TPS54620 LED Driver synced to phase node

First off, the maximum input voltage of the TPS54620 is 17 V..let me know if you made a mistake on your input and output voltages.

Hi Marshall,

Thank you for your interest in TI parts. There are a few issues I see. The TPS54620 has a max input voltage of 17V. Your stated input is 36V, is that the input voltage for these 6 switchers? If so, you will need to select a different device.

The basic architecture of using the PH node as a master CLK is possible, but your instinct regarding issues down the chain is correct. Any device will have jitter, but using a master that contains jitter will result in more jitter in the next link in the chain, repeat. You will likely end up with more jitter in the second or third link than you really want. The jitter will be present in both the frequency and the pulse width. The frequency jitter may actually improve the EMI performance and is akin to FSS (frequency-spread-spectrum, deliberately designed into some devices for just that purpose), but you won't have much control on the amount of jitter.

1) Yes, on the other hand, routing the PH node to places other than the choke pin will definitely degrade the EMI performance. Typically we suggest placing the choke as close to the device as the pick-and-place rules will allow, and limiting the PH node copper to be just big enough to cover the associated pins. The PH node is not the right place to add copper for heat sinking purposes because it is the noisiest node and will corrupt other nodes. If you do use the PH node as a sync CLK, you will want to place the top divider resistor as close to the PH node as you can so as to avoid routing the actual PH node.

2) The PLLs might actually help in this situation since they have a response time, as opposed to a hard sync or logic sync. It may depend on the overall layout. If the layout is not optimal and the noise is high, the PLLs may have trouble locking. If there were no noise or jitter, then they would behave well.

3) Some sort of master CLK is the better way to go, but it doesn't need to be a micro. If you want N-different phases, then you will need a master that is N-times higher, followed by division and phase generation. A master can be designed to be much more immune to ambient noise. A ceramic resonator is a lower cost viable option.

4) The beat frequencies of 6 non-synchronized switchers can be anything. The beats will literally be the difference in frequencies, but with 6 switchers it will likely be complex. They could be above, in or even below the audio band. The tolerance of the un-synced oscillator is such that the differences can be zero to several kHz between two devices. Again, depending on layout, the switchers could end up sync-ing together or partially sync-ing (sometimes synced, sometimes not, some yes, some not. This would not be well controlled).

But back to the first issue, if 36V is the input voltage to these switchers, you will need to select a different device.

MC.

Doh!

    I meant TPS54260! It does 60 volts in.