USB Cams and Motion on Debian


79 posts   Page 2 of 4   1, 2, 3, 4
by ajts » Sat Jun 23, 2012 5:42 pm
Hi guys,

I got my Pi yesterday, very happy about that.

I want to put my Pi to use as a low frame rate (5FPS) video camera (Running on batteries), either with USB webcam or IP camera. Unfortunately I have never touched Linux other than running a Oracle DB on it years ago.

Any advice on how to get started would be greatly appreciated, for example, the name and version of a working combination of Linux distro, camera, driver and recording software..... (MJPEG ideally)

Hope it is OK to post this request here, many thanks.
Posts: 1
Joined: Sat Jun 23, 2012 5:35 pm
by cowpat » Mon Jun 25, 2012 9:53 am
If you haven't seen it already, Jeremy Blythe has a post on this on his blog that my be worth a look;

http://jeremyblythe.blogspot.co.uk/2012 ... otion.html


Personally I would go for the current wheezy-beta image, which should also support any UVC webcam... but feel free to experiment.

I'm trying to run mine on batteries too - but i'm outside wireless range, so at some point I am going to need a real time clock; but as I haven't got my video capture cards working yet, that seems a bit of a moot point.

Another task, because I want to use a big lead acid battery, is a low voltage cutoff protection circuit to avoid running the battery completely flat... i think this may ultimately come from a solar charge controller, but I have seen some circuits on R/C websites for lipo batteries that may work.

By way of an update of the latter - i have a kworld USB2800 producing a green screen too now. Both the EasycapDC60 and the KWorld use the em28xx driver. Interesting the KWorld isn't identified as a KWorld USB2800 and my attempts to correct that (i set 'options em28xx card=8' in a em28xx.conf file in modprobe.d) break it completely.
User avatar
Posts: 90
Joined: Sat Apr 14, 2012 12:13 pm
Location: London
by cowpat » Mon Jun 25, 2012 2:19 pm
This might be of some relevance.
An em28xx device producing a green screen on the shiva plug, (no fix though);

http://www.lavrsen.dk/foswiki/bin/view/ ... x10x101207
User avatar
Posts: 90
Joined: Sat Apr 14, 2012 12:13 pm
Location: London
by mikerr » Mon Jul 02, 2012 4:17 pm
I had no luck with webcams on either debian squeeze or arch,

but have got it working with debian wheezy (2012-06-18-wheezy-beta) :
http://www.raspberrypi.org/archives/1435

apt-get update
apt-get install motion
(successfully installed motion after a while - note it failed when I tried it before doing an update)

then edited motion's config file

su root
nano /etc/motion/motion.conf

change these lines to "off" in that file:

control_localhost off
webcam_localhost off

Start up motion program:

motion -n

then looked at http;//<raspi ip>:8080 and http://<raspi ip>:8081 on the PC

went though several old webcams I had lying around, and found two that worked,
which were the newest ones:

Logitech Pro 9000 (with excellent quality - but I want to keep that for skype on the PC really).
Microsoft NX-6000 (not as crisp but works ok)

both work powered directly from the Pi, but I prefer using a powered hub.

BTW I've added them here:
http://elinux.org/RPi_VerifiedPeripherals#USB_Webcam

anywhere else we should be adding compatibility lists ?
Got a Pi Camera? View it in my android app - Raspicam Remote ! No software required on the pi
User avatar
Posts: 1250
Joined: Thu Jan 12, 2012 12:46 pm
Location: NorthWest, UK
by cowpat » Mon Jul 02, 2012 6:27 pm
Interesting - those should have worked with Arch... that has the UVC module too. (at least the old image did - i haven't tried the latest one). What resolution/frame rate did you get out of the 9000?

I think I mentioned elsewhere that i've got my em28xx cards going now, albeit at a scintillating 320x240 - but it required a kernel compilation, a driver update and a patch; I would hesitate to call it working on both counts.
User avatar
Posts: 90
Joined: Sat Apr 14, 2012 12:13 pm
Location: London
by mikerr » Wed Jul 04, 2012 1:03 pm
They both happily go to 1024x768, but the Pro 9000 is just so much clearer picture and colour:
ImageImage
Got a Pi Camera? View it in my android app - Raspicam Remote ! No software required on the pi
User avatar
Posts: 1250
Joined: Thu Jan 12, 2012 12:46 pm
Location: NorthWest, UK
by cowpat » Wed Jul 04, 2012 1:48 pm
Enviable results... maybe i should go back to using a webcam.

Did you notice what colour palettes those cameras support? - just wondering if that could account for some of the difference in results?
User avatar
Posts: 90
Joined: Sat Apr 14, 2012 12:13 pm
Location: London
by SteveDee » Fri Jul 06, 2012 12:59 pm
I haven't made huge progress here. As my first project is a battery powered, "nocturnal" camera trap, I've put the EzCap to one side and have been concentrating on using a USB webcam (this is because my analogue cameras would require either 7.5Volt or 12Volt, but I want to use a 6V lead-acid battery).

My Advent camera works well with motion on RPi, but its sensitivity to infrared light is poor due to its iR filter. As this is a nice auto-focus camera that the family sometimes use for Skype, I'm reluctant to open it up and remove the filter.

I got quite excited when my son donated his old EyeToy camera (as I think I can remove the filter without wrecking the unit) but unfortunately this has the same instability that I experienced with Squeeze.

So I may either just have to wait for the gspca drivers to be fixed, or buy a cheap camera or two and hope I get one that works with Wheezy AND has an iR filter I can remove.

Below are my test results using guvcview for anyone interested.

On RPi + 2012-06-18-wheezy-beta release:-
=========================================

"Advent" auto-focus camera
ID: 0c45:62c0 Microdia Sonix 2.0 camera
Driver: uvcvideo
Performance: good, stable colour video, but only up to 352x288

Creative
ID: 041e:401c Creative Technology, Webcam NX [PD110]
Driver: gspca_zc3xx (and gspca_main)
Performance: unusable, unstable grey video, which stops after a few seconds

EyeToy
ID: 054c:0155 Sony Corp
Driver: gspca_ov519 (and gspca_main)
Performance: unusable, unstable colour video, which stops after a few seconds


On Lubuntu all 3 cameras work:-
===============================
Advent: good stable colour video (up to 1280x1024)
Creative: good stable colour video (up to 352x288)
EyeToy: good stable colour video (up to 640x480)
Posts: 138
Joined: Thu Dec 29, 2011 2:18 pm
by cowpat » Fri Jul 06, 2012 3:39 pm
It sounds like we have very similar objectives Steve - but I'm hoping to deploy my camera traps in wildlife tunnels for protracted periods; and with that in mind low light and IR sensitive bullet cameras appeal to me because they are really tough. (The colour ones were used for world rally in-car and 'kerb' cameras)

I have stripped down a microsoft lifecam cinema and removed the Ir filter, but it seemed to upset the autofocus so I removed the lens too and fitted a board camera lens. The inner plastic assembly then fits snugly into a short length of 1 1/4" aluminium tube. This is pretty solid, but it's manual focus, and I still have the issue of running the USB data into the tunnel.. and setting up the focus in situ could be an issue.. in a nutshell I would rather be playing with composite video in the outside world.

BTW Other things plaguing me at the moment; are under voltage protection for the battery and real time clocks.
User avatar
Posts: 90
Joined: Sat Apr 14, 2012 12:13 pm
Location: London
by SteveDee » Fri Jul 06, 2012 9:09 pm
cowpat wrote:....in a nutshell I would rather be playing with composite video in the outside world....


Yes, if your camera is remote from your RPi I agree that composite video is the way to go.

Due to todays rain, I found the time to strip down the EyeToy and remove the iR filter. Now its really sensitive, so I may only need 4 iR leds for illumination....plus a working driver.

BTW: I stumbled across Chris Boot's site and tried his 3.2 kernel (http://www.bootc.net/projects/raspberry-pi-kernel/). This didn't help my problems, but it did include an EasyCap driver...however I still couldn't get my device to work.

Under voltage protection: what do you want to protect? I have this vague idea that I may drop the power before the RPi goes doolally, maybe using a latching relay.

RTC: This probably doesn't help you, but it looks like the RPi keeps good time once its running. So I plan to start the camera trap connected to the net, before putting it into position.
Posts: 138
Joined: Thu Dec 29, 2011 2:18 pm
by cowpat » Fri Jul 06, 2012 9:41 pm
I posted a brief summary of the em28xx fix I found to work somewhere - but I put some notes on a blog too;
http://bufobufomagic.blogspot.co.uk/201 ... -sudo.html
These are really just notes for my own benefit, so I may have missed bits out. it may be that Chris Boot has a driver for one of the other easycap devices? The notes above concern the EM2860/SAA711X devices.

Regarding the RTC, I'm thinking about battery changes really. I'd like the operator (probably me!) to be able to switch the battery and restart the system without having to log into it. I've thought about a 3G modem... as uploading images remotely would be a pretty useful bonus, but I'm guessing that would be another 700mA i'd have to find..

From what I've learned thus far. If the voltage gets too low the gel batteries it will, at worst, kill them... it certainly impairs the number of recharge cycles. There are a few circuits out there.. i guess I just have to build a couple and see how they perform. (As a stay at home dad i'm really missing luxuries like bench power supplies... or a bench for that matter!)

I have found some nice wide angle IR LED's http://uk.rs-online.com/web/p/products/654-8334/
I've found three of them are ok in my confined spaces... and they don't give me any hotspots either.
User avatar
Posts: 90
Joined: Sat Apr 14, 2012 12:13 pm
Location: London
by SteveDee » Sat Jul 07, 2012 1:20 pm
Reading your blog I see that this is a serious application, whereas ours is trivial (we just want to plonk our trap by the compost heap and see what's routing around down there). Judging by your need for IP68 I guess you expect your tunnel to flood from time to time (or maybe its permanently flooded?).

Adding wireless will mean supplying more power, so could you run an IP68 ethernet connector to a suitable access point (e.g. vent or inspection hatch) so you could connect your laptop to extract video? I guess you've probably considered all possibilities.

EasyCap + 3.2 Kernel: I get those familiar lines, "Resizing pre_capture buffer to 1 items" & "Watchdog timeout, trying to do a graceful restart"
I'm also puzzled by my USB camera only running on low resolution on the RPi. I did try playing around with memory allocation (via start.elf) but that didn't seem to make any difference.

cowpat wrote:...from what I've learned thus far. If the voltage gets too low the gel batteries it will, at worst, kill them... .


Battery: as I'm a long way from getting a stable picture, I hadn't given any real thought to the supply. Having read this morning about sealed lead acid batteries, I now appreciate [some of] the problems. I'd completely forgotten that way back in the 1970s I'd developed a battery "power" meter for fork-lift trucks. So on reflection, I'm pretty sure that a simple voltage comparitor circuit will not adequately protect your gel battery, because its all about the state of charge rather than terminal voltage. It looks like sealed L-A batteries should only be discharged down to (say) 40% of capacity, otherwise their little lives are dramatically reduced.
The principle of my 1970s system was to run a clock into a counter, and increase the speed of the clock when the demand (current) on the battery increased.
Fortunately with the RPi, there should not be much variation in the battery load. So maybe the RPi could be programmed to disconnect the supply after 'n' hours via the i/o interface. However, I've no idea how you plan to manage your batteries (i.e. access, recharge, swap-over & so on) so maybe this does not help.
Posts: 138
Joined: Thu Dec 29, 2011 2:18 pm
by rleyden » Sat Jul 07, 2012 2:03 pm
SteveDee wrote:I'm also puzzled by my USB camera only running on low resolution on the RPi. I did try playing around with memory allocation (via start.elf) but that didn't seem to make any difference.

My Silicon Motion webcam-SM 371 (090c:b371) only gets usable images at 320x240. Setting frame capture to anything higher results in fractured images of with horizontal bands of grey. On other systems the cam gets beautiful images up to 1280x1024. This includes my Chumby with it's 454 MHz ARM926EJ-S processor and 64 MB ram. I've tried bumping up the RPi cpu clock to 900 MHz without any visible improvement.
Could the problems be the USB drivers? But, then why are some people reporting much better results?
Rich
User avatar
Posts: 54
Joined: Thu Jun 14, 2012 2:17 am
by SteveDee » Sat Jul 07, 2012 2:38 pm
rleyden wrote:...could the problems be the USB drivers?...


I just don't know, and I don't have enough experience of how the system hangs together at the moment. But here is my train of thought:-
- I have 3 cameras that all work on my Lubuntu laptop.
- On RPi the only one that works has a wide range of resolutions, but only works at relatively low resolutions (you have the same problem).
- The drivers used on my laptop & Rpi appear to be the same (same source code), just re-compiled for Arm.

So it seems to me that it probably is not the drivers that are the problem.

Good to hear you tried over-clocking, as its one less thing for me to try.
Posts: 138
Joined: Thu Dec 29, 2011 2:18 pm
by cowpat » Sun Jul 08, 2012 6:23 am
SteveDee wrote:I guess you expect your tunnel to flood from time to time (or maybe its permanently flooded?).


This tunnel shouldn't flood... but some of the tunnels they are planning to build on the site will be 'swim through' tunnels. That said, only the camera is going into the tunnel. My work in outside broadcast has left me with complete paranoia regarding of water ingress. I have looked at the Bulgin IP68 ethernet connectors - but if it's an ecologist collecting the data, I thought it might be easier if they can whip out a USB stick. The battery box won't be IP68 - but will be large enough not to get swamped - I did think about running data out to there... but i'd still like not to have to have a laptop on site.

SteveDee wrote:Adding wireless will mean supplying more power

I wondered about having it scheduled to upload periodically to an ftp site, and then turn itself off. Although it would be nice to be able access the hardware 24/7... the cost of that extra power would be quite significant. - a lot more than a £12 IP68 cat 5 port.

Yes, I have seen "Resizing pre_capture buffer to 1 items" & "Watchdog timeout, trying to do a graceful restart" a few times, particularly with the latest cx132xx driver and hauppage USB Live 2. The linuxtv drivers, with that one amendment, are working for both my easycap dc60, and the kworld usb2800. Even the kworld doesn't come up as a kworld device, as (even if it isn't a knock off) they seem to change the chips in the these devices on a whim.

SteveDee wrote:because its all about the state of charge rather than terminal voltage. It looks like sealed L-A batteries should only be discharged down to (say) 40% of capacity, otherwise their little lives are dramatically reduced.

I think i see the problem; by the time you hit the voltage drop it's too late. What I've read concurs with that - even if deep discharge batteries will mostly seem to function at lower than 40% but the deeper you discharge them the shorter their life. I've looked at solar charge controllers, which have some protection on their output. (although they might just knock off the power at 11v like most of the DIY circuits I've seen).

Isn't there a problem with estimating charge capacity if the battery efficiency is unknown? (although at a fairly low discharge I expect the efficiency to be quite high)
User avatar
Posts: 90
Joined: Sat Apr 14, 2012 12:13 pm
Location: London
by SteveDee » Sun Jul 08, 2012 9:55 am
cowpat wrote:...I thought it might be easier if they can whip out a USB stick...

...so what about an IP67/8 USB socket?

cowpat wrote:I think i see the problem; by the time you hit the voltage drop it's too late. ... just knock off the power at 11v...

I think 11V would be too late. Also, any ref voltage you pick requires an accurate & repeatable circuit to monitor it (e.g. as the ambient temperature changes, so will circuit values, especially resistors).

cowpat wrote:Isn't there a problem with estimating charge capacity if the battery efficiency is unknown? (although at a fairly low discharge I expect the efficiency to be quite high)

Yep, trying to assess the state of charge would be a problem if you are trickle charging via solar cells (what I had in mind was fully charging between use, then estimating the cut-off point).

The trouble with sealed LA batteries is they are full of compromises. Seems to me that the only good reason to use these in preference to traditional LA batteries is to avoid spillage. You may be interested in this: http://www.sterling-power.com/support-faq-2.htm.

I'm tempted to make a battery from a large number of AA Ni-cads...now there's a thought! At least the batteries would be useful after we got bored with the camera trap.
Posts: 138
Joined: Thu Dec 29, 2011 2:18 pm
by SteveDee » Sun Jul 08, 2012 10:59 am
What about Nickel-Zinc batteries: http://www.maplin.co.uk/nickel-zinc-aa- ... ack-624782

I'd probably need 12 batteries for 3-5hrs operation, connected something like this...

Ni-ZN.png
Ni-ZN.png (8.71 KiB) Viewed 20545 times
Posts: 138
Joined: Thu Dec 29, 2011 2:18 pm
by cowpat » Sun Jul 08, 2012 1:36 pm
Yes - there's a USB in the same range. I was thinking of mounting the pi in a small aluminium enclosure, inside a peli case - with a flying lead out to USB stick on a bit of velcro. This assembly to be mounted on the bottom of the peli case - so it's just a case of opening the lid to get to the USB stick. There would obviously have to be panel mount sockets in the peli case for power (unless I use a bigger case and put the battery in situ) and camera - i was thinking micro buccaneer or hirose for those. You can rapidly start spending more on connectors than on the processor, which seems a bit wrong?!

Sorry - I confused the issue throwing in solar power. It is a consideration - but the basic problem with solar is it would just get stolen. I am basically looking at swapping battery packs... given the low cost of the technical hardware (hopefully the raspberry pi - but whatever works) it might make sense to build two and swap the whole thing over.

That article talks a lot of sense. Avoiding spillage is pretty much the issue. We are pretty much just working in a field, and they are going to have to be transported about a lot. It makes it quite expensive though...

What do you make of amazon comments on Nickel Zinc?
http://www.amazon.com/PowerGenix-PGX-4A ... B0029LHXG2

Just the fact the Maplin sell them, and i couldn't find them on RS/Farnell does worry me!

You've really talked me out of lead acid chemistry though!

I looked at tagged NiMh D cells - They aren't so good in the cold (I believe that's the case; I remember reading it somewhere, but can't cite my sources) - and I've have to "modify" the TV remote a bit afterwards - but RS do a nice big one; http://uk.rs-online.com/web/p/d-recharg ... s/5114801/
User avatar
Posts: 90
Joined: Sat Apr 14, 2012 12:13 pm
Location: London
by SteveDee » Sun Jul 08, 2012 5:05 pm
cowpat wrote:Yes - there's a USB in the same range...

I was thinking of a capped IPxx USB socket, so you could insert a memory stick just to "milk" the system of files, and then remove it. Without a user interface you would need some clever code; maybe to be triggered by a push button switch, with an led to indicate file transfer completion.
cowpat wrote:What do you make of amazon comments on Nickel Zinc?

Wow! Imagine your IP68 Pi box inflating like a Zeppelin, then bursting into flames when one of your amphibian operatives lights up a Woodbine!

I guess its back to the drawing board. The tagged cells look very expensive, how many would you need?
Posts: 138
Joined: Thu Dec 29, 2011 2:18 pm
by cowpat » Sun Jul 08, 2012 7:30 pm
SteveDee wrote:
cowpat wrote:Yes - there's a USB in the same range...

I was thinking of a capped IPxx USB socket, so you could insert a memory stick just to "milk" the system of files, and then remove it. Without a user interface you would need some clever code; maybe to be triggered by a push button switch, with an led to indicate file transfer completion.


I don't suppose it would need to be that clever - I'm not sure how to read from the GPIO yet, but there seem to be plenty of tutorials to turn LED's on and off... I think, given the battery limitations, it's going to have be turned off every day/couple of days - so swapping the USB stick won't be a great hardship.

SteveDee wrote:
cowpat wrote:What do you make of amazon comments on Nickel Zinc?

Wow! Imagine your IP68 Pi box inflating like a Zeppelin, then bursting into flames when one of your amphibian operatives lights up a Woodbine!

I guess its back to the drawing board. The tagged cells look very expensive, how many would you need?


The were the biggest I'd seen - not sure if they are the most cost effective but; Hmmm, yes. Just to get to 12v is going to be expensive enough... and that would barely get through the night. I'd have to check the minimum power req for the bullet camera - but I think 8 cells (9.6v) might run it...
and three sets of them would run for a day. That's about £300 I think?


Whilst that might sound ludicrous - the current alternative is pitfall trapping, which means setting traps every morning, and collecting them every night.

There is a plan C; use a time lapse camera. Currently the 'off the shelf' time lapse systems (at least the ones I've looked at) have a minimum interval of 60s - this is too slow to reliably capture most of our amphibians. (I've got a table of amphibian velocities somewhere!)

PIR trail cameras have been tried with long toed salamander, but were shown not to be great for anything much slower or smaller.
User avatar
Posts: 90
Joined: Sat Apr 14, 2012 12:13 pm
Location: London
by SteveDee » Tue Jul 10, 2012 9:12 pm
I managed to borrow another webcam last night:-
ID 046d:08ce Logitech, Inc. QuickCam Pro 5000

...which ran fine at 640x480.

I then did something I should have done weeks ago. I took the lid off my Lego Pi case and measured the voltage at TP1 w.r.t. TP2. It was around 4.75Volts with this camera connected (...I then had to stop to watch "Blackout").

This evening I took some more measurements, conditions as follows:-
- Display connected,
- Keysonic combined keyboard/touch-pad connected,
- attached to network,
- booted to username prompt,
- TP1 to TP2: 4.85V

As above but Desktop displayed: 4.82V
Then with Advent camera connected: 4.79V
Then running guvcview with Advent at 352x288: 4.71-4.75V (varies with video content).

My power details:-
Power Supply for Huawei U8510 Android phone
Model: HW-050100B1W
5.0V, 1.0A

With a second, identical power supply I got very similar results.

I don't know the accuracy of my multimeter, so plan to at least get a second opinion. But I'm sufficiently concerned that I've ordered one of these: http://www.amazon.co.uk/gp/product/B0065JCIPU/

Its a pity that the RPi info here: http://elinux.org/RPi_VerifiedPeriphera ... r_adapters isn't supported by voltage readings.
Posts: 138
Joined: Thu Dec 29, 2011 2:18 pm
by cowpat » Wed Jul 11, 2012 11:40 am
That's interesting. I was using a 5v/1A supply for one pi and this; http://uk.rs-online.com/web/p/products/672-7155/ for the other - so I thought I'd be ok for power.
I'm getting similar readings - approx 4.85v with nothing connecting, then about 4.76 with the capture sticks connected. One of these is plugged in via a powered hub too - that doesn't seem to make any difference.

The output of the regulator is still measuring 5.04v with the capture device in.

I haven't been as scientific about my test. One pi is using hdmi output, the other is using composite video; neither have keyboards, or desktops running - they are both on the network. One of them had a usb hub with a memory stick connected.

My multimeter is a Fluke 111. It's a few years old, but it hasn't fallen out of a christmas cracker or anything. I'd be a bit more systematic, but i'm quite short on time today.
User avatar
Posts: 90
Joined: Sat Apr 14, 2012 12:13 pm
Location: London
by SteveDee » Wed Jul 11, 2012 12:35 pm
With my Advent camera running with low res (guvcview) the USB voltage is only 4.03V. (see measurement points)

I'm going to canabalise 2 USB cables so I can power both the RPi and my camera direct from the power supply, and then see if that makes any difference.

RPi_USB_Voltages.jpg
RPi_USB_Voltages.jpg (46.59 KiB) Viewed 20376 times


Don't understand this bit:-

cowpat wrote:...The output of the regulator is still measuring 5.04v with the capture device in...

...which regulator?
Posts: 138
Joined: Thu Dec 29, 2011 2:18 pm
by cowpat » Wed Jul 11, 2012 2:45 pm
SteveDee wrote:With my Advent camera running with low res (guvcview) the USB voltage is only 4.03V. (see measurement points)

I'm going to canabalise 2 USB cables so I can power both the RPi and my camera direct from the power supply, and then see if that makes any difference.


I've spliced the USB cable for a logitech 270 - to eliminate the need for a powered usb hub. I didn't notice any difference in results. Similarly, where I have added a powered hub I haven't seen an improvement in results. I haven't necessarily been looking at image corrupt, and I may not have tried increasing the image resolution of the logitech 270 after powering it independently, but one of my raspberry pi capture tests has been using a powered hub, and the other hasn't. There are no differences between them it terms of what resolution I can get away with before the image corrupts.

SteveDee wrote:Don't understand this bit:-

cowpat wrote:...The output of the regulator is still measuring 5.04v with the capture device in...

...which regulator?


Sorry - the 5v switching reg i mentioned that powers one of pi's;
http://uk.rs-online.com/web/p/products/672-7155/
I use a 12v supply, via this regulator, to power the raspberry pi.
User avatar
Posts: 90
Joined: Sat Apr 14, 2012 12:13 pm
Location: London
by SteveDee » Wed Jul 11, 2012 5:24 pm
OK, thanks for this info.

It looks like I've just wasted £7.99. I just ripped my Lego box to pieces and took out the board so I could flip it over. Measuring the incoming supply, I can now see that I'm dropping 200mV across the poly-fuse (F3). This pretty much ensures that most RPi users are running towards the lower voltage limit, unless they over-voltage the supply by chance or design. In fact having the USB poly-fuses in series with the main fuse is not a great idea.
RPi_PCB_back_F3.JPG
RPi_PCB_back_F3.JPG (58.69 KiB) Viewed 20356 times

cowpat wrote:I've spliced the USB cable for a logitech 270 - to eliminate the need for a powered usb hub. I didn't notice any difference in results....

Not sure on your arrangement, my plan was to separate the camera & Rpi supply connections at the USB [device input] lead. But I think you're indicating that this won't solve anything.
Posts: 138
Joined: Thu Dec 29, 2011 2:18 pm