Blocking 300MHz noise on a DCDC output

Go To Last Post
17 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi - I have a synchronous boost converter that I'm working on. I'm seeing a ~1V positive spike on my output when my primary FET turns on. It is an approximately 300MHz spike. Any suggestions for how to kill it? It seems to be originating in the switch node, which has a couple volt negative spike at the exact same time. But snubbers on the switch node only seem to dampen the ringing, but not the initial spike. Any ideas?

I'm thinking about putting a ferrite on the output of the DCDC, and having some output capacitance after that. I could also do an RLC filter. I am also going to try slowing down the signal to my primary FET's gate, as much as that pains me. I'm still debating...

Any other ideas?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You would be far better off killing it at its source rather than trying to cover it up after it (the spike) has been created.

There are many possible causes for these gremlins. One of the common overlooked ones is ground "bounce"; its also very hard to track down. Have you used the recommended component layout?

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
You would be far better off killing it at its source rather than trying to cover it up after it (the spike) has been created.

There are many possible causes for these gremlins. One of the common overlooked ones is ground "bounce"; its also very hard to track down. Have you used the recommended component layout?

Jim


Hi Jim - I have done my best to follow the layout guidelines. The ground point for my input filter caps is not super close to the ground of the IC (about 1.5cm), but that's about the only thing that is not exactly by the book according to the datasheet. What do you mean exactly by ground bounce? Do you mean just that the ground plane is developing a voltage across it?

I am measuring my Vout at the end of the board that is away from my input. I have separate ground out and in leads, and so all ground current is going through the board (no loops). Further, my ground point on my probe is right next to the output (where I'm measuring it), so it should be free from the large ground currents from the primary switch and should also be very low inductance (no ringing due to the ground lead).

The IC I'm using is the TPS43000, by the way.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

"Ground bounce" happens when large currents pass through a common conductor that is shared with other circuits. Ground is a common culprit, here. Inductance in this shared conductor causes a short transient voltage.

I have added slots in the ground plane to keep the large currents confined to regions right around the switcher.

It helps to try to trace out where the current is (probably) going. Physically. Never assume that any path is zero impedance; at those speed, there is no such thing.

If there is a catch diode, note where the current must flow when the diode is on (often through an inductor and an output cap). Is that path between the cap ground lead and the diode short. Do other (signal) currents have to pass through that region.

Ditto when the transistor switches on. Current will often flow from the input cap through the switch and an inductor or transformer primary. Where does it go in the ground plane and what else depends on that region.

Hope this helps
Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The LT field engineer came into our shop and changed an electrolytic cap to a low ESD cap and magically made all the noise go away in the switcher we were debugging (it used an LT switcher). Dude earned his money that day.

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You mean a low ESR? :)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

LT, particularly, has some excellent app notes that may help. There are lots of notes and they often don't search easily (how would you know to search for "Power Supplies for Poets"?).

You can find the list here. Just go through the list, open any that looks even remotely useful. and take a look.

http://www.linear.com/designtool...

Given the context, pretty sure non-sparkie Bob meant "low ESR".

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You mean Switching Regulators for Poets. :) I remember reading that one. Most excellent. I also remember talking to Jim Williams, who wrote it. Excellent man. I can highly recommend Switching Regulators for Poets.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes, that is the one. I agree on Jim Williams. One of the best engineers around. Sadly, he is no longer alive, but we do have some of these marvelous, and quirky, things he wrote.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Well, I got 66% of the letters in the three letter acronym right. That's a B in the government schools aint it?

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'm not sure whether the attached LT appnote is available on their site; IIRC I once got it from a LT FAE. I posted a link to it a couple of times before, now as attachment.

Anyway, as the layout guides have been followed this document is not going to be that helpful I think.

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The path from the switch to the diode to the inductor is critical. Keep this as short as possible. The other issue is the diode as when it snaps on it can cause the high frequencies you mention. I had a large peak at 100MHz. A ferrite bead on the diode cathode can do wonders. You see this a lot in flyback converters in consumer gear. For a leaded part, the ferrite bead goes over the component lead. For SMT, you need to add a ferrite bead part.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Right, those second order effects like diode reverse recovery (for a non-Schottky diode) and capacitor ESR can become killers. The fact that they are there usually means that even more attention is required for layout. Those ferrite beads can sometimes be very helpful.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

An update!

I played around with capacitors today. I have three 47uF ceramics on my output, along with two .47ufs and one 100nf.

Adding in a 10nf 0603 ceramic right at the output knocked the peaks I was seeing down significantly. From about 1.1V to 450mV. I played around with other values (on either side of 10nf) and that seemed to be the ideal value.

I had actually expected a 220pf 0603 to be the best value, as, according to Kemet spice, it should have a SRF around 300MHz. But my ringing got a bit worse with that part! Anybody have any idea why that would be?

As for the layout concerns - I don't think the layout is guilty here, but I could be mistaken of course. I've had one engineer from TI look at it and he gave it the thumbs mostly up, and another is supposed to be getting back to me shortly about it.

So, progress has been made, but I'm not out of the woods yet! I think I really need to find a way to reduce the magnitude of the spike on the switch node, but I'm not sure of just how to do that.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Tried another diode?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Maybe your caps aren't low enough ESR, like in Bob's case.

1) Studio 4.18 build 716 (SP3)
2) WinAvr 20100110
3) PN, all on Doze XP... For Now
A) Avr Dragon ver. 1
B) Avr MKII ISP, 2009 model
C) MKII JTAGICE ver. 1

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jayjay1974 wrote:
Tried another diode?

I don't have a diode. It's a synchronous device so it has two FETs.

I do plan on trying out some other FETs though. I'm hoping that a lower Crss FET will increase the frequency of my ringing and make it a bit easier to block out. Hope to get to that tomorrow!