DLP651NE: dithering and flickering question

thanks for your response... it answered partially my questions. I need to know on dithering.

1-A- you said "he actual time between mirror transitions will generally vary between ~8us and ~400us." why 400us. you mean to scan through entire mirrors or each mirror takes 8-400um. 

1-B ==> mirror transition in series or in parallel- all at once-? 

2- does DMD dithers in Off and Flat sate as well.

3- what is dithering range in angle in On-State and Off-Sate ? 

4- what is the dithering speed (frequency) in On-State, Off-Sate.==> is it like 2.5us too?

thanks

Babak

well explained ....thank you... now I understand my problem. Last question  on what we see in our test:

Question 5-
- DMD dithers to avoid sticking of mirrors. correct? 
- And now if DMD is in ON State and since DMD only have two state: What is the duty cycle  of dithering in ON state?  like 99% ON (mirror stay in +24 degree) and 1% Off (-24 degree). and what is the full period of dithering in msec or usec

Samwise, What is the duty cycle  of dithering in OFF state?  like 99% OFF (-24 degree) and 1% ON (+24 degree).

asking as we dump the beam  to empty space so no scattering coming to system but we see some beam comes back to system in dithering frequency... it has a beating pulse.. 
Please let me know what is the Ona nd Off duty cycle for dithering. And can we change the duty cycle. we need to have minimum duty cycle in ON mode  when we are OFF state to have almost zero background.


we want to know this % and if we can change it... 
thanks for response

I am waiting thanks 

Ok now we are talking....
so I graphically plotted above what % in constant On/Off state mirrors flip or swing or what I called it "dither".

please confirm or reject my logic below. This is what I understand form you==>

A- Although mirrors stay 99.5% On/Off states, a pwm sequence always running, loading data and generating mirror clocking pulses. As a result, even if the input data is unchanged,  like I am always On or always Off,   This may cause the mirrors to wiggle or deflect slightly from their on or off positions. like 99.5% On/Off and 0.5% Off/On. 
B- this pulses are approximately generated every ~8us to ~400us, data is loaded and mirror clocking pulses generated

is above what I said correct? thanks for Ye or Ne

is this any different that DLP9500. I am comparing our cheap  DLP651NE to ur more expensive version DLP9500.... thanks 

Babak

thanks Gary... we are almost done......

in short  I am comparing DLP9500 vs DLP651NE... do we see any 99.5% on/off and 0.5% flat/flat (~4~6) case when we are holding DMD in On/Off state? 

we are purchasing currently both DLP9500 and DLP651NE

thanks

Babak

aha... so DLP9500 differs from DLP651NE that the mirrors only move or attempt to move when you tell DMD to do so when Data is loaded, when we want it to (of course following some data-sheet rules or recommendations around how long you should leave them on or off). While in LP651NE  every ~8us to ~400us, data is loaded and mirror clocking pulses generated which causes DLP651NE move roughly half way to flat state. This does not occurs for DLP9500 when mirror are kept at On or Off stated as there is no  PWM sequence. (again this is of course following some data-sheet rules or recommendations around how long you should leave them on or off) . It is as in case of DLP651NE due to nature of Video control board , a PWM is always running in DLP651NE as loading data and generating mirror clocking pulses  every ~8us to ~400us. 

if above is correct I close the case. I need your confirmation to get my narrative correct to report to VP of engineering. 
thanks

if above is true... mirror function in DLP9500 doe snot need to move (of course following some data-sheet rules or recommendations around how long you should leave them on or off). I heard in all DMD mirrors have to move back and forth constantly (On/Off mode to flat state) else mirror stick and will not function... perhaps info was not correct. Now you know why I repeat my question and need your/expert confirmation. than following chats on internet or verbal/tribal knowledge. I need engineering facts. 

thank you

Babak

Hi Gary
We followed what you said ... all about PWM and mirrors refreshing whit DLP651NE. He mentioned due to PWM streaming,  mirrors that are stayed in ON/OFF will stay in ON/OFF state 95% and 0.5% will be in OFF/ON state. Gary also helped us and answered my question and mentioned mirrors transition from +/-12 degree to ~4-6 degree and not completely back to OFF/ON  (-/+12) degree. We conducted a test using laser as follows and we noticed mirrors refresh as Gary said. But it refresh fully from 95% in  +/-12 degree (ON/OFF)  back to fully -/+12 degree (OFF/ON) 0.5%  and not ~-/+ 4-6 degree.

This is causing an issue for us. Can you please advise why mirrors doe full swing 0.5% of the time and not~4-6 degree?

Is there a way  away to make this 0.5% swing rather than -/+12 degree to -/+4-6 degree?

here is what we did: 

 Test of laser on DMD==> to proof when we go from ON state to OFF state, what we call "dithering or swing or wiggling of mirrors" does or does not happens and in what percent and how much.

1- We created a power point presentation with one page all white (to force DMD ON) and one page all black (to force DMD off)

2- first when DMD is powered off -its micromirrors stays flat and it is called flat state

3- DMD is turn on while white page was on monitor connected with HDMI to DMD board. then we go on monitor white page-black page-white page -black page... to move mirrors from ON to OFF state.

As shown in video below, exact same spot that is bright on white page as ON state, will goes dim- NOT OFF- and spot that is dim goes ON bright on black page.

Note: due to temporal coherent of laser, laser creates may spots but position of spots does not change but dims when power point on monitor goes white (ON)  and black (OFF). since position of a bright spot goes dim it means mirrors are wiggling but in full +/-12 degree and not +/-4-6 degree . If mirrors were wiggling +/-4-6 degree position of dim spot would of changed.   

Hi Gary.... we kindly need your input... we are in middle of our project... appreciate for responding above Thanks 

Hi Mr. Gary,
Thanks for response. you did not tell me about once in a 60HZ full tilt/wiggle ...I asked specifically (see through our back and forth communication above), is 0.5% corresponds to +/-12 degree and you mention no it is ~4-6 degree and never mentioned once a 60HZ… ok as long as I know full story I am happy .. thank you.

Our question is focused merely on tilt and wiggling of mirrors...  so there two types of mirror wiggling/tilt/moves... In a high-level description :
1- A non-intentional bounce ~1/2 way to flat (around 4-6 degrees) as a side-effect of how the video application and continuous running of PWM sequence.  The PWM sequence is always loading and resetting the mirrors every ~8us - ~400us, even if the data is the same.  When the data loaded is the same as the current pixel state, the load and reset (clocking pulse) results in a little "bounce" of around 4-6 degrees (roughly half-way to flat) over roughly 1-2us.

2- These 0.5% transitions are intentionally inserted and are full on / full off.  We specifically want all the mirrors to move fully (+/- 12 degrees) at least once per 60Hz frame time, even if the video data is full white or full black for a long period of time.  At the moment of full move (+/-12 degree), the illuminators (RGB illuminator enables / strobes) are disabled / off during this time, so this is not visible on the screen in a video application.

• Please kindly advise if above narrative is correct.
• Please kindly advise if there is a trigger in controller cart that we can use to know the intentional swing for +/-12 degree. We need to use this to trigger off the light source than shooting it somewhere else.
• Is the occurrence of once every 60Hz is controllable? can we make it once 30Hz or once 120Hz.
• Is this occurrence of ever 60Hz going to suddenly change randomly from once 60Hz to 120 Hz or 20Hz… sure there is a little jitter in it maybe every 61Hz as an example…
• What causing every 60Hz occurrence change to some other rat of change value beyond user command.
• Is there any other mirror wiggling that I need to know beside of unintentional transition and forced every 60Hz and User acquiring it?

Thanks for all your support.

Babak

> Thanks for response. you did not tell me about once in a 60HZ full tilt/wiggle ...I asked specifically (see through our back and forth communication above), is 0.5% corresponds to +/-12 degree and you mention no it is ~4-6 degree and never mentioned once a 60HZ… ok as long as I know full story I am happy .. thank you.

I mentioned that the 0.5% on/off pulses are inserted once per frame in the first blue segment.  I assumed a video frame rate was implied.  60Hz is typical, but it could be 60Hz or 50Hz or 120Hz or something else. 

> Our question is focused merely on tilt and wiggling of mirrors...  so there two types of mirror wiggling/tilt/moves... In a high-level description :

> 1- A non-intentional bounce ~1/2 way to flat (around 4-6 degrees) as a side-effect of how the video application and continuous running of PWM sequence.  The PWM sequence is always loading and resetting the mirrors every ~8us - ~400us, even if the data is the same.  When the data loaded is the same as the current pixel state, the load and reset (clocking pulse) results in a little "bounce" of around 4-6 degrees (roughly half-way to flat) over roughly 1-2us.

Yes.  Keep in mind, this "bounce" assumes the data being loaded was the same as the already-displayed data.  If you are displaying full white or full black, this will happen most of the time.  These are same-side transitions (i.e. we are going from a 1 to a 1 or from a 0 to a 0).  Also keep in mind my previous note that this is a video system with dithering, gamma, color processing, etc..  Even with a full white input, you may see some pixels off sometimes.

> 2- These 0.5% transitions are intentionally inserted and are full on / full off.  We specifically want all the mirrors to move fully (+/- 12 degrees) at least once per 60Hz frame time, even if the video data is full white or full black for a long period of time.  At the moment of full move (+/-12 degree), the illuminators (RGB illuminator enables / strobes) are disabled / off during this time, so this is not visible on the screen in a video application.

> Please kindly advise if above narrative is correct.

Yes... but I would say you haven't mentioned the 3rd case... normal operation... every other possible input that is not full black or full white.  In this case

• Full ON/OFF transitions - pwm modulation of the mirrors to create the desired grey/color shade per pixel (typically every ~8us to ~400us).  
• Same-side transitions causing the ~4-6degree "wiggle" when the data loaded to a mirror is the same as its existing state.
• The inserted 0.5% on and 0.5% off pulses I mentioned causing the entire array to move to on and off (illumination is disabled, so these don't display) once per frame.

> Please kindly advise if there is a trigger in controller cart that we can use to know the intentional swing for +/-12 degree. We need to use this to trigger off the light source than shooting it somewhere else.

There are R/G/B LED/Illuminator enables that are used control the illuminators; they are all off when we generate the 0.5% on / 0.5% off pulses.  

> Is the occurrence of once every 60Hz is controllable? can we make it once 30Hz or once 120Hz.

As I mentioned, it is generally once per frame for LED/RGB-based illumination (things without a wheel).  60Hz is an example.  However, it is slightly more complicated....  At frame rates less than or equal 47Hz, we frame-rate convert the input to 2x the rate and run a faster PWM sequence, so you would actually see two sets of those pulse per frame at 30Hz.  

At 120Hz, you would see the 0.5% pulses once per 120Hz frame.

> Is this occurrence of ever 60Hz going to suddenly change randomly from once 60Hz to 120 Hz or 20Hz… sure there is a little jitter in it maybe every 61Hz as an example…

As you change frame rates, the embedded software will switch to a different PWM sequence designed for something closer to that rate.   There are various different bins to support different frame rates.  The typical rates of 48Hz, 50Hz, 60Hz, 120Hz, and 240Hz should all run at "1x" and have pwm sequences that run at that rate.  They should all be fine for +/1Hz off the frame rate (i.e. if your 60Hz is really 61Hz, it should work fine). 

You mentioned 20Hz... 20Hz is not supported.

> What causing every 60Hz occurrence change to some other rat of change value beyond user command.

> Is there any other mirror wiggling that I need to know beside of unintentional transition and forced every 60Hz and User acquiring it?

As I mentioned previously:

***This is a video system with gamma, dithering, color processing, etc.; so even with full white, some pixels may be off for short times due to the various algorithms.  

There are a lot of things going on inside the 7540.  It is a video solution with a lot of different features and processing for video.  If you take a look at the programmer's guide, you may get an video of how much is going on inside.  https://www.ti.com/lit/ug/dlpu082c/dlpu082c.pdf

I believe I have addressed the initial "dithering and flickering" question as best we can for this product.  At this point, we are well beyond the scope of what is supported for the 7540; it is a video display product for video display applications.  We really cannot document every possible case of the mirrors moving, not moving wiggling, etc..  The 7540 does not offer that level of control of the DMD to the end user.

If you need that kind of control for your application, the DLP9500 w/ DLPC410 that you already mentioned is likely a better fit for your application; that solution offers that level of control -- that is what it is for.

Regards,
Gary

Thanks for your response and dearly supporting us beyond.... we are close to reolve this case... few confirmation note need your help on this: 

1- where is this R/G/B LED/Illuminator enables that are used control the illuminators ... i see something here: See data sheet page 20 DLP65NE https://www.ti.com/lit/gpn/dlp651ne and DLPC7540 https://www.ti.com/lit/gpn/dlpc7540 

...please see below is this what you are talking about==> 

2-  does R/G/B LED/Illuminator enables that are used control the illuminators; will catch up with following rates: "As you change frame rates, the embedded software will switch to a different PWM sequence designed for something closer to that rate.   There are various different bins to support different frame rates.  The typical rates of 48Hz, 50Hz, 60Hz, 120Hz, and 240Hz should all run at "1x" and have pwm sequences that run at that rate. "

3- Our team showed me the advanced GUI for this eval board for DLP651NE. There are hundreds of settings there, maybe some can reduce the amount of period of flickering (60 Hz 48Hz... 240Hz) or even just fully command to off state (we could test how much this affects our lifetime). They think there are many cryptic settings with little or no documentation, but there is get and set support, so if we can figure out the command we need then it should be something we can set over USB to control board of DLP651NE. could you please advise. (I know I asked you if there is a way to change the PWM set.. or shut down or change flickering.... I told them 10 times I copy past your notes... they keep asking... thanks )

Thank you

>1- where is this R/G/B LED/Illuminator enables that are used control the illuminators 

The illuminator enables are on the DLPC7540 board.  In the schematic, they are labeled LED_EN_RED/GRN/BLU and are available on J97 or J115:

DLPC7540EVM Board Design Files (Rev. A)

>2-  does R/G/B LED/Illuminator enables that are used control the illuminators; will catch up with following rates

The signals that drive the LED enables are built-in to the PWM sequence; so they will match whatever pwm sequence is running at the time.

The flash image you have should be for an RGB LED system.  I believe it has ~6 cycles @ 60Hz, and one RGB cycle at 100hz or faster.  So, with a 60Hz input, I suspect you will see 2 cycles in 60Hz; so you should see BGRBGR in one frame.  At 100Hz, 120Hz, 240Hz, you would just see BGR. 

The color duty cycles are built for a specific LED configuration to get 6500K; so you will not see equal times for red, green, and blue.  I think you will see ~36.1%/45.16%/18.74% R/G/B. Duty cycles (one or more) are selected at design/build time and are built-in to the flash image.  I believe the GUI will let you read the current duty cycle back.

Note that the the illuminator enables are shifted to account for specific illuminator delays and rise/fall times; so they will not directly with the DMD's on/off time (they will be shifted I believe around 12.2us on the enables and around 3.1us on the disables).

>3- Our team showed me the advanced GUI for this eval board for DLP651NE. There are hundreds of settings there, maybe some can reduce the amount of period of flickering (60 Hz 48Hz... 240Hz) or even just fully command to off state (we could test how much this affects our lifetime).  They think there are many cryptic settings with little or no documentation, but there is get and set support, so if we can figure out the command we need then it should be something we can set over USB to control board of DLP651NE. could you please advise. (I know I asked you if there is a way to change the PWM set.. or shut down or change flickering.... I told them 10 times I copy past your notes... they keep asking... thanks )

The advanced gui exposes most of the commands supported in the embedded software.  It is an engineering interface for communicating with that software.  I think everything in it is in the programmers guide (https://www.ti.com/lit/ug/dlpu082c/dlpu082c.pdf). 

You can turn of some of some of the image algorithms (some of the color processing and gamma).  For what you want, I suspect you should disable all the image algorithms that are available to disable.

I do not believe it has the ability to disable dithering or the pwm seqencer; those are fundamental to normal (and safe) operation. 

You should be able to select a full black or full white test pattern (among a few others, I believe).  The PWM sequence should still run and show the behaviors we have discussed above.

Regards,
Gary