TLC5929: Timing Problem Cascading TLC5929

Hi Michael,

Please note that the output SOUT is driven after a small delay (tD0) following the rising edge of SCLK.

Are you also using a buffer at each board? This may cause the problem. You could refer to the TLC5929-EVM, the boards can be cascaded.

Best regards,

Suchen

Hi Suchen,
thanks for the fast reply. The delay tD0 is clear so far. 

We are using an HCS245 on each board. Your are using additional pullup-resistors on data lines, what ist the reason? What kind of buffer do you mean?

Following timing diagram shows the Data Line on first, second, 12th IC and also clock and strobe. Hope this would be helpful

Best regards,

Michael

Hi Michael,

Michael Brandmaier said:

We are using an HCS245 on each board. Your are using additional pullup-resistors on data lines, what ist the reason?

It's for compatibility. It's not necessary.

I'm worrying the propagation delay and mismatch of different signals when transferring through lines between boards will influence the data recognition. Typically, cascaded LED drivers are on the same board. In that case, we may need to provide a host controller port for each board.

What's more, you could try bypass HCS245 to see if it's caused by its delay and mismatch between channels.

Best regards,

Suchen

Hi Suchen,

to prevent misunderstanding:

We have 72 LED drivers on one board. The timing represents the inputs to the LED drivers. So we have the shift before the 2nd driver, then always the same. If we connect a second board ( one HCS245 at the input and one HCS245 at the output), then we also have a shift of half a bit after the first LED driver, but no more after that.

Bypassing the HCS245 results in the same output.

I can also send you the shematic per E-Mail, if its interesting.

Best regards,

MIchael

Hi Michael,

So, what's the current status? Does the 72 LED drivers on board work normally? Because the circuit seems work normally in the past thread you mentioned.

It's reasonable to have such waveforms for all the SOUTs are driven on the rising edge of the SCLK.

Best regards,

Suchen

Hi Suchen,

The drivers on board work normally. But when we cascading two boards, on the second board all data are shiftet one bit. The waveforms look exactly the same, after the first LED driver data shifted half a bit. Therefore the second board doesnt work normally. You have any idea what the problem could be? It is not necessary, that HCS245 is assembled or not, the waveforms are exactly the same.

I saw simillar problems from the thread with the TLC5917 and wonder, if there is a general problem with this type of drivers?

Best regards,

Michael

Hi Michael,

For the first board, the initial device shifts the signal by a half cycle. This is caused by the output timing of the TLC5929, as explained. The other devices on the same board operate normally. For the subsequent devices on this board, the rising edges of the clock and data signals will be very close, but this does not affect their operation.

Regarding the other boards, the signals go through transmission lines. This may be the cause, as signal recognition could be affected by the time they reach the TLC5929s on the second board.

In this situation, it's better to build a cascaded system where all LED drivers are connected relatively closely, or give each board with an individual controller port.

Best regards,

Suchen