Author Topic: CCD Output conditioning  (Read 367 times)

0 Members and 1 Guest are viewing this topic.

Offline optotester

  • Contributor
  • Posts: 10
  • Country: be
CCD Output conditioning
« on: April 09, 2021, 02:20:03 pm »
Hi all,

I am willing to build a spectrometer and was thinking of using the following project as a basis :
https://hackaday.io/project/56937-stm32f4inputs

Basically, it is a linear CCD sensor connected to STM32 ADC through an opamp because the CCD does not output voltage in the correct range. However, the conditioning stage in this project looks quite complex. AD8027 being a rail-to-rail opamp, is negative voltage supply really required ?

Datasheet states that :
"The output stage is designed to drive 50 mA of linear output current, 40 mA within 200 mV of the rail, and 2.5 mA within 35 mV of the rail."
I intend to do software processing of the signal anyway so if opamp output is 0.1V-3.3V it would be definitely acceptable.

Is that possible to simply power the opamp from GND/+5V ? Or if AD8027 supply voltage really need to be symmetrical, is there any single-supply opamp that would be equivalent ? I would be happy to prototype but with 4Mhz signal breadboarding is not really an option apparently.

Kind regards
« Last Edit: April 09, 2021, 04:57:05 pm by optotester »
 

Offline Andree Henkel

  • Contributor
  • Posts: 25
  • Country: de
Re: CCD Output conditioning
« Reply #1 on: April 12, 2021, 10:38:33 am »
there is a chip especially for buffering CCD output (I have not used it myself):
ADA4800
https://www.analog.com/en/products/ada4800.html
 

Offline Pack34

  • Frequent Contributor
  • **
  • Posts: 712
Re: CCD Output conditioning
« Reply #2 on: April 12, 2021, 06:30:29 pm »
 

Offline Bassman59

  • Super Contributor
  • ***
  • Posts: 1854
  • Country: us
  • Yes, I do this for a living
Re: CCD Output conditioning
« Reply #3 on: April 12, 2021, 09:44:45 pm »
Hi all,

I am willing to build a spectrometer and was thinking of using the following project as a basis :
https://hackaday.io/project/56937-stm32f4inputs

Basically, it is a linear CCD sensor connected to STM32 ADC through an opamp because the CCD does not output voltage in the correct range. However, the conditioning stage in this project looks quite complex. AD8027 being a rail-to-rail opamp, is negative voltage supply really required ?

Datasheet states that :
"The output stage is designed to drive 50 mA of linear output current, 40 mA within 200 mV of the rail, and 2.5 mA within 35 mV of the rail."
I intend to do software processing of the signal anyway so if opamp output is 0.1V-3.3V it would be definitely acceptable.

Is that possible to simply power the opamp from GND/+5V ? Or if AD8027 supply voltage really need to be symmetrical, is there any single-supply opamp that would be equivalent ? I would be happy to prototype but with 4Mhz signal breadboarding is not really an option apparently.

A CCD output is a source follower. You provide a load resistance to set the output voltage, which will be in the range from ground to whatever voltage is applied to the source follower's drain.

This is only a small piece of the puzzle. You need to implement correlated-double-sampling for noise reduction, and then there's the small matter of clocking the CCD to actually get charge out.
 
The following users thanked this post: Pack34

Offline Pack34

  • Frequent Contributor
  • **
  • Posts: 712
Re: CCD Output conditioning
« Reply #4 on: April 13, 2021, 12:18:32 am »
Hi all,

I am willing to build a spectrometer and was thinking of using the following project as a basis :
https://hackaday.io/project/56937-stm32f4inputs

Basically, it is a linear CCD sensor connected to STM32 ADC through an opamp because the CCD does not output voltage in the correct range. However, the conditioning stage in this project looks quite complex. AD8027 being a rail-to-rail opamp, is negative voltage supply really required ?

Datasheet states that :
"The output stage is designed to drive 50 mA of linear output current, 40 mA within 200 mV of the rail, and 2.5 mA within 35 mV of the rail."
I intend to do software processing of the signal anyway so if opamp output is 0.1V-3.3V it would be definitely acceptable.

Is that possible to simply power the opamp from GND/+5V ? Or if AD8027 supply voltage really need to be symmetrical, is there any single-supply opamp that would be equivalent ? I would be happy to prototype but with 4Mhz signal breadboarding is not really an option apparently.

A CCD output is a source follower. You provide a load resistance to set the output voltage, which will be in the range from ground to whatever voltage is applied to the source follower's drain.

This is only a small piece of the puzzle. You need to implement correlated-double-sampling for noise reduction, and then there's the small matter of clocking the CCD to actually get charge out.

CDS is very important. Working with sensors like these are actually very simple. I've worked with a lot of them. Loading the output of the sensor with something simple like a 100k load, followed by an emitter follower, then gaining it will give you the data that you want. The Hamamatsu document I linked details the sources of noise you'll be fighting against. Dark noise exacerbated by the integration time and etaloning can be dominant sources. An optimized optical design around a transmission grating can improve your SNR by an order of magnitude.
 

Offline optotester

  • Contributor
  • Posts: 10
  • Country: be
Re: CCD Output conditioning
« Reply #5 on: April 13, 2021, 03:36:54 pm »
Thank you all for your answers, I forgot to mention that this particular linear CCD sensor (TCD1304) has a sample and hold function:
https://toshiba.semicon-storage.com/info/docget.jsp?did=13709&prodName=TCD1304DG

So it looks to me that the output is already kind of CDS output and that only amplifying/inverting stage is required to be added.
ADA4800 could be a good cost effective solution indeed, I will have a closer look.
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 9196
  • Country: de
Re: CCD Output conditioning
« Reply #6 on: April 13, 2021, 04:19:58 pm »
With a linear CCD one would likely not need such a fast amplifier / buffer. The data are supposed to come out at 1-5 ┬Ás per pixel.
A spectometer often has relatively low light intensity and thus more like a data rate at the lower end. Of cause light sources can be quite different.

Optics can gain quite a bit, e.g focussing the lateral direction.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf