1252 replies to this topic

[hippieua]

At Hippieua's request, I've added this to the post...

Bruce

 

My account on a dropbox was disabled due to high bandwidth, so until I will fix it here is a direct link to all the files/schematics/gerbers I've posted here.

All software and drivers can be found here.

 

Project description:

 

This project is about how to make CCD camera based on ICX453AQ chip. This chip is also used in astro camera QHY8PRO.

• Camera resolution is 3000x2000 = 6Mpx
• Square pixel 7.8 x 7.8um
• 16bit ADC with CDS
• Color RGB Bayer
• 28.4 mm diagonal
• USB 2.0 interface
• Readout time is ±2s

The schematic is very simple, so anyone who knows how to solder and read documentation can easily repeat this model.

CCD Sensor Sony ICX453AQ unfortunately do not have datasheet in open access. So only reverse ingineering helped to start the project.

 

This camera is based on FT2232H chip working in bit-bang mode, so no need to compile firmware or programming mcu. FT2232 alows to create 2 USB slave high speed devices(one for controlling camera, second for downloading image).

 

ADC – AD9826. Complete 16-bit Imaging Signal Processor with gain(0..63) and offset (-127 ..+127) control.

 

Virtical CCD Driver (Sony CXD1267) is controlled through shift register 74HC595.

 

Horizontal Driver is semi software. Based on dual high-speed mosfet driver MAX4428 wich contains one inverting and one non-inverting section. It drives 6 volts to CCD in counterphase. In future releases I think ill replace it with spesialized ISL55111 chip.

 

Camera uses several power supplies (LDO/DC-DC) and runs from 12V PS.

• 2 LP2985-33 for FT2232 and shift registers
• 2 LP2985-50 for  ADC digital and analog supply
• 1 LM2937-50 for DC-DC bias PS and nor-gate
• 1 AD3336ARMZ  with 6 Volt for horizontal register
• 1 TPS65130 for -8/+15 for CCD bias supply

FT2232 have ability to be self powered, but currently it runs from VBUS.

 

The camera works in next sequence

1. Generating bitbang impulses (around 24Mb)
2. Write it to output buffer of FT driver.
3. Start dedicated thread for reading on 2nd part of FT2232
4. Send output buffer
5. Waitng for input to be finished
6. Forming image
7. Goto 2

Cam84 is forth (Cam81/Cam83/Cam8s) major revision of this project and currently just works out of the box. No need for cuting routes or making airwires or any other fixes.

 

Unfortunately it do not have builtin TEC controller yet. So you have to make it yourself. But there is a possibility to control it from ASCOM driver.

 

Materials needed for the project:

1. CCD chip, Removing the CCD from the carrier
2. PCB & Schematic Part 1, PCB Part 2
3. Enclosure
4. UV-IR CUT filter (Astronomik or Baader f.e.)
5. Bill of materials.
6. 2-3 hours of free time.

In the next messages i'll describe in more detail each item.

 

And now couple of photos, just to let you know what i am talking about.

 

 

Bottom Side with sensor:

 

 

Top Side Placed In Gainta Case:

 

 

Single BIAS frame at +5C on chip:

 

 

Average SUM of 38 BIAS frames:

 

 

To be continued …

Edited by noisejammer, 30 May 2016 - 11:30 AM.

[hippieua]

At Hippieua's request, I've added this to the post...
Bruce

My account on a dropbox was disabled due to high bandwidth, so until I will fix it here is a direct link to all the files/schematics/gerbers I've posted here.
All software and drivers can be found here.

 

CCD Sensor

 

Camera uses ICX453AQ sensor. As i said before there is no datasheet in open sources. So we used reverse ingineering. Im not 100% sure, but it seems to work.

 

ICX453AQ sensor pinout:

 

This sensor is easy to get. Just buy used / not working Nikon D40/D50/D70/D70s.  95% that sensor will be in a perfect condition.

Than you can use this manual to get ccd pcb out of the body.

 

 

And that was the easy part.

 

The tricky part is to get sensor out of aluminum platform which is glued with very strong compaund. There are 2 ways: the fast and the safe one

 

Fast is to cut with hacksaw or dremel or smth else. Just make X-like cuts, remove aluminum and then remove remaining glue with a sharp blade. Use sandpaper after all.

 

 

 

 

And the safe method is to dissolve aluminum with NaOH or dissolve compaund  with DMSO(prefered). This will take up to 3 days. Afaik steveastrouk has success with DMSO. Maybe he can tell more about this method.

 

And be sure not to forget about ESD protection of the chip while working in a dry season.

 

After we got the sensor

[hippieua]

PCB and Schematic:

 

 

PDF is available for download HERE

 

PCB is a dual layer.

 

Top:

 

 

Bottom:

 

As you can see it has a lots of jumpers. They are on both power and signals lines. Made for easy debugging and current measurements in case smth goes wrong. In normal state should be closed.

 

Gerber files for production can be found HERE.

• Top layer
• Bottom layer
• Solder mask top
• Solder mask bottom
• Silk top
• Silk bottom
• Solder paste top
• Solder paste bottom
• DrillHolles (EXCELLON)

Note that Design Rule Check was completed with SeedStudio specification.

 

PCB Soldering:

 

First of all you will need to make a USB+Power Connector

Get "2X5P 2.54mm Pitch Dupont Jumper Wire Cable Housing Female Pin Connector", USB Cable and Power supply 12V (Camera will need max 500mA).

 

Connector pinout:

 

Its realy good idea to use selected pin as input for 12V, believe me In case you put the connector upside down, 12V will go to VBUS and will be lowered to 3,3V. So no harm for FT2232. The rest 12V pin can be used for TEC or coolling.

 

The next goal is to solder FT2232(IC7), EEPROM(IC6), U7 for power and all elements(resistors, capasitors, filters) dedicated to these ICs. WR# Jumper is also needed. After you finish, you can plug the board to your computer(no need for external power at the moment). A new device on PC should be detected. Win7+ will automatically download and install latest driver.

Note that i didn't test camera on the latest drivers. I use CDM-2.08.24-WHQL-Certified.zip

 

After you get 2 new devices, double click each of them and uncheck Load VCP.

 

 

Then you will need MPROG Utility and template.

 

Extract template from zip, launch utility, open extracted template and "Program all existing devices"

 

 

After that you need to unplug the board and plug it back. That's all. Now ASCOM driver can work with the board.

 

I assume that ASCOM platform is installed, so now you can install ASCOM driver for the camera.

After installing run your favorite application for imaging(fe Maxim DL)

 

Setup Camera->Choose ASCOM->Advanced->Choose Cam8s

 

 

And make a test shot. If everything is ok you can go to solder all other devices. If no, please double check that WR# jumper is on.

Edited by hippieua, 16 April 2015 - 11:25 AM.

[steveastrouk]

Removing the CCD from the carrier is easy with DMSO.

 

Dismantling the camera to remove the CCD block is simple. I then used full anti-static precautions to remove the existing circuit board from the chip. THEN I bonded all the CCD pins with a single piece of wire, which I could loop around the carrier, so that when the carrier was removed, the chip would come free with the bonding intact.

 

A note on DMSO.

DMSO is a very heavy solvent, its boiling point is obscenely high, and it is very persistent. DO NOT WEAR NITRILE GLOVES if handling DMSO - it dissolves nitrile instantly, and passes DMSO  and the nitrile straight through the skin. If you absorb DMSO through the skin, you will taste garlic for a week.

 

We placed the CCD, immersed in DMSO, in a safe place for around 4 days. At the end of the period, the CCD was completely free of the carrier.

Remove the CCD from the DMSO and wash off the solvent with acetone.

 

Your chip is now ready for the new PCB.

[steveastrouk]

I am prepared to get a run of 25 PCBs, to HippieUAs design run off, if there is enough interest. I will also supply a kit of parts bought in 25 off quantities. There will be no mark-up, and the discount savings on 25 pieces, where available, will be passed to the group.

 

Steve

[hippieua]

PCB Part2:

 

After the successful launch of FT chip, you'll need to solder all the power supplies: U1-U6 (U7 goes with FT2232). 

U1 generates -8V/+15V for the sensor(IC3) & Vertical Driver (IC5)

U2 generates +6V for horizontal driver (IC8)

U3 generates +5V for U1 & IC4

U4 generates +5V for analog power ADC

U5 generates +5V for digital power ADC

U6 generates +3,3V for IC2 & IC5

 

Double check all the voltages before placing ICs. Also good idea to measure current on +12V JP. Check supplies temperature, even by finger tip f.e. All the chips have to be cold. Also a good practice will be to use oscilloscope or multimeter with true RMS to be sure all power rails are correct voltage, steady and noise clean.

 

Next stage is to solder all the rest of chips except the sensor. We don't need CCD at the moment, besides that its really hard to solder it out and its very sensitive to overheat. When you'll finish - put all the jumpers on. And we are ready for pre test.

 

Plug it to PC and turn the power on. If there is no smoke - my congratulations All the supplies have built-in protection, but s**t happens. Check that all chips are not hot. Repeat voltage and current measurements. There should not be dropdowns or smth. If so - unplug and look for short circuit.

 

Assume that you have connected the device and there is no smoke, no heat and no noise coming out from the board -  its time to run the test. Launch you imaging software, connect the camera and start continuous shooting. If you have oscilloscope you can check WR#/SL/SL2/SCK/SDA/S0/S1/S2 signals on jumpers and H1/H2/V1/V2/V3/V4 signals on CCD pads, they all have to be alive with some data. If everything is connected in a right way you should get an image full of noises. You can event touch capasitor C56 with a probe while exposing the frame and you should see a different noise patterns in response to this action.

 

If everything is OK - unplug and solder the sensor. Make couple shots to see if it reacts to the light bulb, put it into closed box to evaluate noise on biases. 

 

If no - recheck signals, pads, joints, short circuits, etc until you find and fix the problem.

 

If you are not good at soldering tiny parts i recommend to watch tutorials on youtube.

 

EEVblog #180 - Soldering Tutorial Part 1 - Tools

EEVblog #183 - Soldering Tutorial Part 2

EEVblog #186 - Soldering Tutorial Part 3 - Surface Mount

 

Good luck!

Edited by hippieua, 16 April 2015 - 12:06 PM.

[hippieua]

Bill of Materials:

 

Download excel file.

 

Transistor Q1 (IRLML2803TRPBF) is used at CCD Out. Actually this is not the best idea to use MOSFET at this place. So probably you can do some experiments with high freq transistors. Here is my wish list:

• BFR92A.215
• BFR93A.215
• 2SC3356
• BFS17A.215
• BC847B.235
• MMBT3904.215
• MMBT3906
• FMMT617TA
• BFQ67

 

Resistors R29, R30 are used in Horisontal Register. Their value effect on the horizontal stripes that may occur. You need to find value experimentaly, from 0R to 100R, or leave unchanged if everything is Ok. May be of different values.

 

Edited by hippieua, 16 April 2015 - 03:15 PM.

[wasyoungonce]

Oh Lord I am thankful for your bounty...yep count me in!

 

I have been watching this progress in the sourcing CCDs for DIY cameras.   Well done indeed and can I get one pleeeze!  I have soldering and SMD expertise.

 

Brendan

Edited by wasyoungonce, 16 April 2015 - 06:20 PM.

[hippieua]

Enclosure:

 

I designed PCB for Gainta G107 enclosure. It's widespread here in Ukraine, but don't know actually about US market.

I have found some on ebay.

 

I wont post here any blueprints about cnc works for the box, because i do not have them. Just look at the pictures

 

Copper finger used to transfer cold from TEC that is located outside of the box to sensor

 

Place for TEC, wires and coldfinger

 

 

 

Hole for sensor. Will be sealed with UV-IR Cut filter

 

 

 

[steveastrouk]

I have 5 takers for kits of bits so far.

[hippieua]

First light Selfie on 50mm canon lens

 

[wasyoungonce]

I have 5 takers for kits of bits so far.

Message sent

 

Brendan

[wasyoungonce]

First light Selfie on 50mm canon lens

 

 

Excellent...but...looks kinda "Cone head"!

 

edit: Just got a cheap working D70s from fleabay...lets hope it's all ok.

Edited by wasyoungonce, 19 April 2015 - 04:40 AM.

[Oberon]

Dumb question...I already have a D70. Why not just use the D70? What does this project do better?

[hippieua]

Dumb question...I already have a D70. Why not just use the D70? What does this project do better?

 

Your D70 in a 14(or maybe even 12) bit camera without temperature stabilization, cooling with very narrow range UV-IR cut filter. That why  

[hippieua]

Enclosure almost finished

 

 

 

 

[steveastrouk]

6 takers so far....

 

 

Dumb question...I already have a D70. Why not just use the D70? What does this project do better?

 

Your D70 in a 14(or maybe even 12) bit camera without temperature stabilization, cooling with very narrow range UV-IR cut filter. That why

 

 

UNcooled.

[Oberon]

OK thx. Missed the cooled bit. Duh!

 

Another question. Why THIS chip? I ask because I was generally disappointed with my Nikon D70 with respect to color and sensitivity (daylight photography) when compared with, say, the compact Panasonic I purchased a year or two later. So I was surprised to see it picked out for this sort of project.

 

Sorry if I appear to be a total newbie on this, as I am in some ways, but you've piqued my interest and I have a bit of catching up to do.

[hippieua]

OK thx. Missed the cooled bit. Duh!

 

Another question. Why THIS chip? I ask because I was generally disappointed with my Nikon D70 with respect to color and sensitivity (daylight photography) when compared with, say, the compact Panasonic I purchased a year or two later. So I was surprised to see it picked out for this sort of project.

 

Sorry if I appear to be a total newbie on this, as I am in some ways, but you've piqued my interest and I have a bit of catching up to do.

 

As i said before not only cooling, but 16bit too.

 

What about the sensor
1. its was tested with lots of QHYs in astrophotography
2. Its easy to get. Despite the fact that new sensor costs like $450, you can easily get used/not working camera for $30 max.

[wasyoungonce]

I have to ask...any chance of removing the sensor Bayer matrix to make it monochrome?  I heard this was reasonably easy on Nikon's.  Of course this will effect drivers of formatting of the image.

 

Brendan

[steveastrouk]

I have to ask...any chance of removing the sensor Bayer matrix to make it monochrome?  I heard this was reasonably easy on Nikon's.  Of course this will effect drivers of formatting of the image.

 

There's certainly a chance. I'll let you try it first. DMSO is such a powerful solvent, I do wonder if it will remove the matrix, but I also wonder what it does to silicon.

[wasyoungonce]

Ahh....yes I'll build it first then attempt to wreck it... The draw back is you loose microlense from the sensor when removing the bayer.

 

Yes I'm sourcing DMSO locally, appears not to hard to get but I did see some with low purity ~35%...would this suffice as it's cheaper?  I can also get 70% and 99% but price goes up.

 

Brendan

[steveastrouk]

Yes I'm sourcing DMSO locally, appears not to hard to get but I did see some with low purity ~35%...would this suffice as it's cheaper?  I can also get 70% and 99% but price goes up.

 

I know we used pure stuff, because we have an analytical lab, but I don't know if less pure stuff works as well.

 

There are now 7 in on the deal, and I'll buy parts when we have 10 of us.

Edited by steveastrouk, 20 April 2015 - 12:57 PM.

[mdavister]

This project looks really interesting.  Great work on the design!  What layout software was used for the PCB?

[Raginar]

What's your cost at for the kits? Or are we locally sourcing the sensors?