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.

LM5122: Voltage not boosting from 12V power supply

Part Number: LM5122
Other Parts Discussed in Thread: CSD18503KCS,

This is in relation to a 12V boost to 24V 20A application for the LM5122MH-NOPB.

I have hot air reflow soldered 2 samples onto 2 circuit boards. On the first board the output voltage slowly rises and stays somewhere between 6V and 7V open circuit but I have seen it rise to 9V once. The output voltage increases as the input voltage increases. On the second board, the output voltage is 10V but also has a high idle current (0.1A) in open circuit. I have tried connecting RES to ground with no change, measured the UVLO pin at 1.5V, and the bench top power supply was 12V with a 500mA current limit set.

I've been so busy lately that this has been low priority, but it seems like these particular boards really need the voltage boosting to be more desirable for customers. I have attached the circuit diagram below and the top and bottom layers of the relevant PCB area. Please note the transistors are not the ones listed on the PCB layer images, they are all TI CSD18503KCS. The inductors are 3.3 milliohm each and the ones I've used on the prototype boards support up to 27A as they were currently easier to obtain.

I'm looking for advice on either the circuit design if there is any outstanding issues, or if it is down to the PCB layout or even some component choices.

 Top LayerBottom LayerTop & Bottom Layers

  • Hi Anthony,

    Thanks for your interest in LM5122. I'll review the design.

    Did you do the start up with no load?

    Thanks,

    Yinsong

  • Hi Yinsong,

    Thanks for your help. Yes it was started with no load. I tried applying a small load (~1W) but there was no change in voltage.

    Thanks

    Anthony

  • Hi Anthony,

    Not sure about the annotation on the schematic but do you put a 3kOhm resistor on the VIN filter?

    Thanks,

    Yinsong

  • Yinsong, yes that is correct.

    Best regards,

    Anthony

  • Hi Anthony,

    Not sure its impact but that is a very large resistor for the input filter. Do you have a very harsh input transient that push you to use 3kOhm?

    Thanks,

    Yinsong

  • Thanks for the observation, Yinsong. The input should be tightly controlled before reaching this circuit, so there will not be any spikes. The resistor values are from the Webench design tool. I've changed the resistor twice to see if there is much of an impact.

    1k Ohm- 13V

    250 Ohm - 13.1V

    It appears like it has a limited impact to the circuit performance, it is now boosting, producing 1V more than the input and 3V more than previously. There is an OC current of around 350mA now though. Perhaps some other resistor values are too high as well?

  • Hi Anthony,

    I'm sorry but I think that is a typo in Webench. Please try 3.3Ohm or even 0 Ohm since it is tightly regulated already.

    Thanks,

    Yinsong

  • Hi Yinsong,

    I have just tried 0 ohm and 3.3 ohm. The output voltage has dropped to 10V on open circuit with a current draw from the power supply of 0.9A. The transistors at Q2A1 and Q2A2 heat up quickly from the current. I have re-run the webench tool with slightly different parameters and tried changing some values to match, the changes made no difference. 

    Thank you for all your help and suggestions but I'm a bit baffled really. I would like to know if there are any alternative chips that would be able to do the job. 

  • Hi Anthony,

    I fully understand your concern and I really appreciate your patience. Debugging the hardware can be time and energy consuming, but sometimes it could turn out to be a small issue that was ignored.

    By looking at your schematic carefully, it looks like your synchronous rectifier Q1 is reversed. Could you double check if this is a schematic symbol issue or not?

    Thanks,
    Yinsong

  • Yinsong,

    You are correct. I must have missed the error in Q1, I thought I had double checked everything. Changing pins 2 and 3 on the transistor has boosted the voltage to 19V OC which is a big step in the right direction. The idle current is 2mA which is what I would have expected.

    One problem remains however, the voltage peaks at 19V then lowers 14V, then jumps back to 19V. My multimeter does not update very fast so I am leaning towards it looking like a sawtooth wave on an oscilloscope.

    I have checked on an oscilloscope and confirmed  it is a sawtooth wave. Sadly my picoscope can only go up to 20V.  

    The below chart is with a 12V input. Results show it drops 4.8V in 2.5 seconds and the peak that was read on the multimeter is an undesirable spike. I greatly appreciate the efforts you have went through to help me so far. Do you have any further suggestions to clean up the signal?

    Thank you.

    Anthony

  • Hi Anthony,

    The 2.4 seconds gap actually matches well with the 470nF restart cap. Although it shouldn't enter current limit given that you are running at no load, but could you check the RES voltage?

    Thanks,

    Yinsong

  • Yinsong,

    The voltage from RES to ground is 4.3V and decreases to just over 3V. It appears to be a similar pattern to the output voltage but I am unable to confirm the timing without an oscilloscope.

    I have tried 0 ohms and 3.3 ohms on the Vin filter, there is an output difference between the two values of a few volts.

    As per a previous suggestion, since you corrected the transistor problem, I have tried connecting RES directly to ground and bypassing the capacitor. This appears to stabilise the voltage around 24.3V which I am very happy with. That particular meter reads a little on the high side though so I will connect it up to the oscilloscope on Monday, along with trying a load just to see if it holds up under stress.

    Enjoy your weekend!

    Thanks a lot for your help!

    Anthony

  • Hi Anthony,

    Yes, not sure why the RES gets charge up high, but whenever the RES goes higher than 1.2V, the SS is being pulled low and both LO and HO disabled.

    Please take a close look at the current sense pins as right now you do not have the overload protection.

    Just let me know if you have further questions.

    Thanks,

    Yinsong

  • Yinsong,

    CSN and CSP are both 13.9V to ground, CSN to CSP is jumping between 4-12mV on open circuit which seems to be in spec for the zero cross detection threshold and way below the current limit detection threshold.

    If the only feature that is lost due to the RES grounding is overload protection then it is something I can live with, there are isolating blade fuses in the circuit that result in a current limit and have been tested to trip instantly on short circuits. This circuit board doesn't go into anything with large safety requirements like automotive applications, so the fuses should suffice for the application.

    I have tested the circuit with a 60 ohm load (0.4A) and the voltage dropped to 16V. Is there something else I've missed? This shouldn't have been too high a load for the circuit as it should have been able to cope with 10A easily. Is this down to the lack of overload protection possibly having an effect on the feedback loop? I can't see a direction relation between RES and FB in the block diagram other than the link you mentioned to SS.

  • Hi Anthony,

    Could you check the COMP in the 16V case? The COMP should be high.

    Thanks,

    Yinsong

  • Hi Yinsong,

    The COMP pin to ground is 5.3V in this case.

    Thank you.

  • Hi Anthony,

    The COMP is high means that the feedback loop is ok. However, the FETs should be switching full duty cycle unless it is disabled by the current limit. Could you get the differential current sense waveform under the 16V condition if possible? It will also be helpful to see the SW waveform to make sure the duty cycle.

    Thanks,

    Yinsong

  • Hi Yinsong,

    I have tried the circuit from the actual DC supply it will be using, along with a 1.8 ohm heater. From what information I could gather, that brought the output voltage down to 11V with the load attached, and it was 24Voc. 

    Unfortunately the the CSP-gnd and SW-gnd were outside the 20V max input for my oscilloscope and it appears that the scope itself has been fried, neither channel on it are able to read anything now. From what I could see beforehand, both were over 20V under load and very noisy even though the 11V at the heater was a crisp DC, this may be due to the limits of the oscilloscope though. The transistors at Q2 were cold as if they had not been running, and Q1 was running hot to the touch.

    Apologies but I am unable to gather any further info aside from on the multimeter.

    Regards,

    Anthony

  • Hi Anthony,

    Understood. To me the current limit is still the most possible reason considering two phenomena:

    1. output voltage drop with adding load

    2. Hiccup is triggered.

    Could you double check the value of the R sense?

    Thanks,

    Yinsong

  • Hi Yinsong

    Rsense is 1 milliohm.

    Thanks

  • Hi Anthony,

    Have you ever tried with another IC?

    Thanks,

    Yinsong