to get pixy to reset by plugging + and - leads into a power supply and trying to vary voltage and quickly connect/disconnect it from power supply. That didnt seem to trigger any resets. If I had to guess, the issue is caused by either a lack of/ over current going to pixy. I didnt play with the current supply cause I didnt want to fry my pixy.
Also, as soon as I plugged in USB power to arduino MEGA, either BLUE or RED led would light up and stay on Pixy .I had my lens covered.
I’m also having the signature loss problem. Attempting to use Pixy with an Arduino Uno using firmware version 1.0.2beta and PixyMon for Windows version 1.0.2beta. I think I’m seeing a trend for the duration of time I leave the Pixy un-powered. If I set the signature via PixyMon and a USB cable, then immediately transfer it to the Arduino, the signatures are retained. However, if I let it sit un-powered over night, I notice all signatures are lost the next day when I turn it on. Anyone else notice this?
Hello bit head!
Just out of curiosity, how do you talk to your Pixy? are you using ISP pins (that 6 wire cable)? UART, or I2C? , If you are not using ISP, how are you powering your pixy (whats the source, and to which pins on pixy do you connect it to?)
I first started using Pixy with Arduino Fio. My pixy had no problems retaining color signatures for at least 3 days. I just switched to Arduino Mega, and will post another response if I observe the same thing in my setup.
I use the ISP pins to connect the Pixy to the Arduino. The Arduino is currently powered via the USB cable connected to a PC.
Im actually now having an issues where I lose signature data few minutes after my robot runs around and does stuff. My setup is the same as before except now Im using Arduino Due instead of Mega. This happens shortly after (<5 minutes) I taught my pixy some color signatures and color codes.
If this is a hardware issue, I suppose that it cannot be fixed easily. However, it would be nice if one could reprogram pixy on-the-fly 
Scott has any progress been made on this issue?
We’re going to start flying again in about a month’s time and I’d like to have a reliable Pixy setup.
Thanks
Hi Sean,
I’m not sure that we were able to reproduce this. I’ll ask around and see if there are any suggestions for preventing this from happening.
Scott
Hi Scott
Vadim mentioned above that he had a reproducible case, did you guys try his procedure to see if you could reproduce it?
1st prize would be to figure out the root cause and come up with a fix for it.
2nd prize I guess would be to provide an API for our microcontrollers to download the config info, signatures etc. from the Pixy for us to compare with saved versions on our microcontrollers and provide an API for us to update the config info on the Pixy. That way our microcontroller can detect the Pixy losing it’s configuration data and reprogramming it without having to connect a laptop via USB to reprogram it.
Cheers