TLC5917: TLC5917 power up problem in series

Hi Ryoko,

Here are some things I need to confirm:

1.  Can you power the final product design with a DC power supply to see if the situation still appears?
2.  For the 2nd TLC5817 with this situation at the power up, can it receive new data and recover to normal output state?
3.  Are data sent over communication lines during power up?
4.  Does this situation happen occasionally, or all the time? Does this happen to a single DUT, or all the DUTs? 

Best Regards,

Steven

Hello

Thanks for the reply

Here are some answers to your questions:

Yes, we are using the final product design, tested with a very fast startup bench supply which works fine.

But the product is used with customers supply, which is usually a DC switching supply with other devices connected. Hence it has a slower startup to full VDD.

This seems to be when the problem happens.

There is no data sent during power up, we have a delay of about 500ms before data starts, this is definately after

the power supply is stable.

The situation happens almost every time, probably at least 95%

Sorry I dont understand what you mean by DUT, if the LED drvier then It only happens on the 2nd TLC5917

Thanks

Hi Ryoko,

1. DUT = Device Under Test. Sorry for any misunderstanding it may cause.
2. Have you tried holding OE high during power-up? Will the phenomenon still appear? I would recommend holding OE high during power-up since the data latch may be at an unknown state at that time.

Best Regards,

Steven

One supplement:

Can you kindly answer below question?

Steven Li said:

 For the 2nd TLC5817 with this situation at the power up, can it receive new data and recover to normal output state?

Best Regards,

Steven

Hello

Sorry we dont have control of the OE pin, it is tied to Ground

After power up the 2nd TLC5917 is always scrambled, it never seems to recover

Thanks

Hi Ryoko,

When will the scrambling start? Before the data transmission?

Can you give me an oscilloscope capture of your power rail, output pin, and communication lines to the 2nd device?

Best Regards,

Steven

Hello

The 2nd TLC5917 is scrambled on the first data transmission

I cant give you a scope picture at the moment, probably tomorrow

Which output pin are you referring to ?

Hi Ryoko,

1. I refer to the flickering / wrongly selected output pin of TLC5917.
2. Please provide scope captures at power-up and at the first data transmission.

Best Regards,

Steven

Hello

It is all of the LED driver output pins on the 2nd TLC5917 that are not outputting the expected data

We only have a 2 channel scope, so will have to make a few different scope shots.
Also the power up is a long time [ in SPI terms ] before the data transmission, so

to get both of these in one scope shot will not show much SPI detail

Hi Ryoko,

No. Just one malfunctioning channel of TLC5917 is ok. 

Two or three captures should be enough: 

• VDD pin of 2nd TLC5917 at power-up
• CLK, SDI, OUTn pin of 2nd TLC5917 at first data transmission when blinking / malfunctioning starts.

BR,

Steven

Hello

Here are the scope shots of your requested items:

TLC5917 VDD Pin Power up

TLC5917 SDI pin Blue trace, TLC5917 LE pin Red trace

TLC5917 OUT8 pin [ connected to LED ]

TLC5917 CLK pin

Thanks

Hi Ryoko,

Thanks for your feedback. See my comments below:

1. Why does the power-up curve look so weird? How do you power the chip?
2. What is your VLED?
3. From your scope capture, I guess you want to achieve a shifting LED effect. But the channel 7 (The "OUT8" you wrote above) does not turn on/off as expected. Is my understanding correct? If so, have you checked whether the communication signals satisfy the timing requirements of TLC5917?
4. Could you check the SDO of the 2nd TLC5917? Does it shift out the correct data as expected?

Best Regards,

Steven

Hello

Thanks for the reply.

Yes the power up is indeed not idea, but unfortunately that is how the power supply seemsto work on customers units. We build modules that are used in customer racks, which have a wide range of power supplies. The scope trace is from a popular brand, which uses cheap DC-DC converters for the power.

Unfortunately there is nothing we can do to change this, as it is big market, and we are a small company !

We have put in a lot of decoupling, and inductors on the power side to isolate it from the worse stuff, but this is about as good as it gets :-(

Just to clarify as before, when powered from a fast rising power up, from alternate bench supply the circuit functions as expected.

The LEDs we are using are 150ma forward current from Lumix SSL-LX3059IGW-CA, anode is at 5v

The scope trace shows two 8bit transfers. The first byte shifts into and out of the first TLC5917 and into the 2nd TLC5917, the second byte resides in the first TLC5917.

There appears to be no coherrant activity on the LEDn pins of the 2nd TLC5917

We can scope the SDO onf the 2nd TLC5917, and get back to you.

Thanks

Hi Ryoko,

Thanks for your feedback.

1. Please check the SDO of the 2nd TLC5917 to see if it shifts out the correct data as expected.
2. If not, please check whether the communication signals satisfy the timing requirements of TLC5917. (You can give me a zoom-in scope capture of the SDI and CLK pin of the 2nd TLC5917.)

Best Regards,

Steven

Hi

Sorry for the delay, it was hard to scope the tiny pins on a TSOP  !

Here is the SDO output of the 2nd TLC5917 when the chip works ok [ using fast power up ]

Blue trace is LE, Red trace is SDO Pin

Here is the SDO output of the 2nd TLC5917 when the chip doesnt work [ slower power up ]

Here is the SDI in on the 2nd TLC5917 [ from the SDO out of the 1st TLC5917 ]

Blue trace is CLK, Red trace is SDI Pin

Here is the SDI in on the 1st TLC5917 [ from the SDO of PIC MCU ]

Blue trace is CLK, Red trace is SDI Pin

Thanks

Hi Ryoko,

Ryoko Tanaka said:

Blue trace is CLK, Red trace is SDI Pin

I guess the blue trace is SDI and the red trace is CLK. Right?

There seems to be an abnormal timing mismatch in the third capture.

Could you repeat this testing procedure on the 2nd TLC5917 on a fast power up and show me the scope capture?

Best Regards,

Steven

Hello

Oops, sorry yes I mislabled the trace ! you are right Red is CLK, Blue is SDI/SDO

I think the timing of the pulse out of the first TLC5917 is the same

on either setup [ good power up, or bad power up ]

I'll check now and get back to you

Thanks

Hello

Here is a scope trace of the SDI input pin of the 2nd TLC5917, with the bad power up [ 2nd TLC5917 not working correctly ]
CLK is Red trace, SDI is Blue trace

Here is a scope trace of the SDI input pin of the 2nd TLC5917, with the good power up [ 2nd TLC5917 IS working correctly ]
CLK is Red trace, SDI is Blue trace

thanks

Hi Ryoko,

Sorry. I find I read the previous 3rd capture wrong. Sorry for taking your time. I sincerely apologize for that.

I don't see any timing problems from the SDI&CLK captures. 

1. Does the VLED and VDD shares the same power rail? If so, could you kindly monitor if there is a glitch on that power rail?
2. The root cause is uneasy to capture from inspection. I will discuss with some senior experts from my team tomorrow morning about this case. 

Best Regards,

Steven

Hello

Yes the VLED shares the same power source as VDD.

Here is a scopt picture to show the VDD line and LE line
VDD in Red, LE in Blue

Hi Ryoko,

I am really sorry. Our experts are out of office today. I have arranged a talk with them tomorrow. I will come back to you within 24 hours.

BR,

Steven

Hi Ryoko,

Can we move this thing to email? If it is ok for you, please contact me at steven-li@ti.com

Best Regards,

Steven