Other Parts Discussed in Thread: LM5109, LM5111
Compared to the LM5106, the LM5109 has two inputs - one for low side and one for high side. This means that it is "easily" possible to achieve three different drive options with the LM5109 : only drive the Low Input, only drive the High Input and possibly (with dead time added - which may not be quite as "easy") drive both. In essence, the LM5109 can be used as a Low Side, High Side and Half Bridge driver, it's just down to how you configure the signal source.
As an example of how and when this could be useful, consider for a moment a load which is connected to something other than the half-bridge's VBUS - you may want to control the Low Side transistor of the half bridge, and you may be happy for the load (if inductive) to fly back up to VBUS (forward biasing the high side transistor's body diode). What you do not want to do, though, is to force the controlled end of the load up to VBUS when it should be "off". Moreover, even if it was OK to drive that side of the load up to VBUS, at 100% HS duty you would eventually discharge the bootstrap cap - you would need to cycle the output to recharge it and you may not want (or be allowed) to.
In contrast, when used in its intended way, the LM5106 can only act as a Half Bridge driver [so long as it is ENabled, either HO or LO will be on] - it "could not" be used as per the example above, since you never want to turn on the high side transistor (content with it acting as a flyback diode to VBUS). This then begs the question : would it be "permissible" to reverse the allocation of the ENable and IN pins ? [In case someone is wondering "why not use an LM5111 if you only need low side drive?", consider flexible hardware which is not application specific and/or where you cannot predict what sort of load you will need to drive, ergo the output stage must cope with as many scenarios as is realistically feasible].
Using the same situation as above, we see we need to generate an LS PWM signal and we do not want to drive the HS. This is easily achievable by driving IN to 0 (so LM5106 will drive the LO) and then applying the PWM signal to the EN pin. There will be around 36ns of delay as per datasheet both on turn on and off, so it won't distort the applied duty. It doesn't need to worry about dead time because the IN pin will not be changing in this drive mode.
It would appear, at least per the information in the datasheet, that an LM5106 can do "everything" an LM5109 can but better in one regard - the signal source does not need to insert dead time. That can be of some benefit where adding such dead time is hard / computationally expensive or impossible in the signal source. Whilst this potential drive paradigm still requires two PWM capable pins, it sacrifices the ability to change the dead time in the signal source (if the source is even capable of that) for the ability to drive a half bridge without needing to add the dead time at the source.
My question here is whether anyone can see any reason why this would not work ? The datasheet doesn't consider this drive paradigm (understandably so) yet states nothing that suggests it wouldn't work. The Simplified Block Diagram looks like it should work. But a Simplified Block Diagram is just that, and may not cover nuances which would preclude this drive method.
Best regards,
Patrick Herborn.
Appendix : For full bridge operation with 2 LM5106s it is still possible to pre-charge the bootstrap caps by doing a single cycle with both INs at 0 and ENabled. When using only one LM5106 switching only the high side, the ground connected load would provide the charge path for the bootstrap anyway.
