Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

ZWO Camera Specs: Gain versus Gain

  • Please log in to reply
23 replies to this topic

#1 Farzad_K

Farzad_K

    Mariner 2

  • *****
  • topic starter
  • Posts: 226
  • Joined: 07 Oct 2016
  • Loc: Hillsboro, Oregon, USA

Posted 26 October 2020 - 01:03 PM

Hi Folks.

 

I have discussed this question with ZWO and I don't have a good answer from them that I can understand.

 

The performance graphs show the horizontal axis as Gain (0.1 db) and in one of the graphs there is a vertical axis also called Gain in units of electrons per ADU. I know the Gain shown on the horizontal axis is the value I set in the camera driver, and I know what the Gain e/ADU is.

 

Can someone help me understand the horizontal axis?

 

Thanks in advance.

 

Farzad

Attached Thumbnails

  • 2020-10-26_10-55-35.jpg


#2 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 21,894
  • Joined: 27 Oct 2014

Posted 26 October 2020 - 01:09 PM

Don't quite understand the question.

 

Gain on the horizontal axis is simply the (artificial) number to input into the software, it has no other meaning or purpose for you.

 

The vertical axes tells you what the number means.  What is the full well capacity at that number.  How many electrons you get when one photon is collected.  What is the dynamic range at that number.  What is the read noise at that number.

 

The vertical axes are what define the horizontal axis for you.  They provide the meaning.


Edited by bobzeq25, 26 October 2020 - 01:13 PM.


#3 gatsbyiv

gatsbyiv

    Viking 1

  • *****
  • Posts: 921
  • Joined: 29 Mar 2009
  • Loc: Doylestown, PA

Posted 26 October 2020 - 01:14 PM

The gain/gain chart is telling you for any given software gain setting (horizontal axis) how many electrons per ADU that setting corresponds to (vertical axis). 

 

E.g., at a gain setting of 139, it takes one electron to increase the output by one ADU.  At gain setting of 0, it takes 5 electrons in the well to increase the output by one ADU.  



#4 AZ Maverick

AZ Maverick

    Viking 1

  • *****
  • Posts: 548
  • Joined: 02 Nov 2017
  • Loc: Prescott, Arizona

Posted 26 October 2020 - 01:29 PM

Perhaps this blog will help explain 'gain/iso/offset' a little bit.

https://cloudbreakop...by-pixel-part-3


  • Farzad_K likes this

#5 Farzad_K

Farzad_K

    Mariner 2

  • *****
  • topic starter
  • Posts: 226
  • Joined: 07 Oct 2016
  • Loc: Hillsboro, Oregon, USA

Posted 26 October 2020 - 01:31 PM

Don't quite understand the question.

 

Gain on the horizontal axis is simply the (artificial) number to input into the software, it has no other meaning or purpose for you.

 

The vertical axes tells you what the number means.  What is the full well capacity at that number.  How many electrons you get when one photon is collected.  What is the dynamic range at that number.  What is the read noise at that number.

 

The vertical axes are what define the horizontal axis for you.  They provide the meaning.

The labeling of horizontal axis as Gain was confusing to me especially on the graph that also has gain on the vertical axis. You are saying the gain label on the horizontal axis could be as meaningful as saying "clicks". When I set that control in the ASCOM driver to 139, the gain is 1.0 electron per ADU.

 

Thanks  


  • bobzeq25 likes this

#6 Farzad_K

Farzad_K

    Mariner 2

  • *****
  • topic starter
  • Posts: 226
  • Joined: 07 Oct 2016
  • Loc: Hillsboro, Oregon, USA

Posted 26 October 2020 - 01:35 PM

The gain/gain chart is telling you for any given software gain setting (horizontal axis) how many electrons per ADU that setting corresponds to (vertical axis). 

 

E.g., at a gain setting of 139, it takes one electron to increase the output by one ADU.  At gain setting of 0, it takes 5 electrons in the well to increase the output by one ADU.  

Thanks for input.

 

Having the same label on both axes of the graph contributes to the confusion when you are thinking "gain". If the horizontal axis is simply a graduated line of adjustment of a knob that could be called something as simple as "click" then all is good. When I set the number to "139" the sensor gain is 1.0 electron per ADU.


Edited by Farzad_K, 26 October 2020 - 01:35 PM.


#7 gatsbyiv

gatsbyiv

    Viking 1

  • *****
  • Posts: 921
  • Joined: 29 Mar 2009
  • Loc: Doylestown, PA

Posted 26 October 2020 - 02:15 PM

Thanks for input.

 

Having the same label on both axes of the graph contributes to the confusion when you are thinking "gain". If the horizontal axis is simply a graduated line of adjustment of a knob that could be called something as simple as "click" then all is good. When I set the number to "139" the sensor gain is 1.0 electron per ADU.

It might have been clearer if they labeled the horizontal axis "Camera gain setting (0.1 dB per unit)." 



#8 Farzad_K

Farzad_K

    Mariner 2

  • *****
  • topic starter
  • Posts: 226
  • Joined: 07 Oct 2016
  • Loc: Hillsboro, Oregon, USA

Posted 26 October 2020 - 05:12 PM

It might have been clearer if they labeled the horizontal axis "Camera gain setting (0.1 dB per unit)." 

I am reading the horizontal scale as being 0.1dB per tick on the scale. At 50, it is 5.0 dB; at 200 it is 20.0 dB, and at 117 (for the 294) it is 11.7dB.

 

There is a reason they are using decibels (dB) for this and I can't remember where I was reading about it but it comes from audio industry/technology. We know decibels is the measure of sound and its loudness, so could the horizontal scale be like the volume knob on an old radio? Turn it down and four things change with every turn of that knob: Read Noise, Dynamic Range, Gain, and Full Well. So could Gain on the horizontal axis be "increase/decrease" on the four characteristics of the sensor, not to be confused with the actual gain of the sensor?

 

I wonder how other camera manufacturers communicate that with their users.

 

Here is an interesting link.



#9 Farzad_K

Farzad_K

    Mariner 2

  • *****
  • topic starter
  • Posts: 226
  • Joined: 07 Oct 2016
  • Loc: Hillsboro, Oregon, USA

Posted 26 October 2020 - 05:21 PM

https://en.wikipedia...n_(electronics)



#10 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 21,894
  • Joined: 27 Oct 2014

Posted 26 October 2020 - 05:55 PM

I am reading the horizontal scale as being 0.1dB per tick on the scale. At 50, it is 5.0 dB; at 200 it is 20.0 dB, and at 117 (for the 294) it is 11.7dB.

 

There is a reason they are using decibels (dB) for this and I can't remember where I was reading about it but it comes from audio industry/technology. We know decibels is the measure of sound and its loudness, so could the horizontal scale be like the volume knob on an old radio? Turn it down and four things change with every turn of that knob: Read Noise, Dynamic Range, Gain, and Full Well. So could Gain on the horizontal axis be "increase/decrease" on the four characteristics of the sensor, not to be confused with the actual gain of the sensor?

 

I wonder how other camera manufacturers communicate that with their users.

 

Here is an interesting link.

It's better to think of it as ticks.  The dBs have zero meaning to you.  That's the purpose of the vertical axes, to provide meaning.

 

The graph that confused you translates gain in dB (no meaning) to gain in ADU per photon measured (aka electrons).  Meaning.

 

Minor point.  dBs are used for measurements of any signal, not just sound.


Edited by bobzeq25, 26 October 2020 - 05:59 PM.


#11 AhBok

AhBok

    Mercury-Atlas

  • *****
  • Posts: 2,803
  • Joined: 02 Dec 2010
  • Loc: Lakeland, TN

Posted 26 October 2020 - 08:40 PM

Like Bob, I consider the horizontal scale to be rather arbitrary and not in decibels since decibels are a logarithmic function and not linear. What is important is that 139 gain on this camera has a gain of .1 dB which is just a tic above 0 dB. 0db gain is unity gain (input = output). Or, more precisely 1 electron produced by a pixel = 1 ADU.


Edited by AhBok, 26 October 2020 - 08:55 PM.

  • bobzeq25 likes this

#12 TelescopeGreg

TelescopeGreg

    Vanguard

  • -----
  • Posts: 2,308
  • Joined: 16 Jul 2018
  • Loc: Auburn, California, USA

Posted 26 October 2020 - 10:38 PM

The only thing that is important about the horizontal scale is that the numbers represented there are what you type into your astro software.

 

That said, one question:  For some cameras (e.g. the ASI2600MC, ASI533MC), there is a vertical section on the Read Noise graph.  The 2600 has this at "100" on the X-axis.  If one wants the lower setting, is that 100 or 101?  The text suggests that 100 would grant the lower noise and higher dynamic range setting, but that's not clear in the graph itself.  And, if for some reason you wanted the higher value, would "99" get you there?  At least for the ASI294MC they label this point, but not the other two.



#13 Der_Pit

Der_Pit

    Surveyor 1

  • *****
  • Posts: 1,643
  • Joined: 07 Jul 2018
  • Loc: La Palma

Posted 27 October 2020 - 09:27 AM

Like Bob, I consider the horizontal scale to be rather arbitrary and not in decibels since decibels are a logarithmic function and not linear. 

Yes, dB are logarithmic.  That's why the e¯/du curve is not linear, and the full well graph is linear when using a log Y-scale.....

So I don't really get your argument confused1.gif



#14 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 21,894
  • Joined: 27 Oct 2014

Posted 27 October 2020 - 10:25 AM

Yes, dB are logarithmic.  That's why the e¯/du curve is not linear, and the full well graph is linear when using a log Y-scale.....

So I don't really get your argument confused1.gif

The argument is simple.  The _vast_ majority of people are better off considering the horizontal axis as simple ticks, for inputting into software.  There's nothing to be gained by considering dBs, logarithmic versus linear, etc.  Meaning is had from the vertical axes. 

 

Considering the horizontal axis in detail is wasted time and effort.


Edited by bobzeq25, 27 October 2020 - 10:26 AM.

  • AhBok likes this

#15 Der_Pit

Der_Pit

    Surveyor 1

  • *****
  • Posts: 1,643
  • Joined: 07 Jul 2018
  • Loc: La Palma

Posted 27 October 2020 - 10:44 AM

Oh, that's fine, just using them as ticks, ignoring what they actually are.  But claiming they are not decibel is something different.  I'd assume ZWO knows quite well about the internals of the ADC and what the numbers mean....


Edited by Der_Pit, 27 October 2020 - 10:46 AM.

  • AhBok likes this

#16 AhBok

AhBok

    Mercury-Atlas

  • *****
  • Posts: 2,803
  • Joined: 02 Dec 2010
  • Loc: Lakeland, TN

Posted 27 October 2020 - 10:57 AM

Oh, that's fine, just using them as ticks, ignoring what they actually are.  But claiming they are not decibel is something different.  I'd assume ZWO knows quite well about the internals of the ADC and what the numbers mean....

Oops. You are correct. The logarithmic function is in the plotted curve, not the scale. My mistake. That said, Bob’s explanation is what is important.


  • bobzeq25 likes this

#17 Farzad_K

Farzad_K

    Mariner 2

  • *****
  • topic starter
  • Posts: 226
  • Joined: 07 Oct 2016
  • Loc: Hillsboro, Oregon, USA

Posted 27 October 2020 - 10:59 AM

Oh, that's fine, just using them as ticks, ignoring what they actually are.  But claiming they are not decibel is something different.  I'd assume ZWO knows quite well about the internals of the ADC and what the numbers mean....

So decibels or not, simpler to ignore them or not, what is your personal take on the meaning of the values on the horizontal axis? We are all here in this field because of curiosity. When a number and unit is shown we want to know what they are and how they came to be and what their effects are. Life goes on one way or another.



#18 Farzad_K

Farzad_K

    Mariner 2

  • *****
  • topic starter
  • Posts: 226
  • Joined: 07 Oct 2016
  • Loc: Hillsboro, Oregon, USA

Posted 27 October 2020 - 11:03 AM

The only thing that is important about the horizontal scale is that the numbers represented there are what you type into your astro software.

 

That said, one question:  For some cameras (e.g. the ASI2600MC, ASI533MC), there is a vertical section on the Read Noise graph.  The 2600 has this at "100" on the X-axis.  If one wants the lower setting, is that 100 or 101?  The text suggests that 100 would grant the lower noise and higher dynamic range setting, but that's not clear in the graph itself.  And, if for some reason you wanted the higher value, would "99" get you there?  At least for the ASI294MC they label this point, but not the other two.

What I have learned recently, which is the reason I am zeroing in on these graphs again, is that the values shown are not exact for every sensor. The best thing is to measure the performance and behavior of the camera yourself. 

 

There are some resources out there for doing this if you need them.



#19 Der_Pit

Der_Pit

    Surveyor 1

  • *****
  • Posts: 1,643
  • Joined: 07 Jul 2018
  • Loc: La Palma

Posted 27 October 2020 - 11:10 AM

The horizontal axis is something like the 'throttle' of the ADC, i.e., how much the signal is amplified.  The units do have a specific (electronic) meaning and are (should be?) calibrated, but Bob and Randy are of course right that they just give you the number that you have to dial in to get your wanted result (the wanted e¯/du rate).


Edited by Der_Pit, 27 October 2020 - 11:14 AM.

  • Farzad_K likes this

#20 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 21,894
  • Joined: 27 Oct 2014

Posted 27 October 2020 - 02:04 PM

So decibels or not, simpler to ignore them or not, what is your personal take on the meaning of the values on the horizontal axis? We are all here in this field because of curiosity. When a number and unit is shown we want to know what they are and how they came to be and what their effects are. Life goes on one way or another.

Understood, but you're wasting time and effort that could be more profitably spent on something else, like learning to process better.  <smile>



#21 Farzad_K

Farzad_K

    Mariner 2

  • *****
  • topic starter
  • Posts: 226
  • Joined: 07 Oct 2016
  • Loc: Hillsboro, Oregon, USA

Posted 27 October 2020 - 08:14 PM

Understood, but you're wasting time and effort that could be more profitably spent on something else, like learning to process better.  <smile>

Yeah, it could appear like time is being wasted but if you have a splinter in your mind you will be less effective. This is a high-intensity mind work and I will not go through it mindlessly.

 

I actually found a good explanation of it here. I still need to better understand the video after I get my hearing trained a bit, but the issue isn't the "gain" on the horizontal axis, it might be the "gain" on the vertical axis that may need to be called something more descriptive - they both need to actually. Not sure yet but I will get there. 


  • Der_Pit likes this

#22 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 21,894
  • Joined: 27 Oct 2014

Posted 27 October 2020 - 10:06 PM

Yeah, it could appear like time is being wasted but if you have a splinter in your mind you will be less effective. This is a high-intensity mind work and I will not go through it mindlessly.

 

I actually found a good explanation of it here. I still need to better understand the video after I get my hearing trained a bit, but the issue isn't the "gain" on the horizontal axis, it might be the "gain" on the vertical axis that may need to be called something more descriptive - they both need to actually. Not sure yet but I will get there. 

The gain on the vertical axis (electrons (captured photons) per ADU) is the usual definition of gain, since it's actually useful for things like determining proper subexposure.  For example, see this chart.  Gain is either "variable" or it's given in electrons per ADU.  the standard definition.

 

https://www.atik-cam...ication-tables/

 

I know of no use for the gain on the horizontal axis except as a software input to your data capture program, which is probably why I rarely see it at all.  So, if you're going to leave off the word gain somewhere, it should be on the horizontal axis.  Not sure what you'd replace it with, though.  "amplification"?



#23 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,455
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 27 October 2020 - 10:31 PM

Hi Folks.

 

I have discussed this question with ZWO and I don't have a good answer from them that I can understand.

 

The performance graphs show the horizontal axis as Gain (0.1 db) and in one of the graphs there is a vertical axis also called Gain in units of electrons per ADU. I know the Gain shown on the horizontal axis is the value I set in the camera driver, and I know what the Gain e/ADU is.

 

Can someone help me understand the horizontal axis?

 

Thanks in advance.

 

Farzad

 

Horizontal axis is gain in Decibels (dB). For ZWO cameras, this corresponds directly with their gain "setting". So, if you use a setting of 139, or 13.9 dB of actual analog gain, that is "unity gain", and therefor your actual conversion ratio is 1e-/ADU.

 

The vertical axis in the second chart is the conversion ratio of electrons (e-) to ADU (analog to digital units, the numeric value saved for the given pixel in an image as a result of converting charge/voltage to a digital value), at a given gain. 

 

So, in the second chart, you are mapping GAIN SETTINGS/GAIN DB to the CONVERSION RATIO of electrons to digital units. 

 

Both values are important. One tells you a useful value, e-/ADU, with which you can calculate other useful things (i.e. swamp factor, so you can understand how much signal you need to measure in a sub (ADU) to know you are swamping read noise (which is usually reported in e-). The other tells you what camera gain setting to use in order to achieve that conversion ratio (gain) in e-/ADU. 


Edited by Jon Rista, 27 October 2020 - 10:35 PM.

  • RazvanUnderStars, Farzad_K and Der_Pit like this

#24 Farzad_K

Farzad_K

    Mariner 2

  • *****
  • topic starter
  • Posts: 226
  • Joined: 07 Oct 2016
  • Loc: Hillsboro, Oregon, USA

Posted 27 October 2020 - 10:52 PM

Horizontal axis is gain in Decibels (dB). For ZWO cameras, this corresponds directly with their gain "setting". So, if you use a setting of 139, or 13.9 dB of actual analog gain, that is "unity gain", and therefor your actual conversion ratio is 1e-/ADU.

 

The vertical axis in the second chart is the conversion ratio of electrons (e-) to ADU (analog to digital units, the numeric value saved for the given pixel in an image as a result of converting charge/voltage to a digital value), at a given gain. 

 

So, in the second chart, you are mapping GAIN SETTINGS/GAIN DB to the CONVERSION RATIO of electrons to digital units. 

 

Both values are important. One tells you a useful value, e-/ADU, with which you can calculate other useful things (i.e. swamp factor, so you can understand how much signal you need to measure in a sub (ADU) to know you are swamping read noise (which is usually reported in e-). The other tells you what camera gain setting to use in order to achieve that conversion ratio (gain) in e-/ADU. 

Thanks a lot. I think this is going to help a lot.

 

Farzad




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics