More precise sensor readout measurement

[Horshack]

 

[ThrillaMozilla]

I think I would just put the camera on a phonograph turntable and photograph vertical lines. No electronics or construction needed.

I look forward to someone posting a nice table of current cameras by any method.

He has covered approx 10 cameras.

Well I sure don't know where that is.

There's a figure going around the internet that isn't up to date. Also, the source of the data is unknown and the readout modes are unspecified.

OK, on this thread I count about 6 or 7. That's very nice, and I look forward to seeing people fill in the data with more of the current cameras.

 

[Ephemeris]

I think I would just put the camera on a phonograph turntable and photograph vertical lines. No electronics or construction needed.

I look forward to someone posting a nice table of current cameras by any method.

He has covered approx 10 cameras.

Well I sure don't know where that is.

There's a figure going around the internet that isn't up to date. Also, the source of the data is unknown and the readout modes are unspecified.

In this thread. Horshack is posting here and added a Leica today.

 

[Ephemeris]

I think I would just put the camera on a phonograph turntable and photograph vertical lines. No electronics or construction needed.

I look forward to someone posting a nice table of current cameras by any method.

He has covered approx 10 cameras.

Well I sure don't know where that is.

There's a figure going around the internet that isn't up to date. Also, the source of the data is unknown and the readout modes are unspecified.

OK, on this thread I count about 6 or 7. That's very nice, and I look forward to seeing people fill in the data with more of the current cameras.

Well it's Horshacks thread so it would have to be using that same technique. The technique initially the interest as others, including myself have home brew concepts.

The technique looks sound and reliable.

I think 11 cameras now.

 

[John Vickers]

Here's a Z6 photo of a 60 Hz AC LED light bulb with a < 100% brightness PWM of 1920 Hz, where the effect of the lens correction applied to a sensor-created projection is more obvious. To restate, the bands are entirely the creation of the rolling shutter readout on a cycling light source and are not a projection of the lens, which means applying distortion correction to the horizontal sensor bands actually creates distortion rather than correcting it.

Animation: Lens correction on sensor-projected rolling shutter bands

I've been using RawDigger to create images without lens correction.

 

[John Vickers]

Cheap signal generator this time. My main focus was shutter speed measurements.

If you set "1/60" shutter speed, what exposure time do you expect?

Started by JohnVickers on May 23, 2023 in the Technical Discussions category. 87 replies, last one from July 1, 2023.

dprevived.com

Example shutter travel time AKA rolling shutter measurement for a mechanical shutter:

New ways with shutter travel time measurement

Started by JohnVickers on May 14, 2023 in the Technical Discussions category. 0 replies, last one from May 14, 2023.

dprevived.com

 

[dc064]

I made a measurement of the A9 as reported here: https://www.fredmiranda.com/forum/topic/1663827/3#15358329 . I got 1/140 sec compared to your 1/150.5 sec. I estimated accuracy of 1% or better. Your quoted significant figures seems to imply 0.1% accuracy or better. The measurements differ by 7%.

 

[Ephemeris]

I made a measurement of the A9 as reported here: https://www.fredmiranda.com/forum/topic/1663827/3#15358329 . I got 1/140 sec compared to your 1/150.5 sec. I estimated accuracy of 1% or better. Your quoted significant figures seems to imply 0.1% accuracy or better. The measurements differ by 7%.

How is your reference data validated? That is the PWM profile and frequency of the light source?

Sorry if it's within one of your posts but I couldn't spot it.

 

[Horshack]

I made a measurement of the A9 as reported here: https://www.fredmiranda.com/forum/topic/1663827/3#15358329 . I got 1/140 sec compared to your 1/150.5 sec. I estimated accuracy of 1% or better. Your quoted significant figures seems to imply 0.1% accuracy or better. The measurements differ by 7%.

I estimate my accuracy of 1% or better. Sounds like we're at an impasse

FYI Jim measured 1/160.

 

[dc064]

Yeah, it's hard to see how they could be that far off. My measurement depends on the LED frequency being 240Hz = 4x60Hz. That is the LED pulses are phase locked to the line frequency. I find it hard to believe they are not. I am not sure how well you are able to set the frequency in your set up, but regardless 7% is a lot. I also find it hard to believe there is that large a variation camera to camera. So I am stumped to explain the difference, but I suppose it is not a big issue for assessing the impact on images. Still it seems like all these methods should produce more precise measurements than the differences we are seeing.

 

[Ephemeris]

Yeah, it's hard to see how they could be that far off. My measurement depends on the LED frequency being 240Hz = 4x60Hz. That is the LED pulses are phase locked to the line frequency. I find it hard to believe they are not. I am not sure how well you are able to set the frequency in your set up, but regardless 7% is a lot. I also find it hard to believe there is that large a variation camera to camera. So I am stumped to explain the difference, but I suppose it is not a big issue for assessing the impact on images. Still it seems like all these methods should produce more precise measurements than the differences we are seeing.

So you are basing it on an unvalidated and described light source?

I think the key to measurements is it needs to be calibrated, validated and ratified.

 

[Horshack]

Yeah, it's hard to see how they could be that far off. My measurement depends on the LED frequency being 240Hz = 4x60Hz. That is the LED pulses are phase locked to the line frequency. I find it hard to believe they are not. I am not sure how well you are able to set the frequency in your set up, but regardless 7% is a lot. I also find it hard to believe there is that large a variation camera to camera. So I am stumped to explain the difference, but I suppose it is not a big issue for assessing the impact on images. Still it seems like all these methods should produce more precise measurements than the differences we are seeing.

The logic I'm using has been externally verified with an oscilloscope to show the LED is toggling at frequency its being programmed to. I slightly improved the code by removing some of the latency in accessing the GPIO. I'd be happy to share the code with you, which you can run on this $16.99 Ardunio board that I'm using.

 

[Horshack]

I made a measurement of the A9 as reported here: https://www.fredmiranda.com/forum/topic/1663827/3#15358329 . I got 1/140 sec compared to your 1/150.5 sec. I estimated accuracy of 1% or better. Your quoted significant figures seems to imply 0.1% accuracy or better. The measurements differ by 7%.

I noticed in your FM thread that you also posted A7rIV results of 1/9.25. I just tested my A7rIV and get 1/9.94, which is approximately the same 7% differential of your A9 results to mine. This indicates our discrepancy is methodology-specific rather than camera-specific.

Perhaps the LED strip you're relying on for the measurement is cylcing slightly faster than 240 Hz, from a fractional PWM being applied. That would produce slightly more bands than 240 Hz, which would yield the slightly lower result you're seeing.

FYI Jim measured the A7rIV at 1/10.

 

[Ephemeris]

Yeah, it's hard to see how they could be that far off. My measurement depends on the LED frequency being 240Hz = 4x60Hz. That is the LED pulses are phase locked to the line frequency. I find it hard to believe they are not. I am not sure how well you are able to set the frequency in your set up, but regardless 7% is a lot. I also find it hard to believe there is that large a variation camera to camera. So I am stumped to explain the difference, but I suppose it is not a big issue for assessing the impact on images. Still it seems like all these methods should produce more precise measurements than the differences we are seeing.

The logic I'm using has been externally verified with an oscilloscope to show the LED is toggling at frequency its being programmed to. I slightly improved the code by removing some of the latency in accessing the GPIO. I'd be happy to share the code with you, which you can run on this $16.99 Ardunio board that I'm using.

How was that verified with a scope?

Understand the capability of that measurement would appear important.

It's an area of potential error, but perhaps other areas also exist.

 

[Horshack]

Yeah, it's hard to see how they could be that far off. My measurement depends on the LED frequency being 240Hz = 4x60Hz. That is the LED pulses are phase locked to the line frequency. I find it hard to believe they are not. I am not sure how well you are able to set the frequency in your set up, but regardless 7% is a lot. I also find it hard to believe there is that large a variation camera to camera. So I am stumped to explain the difference, but I suppose it is not a big issue for assessing the impact on images. Still it seems like all these methods should produce more precise measurements than the differences we are seeing.

The logic I'm using has been externally verified with an oscilloscope to show the LED is toggling at frequency its being programmed to. I slightly improved the code by removing some of the latency in accessing the GPIO. I'd be happy to share the code with you, which you can run on this $16.99 Ardunio board that I'm using.

How was that verified with a scope?

Understand the capability of that measurement would appear important.

It's an area of potential error, but perhaps other areas also exist.

I assume with the probe on the PCB pad for the LED.

 

[Ephemeris]

Yeah, it's hard to see how they could be that far off. My measurement depends on the LED frequency being 240Hz = 4x60Hz. That is the LED pulses are phase locked to the line frequency. I find it hard to believe they are not. I am not sure how well you are able to set the frequency in your set up, but regardless 7% is a lot. I also find it hard to believe there is that large a variation camera to camera. So I am stumped to explain the difference, but I suppose it is not a big issue for assessing the impact on images. Still it seems like all these methods should produce more precise measurements than the differences we are seeing.

The logic I'm using has been externally verified with an oscilloscope to show the LED is toggling at frequency its being programmed to. I slightly improved the code by removing some of the latency in accessing the GPIO. I'd be happy to share the code with you, which you can run on this $16.99 Ardunio board that I'm using.

How was that verified with a scope?

Understand the capability of that measurement would appear important.

It's an area of potential error, but perhaps other areas also exist.

I assume with the probe on the PCB pad for the LED.

That doesn't demonstrate the setup or how much error there is in that part of the measurement system.

 

[Horshack]

Yeah, it's hard to see how they could be that far off. My measurement depends on the LED frequency being 240Hz = 4x60Hz. That is the LED pulses are phase locked to the line frequency. I find it hard to believe they are not. I am not sure how well you are able to set the frequency in your set up, but regardless 7% is a lot. I also find it hard to believe there is that large a variation camera to camera. So I am stumped to explain the difference, but I suppose it is not a big issue for assessing the impact on images. Still it seems like all these methods should produce more precise measurements than the differences we are seeing.

The logic I'm using has been externally verified with an oscilloscope to show the LED is toggling at frequency its being programmed to. I slightly improved the code by removing some of the latency in accessing the GPIO. I'd be happy to share the code with you, which you can run on this $16.99 Ardunio board that I'm using.

How was that verified with a scope?

Understand the capability of that measurement would appear important.

It's an area of potential error, but perhaps other areas also exist.

I assume with the probe on the PCB pad for the LED.

That doesn't demonstrate the setup or how much error there is in that part of the measurement system.

I'm not sure what sources of error you mean. It's an oscilloscope sampling a very low-frequency level-based input into an LED.

 

[Ephemeris]

Yeah, it's hard to see how they could be that far off. My measurement depends on the LED frequency being 240Hz = 4x60Hz. That is the LED pulses are phase locked to the line frequency. I find it hard to believe they are not. I am not sure how well you are able to set the frequency in your set up, but regardless 7% is a lot. I also find it hard to believe there is that large a variation camera to camera. So I am stumped to explain the difference, but I suppose it is not a big issue for assessing the impact on images. Still it seems like all these methods should produce more precise measurements than the differences we are seeing.

The logic I'm using has been externally verified with an oscilloscope to show the LED is toggling at frequency its being programmed to. I slightly improved the code by removing some of the latency in accessing the GPIO. I'd be happy to share the code with you, which you can run on this $16.99 Ardunio board that I'm using.

How was that verified with a scope?

Understand the capability of that measurement would appear important.

It's an area of potential error, but perhaps other areas also exist.

I assume with the probe on the PCB pad for the LED.

That doesn't demonstrate the setup or how much error there is in that part of the measurement system.

I'm not sure what sources of error you mean. It's an oscilloscope sampling a very low-frequency level-based input into an LED.

Your measurement system has error.

I'm suggesting analysis of those errors is needed to provide a robust answer.

I can think of ways the scope can be used and setup incorrectly where the value will be incorrect.

It's not open how this was done within the thread.

 

[Horshack]

Yeah, it's hard to see how they could be that far off. My measurement depends on the LED frequency being 240Hz = 4x60Hz. That is the LED pulses are phase locked to the line frequency. I find it hard to believe they are not. I am not sure how well you are able to set the frequency in your set up, but regardless 7% is a lot. I also find it hard to believe there is that large a variation camera to camera. So I am stumped to explain the difference, but I suppose it is not a big issue for assessing the impact on images. Still it seems like all these methods should produce more precise measurements than the differences we are seeing.

The logic I'm using has been externally verified with an oscilloscope to show the LED is toggling at frequency its being programmed to. I slightly improved the code by removing some of the latency in accessing the GPIO. I'd be happy to share the code with you, which you can run on this $16.99 Ardunio board that I'm using.

How was that verified with a scope?

Understand the capability of that measurement would appear important.

It's an area of potential error, but perhaps other areas also exist.

I assume with the probe on the PCB pad for the LED.

That doesn't demonstrate the setup or how much error there is in that part of the measurement system.

I'm not sure what sources of error you mean. It's an oscilloscope sampling a very low-frequency level-based input into an LED.

Your measurement system has error.

I'm suggesting analysis of those errors is needed to provide a robust answer.

I can think of ways the scope can be used and setup incorrectly where the value will be incorrect.

It's not open how this was done within the thread.

You think a scope analyzing a single, low-speed signal is prone to error? Analyzing the wrong wire?

 

[Ephemeris]

Yeah, it's hard to see how they could be that far off. My measurement depends on the LED frequency being 240Hz = 4x60Hz. That is the LED pulses are phase locked to the line frequency. I find it hard to believe they are not. I am not sure how well you are able to set the frequency in your set up, but regardless 7% is a lot. I also find it hard to believe there is that large a variation camera to camera. So I am stumped to explain the difference, but I suppose it is not a big issue for assessing the impact on images. Still it seems like all these methods should produce more precise measurements than the differences we are seeing.

The logic I'm using has been externally verified with an oscilloscope to show the LED is toggling at frequency its being programmed to. I slightly improved the code by removing some of the latency in accessing the GPIO. I'd be happy to share the code with you, which you can run on this $16.99 Ardunio board that I'm using.

How was that verified with a scope?

Understand the capability of that measurement would appear important.

It's an area of potential error, but perhaps other areas also exist.

I assume with the probe on the PCB pad for the LED.

That doesn't demonstrate the setup or how much error there is in that part of the measurement system.

I'm not sure what sources of error you mean. It's an oscilloscope sampling a very low-frequency level-based input into an LED.

Your measurement system has error.

I'm suggesting analysis of those errors is needed to provide a robust answer.

I can think of ways the scope can be used and setup incorrectly where the value will be incorrect.

It's not open how this was done within the thread.

You think a scope analyzing a single, low-speed signal is prone to error? Analyzing the wrong wire?

Do I think a measurement is likely to contain an error? Yes.

What is that error, the reactions are not providing that information.