[fixed] LT3435 switching reg with 2V ripple on output

[dnotq]

I am trying to troubleshoot a 5V regulator circuit based on the LT3435 switching regulator.  The output has an approximately 2V ripple that is causing havoc with down-stream circuits.

Datasheet is here: https://www.mouser.com/datasheet/2/609/3435fa-1271903.pdf
Input: ~23V
Output regulation: 5V

The circuit seems to mostly follow the "typical application" from the datasheet.

Scope test points are either side of the output inductor.  Note the scale of the yellow trace is 10V/gradient, purple is 2V/gradient. Hopefully this is a common failure mode for a switcher... I'm not sure where to begin testing and any insight would be greatly appreciated.

Thanks,
Matthew

[floobydust]

Check the usual - MLCC capacitances are "real" sized, they look way too small, and the PC board layout looks not so good.

Sorry but I don't use LT switcher IC's because of their stupid pinouts for the high current pins - here SW is on either side of VIN which means a mess is required.
We had one project fail using a LT (SEPIC) switcher IC, designer got fired too because he couldn't get it stable. It was the stupid pinout along with the PCB designer's inexperience but no time to do another board and build. Who would design an IC forcing use of vias on VIN or SW

The PCB layout does not look very good, the inductor and diode are a mile away, making a large loop area.
Two small vias for SW, pin 8 GND goes where?, the GND for C121, C113 (only 2 vias?) are a mile away from the high current path. C126 GND is grounded at the noisy node etc. So I would say the board layout is suspect.

Refer to the EVAL board layout, it's usually 4-6 layer and shows what is needed to get the part to work.

[dnotq]

I did notice the layout recommendations from the datasheet were not taken into consideration.  It is not my board, just one I'm trying to repair, so changing the IC and/or layout is not really an option.  I am also going on the assumption that the circuit did actually work *better* at some point in the past... At least I hope these were not released into production with that kind of ripple on the output!?

The board appears to be designed by LG (see photo), and is 1-of-3 boards in the Apple Cinema 30" display, circa 2008.  All three boards are LG in this monitor, and the panel is Philips; seems Apple only designed the case, which surprises me (they usually do their own PCBs).

... Check the usual - MLCC capacitances are "real" sized, ...

I'm not sure what you mean here.  They are all 0603, except for C129 which is 0804.

I'm starting to consider just buying a small 5V@3A switcher and wiring it in, rather than trying to fuss with fixing this jacked up circuit.

[floobydust]

I would check the capacitors (electrolytics included) for low value.
MLCC's also age, giving even less capacitance than expected, including the drop due to DC bias voltage.
The input cap C113 and maybe C121, and whatever is under the sticker.

[dnotq]

Removed and measured input caps: C113=0.1uF and C121=0.01uF.  Those seem to be entirely the wrong range, according to the datasheet a good choice would be a 4.7uF ceramic cap.

The electrolytic caps on the output (C123, C124, C125) are all the same: 100uF 16V.  I'll pull those next and check them.  The datasheet specifies they should be low ESR tantalum caps, but this circuit uses electrolytic caps.  I have no idea if there is any way to know that they are low ESR, so maybe I just replace them with new caps to be safe.

I tracked the circuit and made a schematic.  No traces on the bottom, the vias just go to test points on the bottom.  All values filled in were measured.

[capt bullshot]

The electrolytic caps on the output (C123, C124, C125) are all the same: 100uF 16V.  I'll pull those next and check them.  The datasheet specifies they should be low ESR tantalum caps, but this circuit uses electrolytic caps.  I have no idea if there is any way to know that they are low ESR, so maybe I just replace them with new caps to be safe.

Assuming the circuit once worked as desired, by looking at the scope picture I'd guess: C123, C124, C125 lost most of their capacity. Replacing them would be it.

[dnotq]

Fixed!

I replaced C123, C124, C125 with new low ESR electrolytic caps, as well as replacing C121 with a 4.7uF 100V X7R ceramic cap as recommended in the datasheet.  Not sure if C121 was necessary, but the 5V output is stable now and the monitor is working correctly (no more flickering display problem).

Edit: C121 is connected to a large top and bottom ground plane and was a serious PITA, I don't recommend replacing it unless necessary (it is too easy to break something else while trying to put enough heat into the ground plane to get a good solder joint).

Thanks everyone for the feedback!

[Therm1]

I'm going to attempt this repair as I have a blinking 30 inch monitor (also fails to turn on 70% of time unless I unplug and replug in).  I'm not afraid of pulling the monitor apart, as I've done it several times (good video on YouTube.  ).    Can you tell me what part number / brand you would recommend or if any of these from Mouser would be appropriate (not sure if the length will be an issue).  You mention having a data sheet, but I didn't see it attached. I love these monitors, but it's hard getting a good one that works properly.  I have 3 of these.  Cheap way for a lot of screen real estate with zero glare. 

https://www.mouser.com/c/passive-components/capacitors/ceramic-capacitors/mlccs-multilayer-ceramic-capacitors/multilayer-ceramic-capacitors-mlcc-smd-smt/?capacitance=4.7%20uF&case%20code%20-%20in=1210&dielectric=X7R&voltage%20rating%20dc=100%20VDC

[Therm1]

This is what I settled on after pulling it apart.  I replaced the caps, but when I place the white panel with a 1/4" of the frame, the power button doesn't work.  If the panel is far enough away no issues.   Not sure what do at this point.  I can figure out how to switch parts, but don't have the knowledge to troubleshoot the circuit.  Maybe I should attempt to change C121, but I'm hesitant to do so based on the recommendation for the repair.  I did measure the other 3 caps and they measured about 80uF each instead of 100uF