data/text/DLP2/888430

LMZM23600: Output voltage too low

Part Number: LMZM23600 We are using the LMZM23600V5SILT on a prototype to generate a 5 volt output from a 20 volt input. With an approximately 2 mA load (the minimum feasible at this time), the output on the prototype is slightly less than 3 volts; adding a 100 ohm resistor load reduces the output to around 2 volts. With the minimal load, the output voltage appears stable on the oscilloscope; with the 100 ohm load, the output is erratically varying by 250 mV. The circuit diagram is the same as Figure 37 with the exception that feedback is directly connected to the output without any resistors. Cin (C1102) is 10 uF 50 v X7R and Cout (C1101) is 47 uF 16 v X5R. The board layout is approximately the same as that shown in Figure 82 of the data sheet. The differences for the layout are: 1. Footprint pad size is 0.45 x 0.8 (recommended footprint has 0.3 x 0.65) 2. FB and OUT are both connected directly to a component side power plane 3. Under component ground plane fill has 0.225 gap to component pads. Are these differences likely to account for the problem? Are there exposed pads on the underside of the component that could be shorting to the ground fill under the component? Or does it seem more likely that either the component is bad or there is a solder bridge between pins? Thank you for your help.

Hi J A Valina, Can you probe the SW node to check the switching frequency of the part? This can be done by probing the top side of the exposed inductor. Please be careful when probing the exposed terminal and make sure not to accidentally short it to GND. Doing so will damage the part. Additionally can you read out the voltage measurements on each pin? As for the PCB layout, generally I would recommend having a small thin trace connecting feedback to VOUT. The feedback pin is generally considered a sensitive node since it can couple noise if there is a large enough plane. This may result in output voltage regulation issues. Note that Figure 82 also shows the thin feedback connection to VOUT power plane. Another thing you can do is try to mechanically press on the part and check regulation. This will exercise if there is any pin connectivity issue which would point to solder voids/ no connects. You can reflow the part off your prototype board and examine both the part and your board to see if there are any solder bridges. Regards, Jimmy

Hi Jimmy, Here's the SW node trace: It appears to be oscillating at 754 kHz (the frequency shown in the image above is incorrect). I cannot probe most of the pins since the pads are concealed by the module, but Vin is 20.43 v and Vout is 2.94 v. Pressing down on the module had no effect. I will alter the PCB layout of the FB pin for the next prototype. Thanks for your help! Jay

Hi Jay, If you are using a fixed output power module (LMZM23600V3 or LMZM23600V5) then the switching frequency is expected to be typical at 750kHz. However, the switch node doesn't look like a square-wave. As previously mentioned, please try to reduce the parasitic inductance on FB node by having a thin dedicated trace to VOUT for fixed output option. Let me know if you have any additional questions on this. I will close this thread for now. Regards, Jimmy

Hi Jimmy, Do you think that having the solid connection from FB to VOUT is the cause of the problem? It seems to me that connecting FB to VOUT by a thin trace would increase rather than decrease the inductance. Thanks for your help! Jay

Hi Jay, Let me redact what I said previous, for some reason during the explanation I was thinking adjustable when you are using fixed. Actually when you are using the fixed version, the feedback resistors should be internal. This means the sensitive node is protected and accounted for internally. I don't think I asked this, how many of your boards experienced this failure? Also was the part hand soldered onto your prototype or did it have a control reflow profile? Again the switch node is abnormal and should be a clean square wave. Can you do a ABBA test on two prototype boards to see if the failure follows the part? Regards, Jimmy

Hi Jay, Can I get an update on the questions above? What is the progress of the ABBA test? Does the failure move with the unit? Additionally what are you power downstream? If possible try to isolate the circuit by disconnecting the downstream system and retest troubleshooting with either a load resistor or E-LOAD to check operation of the LMZM23600 alone. Regards, Jimmy

Hi Jay, Can you give an update to the ABBA testing? Have you tried continuity test? You can probe the voltage measurements between all pins and ground to get an understanding if there is any shorts or abnormal voltage. Also I have a few more questions about this condition: Just to reconfirm you application condition: VIN = 20V; VOUT = 5V; IOUT = ~50mA; Tambient = 25degC (room temperature) What peak temperature was the device reflowed at? Was the device hand soldered or did it go through a controlled reflow process? Can you provide the date and lot code for these parts? This information is usually detailed in the shipping label on the package. How many failures has this been observed on? I'm guessing this failure is seen on the same site on different boards? Regards, Jimmy

Hi Jay, Any updates on the testing and feedback for my questions? Again I'd like to reiterate that the switching waveform that you measured is not characteristic of a switching converter (generally square wave). I'd like to see the same test done on a new unit on the same PCB layout to confirm issue is preserved. Regards, Jimmy

Hi, I have not heard back from you for some time and will assume you were able to close this issue on your end. Because of this, I will be closing this thread for now. If you have any further questions on this, please reply back to this and the thread will reopen for further discussion. Regards, Jimmy