Periodic Oscillation of LED Intensity

Hello Joe,

Can you please share details about how are configuring the EVM?

1. Operating Mode? Video or Pattern Display

2. Kind of image displayed on the DMD?

3. LED current settings?

Regards,
Sanjeev

Hi Sanjeev,

The mode is pattern display, and the sinusoidal oscillations are present when the image source is the HDMI input, and also for the images stored in flash memory. The video displayed via HDMI is binary black and white.

The LED current setting is the default, but the oscillations are still present when changing the LED current, and the peak to peak of the detected oscillations is constant, independent of LED current. (only source of light is the DLP, room is dark)

Best,
Joe

Hi Joe,

To understand better; please provide below information -

Can you attach the GUI setting snapshot?

Can you also attach the Photodetector readings plots to show oscillations in the captured values?

Regards,
Sanjeev

Below is a voltage/v against time/s plot of the photodiode detector reading. The large spike near 0.07s is probably the DMD clearing after a pattern exposure.

Here is a screenshot of the UI, and the static black and white image being displayed via HDMI.

Since you have configured the source as external RGB video input, let me know what is the incoming source rate?

A pattern display time of 1000000us requires the source to be @ 1Hz. The DLPC350 works b/w 10Hz to 120Hz so I suggest you to look at reducing the pattern exposure times. To begin with experiment for 60Hz fps, for this you can set value to like 16000us.

Regards,
Sanjeev

Hi Sanjeev,

I'm Joe's colleague who is also working with the DLP LightCrafter 4500. We are running the DLP as if it were a second screen to the computer from which we are controlling it from, and this second screen is set with a refresh rate of 60Hz (and resolution of 912 x 1140). I have tried your suggestion of setting it to 16000us, for both input sources (flash and video mode) but it does not change this oscillation. I attach the screenshots below. Any idea of where these oscillations may be coming from?

Video mode:

Flash input mode:

Best Regards

Cheryl

Hello Cheryl,

Sorry for the delayed reply; i was out for sometime.

There are two things in the w/f for either case -

A)  vertical lines going all the way to 0.  This is clearly happening after DMD exposure completion for example in 100ms at 16.67 ms you will find 6 dips 100/16.67.

B) Sinsuoidal behavior - I am guessing it is more to do with the LED driver and LEDs analog output, DLPC350 or DLP4500 not coming into the picture for this variation. Looks like for the same current and exposure duration, the intensity building for three iterations and falling after that in cyclical fashion. Just to rule out the Detector you can look at the analog output of the detector, when it is looking at LED powered directly i.e., without modulating.

Regards,

Sanjeev

Hi Sanjeev,

No worries! Thank you for the reply!

We have already ruled out the possibility that it could be due to the detector as the oscillation was not observed when the detector was used to measure another light source. In addition, we have changed the detector during our experiment (for other reasons), and the same oscillation was still observed. Is there nothing we can do to minimise or even remove this oscillation? Because it means that since a constant intensity is important for us, we can't run the DLP at 60Hz, but rather we have to run it at 1/0.06 = 16 Hz, which is much slower than what the DLP can really do.. which is quite a shame :(

Best Regards

Cheryl

Hello Cheryl,

I think we have a way i.e., to keep LED constantly ON in Pattern Sequence Mode.

Please do the following -

Under the GUI, LED Driver Control-> LED Selection -> select Manual [Radio Button], select Green [ Check box] then click on the Set [button].

This will keep the selected LED ON all the time, now please do the measurements.

Let us know if it helps.

Regards,

Sanjeev

 

Hi Sanjeev,

Unfortunately, we have tried this and the oscillation still persists..

Best Regards

Cheryl

Hi Cheryl,

Okay; to be sure it is still not from LED driver, you can disconnect the LED connectors from the board and provide external power to the LEDs, keep them ON all the time, make sure you are not overrunning of the LEDs causing damage.  Check for the oscillations.

Regards,

Sanjeev

Hi Sanjeev,

Do you know of anyone else who has experienced the same problem? I tried to google for this but did not see anyone who reported seeing the same oscillations. Do you think it is possible that our DLP is defective? Or is this normal for DLPs?

Thank you.

Best Regards
Cheryl

Hi Cheryl,

We haven't come across so far. Or it may be that the users may not have looked into the this aspect. I don't think DLP is defective. The periodic oscillation of intensity is some other factor not from DLP/DMD for sure.

Did you get a chance to drive the LEDs from external power supply?

Regards,

Sanjeev

Hi Sanjeev,

I am just about to try that. Fizix gave me some advice about the specifications of the LED driver in another post:

(http://e2e.ti.com/support/dlp__mems_micro-electro-mechanical_systems/f/924/p/466997/1679907#1679907

I will try it and keep you guys updated in that post. Thank you very much for your advice!

Best Regards

Cheryl

Hi Cheryl,

We have noticed this same issue. Please let us know if you make any headway.

Thanks,

Mark

Hi Sanjeev,

I think I found the reason for the oscillations. I believe that it's due to the voltage input to the LEDs being modulated by PWM. When I set the LED current to 255 (maximum), the oscillations disappear, as would be the case if PWM was causing the oscillations. I attach the measurements below. The data has only been filtered for the spikes in the data. 

I have a follow-up question though. The manual says that :

"Exceeding more than 4.3 A for continuous or simultaneous LED operation can damage the LightCrafter 4500 LEDs."

Hi Cheryl,

Thank you for sharing your experimental results. That proves the LED driver associated circuit causing issues at lower PWM current settings.

We haven't characterized the LEDs on the light engine for long duration tests. In general use case the LEDs turned ON/OFF during normal operation.

You may want to check with the LED manufacture, details can be found from the link ( https://e2e.ti.com/support/dlp__mems_micro-electro-mechanical_systems/f/924/t/278008 ). I am guessing there may be need for thermal cooling for the LEDs if they are operated in continuous mode. So better to check with the LED manufacturer.

Regards,
Sanjeev

Hi,
I'm Mark's colleague (post above). As Mark said we have seen the same 15Hz oscillation which goes away if the LED currents are set to maximum. After a good amount of head-scratching, we think that this 15Hz signal is not real, but an aliasing artifact from digital sampling of the photodiode signal. When we increase the sampling rate on our ADC (originally 10Khz), use a faster digital oscilloscope, or an analog oscilloscope, the 15 Hz oscillation disappears. With sampling rates greater than ~50Khz there is no 15 Hz oscillation, but there is an oscillation around 10-15Khz. This fast oscillation does appear to be real, and also disappears at maximum LED currents. For our purposes this is oscillation is way too fast to be a problem.

Cheryl - I suggest looking at the photodiode signal with a faster sampling frequency or with an analog oscilloscope to see if your 15Hz oscillation disappears and a faster (~10-15Khz) oscillation appears. If you check this, would you please let us know the results? It would give us some more peace of mind that the 15Hz signal is just from aliasing this faster oscillation.

Best,
Max

Hi Max,

Thank you for sharing more insights into the problem. We appreciate your contribution to the forum.

Regards,
Sanjeev