USB-ed smartphone camera precautions

[LesioQ]

Guys,

I have a feeling that noone here has fully expressed the vulnerability of smartphone add-ons when used as 'wired' PC accessories.
My sad and short expirience is related to Seek Thermal XR which died on me after some 4 hours of exciting operation with Windows PC.

I was playing with JoeC's great capture software, set up the camera atop the LCD monitor (as typical cams are placed),
and noticed that a few times when detaching the cam from USB (diy) cable for focusing, I got a small static spike.
Three times this caused lost link to live picture, so a reconnect was necessary. Those were the warnings.
The fourth time did the job. The cam gets identified (also in a smartphone) but the image is blank.

So now the camera is dead. Most likely the issue here is the fact that the cam is grounded by USB ground cable to the PC and further,
so it's a kind of extended lightning post. A smartphone environment being usually insulated from ground appears to be much safer for it.

Maybe this vulnerability was adressed by manufacturer (Seek) when additional conducting ruber post was added between camera case halves,
promoting grounding of the whole magnesium case, but mine was the old one.

So - take care of Your Seek Thermal Cameras when USB-ed.

[Micz]

That`s interesting. What cables adapters do you use to connect camera to PC ?

PS. mam na sprzeda? now? seek thermal pro

[Spirit532]

My camera only stops functioning when I send it malformed packets over USB, otherwise it seems completely fine.
It could just be your PC, check if the case is properly earthed.

[LesioQ]

The cable used was a generic USB-A extension (1meter) plus short DIY 4-wire non-shielded USB-A to micro USB socket.
Micro USB socket shell was floating (unconnected to ground).

I suspect that some part of camera housing (either of two) was not grounded, accepted the static charge and forwarded it not to the ground, but somewhere else.

Maybe if I could reload the camera firmware - things would revert to operational stage, as the cam is recognized, but shutter never starts to move.
But I have no idea how to force Seek to upload firmware when no new version is awaiting.
Or I could try ISP if testpoints are available inside ...

[Spirit532]

But I have no idea how to force Seek to upload firmware when no new version is awaiting.

Downgrade the app, run it without an internet connection. It should try to "upgrade" the firmware to an older version.

[LesioQ]

My measurements show that the camera itself has it's magnesium case connected to USB shell, but NOT to the USB ground.
This means that my DIY cable should also have (shield?) interconnection between USB connectors shells, which it lacked.

As for the firmware - I tried 1.9.1 apk, but noticed no version downgrade (neither Camera Software being 1.3.0.0 nor Image Processing Software being 2.1.0.0.).

[LesioQ]

One more interesting followup - as I'm digging a lot inside my Seek Thermal XR (the one working ...) I decided to protect the bonding wires going from PCB to VOX sensor.
I put a nice drop of UV curing glue, let it sink under the bonding wires, then cured for 1 minute with 405nm laser pointer then left for 20 minutes under blacklight fluorescent lamp.

The result ? Nice trasparent reinforcement of bonding wires ... and ... new frame rate of .... 0,2Hz. Yes, that is 5 seconds per frame, actually only single frame per shutter move,
both @ PC (JoeC) & Android.
The VOX sensor was covered with a piece of white paper during glue curing, but still this dose of UV must have done something bad to it. Should have used Aluminum foil instead ...

So now I have two sick Seeks 

 

[Spirit532]

Never, EVER encapsulate bond wires like that.
One slight drop and the whole glue setup will shift by an imperceptible amount, shearing the extremely fragile wires right off the die, and possibly the board.

It has to either be very soft selastic, or the whole device has to be encapsulated(through and around the board, around the sensor) with very rigid epoxy(like the one SMD ICs are covered in), to ensure no movement can occur.

Also, your glue may be conductive, and the UV light, if reflected into the sensor, could have severely damaged the ROIC and pixels themselves, because MEMS.

Send me your dead seeks, maybe I'll find a use for them 

[LesioQ]

The glue is not conducting, completely transparent, and not glass-hard, the bond connections are still visible & OK.
I bet that UV was the source of problems, and this makes me wonder 'how' ... ain't no EPROM ! 

Oh yes aside of frame rate drop, the temperatures reported went up by c.a. 20deg C.

[Spirit532]

The sensor likely has quite a few bytes of internal information, probably set at the factory, and the UV could easily have knocked out a few bits here and there.

[LesioQ]

SeekOFix software shows that only 1st frame after each calibration is displayed, however histogram shows some slight variations in between.
First two or so seconds after SeTh bootup (? or connect), when the shutter operation is more frequent, so is the frame refresh.

Makes me wonder if SeTh firmware is making some registry updates to the sensor itself ...

[amanda1]

I use usb 3.0 camera, I did not have any problems .. Try to reinstall the drivers, maybe something with them. this is just my opinion, if I understood you correctly...

[Uho]

1. Have you checked the voltage on the chips( FAN5362 )? The voltage regulator may not work.
2. Check the USB protection chip( ECMF02-4CMX8 ). Perhaps it is a breakage or a short circuit.
I do not think that the problem is in the firmware.

[LesioQ]

When playing with my partially-broken Seek XR and ThermoVisionJoeC 1.8.0.0 I noticed that when I switch to manual NUC -
the internal shutter action is kind of logically inverted, thus causing ghosting.

This is an interesting camera failure. In JoeC software - 'shutter open' - closes it, and reverse 
On the other hand - if I click "shutter Open", then "big NUC", then "shutter closed" - I get good live image - no ghosting, proper frame rate.

When XR's on Android - the sensor is being read with shutter closed and calibrated with shutter open.
This results in inverted palette, ghosting and low usable frame rate (1 full refresh per shutter cycle).

Any suggestions ?
As a reminder - this happened to my XR upon direct UV exposure months ago.

[LesioQ]

I made some measurements of shutter operation.
It is driven bipolar, i.e. when request is "Shutter Closed" - it's receiving 280mV constantly, then when "Shutter Open" is sent - this voltage is swapped to minus 280mV.
In other words - shutter is under current constantly, only current flow direction is inverted.

When camera is disconnected - the shutter is closed 'by means of natural magnetism' 

So, since I had noticed inverted operation of shutter in my camera - I swapped the coil winding ends on the shutter assembly -
so it's working the opposite way, resulting in original frame rate and positive palette range! 

That's a remedy I was expecting, but how the shutter switched it's expected action ?!   
Piotr.K

[pauledd]

There is a  "SET_SHUTTER_POLARITY" besides of the "TOGGLE_SHUTTER" command... Maybe the polarity got somehow
messed up via setting it on the software side, but I would assume that this would revert to the stock state after cutting power from the device 

[Ben321]

My measurements show that the camera itself has it's magnesium case connected to USB shell, but NOT to the USB ground.
This means that my DIY cable should also have (shield?) interconnection between USB connectors shells, which it lacked.

As for the firmware - I tried 1.9.1 apk, but noticed no version downgrade (neither Camera Software being 1.3.0.0 nor Image Processing Software being 2.1.0.0.).

The shield part of the USB cable is grounded on the PC end. If the case of the device is metal, and is connected to the shield part of the device's USB connector, the case will be grounded.