Firstly, thank you for sending for these instructions.
For your records here are the terms and conditions you have agreed to.
Right that bit over, here is the info!
I have divided this mod up into 3 stages. Stages 1 and 2 are pretty simple with stage 3 being for those who like to push the envelope!
1, Replacing a
ICX098 with a ICX424 (SC3.1)
The first stage is to buy a ICX424AL (B/W) or ICX424AQ (Colour CCD untested in this mod). The only place I think the CCD's can be bought in single quantities in Framos http://www.framos.de/www.dir/en/start_en.html
On
the face of it
replacing a 098 with a 424 is not as simple as removing one chip and sticking the new one in its
place! For one thing the ICX424 has more pins!!!
The
datasheets for the 2 CCD's are can be downloaded by following these links. ICX098
and the ICX242.
Here is a quick list of the differences that need to be addressed.
i.
ICX424 has 5V drives to horizontal and reset gate compared to 3V for ICX098
ii.
ICX424 has 3 vertical
clocks vs 4 for the IXC098
iii.
ICX424 needs -7.5V for
vertical clocks vs -5.5V for ICX098
iv.
ICX424 has charge
transfer pluses (CTP's) on V2 and V3 compared with V2a and V2b for ICX098
v.
ICX424 has slightly
different clock shapes at frame transfer to ICX098.
vi.
ICX424 has extra
ground pin and needs a separate substrate voltage and amp gain termination.
The
solutions to these issues are:-
Increase horizontal
clocks to 5V without adding to much delay that the timing is messed up (less
than 10ns).
Transfer the V3 signal
to the vertical driver to the V2b input, keeping the transfer signal on V2b,
then use the hybrid clock to as V3 for the ICX424.
Use charge pump and z-ziode
to generate the -7.5V supply and replace the -5.5V in the webcam.
3 logic chips can be
used to reshape the ICX098 clocks to look like those in the icx424 datasheet and
put V3 at 0V as required for CTP .
Because of the extra
pins etc a capacitor and a couple of extra connections are needed.
Over
Christmas / new year 2002/3 I managed to build circuits addressing the above
issues and was well please when to find the ICX424 would run happy in a webcam.
As you can probably imaging this mod is a bit on the complex side.
But
there is good news. When one part of the circuit started acting up I
removed it and found the CCD would still run. I then removed other parts
one by one to see how simple the mod could be and still work. The answer
is very simple indeed!!!!!
I have
found the ICX424 will run with
3V clocks to
horizontal and reset gates
-5.5V vertical clocks
Only needs one
transfer pulse on either V2 or V3.
In other
words the ICX can be added to a webcam with only the addition of a single
capacitor!!!!!!!!!!!! :)
(It needs to be said though that the CCD is no longer being run within
spec. Also the higher frames rates may not be usable)
ICX424 Pin |
TO |
1 |
Pad 2 |
2 |
Pad 3 |
3 |
Pad 1 |
4 |
No Connection |
5 |
Pad 9 |
6 |
1000pF Cap to Pad 6 |
7 |
Pad 6 |
8 |
Pad 7 |
9 |
Pad 8 |
10 |
Pad 8 |
11 |
Pad 9 |
12 |
Pad 10 |
13 |
Pad 5 |
14 |
Pad 12 |
15 |
Pad13 |
16 |
Pad 14 |
Also see
fig 1.
If you
want to solder a ICX to a webcam board then try aligning up pin 16 with pad 14
and you will find most of the pins on one side of the CCD line up OK.
Top view of ICX424 on a QC4000 PCB |
Pins 1-8 |
Pins 9-16 |
A ICX424
equipped webcam is very sensitive without further modification. If you
need a camera for use as an electronic finder or guider then further
modification may be unnecessary.
2
Long exposure Modifications (SC3.2).
More good news regarding simplicity. The ICX424 webcam can be modified to take long exposures using any of the existing basic (SC1) webcam mod circuits (Links here and here). As the ICX 424 doesn't support separate field read out the self guide mod will still produce only whole images ie no self guide frames. See later about amp switching.
A
dead bug mod to the qc4000
3 2x1
Binning (SC3.3).
Well its
all been a bit too easy up to now! If you fancy a bit more of a challenge
then this is for you.
By
adding two rows of pixels together in the CCD's horizontal register the camera
can be made about twice as sensitive at the expense of some vertical resolution.
To achieve this effect two approaches are possible. Either clocking the
vertical clocks twice each row, or stopping the horizontal clocks every other
row. It's the latter method that is used here.
In use a
flipflop keeps count of odd or even lines. This counts the signal to the
vertical driver (V2b) that occurs once per line at a time when the horizontal
clocks are inactive. In order to sync the counting to the top of each
frame the flipflops reset is connected to the webcams 'vertical' signal.
The system is such that when the non imaging parts of the CCD are cleared the
horizontal clocks are allowed to run regardless so clearing charge from the CCD.
A NAND gate is used to either let the clock pluses pass or to put the clocks
into there inactive state. In order to meet the very tight timing
requirements H2, which is slightly advanced to H1, is used as the only input.
So H1 is formed form the inverted H2 which is then inverted again to become H2.
There is a subtle complication using logic devices to generate the
horizontal clocks. The signals are generated as logical high or low while
the CCD is an analogue device. The effect is the logic signals are quite
dirty contain oscillations at transition which are best reduced. Using
ferrite inductors on the lines the signal quality can be improved resulting in
less readout noise.
Now the
only additional problem is the optical black and dummy pixel clamps to the ADC.
These signals set up the offsets to the voltage presented to the ADC. If
these clamps happen without the horizontal registers being clocked then the
video voltage values are not valid and will result in the picture being
digitised in a much compressed range. By also putting these clamps though
a NAND filter we can strip out those that occur when the horizontal clocks have
been stopped. The timing requirements here are not as tight as for the
horizontal clocks themselves.
As the
webcams chipset works at 3.3V and we need high speed clean signals its best to
use a logic chips from a 3.3V family. There are several surface mount
possibilities but less choice in the easier to handle DIL. The AC family
is sold in DIL and is suitable for use here.
Here is the schematic for the circuit.
Points marked * are good places to add inductors. 5V can come from the USB supply and power a 3.3V regulator. Use decoupling caps across all logic chips.
Pin 96 SAA8116 (ToUcam PCB) |
Pins for the clamps (ToUcam PCB) |
When a 2x1 binned image is retrieved from the webcam it will have alternate image and black lines. The latest version of desire can restore these on the fly or programs like paintshop pro can be used to de interlace the image.
NB Although this circuit allows the camera to produce a picture that is nearly as noise free as the original its still does add some noise to the image. Obviously stacking images will reduce this further. However if you plan to use the camera mainly at hi resolution you would be best not to go beyond stage 2.
Amp Switching.
The amp glow from the ICX424 is rather less than that from the ICX098 used in the webcams. Also as the CCD is much more sensitive there is not much to be gained from exposures lasting a minute or more. For these reasons I recommend you try the camera without a amp switch. If later you would like to add one then the 'standard' methods of dropping the voltage across the CCD will work. The ICX424 does need more volts to keep functioning then the ICX098 (the circuit below leaves the ccd running at 5.3V during integration). The drawback of adding a circuit to reduce amp glow is adding electrical noise to the image. In an attempt to minimise this I have used a relay to keep the analogue and digital sides to the webcam circuitry as separate as possible.
The circuit is here
And results from 120sec and 15 sec tests are here and here
Dark are dark frames and lights are pictures of a super glue bottle (sideways). As the darkest corner of the room was beside the PC exhaust fan and a wall the camera was quite warm. As you can see there is still a bit of glow on the top of the frame. I would guess this is the reset gate transistor which is still running.
Also see the circuit used by Matthias Meijer here
More information on amp switching can be had by following these links