CCD camera design problems

I designed and built my first CCD camera, using a TEC-cooled Hamamatsu
sensor and a TI/Burr-Brown CCD Signal Processor VSP3210. There are
three four-layer boards set in a U shape inside an aluminum box. The
timing comes from a PLD state machine and the whole camera is
connected via USB by way of a Cypress Easy USB controller. The
problem I am having is I get all this garbage noise in the digital
output. I can get an indication of light or dark but the noise is
overwhelming. The noise is usually just there when the sensor is
exposed to light. The thing doesn't behave at all like it should.
The timing looks like it should from the CCD and 3210 data sheets. I
did find out I needed to write the setup twice to the 3210 in order to
even get a signal through. I don't see the extreme noise on a digital
scope probling the CCD output, so I suspect I may not be sampling at
the correct places, it has a built-in double correlated sampler. All
the CCD voltages, both bias and clocking are correct. It is a cute
little camera and runs at -20C very steady, (I have a very good
cooling setup that is SO simple!) so I am anxious to get it working
and eventually build more cameras using different CCD's, including
back-thinned ones. Any help would be appreciated. I have access to a
four-channel digital scope and an 8-channel logic analyzer. Not a lot
of other test equipment here, on a budget!

Thanks,

Doug Hanz

Hi:

You need to verify your propagation delays. Clock drivers like the EL7202
introduce a 30 ns delay.

On the digital side you must make sure you respect the datasheets RD/WR
timings for your signals.

You may also have EMI related problem due to large loop areas for the return
signals. Check your three PCB design for problematic signal return paths.

You may continue this discussion offlist with me.


Pierre Cloutier




In article ,
says...

I designed and built my first CCD camera, using a TEC-cooled Hamamatsu
sensor and a TI/Burr-Brown CCD Signal Processor VSP3210. There are
three four-layer boards set in a U shape inside an aluminum box. The
timing comes from a PLD state machine and the whole camera is
connected via USB by way of a Cypress Easy USB controller. The
problem I am having is I get all this garbage noise in the digital
output. I can get an indication of light or dark but the noise is
overwhelming. The noise is usually just there when the sensor is
exposed to light. The thing doesn't behave at all like it should.
The timing looks like it should from the CCD and 3210 data sheets. I
did find out I needed to write the setup twice to the 3210 in order to
even get a signal through. I don't see the extreme noise on a digital
scope probling the CCD output, so I suspect I may not be sampling at
the correct places, it has a built-in double correlated sampler. All
the CCD voltages, both bias and clocking are correct. It is a cute
little camera and runs at -20C very steady, (I have a very good
cooling setup that is SO simple!) so I am anxious to get it working
and eventually build more cameras using different CCD's, including
back-thinned ones. Any help would be appreciated. I have access to a
four-channel digital scope and an 8-channel logic analyzer. Not a lot
of other test equipment here, on a budget!

Thanks,

Doug Hanz

you probably need to build some testability into the camera, if it's not too
late . Separate the main blocks and find ways to exercise them . You have
too many unknowns daisy chained right now. CCD seems to work but unsure of
timing and what it outputs, then your double correlated A/D sampling which
you didn't probably test in any way to see it really is doing what it's
supposed to do . Where exactly do you get the "garbage noise in the digital
output " ? I hope it's not on the USB interface, because there are too many
steps from your CCD to USB. If you're getting bad USB data , how are you
reading it ? Did you make a prototype and is this your first PCB layout or
is this the first prototype ? You might have all sorts of unexpected
problems if this camera never worked . Problems might range from slight
timing violations to reversing some bits in your PCB layout , power supply
glitches , ground loops , USB firmware , or who knows what else. Are all the
parts 5V ? Did you interface 3 , 3.3 , 5v parts correctly ? How are the
ground planes and power distribution ? Voltage regulation ? Having to write
the setup twice is definitely a red flag for something else wrong , unless
this is a documented bug of the 3210 .

Best Regards,
Matt Tudor

"Doug Hanz" wrote in message
m...
I designed and built my first CCD camera, using a TEC-cooled Hamamatsu
sensor and a TI/Burr-Brown CCD Signal Processor VSP3210. There are
three four-layer boards set in a U shape inside an aluminum box. The
timing comes from a PLD state machine and the whole camera is
connected via USB by way of a Cypress Easy USB controller. The
problem I am having is I get all this garbage noise in the digital
output. I can get an indication of light or dark but the noise is
overwhelming. The noise is usually just there when the sensor is
exposed to light. The thing doesn't behave at all like it should.
The timing looks like it should from the CCD and 3210 data sheets. I
did find out I needed to write the setup twice to the 3210 in order to
even get a signal through. I don't see the extreme noise on a digital
scope probling the CCD output, so I suspect I may not be sampling at
the correct places, it has a built-in double correlated sampler. All
the CCD voltages, both bias and clocking are correct. It is a cute
little camera and runs at -20C very steady, (I have a very good
cooling setup that is SO simple!) so I am anxious to get it working
and eventually build more cameras using different CCD's, including
back-thinned ones. Any help would be appreciated. I have access to a
four-channel digital scope and an 8-channel logic analyzer. Not a lot
of other test equipment here, on a budget!

Thanks,

Doug Hanz