• State
  • Locked Locked
  • Replies 3 replies
  • Subscribers 63 subscribers
  • Views 769 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LM3429 issues, moving to a 4 layer PCB questions

Craig Easdale19
Expert 1425 points
Other Parts Discussed in Thread: LM3429

My very compressed layout for an LM3429 controller board has been pretty full of issues from day 1. I've spent countless hours now trying to debug and make some fixes and eventually got to the point where it's time to give up on 2 layers and move to 4. From what I understand this will make a significant difference.

I used to have a split ground plane, (digital uC ground and Driver ground) but this seemed to cause more issues with current return paths. I've opted for one constant shared ground plane now.

I've done my best to address the very helpful feedback given in this previous post, notably:

  • Moved the HSP trace away from the switch node
  • Put the ground plane in (this was the key point
  • Beefed up the connections and traces for high current paths
  • Added a Vin filter since my input capacitor is far away (not sure if I should also keep the 100nF decoupling cap I had placed there before)

I'd really appreciate if someone could give this new layout a quick look over and let me know if there are going to be any potential issues. 

One of my concerns is that I'm not totally sure if it's okay to separate the power plane (microcontroller vs driver power). This is my first 4-layer design and I'm not sure if noise might couple across them and cause issues on the digital side of things.

Another issue I have seen and as of yet been unable to reliably fix, is that at low voltage input (<8v) when the button is pressed to change the brightness, the noisiness from the driver causes it to 'press' multiple times. I have a decoupling capacitor on there but it doesn't seem to make a difference at all. Will the ground plane potentially fix this or is there something else I can do?

Layout is below. The internal layers still show the top layer, but it's still clear. The ground plane is just one joined plane and the power plane is split for VCC (7-24V) and 3.3V

I haven't included the full schematic, but I can provide it privately if it's needed. Driver schematic:


  • 0
    TI__Mastermind 25500 points
    I think that layout looks fine as long as there is a solid ground plane as the second layer. Having a ground plane will improve things dramatically. If noise in the switch/pot is an issue a ground plane will help that as well as it will quiet everything up quite a bit.
  • 0 in reply to Clinton Jensen
    Expert 1425 points
    Do you think the same will apply to the button issue I'm experiencing?

    Also, do you think there could be any ill-effects from splitting the Power plane the way that I have?

    Sorry if these seem like silly questions, I've just never had to deal with a ground plane in a noisy scenario like this and I am pretty clueless in terms what benefits it might provide and to what magnitude.
  • 0 in reply to Craig Easdale19
    TI__Mastermind 25500 points

    It will likely help the button situation, but I can't say for sure. Mechanical buttons can be notoriously noise. But at a minimum using a capacitor to reduce noise on a board with a good ground plane and less noise can only help.

    I don't think there is anything wrong with having a power plane. It's pretty common in multilayer boards since it can make routing much easier. Just keep bypass capacitors in the right places for each IC (close) and make sure the ground plane is the one beneath the top layer and not the power plane. 

    No silly questions, layout for switching regulators isn't a piece of cake. Just remember, the purpose of the ground plane is to provide return current paths that can be direct and will not interfere with noise sensitive nodes. It will not be a DC return path which takes the path of least resistance, it is a high frequency signal that will follow the path of least impedance. That means it follows the forward path but backward. If there is a good ground plane to provide this noise is very low. If it is forced into some sort of loop by say a cut ground plane or some trace it will open a loop and radiate the noise into everything around it.