On 05/02/11 10:09, Mike Harrison wrote: > High frame-rate sensors are always going to be niche and hence expensive = .. > AFAIK the Cypress one is the only 'affordable' one available - ISTR readi= ng this product line has > just been bought by another manufacturer so don't be surprised if it disa= ppears in the transition. Hmm, interesting. It's still listed as a current product on Cypress's=20 website. The Micron image sensor group was spun off a bit ago -- as Aptina=20 Imaging if memory serves. They don't really make anything particularly=20 fast, though (they had a ~200fps sensor a few years back, but it's NLA). > Could be worth talking to a Cypress franchised disti as for specialised s= tuff it is often the case > that parts can be had way cheaper than places like Digikey (e.g. FPGAs ca= n be 30-50% of the online > disti pricing) , although you may want to imply that you are designing so= mething for production, not > a 1-off to get them interested.. I've had bad luck with the franchised distributors. My conversations=20 with them usually go along the lines of: "I want two of these parts to evaluate. Send me a quote." "Bugger off." Price-wise, though.... - DigiKey: ordered on request, ETA back end of April. =A3216 plus VAT,=20 and free Spanish Inquisition treatment. - Mouser: none in stock, 16 on order, =A3219 +VAT, ETA early May then=20 three more at the end of May. =A3201 if I can get five interested folk=20 together and do a bulk-order (hmm...). - Avnet -- no stock in Europe, =A3370 (!) if supplied from the US, plus= =20 VAT and duty; =A3244 if supplied from Asia, and they supply from multiple=20 warehouses. VAT, duty and such to be added on top... So it seems the franchised distis aren't all they're cracked up to be...=20 (and Avnet are especially bad in my experience). The thing I'd be most worried about is screwing up the design or=20 assembly of the sensor board. Ruining a =A3250 image sensor would not be a= =20 great move. > Pretty much all image sensors have a clock+parallel data interface and I2= C or SPI control, although > for high-speed you tend to have multiple parallel interfaces for bandwidt= h. > I'd expect any newer devices to move to LVDS, but shouldn;t be hard to in= terface to a FPGA. Yeah, the Lupa-1300 (the 300's successor) has 12 LVDS differential lines=20 and a sync output. Tying that to an FPGA would bring you neatly into=20 Virtex / Stratix territory.... eurrgh. Bloody fast frame rate though,=20 especially if you enable windowing. > You may want to look at the sensors used in the Casio consumer digital ca= mas that can do high-speed, > if they are available. Or just buy one of the Casios... > http://exilim.casio.com/products_exf1.shtml I was looking at the Fuji HS10, but while it'll hit 1000fps, it only=20 manages that with a window size of 224x64. It's a colour sensor so=20 low-light sensitivity is, pardon my French, crap. > If you're interested in low light you definitely want a mono sensor as t= eh colour stripe filters > on colour ones lose a lot of light, and for high speed you need a LOT of = light..! Which neatly rules out the Casio and Fuji sensors. > Here's a thread on 4HV from someone who built a camera around the Cypress= : > http://4hv.org/e107_plugins/forum/forum_viewtopic.php?38831 That's more or less what I want to do, though with some form of cable or=20 fibre link between the camera and acquisition unit. --=20 Phil. piclist@philpem.me.uk http://www.philpem.me.uk/ --=20 http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .