1187 replies to this topic

[BGRE]

Hi again,
 
I hope it is okay when I ask a technical question relating to the electronics design of the camera!
 
I'm actually working on a CCD-Camera with the KAF1001 from OnSemi and I'm also using the AD9826 for sampling the video data (CDS-Mode). My schematics for the AD-conversion are pretty much the same then for the ICX453AQ-chip.
 
I'm having issues with the AD9826: When I connect the ADC to the output buffer circuit (after the KAF1001 output stage) the pixels in the beginning of every line have a higher intensity. This is shown in the camera picture and on the scope.
But when I disconnect the AD9826 from the output buffer the effect is totally gone - the KAF1001 outputs normal pixel values.
 
=> see attachments.
 
My questions:
 
- How do you clock the AD9826? Continuously or triggered? Do you clock the ADC only when you want to have an image (snapshot, triggered) or do you have CDSCLK1, CDSCLK2, ADCCLK running all the time?
- Did you ever had something similar when connecting your CCD with the AD9826 and if yes - how do you solve it?
 
Would be very great to have a response
 
Thanks so far!
Juja

What is the KAF1001 output amplifier connected to?
Details such as pulldowns etc required.
Which output amplifier did you use?

Edited by BGRE, 20 October 2016 - 04:49 AM.

[vakulenko_sergiy]

My questions:

 

- How do you clock the AD9826? Continuously or triggered? Do you clock the ADC only when you want to have an image (snapshot, triggered) or do you have CDSCLK1, CDSCLK2, ADCCLK running all the time?

- Did you ever had something similar when connecting your CCD with the AD9826 and if yes - how do you solve it?

 

1) Triggered, starts when image reading initiated from PC.

2) Yes, these is same issue on some cam8* cameras.

[Juja]

Hi again,

 

attached is a simplified schematic of my actual circuit...

 

I think I already figured out what is causing the issue: The AD9826 is not designed to use it in Snapshot-Mode. The time constant of the clamp network requires many pixel cycles to settle initially. So when you run the AD9826 with a longer "dead" duration the 4V (or 3V, depends on the internal register setting) clamping voltage needs some time to get stable.

 

Analog Devices recommends to run the AD9826 always with continuous clocking for CDSCLK1, CDSCLK2, ADCCLK, so I will try now with dummy clocks while "dead" time. I will come back to you with the results...

 

Here you can find a thread relating to this in the Analog Devices Forum:

 

https://ez.analog.com/message/251351

 

So maybe you can improve the CAM8 in the future

Attached Thumbnails

• 

Edited by Juja, 20 October 2016 - 06:37 AM.

[archimede]

 

hello everyone, I finally solved the problem of the vertical lines on the photos, I attach a bias to see now the image bias uniformity, was a disturbance created by the continuous switching of tps65130, its outputs + 15volt and -8volt

[Toups]

Hi again,

 

attached is a simplified schematic of my actual circuit...

 

I think I already figured out what is causing the issue: The AD9826 is not designed to use it in Snapshot-Mode. The time constant of the clamp network requires many pixel cycles to settle initially. So when you run the AD9826 with a longer "dead" duration the 4V (or 3V, depends on the internal register setting) clamping voltage needs some time to get stable.

 

Analog Devices recommends to run the AD9826 always with continuous clocking for CDSCLK1, CDSCLK2, ADCCLK, so I will try now with dummy clocks while "dead" time. I will come back to you with the results...

 

Here you can find a thread relating to this in the Analog Devices Forum:

 

https://ez.analog.com/message/251351

 

So maybe you can improve the CAM8 in the future

 

Cam84_v41 has jumpers for S1 and S2 and since S1 and S2 are used to derive the signals for CDSCLK1, CDSCLK2, ADCCLK via IC2 and IC4, I am thinking that a little external logic to keep toggling S1 and S2 might be doable without actually modifying the board.by using the existing jumpers to modify the S1 and S2 signals during the capture time.  The wild card may be the RG signal which is also derived from S1 and S2.

 

Of course, introducing extra signals during the capture could adversely affect the noise performance.

 

I need to finish the cooling and mounting of my camera before I think about improving it.

Edited by Toups, 24 October 2016 - 07:57 PM.

[Juja]

Hi Toups,

 

I was editing my FPGA code to have dummy clocks for CDSCLK1, CDSCLK2 and ADCCLK (so the ADC is continuously clocked now ) and the effect is totally gone.

To proof if your clamping voltage is stable you can just check in DC on the OFFSET-Pin (Pin25) of AD9826 if you have constantly ~4V. As long as your multimeter is showing a jumping voltage you might have a problem in the beginning of every frame because your reference sample is not correct.

Edited by Juja, 25 October 2016 - 04:43 AM.

[Juja]

1B_2016-10-11_21-47-00_Bin1x1_0.0001s__na.jpg

 

hello everyone, I finally solved the problem of the vertical lines on the photos, I attach a bias to see now the image bias uniformity, was a disturbance created by the continuous switching of tps65130, its outputs + 15volt and -8volt

 

Hey archimede, do you have any solution (e.g filtering) to get rid of the switching frequency of the regulator?

[BGRE]

1B_2016-10-11_21-47-00_Bin1x1_0.0001s__na.jpg
 
hello everyone, I finally solved the problem of the vertical lines on the photos, I attach a bias to see now the image bias uniformity, was a disturbance created by the continuous switching of tps65130, its outputs + 15volt and -8volt

 
Hey archimede, do you have any solution (e.g filtering) to get rid of the switching frequency of the regulator?

Try using something like the Murata BNX002.

[archimede]

 

1B_2016-10-11_21-47-00_Bin1x1_0.0001s__na.jpg

 

hello everyone, I finally solved the problem of the vertical lines on the photos, I attach a bias to see now the image bias uniformity, was a disturbance created by the continuous switching of tps65130, its outputs + 15volt and -8volt

 

Hey archimede, do you have any solution (e.g filtering) to get rid of the switching frequency of the regulator?

 

hello, I've used a simple electrolytic capacitor 470 mf for each output line of the tps65130, one on + 15volt and one on-8volt, that's all ..

[BGRE]

electrolytics aren't very effective filters at the ~ 1.25MHz switching frequency of the tps65130

[steveastrouk]

electrolytics aren't very effective filters at the ~ 1.25MHz switching frequency of the tps65130

Put a 100 nF ceramics across the electrolytics

[BGRE]

electrolytics aren't very effective filters at the ~ 1.25MHz switching frequency of the tps65130

Put a 100 nF ceramics across the electrolytics

Whilst that helps, it would be far more effective if a small ferrite bead were placed in series with the supply output.
However ensuring that any consequent resonances are well damped requires appropriate design choices.

[archimede]

[Attachment = 770,534: adesso questa CCD inizia a lavorare bene, sono 15 immagini di 600 secondi sommate, 2 ore e 30 minuti di integrazione con scuro 15 + 15 + 15 pregiudizi piatta, lavorazione con deepskystacher, telescopio riflettore Newton 250 millimetri f1200mm ammasso NGC 869-864]

Attached Thumbnails

• 

Edited by archimede, 30 October 2016 - 05:41 AM.

[Superdude68]

Thanks,  here is my email superdude.com and again thanks

[gehelem]

hi, i'm still following far behind (i'm just about to begin cam86),

but i'm happy to share my first sucess with my cam84 driver for indilib :

Edited by gehelem, 01 November 2016 - 12:17 PM.

[Toups]

hi, i'm still following far behind (i'm just about to begin cam86),

but i'm happy to share my first sucess with my cam84 driver for indilib :

I assume the latest code has been uploaded to https://github.com/g.../indi_cam84_ccd

[gehelem]

 

hi, i'm still following far behind (i'm just about to begin cam86),

but i'm happy to share my first sucess with my cam84 driver for indilib :

I assume the latest code has been uploaded to https://github.com/g.../indi_cam84_ccd

 

Yes, you must have libftd2xx library installed, and you have to set specific udev rules (see the .rule file)

[kalalaa]

I am really happy with the Cam84. I made HaRGB with it, and it's cool, but i think i need a monochrome camera. Can somebody make a monochrome CCD to me? I will buy it and send it to somebody who can debayer it.

The photo made with Pentacon 300 f4 lens. Ha: 15X10min RGB: 15X5min.

 

[Superdude68]

 

 

 

 

Has anybody have Gerber fields for the tec board and thermal controller please. Hello everyone and to remove the sensor from the base I used Chemset AP 1105 epoxy stripper that only took over night and all the epoxy was completely gone..

I have a TEC controller/usb hub design I built but I'm still testing it. It seems to work fine now but I need to change a few things on the schematic and then the pcb. After that I will share my gerbers.

 

Have you had any luck with the tec yet?

 

 

I have made some changes and it works ok now,  I can offer the gerbers and bom up but I won't take responsibility if it messes anything up on the camera or host computer. 

 

 

Gerbers and BOM for the AUX Board with TEC controller and USB Hub.

https://drive.google...WhIR3BRcUE/view

https://drive.google...EVvZ05meEE/view

 

Use at your own risk, not been fully tested but as yet I have not had it fail on me.

 

The Atmel chip will use the firmware that Vakulenko wrote.

 

See here for the link to his firmware and page.

 

http://astroccd.org/...03/tec_control/

 

Please could any one explain to me what everything is for on the tec board because i have just finished my one but haven't powered it up yet, so can anybody help.

PS i have no electronics know how so need help.

Thanks 

[Cyntrifical]

Been following this post, actually main reason I joined CN other than my telescope build, but anyone in the know would it be better to acquire a new icx453aq or just salvage from broken nikon d70s? I got a lead on new sensors from new strength electronics (china) for new packaged for $140 per pc and the price goes down the more sensors you purchase, but if it doesn't matter then a used nikon will be my sourcing.

[gregj888]

I was able to get a dead camera for about $32.  Sensor has been fine, but I'm somewhat experienced.

 

YMMV, and I think several have had troubles removing sensors without damaging them.

[vakulenko_sergiy]

Cyntrifical

As for me these is no difference where to get the sensor - from some used/broken DSLR or new one from replacement kit to DSLRs.

 

gehelem

Good new, many thanks for your work.

Do you mind if I publish on astroccd that there is INDI driver + link to your github?

[gehelem]

(...)

gehelem

Good new, many thanks for your work.

Do you mind if I publish on astroccd that there is INDI driver + link to your github?

Yes of course !
But please note it might be a little to early, as my code is very buggy and needs a lot of cleaning.
The big point is the recent change from D2XX to libftdi in order to keep compatibillity with other usb-serial converters (wich is lost under linux if you use D2XX)
I was quite happy with the result of the D2XX version, with only a few features missing.
But libftdi seems really more tricky, i'm actually messing around with clocks/timings/baudrates/latency problems.
The reading step is the key, i can't manage to synchronise correctly buffer writing with reading (proc "readframe" in the original code), i always loose some lines (that step used to work like a charm with the D2XX version...)
I must say that your message comes just in time, as I am a little discouraged :
I think it's time to request a little help, just to better understand the code and its purpose.

I think the good place to do that would be my github repository, wich i keep updated each time i have a significant progress. 

My humble goal would be to get it integrated to Indi    (of course with your permission)

Anyway : https://github.com/g.../indi_cam84_ccd

And of course, for those who don't know Indi on linux :http://indilib.org

Gilles

Edited by gehelem, 16 November 2016 - 08:32 AM.

[vakulenko_sergiy]

I think that there is a right time - software works and some publicity does not hurt. .

Hope that bug's will be fixed with common effort. Thanks!

[Toups]

I was able to get a dead camera for about $32.  Sensor has been fine, but I'm somewhat experienced.

 

YMMV, and I think several have had troubles removing sensors without damaging them.

I followed the suggestion of using a Chip Quik desoldering kit and found it helpful.

 

Plus breaking down and buying a "proper" soldering iron, namely a Weller WES51 certainly helped.  (Spent too many years working with cheap pencil irons.)

 

Also beware, some cameras are available because the sensor is already dead.  I had one truly dead sensor that had an internal short and it may have damaged U1 (which caused another long story and a painful repair previously documented on this site http://www.cloudynig...mera/?p=7140675 and the following post). 

 

I thought I had another dead sensor but it turned out to be an issue with my circuit board misbehaving at higher baud rates .  I wonder if this is related to a report on astroccd.org about cam85 design using a different chip due to speed issues with the MAX4428 used in the cam84.