Boost converter noise messes with LDO voltage regulator.

[step_s]

Hi EEVblog.

I have a lithium cell connected to an SX1308 boost converter setup, with 10uF ceramics on input and output, and a 100uF eletrolytic on output.
The cell is also connected to a MIC5504 3.3V LDO regulator, with the 1uF caps on input and output as stated in the datasheet.
The regulator works when the converter is off, but as soon as it's turned on, the regulator goes crazy, and spits out the input voltage. .
I figure it's a noise issue, but how would i separate the regulator from the converters noise?

Hope you can help.

[Ice-Tea]

1) This is why you may want to consider a somewhat decent scope: without it, it remains a bit of a guessing game...
2) Another big unknown: setup (ie: ground plane? flying leads? breadboard? what?)
3) That said: play around a bit with your decoupling caps. Your 10uF may be spot on for your 1.2MHz switching frequency or it may be as good as inserting a small twig (depending on its impedance characteristic). But add a 10uF on both inputs. And I'd say one or more 100nF caps or something..

[step_s]

@Ice tea
If you are talking about oscilloscope, I don't really know what to look for in the readings?
There is almost no noise on the input of the regulator, and i see no change in the input when the converter is on vs. when it's off.

The high power boost converter is all connected on a proto-board, where it is soldered in place. The regulator is sitting on a breadboard that has a direct connection to the battery leads.

I have tried 0.1uF, 1uF, 10uF, 100uF on all inputs and outputs, but no change. Tantalum or eletrolytics.

[izx]

1. Does the LDO have any load? Give it a 100 ohm resistor to feed.
2. Is the LDO output after it goes crazy really just pure DC? Did you look at it with a scope?
3. Try AC-coupling your scope and then look at the input with the converter on/off.

[step_s]

@izx
I have tried to load it down a bit, and have now a 1k resistor on the output.
The oscilloscope I have is the worst of the worst, so sorry for that, but I have added the measurements as pictures here.
Green is input, and yellow is output on the LDO.
The noise is there, but it doesn't seem massively excessive?

This measurements are with the boost converter running a 5V load at 500mA. When running lower loads, the problem seems to go away with a lower value load resistor on the output of the LDO.

[uncle_bob]

Hi

What you have is noise from a poor ground. The switcher is putting noise spikes (current spikes) into your ground and it rings because of it. Since the ground is in common with the LDO, the noise shows up on the LDO output. In some cases the current spikes are large enough to go right through the shield on a scope probe.

Often the only real answer is a ground plane under the switcher and coils to isolate both the input and output side from the rest of the circuit.

Bob

[step_s]

@uncle_bob
I thought about that, but how would I go about isolating the ground?
It's all on a protoboard, so making a "ground plane" isn't really easy, since it's not a normal PCB layout.
Thank you.

[Delta]

I think you'll get more help if you post a schematic and photos of your setup, mate.

[step_s]

@Delta
Made the setup more simple by using a PCB with the SX1308 and components on. As you can see on the picture (awesome paint skills!), it's quite simple, and the converter has ground plane on the PCB.
This setup "works" with 200mA loads on the micro USB output, and the regulator puts out 3.4V (intended 3.3V), but at higher loads it's going a lot higher than that.
This is with a 1k resistor as load on the regulator output.

The schematics are on the datasheets for the SX1308 and MIC5504.
Hope this helps!

I was thinking about a snubber across the SX1308 switch, to prevent the negative voltage spikes, any thoughts?

[Ice-Tea]

You mentioned you tried sevetal caps (tantalum and aluminium) but the LDO is spec'd with 1uF ceramic. Tried that? With additional 100nF ceramic?

[uncle_bob]

@Delta
Made the setup more simple by using a PCB with the SX1308 and components on. As you can see on the picture (awesome paint skills!), it's quite simple, and the converter has ground plane on the PCB.
This setup "works" with 200mA loads on the micro USB output, and the regulator puts out 3.4V (intended 3.3V), but at higher loads it's going a lot higher than that.
This is with a 1k resistor as load on the regulator output.

The schematics are on the datasheets for the SX1308 and MIC5504.
Hope this helps!

I was thinking about a snubber across the SX1308 switch, to prevent the negative voltage spikes, any thoughts?

Hi

Ok, the problem is that you have no proper ground. Your bypass caps have enormous inductance in series with them. You need it *all* tied to a proper ground plane to take care of the sort of spikes you are seeing.  On a switcher input or output, anything over about 1/16" is a big long distance to get to ground. Anything short of a solid copper plane does not count as ground in this case.

Bob

[jitter]

Grounding may be one issue, but the following design considerations for the MIC5504 must also be taken into account:

Input Capacitor
The MIC5501/2/3/4 is a high-performance, high-bandwidth device. An input capacitor of 1uF is required from the input to ground to provide stability. Low-ESR ceramic capacitors  provide optimal performance at a minimum of space. Additional high-frequency capacitors, such as small-valued NPO dielectric-type capacitors, help filter out high-frequency noise and are good practice in any RF-based circuit. X5R or X7R dielectrics are recommended for the input capacitor. Y5V dielectrics lose most of their apacitance over temperature and are therefore, not recommended.
Output Capacitor
The  MIC5501/2/3/4 requires an output capacitor of 1uF or greater to maintain stability. The design is optimized for use with low-ESR ceramic chip capacitors. High ESR capacitors are not recommended because they may cause high-frequency oscillation. The output capacitor can be increased, but performance has been optimized for a 1uF ceramic output capacitor and does not improve significantly with larger capacitance. 
X7R/X5R dielectric-type ceramic capacitors are recommended because of their temperature performance. X7R-type capacitors change capacitance by 15% over their operating  temperature range and are the most stable type of ceramic capacitors. Z5U and Y5V dielectric capacitors change value by as much as 50% and 60%, respectively, over their operating temperature ranges. To use a ceramic chip capacitor with Y5V dielectric, the value must be much higher than an X7R ceramic capacitor to ensure the same minimum capacitance over the equivalent operating temperature range.
No-Load Stability
Unlike many other voltage regulators, the MIC5501/2/3/4 remains stable and in regulation with no load. This is especially important in CMOSRAM keep-alive applications.

My guess is, is that the noise from SX1308 may not be the real cause of the problem, but it might cause a marginally stable setup of the MIC5504 to become unstable.
Chances are, if you were to put the correct type ceramic 1 uF cap on the output, instead of the electrolytic, it might be much less susceptible to noise.
Please note that this particular LDO does not require a minimum load for stability.

[step_s]

Hey guys. Finaly got home to play around some more.
Thank you to ice-tea and jitter for the point about the ceramic cap. This was the cause.

Thanks a lot for reading the datasheet for me. . . 

[jitter]

Hey guys. Finaly got home to play around some more.
Thank you to ice-tea and jitter for the point about the ceramic cap. This was the cause.

Thanks a lot for reading the datasheet for me. . . 

You're welcome and don't feel too bad. I have seen this sort of thing happen to experienced designers just as well.
Getting the output cap's ESR wrong is a recipe for disaster when it comes to LDOs. Always read the design rules in the datasheets because old-school LDOs will be designed to need a somewhat higher ESR that electrolytic and tantalum caps will have. Now there's a tendency to use ceramics with (much) lower ESR, and more modern LDOs have been optimised for this.

If you want to delve deeper into this phenomenon, there are lots of papers on LDO stability you may want to check out.

Oh yeah, another interesting, but not so well known issue, is with the capacitance changing with applied voltage on some types of ceramic capacitors. It is why the datasheet recommends to use X5R/X7R types. Dave also did a video on this in EEVblog #626.

[step_s]

Cheers Jitter!
As always Dave has a video on this. I will have a look