[kaseki]

Most electronic shutters have very long rolling times. In essence, electronic shutters act like an electronic version of the moving slit of a mechanical shutter. The front electronic curtain resets the pixels by draining any charge in them. The rear electronic curtain reads the light level in the pixels. It's the pixel reading process that limits the speed of the curtain and can be especially long on larger sensors with high pixel counts. You can design faster read-outs but that sacrifices the image quality.

Unless the sensor has a true global shutter, an electronic shutter is generally incompatible with flash. (Note that even the Sony A9's electronic shutter is incompatible with flash.)

Once upon a time, RCA as I recall had a sensor architecture with which the image array could be simultaneously flushed, a new image taken, and then the pixel storage capacitors' charges transferred all at once to a secondary storage array under them. This second array could then be read out "leisurely." This architecture should be flash compatible, but not necessarily compatible with the technology that is being used for present-day sensors.

[LensCleaner01]

The ability to meaningfully use leaf shutters is available to the Pentax, but too few lenses include them to support that as a central use case. I have the 75 and only once needed it for my 645NII, though I used it often because it was a decent lens optically.

thats not what i meant i meant to build in a leaf shutter into my pentax as upgrade (as demaged rolling shutters can be replaced) or a faster version of a rolling shutter

thanks photoptimist i very like your technical answers why the flash cant be used with the electronic shutter ?

the readout yes it happens line wise probaly the question would be how fast that is per line

i have a bit electronic knowlegue that problem is known i dont know the exact english name for it but its "a-syncron" what means per line
a "syncron" read can read more lines or electric statuses at the same time
its more cost to build a syncron reader over a electric circuit diagram

but it might be a idea for future sensors at the end it can be mass produced once the efford was taken

---------- Post added 09-11-18 at 08:52 AM ----------

i got a idea for the space problem the discs today are very big
normal HDD´s already have 14 terabytes
"14000GB Toshiba Enterprise Capacity MG07ACA14TE "

that makes 14000000 MB
SSD also have overcome 15 terabytes:
"Samsung SSD PM1633a 15.36TB"

thats a enough space for raw mode pictures :-)

[photoptimist]

Once upon a time, RCA as I recall had a sensor architecture with which the image array could be simultaneously flushed, a new image taken, and then the pixel storage capacitors' charges transferred all at once to a secondary storage array under them. This second array could then be read out "leisurely." This architecture should be flash compatible, but not necessarily compatible with the technology that is being used for present-day sensors.

Good memory!

You are talking about "frame transfer CCD." And it was exactly as you said. The chip was twice the size of the "sensor" and had twice as many CCD "pixels" as the camera advertised. One half of the chip was exposed to the light and the other half was covered. At the end of the shutter time, the accumulated charges in the top half of the array were quickly shuffled down to the covered lower half where they could be read-out more at leisure.

Frame transfer CCD has better quantum efficiency that competing interline-transfer CCDs (which had tiny photodiodes nestled between the opaque columns and rows of CCD electronics). But frame transfer CCDs required double the silicon and suffered from smearing of blown highlights.

[LensCleaner01]

when speeding up a object lets say a bullet over a explosion it speeds up pretty fast
when speeding up due pressure or force like a rad rotating mechanism or compressing air or a reflexing material
i can think that of a mechanical shutter ?

knowing about a rail gun from some experiments you can carry or speed up something that you useally cant move so fast because it would crush

going back to a shutter here what is if we use a neodym magnete in sence of a rail gun in small size so we reach a better shutter y-sync speed ?
----
thinking about this having a second blade coming up to the middle also would double the y-sync speed , if they also come from left and right it would fix that even better - but that sounds like a leaf shutter to me
----
if the answer is that on a flash and the electronic shutter would just recharge the sensors pixels

and the electronic shutter if the pixel sensor drains the charge - why would that have to do it again ? the sensor pixel just gets marked as "ignored" "used" or "not overstepp" again

1/125 is just bad to be honest its not only my meaning , it also has a other positiv effect when the pixels are charged the close down also happens faster that means less time or motion blur also comes for this "next part" of y-sync speed
1/30 open 1/125 to close it up , thats also something that would count

-----
the thing with the light 380-700 nm (photon size bounce) is that the light is coming also in form of a wave lets say our sensor pixel got 380 nm so if the sensor pixel is flat that means the light could bounce to a other pixel according the position was not in the middle of the 380 nm sensor pixel part
then some part is on the other sensor pixel this means also a smaller size can make sence then this information is also covered and 2 positions can be combined
the bounce size or wavesize is known but the position it hits is not , its probaly about the quarks
---
what i meant with the hasselblad was not the mirror it was rather that the megapixels go up the old phase 1 from 2011 also already had 80 megapixels now 400 megapixels its 2018 after all
the viewfinders have a focus seeker and a depth of field seeker it marks them useally with blinking also it can warn with blinking over overlight and show a histogramm ect. the pecision is also better because you can zoom in 16x that allows you to set a very precise focus
the only advantage i can see at the moment in real time depending on the electronic lcd display its a bit delayed - but i accept new information here or more its even wanted
but that got better a bit framerate or refreshrate, reactiontime would be a keyword there is room up to improve here


how fast the y-sync speed of the electronic shutter for the pentax 645z really is ? on the video mode you can set 1/500 but i have no idea what the sensor really do here

also a idea would be a flash with LI-ION batterys this would offer more power with less weight

---

Last edited by LensCleaner01; 09-12-2018 at 01:16 PM.

[rdenney]

Can’t respond to all that, but the H6 didn’t go from 80 to 400 megapixels, it went from 80 to 100 megapixels, and then added pixel shift.

The traditional sync speeds for medium format focal-plane shutters were 1/25 or 1/30. 1/125 reflects weight reduction and improved control mechanisms. During that same period, 35mm SLRs went from 1/60 to 1/250–the same two stops.

We’d be better off with a flash that scanned like the focal-plane shutter.

Rick “perfection is the enemy of excellence” Denney

[SylvainB]

Can’t respond to all that, but the H6 didn’t go from 80 to 400 megapixels, it went from 80 to 100 megapixels, and then added pixel shift.

Rick “perfection is the enemy of excellence” Denney

Not exactly, but you got the idea anyway.
Phase One had 60MP CCD sensors, then 80MP CCD, then 100MP CMOS, and now 150MP CMOS (to stick to full frame sensors).
Hasselblad had the 60MP CCD, 2 different multishot (=pixelshift) versions of 200MP derived from 50MP crop sensors, then they moved to 100MP CMOS which gives a 400MP multishot version now.

[torashi]

Why can't it just have an electronic shutter? No moving parts, no excessive acceleration/deceleration for sync at higher speeds.
Another idea could be, if and only if size weren't a problem, it could bear a rotary shutter. Like the titanium one on a Pen F/FT/FV.

[torashi]

Why can't it just have an electronic shutter? No moving parts, no excessive acceleration/deceleration for sync at higher speeds.
Another idea could be, if and only if size weren't a problem, it could bear a rotary shutter. Like the titanium one on a Pen F/FT/FV.

[kaseki]

Why can't it just have an electronic shutter? No moving parts, no excessive acceleration/deceleration for sync at higher speeds.
Another idea could be, if and only if size weren't a problem, it could bear a rotary shutter. Like the titanium one on a Pen F/FT/FV.

Many devices conceived over several decades have "lusted" for some type of electronic shutter. Most of these depend on electrostatic (ex: liquid crystal) or magnetic effects, and tend to have spectral band-pass and/or angular acceptance issues. Electro-mechanically-driven (piezoelectric, say) clacking prisms (Pechan or right-angle pair) can make use of closing or opening the internal reflection surface and have good spectral characteristics, but angular acceptance is still limited, and in any case one has a seriously large mass for anything approaching a MF focal plane in size.

[photoptimist]

Why can't it just have an electronic shutter? No moving parts, no excessive acceleration/deceleration for sync at higher speeds.

The central challenge is that it's impossible to actually turn off a silicon sensor. It's basically a big solar cell subdivided into millions of little pixels. Even with the camera off and the battery removed, if light hits the sensor, it generates a charge. Only a mechanical shutter (or the kind of light valve shutter mentioned by kaseki) can turn off the sensor by cutting off the light. The challenge for an electronic shutter is in managing the constantly accumulating photocharges in the pixels.

Starting an exposure electronically (an electronic front curtain) isn't too hard. You just short-out the pixel to zero-out the charge* and then let it start accumulating electrons again.

Ending the exposure electronically is the hard part because there's no way to turn-off a pixel other than to cover it with a mechanical shutter. The accumulated charge must be either read-out or moved someplace safe for read-out. It is very hard to do the read-out or charge transfer quickly without degrading DR or massively increasing the cost of the sensor and electronics. Reading out 50 million pixels within a 1/200 sec flash sync time implies measuring 10 billion pixels per second. Top-performance 14-bit ADCs max out at about 400 million samples per second. One could use 25 of these high-performance chips but the result would be a huge camera that sucks power (the 25 chips would need 63 watts of power).

*Note: there are some dynamic range issues with front-curtain related to the resistance to ground. If the light flux is high enough, there will be some residual charge at the start of the exposure that reduces the DR.

[rfkiii]

They’ll go bigger. That’s the only way they can maintain a leadership position. And they’ll keep the price under $10K.

This would be awesome but they'll leave some of their best lenses behind. I might be tempted to buy one more DSLR.

[MarkJerling]

The central challenge is that it's impossible to actually turn off a silicon sensor. It's basically a big solar cell subdivided into millions of little pixels. Even with the camera off and the battery removed, if light hits the sensor, it generates a charge. Only a mechanical shutter (or the kind of light valve shutter mentioned by kaseki) can turn off the sensor by cutting off the light. The challenge for an electronic shutter is in managing the constantly accumulating photocharges in the pixels.

Starting an exposure electronically (an electronic front curtain) isn't too hard. You just short-out the pixel to zero-out the charge* and then let it start accumulating electrons again.

Ending the exposure electronically is the hard part because there's no way to turn-off a pixel other than to cover it with a mechanical shutter. The accumulated charge must be either read-out or moved someplace safe for read-out. It is very hard to do the read-out or charge transfer quickly without degrading DR or massively increasing the cost of the sensor and electronics. Reading out 50 million pixels within a 1/200 sec flash sync time implies measuring 10 billion pixels per second. Top-performance 14-bit ADCs max out at about 400 million samples per second. One could use 25 of these high-performance chips but the result would be a huge camera that sucks power (the 25 chips would need 63 watts of power).

*Note: there are some dynamic range issues with front-curtain related to the resistance to ground. If the light flux is high enough, there will be some residual charge at the start of the exposure that reduces the DR.

Wow. That is very interesting. I had no idea! Thank you for posting!

[torashi]

The central challenge is that it's impossible to actually turn off a silicon sensor. It's basically a big solar cell subdivided into millions of little pixels. Even with the camera off and the battery removed, if light hits the sensor, it generates a charge. Only a mechanical shutter (or the kind of light valve shutter mentioned by kaseki) can turn off the sensor by cutting off the light. The challenge for an electronic shutter is in managing the constantly accumulating photocharges in the pixels.

Starting an exposure electronically (an electronic front curtain) isn't too hard. You just short-out the pixel to zero-out the charge* and then let it start accumulating electrons again.

Ending the exposure electronically is the hard part because there's no way to turn-off a pixel other than to cover it with a mechanical shutter. The accumulated charge must be either read-out or moved someplace safe for read-out. It is very hard to do the read-out or charge transfer quickly without degrading DR or massively increasing the cost of the sensor and electronics. Reading out 50 million pixels within a 1/200 sec flash sync time implies measuring 10 billion pixels per second. Top-performance 14-bit ADCs max out at about 400 million samples per second. One could use 25 of these high-performance chips but the result would be a huge camera that sucks power (the 25 chips would need 63 watts of power).

*Note: there are some dynamic range issues with front-curtain related to the resistance to ground. If the light flux is high enough, there will be some residual charge at the start of the exposure that reduces the DR.

Thanks for the explanation. I guess the legacy LF lenses are the only option for higher sync speed. And they only go to 1/500th s (that is, if all the mechanics are working correctly).

[LensCleaner01]

The accumulated charge must be either read-out or moved someplace safe for read-out. It is very hard to do the read-out or charge transfer quickly without degrading DR or massively increasing the cost of the sensor and electronics

yes a cache or buffer, thanks for this one.

a silizium wafer can or is made of layers
that layers are put over themselfs so they would be room for more connections or buffer

going to the maybe other solution:
a multiple readout in a different way is also a old electronic problem if you have a array of electric 0 and 1´s or a electrical charge status it can be read out using a synchron readout that need more connections
the thing is that this is only "a plan" as electronic shemetics
once drawed it can be mass produced

a other idea would be a higher value for the bus transfer i dont know if its 256 bits or something like that if the small 14 bit transfer is generally to slow for other things

[photoptimist]

The accumulated charge must be either read-out or moved someplace safe for read-out. It is very hard to do the read-out or charge transfer quickly without degrading DR or massively increasing the cost of the sensor and electronics

yes a cache or buffer, thanks for this one.

a silizium wafer can or is made of layers
that layers are put over themselfs so they would be room for more connections or buffer

going to the maybe other solution:
a multiple readout in a different way is also a old electronic problem if you have a array of electric 0 and 1´s or a electrical charge status it can be read out using a synchron readout that need more connections
the thing is that this is only "a plan" as electronic shemetics
once drawed it can be mass produced

a other idea would be a higher value for the bus transfer i dont know if its 256 bits or something like that if the small 14 bit transfer is generally to slow for other things

A digital buffer is easy but does not solve this problem. A faster bus is easy but does not solve this problem. The limit is not on the digital side, it's on the analog side.

An analog buffer is much harder to build because even a single added or lost electron makes the image noisier than current state-of-the-art sensors can achieve.

The real problem is in the conversion of the analog levels to digital values. That problem is solvable (with 25 chips and a 63+watt power supplier) but not in a small battery-powered device.