Multiplex brain teaser

[Mr RB]

If you flash a LED at 100 hz it appear to be be constantly lit up, as it's above the frequency response of the human eye or optic nerve or whatever. Your senses are not fast enough to see the flashing.

But if you flash 4 LED displays, ie 4 digits in a 25% duty standard digit multiplex, and each one is flashing at exactly 100 Hz and 25% duty cycle (which is above the human ability to see) then why can you see very clearly that the whole 4 digit display is flashing??

I have my own theory, but I'd really like to hear other people's theories.

 

[Diver300]

It's because you move your eyes.

Each cell on the retina can only see up to about 40 Hz, but that varies quite a lot, and is higher at the edge of vision and it's higher when you are tired.

However, there are thousands of cells on the retina, and we move our eyes around a lot. While we are moving our eyes, stationary light sources are seen by a line of retina cells. We all learned that is normal as we learned to use our eyes as babies.

If a light source, especially on a dark background, is flashing at up to about 1 kHz, it's quite easy to move one's eyes fast enough that the stationary but flashing light source is not seen by a continuous line of retina cells, but a line with gaps in it. In the case of a 4 digit display, when you move your eyes up and down, the digits appear to be out of line. If you move your eyes from side to side, the width of the display appears to change, because the digit sequence is moving either in the same direction as your eye is moving or in the opposite direction.

That is the sort of thing that we have evolved to be twitchy about, as it just like something moving, possibly to attack, just when we thought it was stationary and dead.

You can see the effect clearly on a photo taken at about 1/10th second shutter speed, but with serious camera movement. The tail lights of the car appear as a separate row of dots. The streak of white is a filament light. The mains is 50 Hz so the street lights flash at 100 Hz, but there is hardly any off period, so the orange line that appears wider an narrower is a sodium street light.

This was in the UK, where flashing lights on cars are illegal, but it isn't enforced. The car manufactures often use LED drivers that suggest PWM works and 100 Hz isn't visible. As you say, it's not visible when looking directly at it, but when there is movement it is visible.

**broken link removed**

 

[transistor495]

Can't believe it because monitors set at >70Hz are very stable without any flash.


So I could not build up any theories in this regard...

 

[Mike - K8LH]

If you flash a LED at 100 hz it appear to be be constantly lit up, as it's above the frequency response of the human eye or optic nerve or whatever. Your senses are not fast enough to see the flashing.

But if you flash 4 LED displays, ie 4 digits in a 25% duty standard digit multiplex, and each one is flashing at exactly 100 Hz and 25% duty cycle (which is above the human ability to see) then why can you see very clearly that the whole 4 digit display is flashing??

I have my own theory, but I'd really like to hear other people's theories.

Can I assume you're talking about "flicker"?

If you're lighting each display in a four digit display at 100-Hz (one digit displayed every 2.5-msecs within a 10-msec period), you shouldn't see any flicker.

 

[Gauthamvg]

i checked this practically, i cant see the flashing in single LED but i did in 4 digit. but i don't know the reason for that. My faculty member scolded me for doing this silly experiment in my lab . But its a tricky question

 

[Diver300]

Can't believe it because monitors set at >70Hz are very stable without any flash.


So I could not build up any theories in this regard...

It depend how sensitive you are, and what type of monitor is used.

CRT monitors flash at the refresh rate. Take a photo of one at 1/125 sec and only part of it is lit. When I was using one of those, I could tell the difference between 75 Hz and 100 Hz, and would chose 100 Hz even if I had to have fewer pixels.

LCD monitors do not flash at the refresh rate. They stay lit the whole time, and the refresh rate is simply how often the computer can change what is on screen. That is why most do not support high refresh rates, as it is pointless. The LCD monitors that I use now are refreshed at 60 Hz and they don't flicker.

 

[Diver300]

If you're lighting each display in a four digit display at 100-Hz (one digit displayed every 2.5-msecs within a 10-msec period), you shouldn't see any flicker.

I am getting old. I need reading glasses. I can't hear the bats any more.

But given a dark background, I can spot the difference between 100 Hz LEDs and DC powered LEDs without any problems. To me, it's a bit like the difference in feel between a smooth bit of copper pipe and a smooth bit of copper pipe with 240 V mains on it.

I really can't stand LEDs that are lit at 100 Hz if there is a significant off period every 10 ms. I have had to add capacitors to several sets of Christmas lights so that I can be in the same room as them.

It's only where the off period is really off that I have a problem. The capacitors that I have to add are often nowhere near large enough to keep the LEDs at full brightness for the 10 ms, but as long as they don't go out, the line of light that I see as I move my eyes past them is unbroken, and I don't get upset by them.

 

[Mr RB]

Nice info Diver300, but I'm not talking about any eye movement or using the sensors at the vision periphery that detect movement.

Gauthamvg had the right idea, actually do the 4 digit test (as I originally did, accidentally, which started me thinking about it). You can't see any single digit flashing at 100Hz (if you cover up the other 3 digits) but when viewing the whole 4 digit display and not moving your eye the 25Hz phase beat frequency is really visible, and annoying.

I'll wait for more takers before saying my thoughts on it.

 

[Diver300]

On an electronic note, I don't see where any 25 Hz is coming from. You have 4 digits flashing at different phases, but they are all at 100 Hz. There shouldn't be anything lower frequency happening at all.

I think that eye movement comes into what you can see. It isn't possible to hold your eyes completely still.

Fixation (visual) - Wikipedia, the free encyclopedia says :-
Visual fixation is never perfectly steady: fixational eye movement occurs involuntarily. That allows our visual system to ignore non-changing signals, so that we don't see flaws and damage in the retina, including the blind spot.

There is an effect called Saccadic masking. Saccadic masking - Wikipedia, the free encyclopedia
I think that mine is below average, so I see more while my eyes move which is why I can see lights as flashing more easily than many people.

 

[Mike - K8LH]

eye the 25Hz phase beat frequency is really visible, and annoying.

I'll wait for more takers before saying my thoughts on it.

Perhaps you're seeing some visual artifact due to your 50 Hz incandescent room lighting?

 

[john1]

The central pit is actually very small, and it is most concentrated at the middle of that.

The areas outside the central pit are dealt with in a more casual manner, and other inconsistencies are often also visible.
Such as the second-hand of a digital clock may seem to miss a movement if one is looking near the clock, but not directly at at. I think this is some brain function which kind of estimates the vision because the true information does not always get seen by areas just outside the central pit.

I have not tried this particular effect that you mention, but my guess is that it is related to the area around the central pit, still very small.
I would guess that the concentrated area of the central pit covers about the visual size of a new penny at around six feet. The higher concentrations are much smaller.

Looking at four LEDs in a row could easily span way outside even the edges of the central pit.

If you draw back, so that the row of four falls on a more concentrated area of retina, do they assume a more continuous appearance ?

If so, you may be able to estimate the areas of high resolution of your eyes.

Regards, John

 

[Mr RB]

It's not related to eye movement. Although that was an interesting read on saccadic masking.

It's nothing to do with mains frequency, the 25Hz beat frequency is real and obvious. The complete phase cycle of multiplexing 4 displays occurs at 25Hz, which can be seen easily.

Sorry John1 I didn't try different distances but the effect is so obvious and doesn't seem related to the position of the eye so I'm not sure it would matter.

 

[Gauthamvg]

Mr RB, as you said if 4 led are flashing at 100 hz but when multiplexing, their frequency is phase shifted for 25 hz
i.e 2nd led will lit after 25% of the First led's cycle. for example, if all the time taken by the all the 4 LEd in one cycle is 4 seconds means the second led will lit 1 sec after the first (consider lit time as 0) so the frequency taken for i led is 1/4 the the actual. i.e 25 hz which can be easily detected by humans.
This can be explained easily by slow motion video. i'll try to take that

 

[Diver300]

But if you flash 4 LED displays, ie 4 digits in a 25% duty standard digit multiplex, and each one is flashing at exactly 100 Hz and 25% duty cycle

It's nothing to do with mains frequency, the 25Hz beat frequency is real and obvious. The complete phase cycle of multiplexing 4 displays occurs at 25Hz, which can be seen easily.

Which one is it? In the first post of this thread, you said each one flashes at 100 Hz. Now you say the whole cycle occurs at 25 Hz. Those are different.

Just because there are 4 separate phases does not change the frequency. As an example, people spread around the world see dawn at different times. But place 4 people each 90 degrees longitude apart, and ask them all to time how long from one dawn to the next, they will all say 24 hours. Just because the are 4 of them, doesn't mean something magical has happened to the experience for any one, just because the others exist.

 

[Mike - K8LH]

Roman, you lost me too. I asked earlier if the four digits were lighted for 2.5-msecs each for a total 10-msecs (100-Hz refresh rate). Is this the case or are you talking about something else?

 

[unclejed613]

is the 100hz clock driving a 4-digit mux? if so, each digit is lighting up at 25hz. the strobe line on each digit might be running at 100hz, but each digit is only turned on every 4th pulse.

 

[Mr RB]

The digits each flash at exactly 100Hz (10mS period). Each of the 4 digits is on for 2.5mS, ie 100Hz * 25% duty.

The big question, is; Each of the digits is flashing at 100Hz, and if any digit is viewed totally alone it cannot be seen to flash. But when you view the entire 4 digit display you can easily see the display flashing... Why?


For anybody that wants to play with one and see for themselves, it only needs 4 LEDs and minimal code;

// assuming 4 LEDs on PORTB 0-4
while(1)
{
  PORTB = b'00000001';
  Delay_us(7200);
  PORTB = b'00000010';
  Delay_us(7200);
  PORTB = b'00000100';
  Delay_us(7200);
  PORTB = b'00001000';
  Delay_us(7200);
}

Hmm, I just tested with 4 red LEDs spaced 7mm apart and its nowhere near as effective as with 4 green 7seg displays (5/8" tall). I had to drop the freq very low to about 35Hz or 40Hz (see code above) to try to get the same effect...

I guess the larger display activates more optic sensors than 4 tiny leds... Or maybe the green light sensors in the eye work a bit differently. I don't have the green display handy, this was a many weeks back when I was doing a project noticed the effect and I'm sure it was much higher than 35Hz freq.

Anyway my theory was that the eye can see each digit flashing (just) but more as a small ripple in the brightness than as a "flash" as it is still just under the freq where the eye sensors filter it out. Then the brain acts as a second filter to reduce the small flash ripple to get rid of noise.

So there's basically 2 systems; the eye sensors (hardware + integrator) which can still just detect the ripple, and the brain (software) that deliberately filters out the one single flashing object to make it a constant light.

But when there are 4 flashing objects in sequence, the eye sensor still detect the same very small ripple in each object, but now the brain doesn't do the second filter, or even does the opposite and pays attention to the 4 *different* flashing objects... Like the brain identifies the very small 1234 ripple as "movement!" and focuses on it. That was my theory anyway.

It could be just as likely that Diver300 was right and it might be simply because the larger display size of the 4 digits within the viewing area of the eye means that part of the display is in the field of view of the outer eye sensors that are better at detecting movement.

 

[transistor495]

I tried my best to understand what's happening here. Let me explain it in my own words and plz correct me if I'm wrong.

*) So what I got is multiplexing the 4digit display with 100Hz means each digit is lit for only 1/4 sec of the whole 100Hz pulse and one by one.

*)ie, each digit is theoretically provided a 100Hz pulse but it's divided into 4digits by 25%.

*)Important point: So When viewing the 4digit display as a 'whole' our eye+brain assumes that it's running at 25Hz and there comes the 25Hz phase beat frequency Mr RB mentioned....

*)But in the case of masking 3digits and viewing only 1digit -our brain understand that it's running
at 100Hz but actually it is continuously picking up 25% of 4 consecutive pulses.

*****theory ends*****
Someone need to kick me outta here if I'm utterly wrong BTW

 

[Diver300]

Firstly, I don't think that there is anything happening at 25 Hz at all.

Here is my simple theory. We are sensitive to movement, and our peripheral vision is more sensitive to movement than the central spot.

With 4 LEDs, they are too far apart to all of them at the same time with your central vision. The multiplexing from one side to the other appears as movement, which we are sensitive to in the peripheral vision.

(Reference. The Rods and Cones of the Human Eye says "You can detect motion better with your peripheral vision, since it is primarily rod vision")

Two other points:-

Our sensitivity to movement had obvious evolutionary advantage, especially in the peripheral vision as it increases the chances of us being able to see something creeping up on us.

If we take a wheel spinning at 6000 rpm (100 Hz) that has a blob of paint or something on it, we might be able to see that it is spinning. That would be like having the 4 LEDs

If we hide most of the wheel behind a screen so that we can only see 1/4 of it, the blob will appear for 1/4 of the time, in a smaller area. That will be like having the one LED.

It doesn't really surprise me that it would be easier to see that the wheel is spinning if all of it is view than when only a small part is in view.

 

[Mr RB]

Diver300-

Firstly, I don't think that there is anything happening at 25 Hz at all.

Sorry, that was my fault! I was going on memory and remembered something about 100Hz and 25Hz beat frequency, sorry everyone that 25Hz thing was my error. I've edited above with the new freq code. I can't remember the exact multiplex freq I used on that display it would have just been a fluke it was at the right freq to cause that effect, then I increased the freq a lot to fix it and finished the project.

Diver300-

If we take a wheel spinning at 6000 rpm (100 Hz) that has a blob of paint or something on it, we might be able to see that it is spinning. That would be like having he 4 LEDs

If we hide most of the wheel behind a screen so that we can only see 1/4 of it, the blob will appear for 1/4 of the time, in a smaller area. That will be like having the one LED.

It doesn't really surprise me that it would be easier to see that the wheel is spinning if all of it is view than when only a small part is in view.

I like your wheel analogy! It's still interesting to think whether this is just an eye sensor (hardware) phenomenon or eye+brain (hadware+software) phenomenon...