LM5145-Q1: Inquiry Regarding Output Voltage Behavior When Disabling LM5145-Q1 Buck Converter

Hi Makek,

If it's at no load, there is a small leakage current from BOOT to SW that will flow to the output and cause Vout to rise.

Regards,

Tim

Hello Tim,

The 5V loads are connected but off. A sensing circuit is in place to monitor 5V voltage rail.

Regarding the leakage current, what would be the best approach to minimize it? Should I consider changing the bootstrap capacitor? Currently, I am using a 0.1µF capacitor in series with a 0-ohm resistor.

I've attached the waveforms captured during testing, and they match the design calculations.

Best regards,
Malek

at switching node:

at gate of high side MOSFET:

at gate of low side MOSFET:

When enable 5V PS

When disable 5V PS

A small preload on the output will minimize the resultant voltage. Changing the boot cap will likely have zero effect on the leakage current though.

--

Tim

Dear Tim,

Thank you for your support.

I am still facing the same issue. I tried connecting different values of resistors (ranging from 10k to 100k) to the output, but there has been no change when the power supply (PS) is off.

Also, There are three small loads in the system, always connected:

1. 23.83 kΩ feedback resistor of the converter.

2. 67 kΩ sensing resistor.

3. A 2V LED in series with a 1 kΩ resistor and a 2V Zener diode, used to indicate that the 5V PS is working.

By the way, I noticed the voltage increases with the input voltage, reaching around 1.6V to 1.7V.

On the positive side, the 5V loads are off, but I find it unusual to have a residual voltage when the system is off. I’ve designed several converters before, and this is the first time I’ve encountered this issue. The concept of a preload is also new to me. Could this be the same issue described in application note below?

https://www.ti.com/lit/pdf/snvaa82

Or what exactly do you mean by small preload?

Thank you again for your help, and I look forward to your thoughts on this.

Best regards,

Malek Hattab

Hi Malek,

As mentioned, there is some boot leakage current into SW for this contorller. I need to check the expected amplitude. The resistances you mentioned are quite low - you may need a few kΩ of resistance from VOUT to GND to limit the increase in voltage (depending on how low you need the actual voltage). Is it an issue of the voltage increases?

Regards,

Tim

Hello Tim,

I tested resistors of various values (ranging from 0.5kΩ to 10kΩ), but encountered the same issue each time.

For instance, when I connected a 0.8kΩ resistor, the output voltage only dropped by 0.1V. I also measured the voltage at the BST capacitor, and it was the same as the output voltage.

This is not an issue for our application as long as the 5V loads are off and not operating. We can accept the situation if we are certain that the loads will never be turned on.

Currently, there are two 5V 30W LED displays connected, and when I physically disconnect the connector, the output voltage increases to 2V. Additionally, when I apply a short circuit to the output terminals, the voltage drops to zero, but once the short is removed, it rises immediately to 1.5V.

Do you have any other recommendations or insights on this behavior?

Best regards,

Malek

Hi Malek,

Can you measure the current flowing the the output? It should be in the uA range. Send on the schematic and a completed LM5145 quickstart calculator.

Regards,

Tim

Hello Tim,

I’ve encountered a strange behavior that I wanted to bring to your attention.

I provides two 5V power (PWR) connectors and one control connector on the board. The control signals are 5V tolerant, coming from an Octal Bus Transceiver (SN74AHCT245PWR) powered by 5V.

Here’s what I’ve observed in different scenarios:

1. When neither the power nor the control cables are connected: The output voltage is around 2.6V, with no current flowing through the PWR cable.
2. When only the control cable is connected: The output voltage drops to around 2V, and again, no current flows through the PWR cable.
3. When the PWR cable is connected (with or without control cable): The output voltage is around 1.5V, and about 19.35mA flows through the PWR cable.

Additionally, I should mention that I’m using two power supplies (5V and 3.3V) on the board, both powered by the same input voltage. 3.3V PS is designed using LMR38010SQDDARQ1.

The 5V PS is enabled by a 3.3V MCU signal.

When I tried to manually enable the 5V power supply using a screwdriver (without enabling the 3.3V power supply), the 5V supply generates 5V when turned on but 0V when turned off.

So now, I’m wondering if there could be reverse current flowing from the 3.3V rail to the 5V rail, potentially causing these issues. What are your thoughts on this?

This is the first time I’m using the SN74AHCT245PWR. In the datasheet, I solidly connecting OE to GND all time. Could this be causing the problem?

I've attached the 5V power supply schematic and the completed LM5145 Quick Start Calculator for your reference.

Also, I suspect that the Excel calculator may have swapped the pole values in the Actual P/Z Frequencies field. According to the datasheet, the first pole should be placed at Fsw/2, and the second pole should be placed at Fesr.

As I am using two types of capacitors with different characteristics at the output, I double-checked my calculations using the Power Stage Designer from TI and simulated the converter with TI PSpice. Please take a look at the loop calculation in the image below.

As mentioned previously, the performance of the power supply is excellent during startup and steady-state operation across the entire input range. The overload up to 16.7A for 15 mins and short-circuit protection for 1 minute features were tested and passed, though there was some audible noise when in close proximity to the board during short-circuit condition.

The issue only occurs when the 5V power supply is disabled.

Thanks in advance for your feedback.

Best regards,
Malek

5V 12A DC DC.xlsm

5V PS.pdf

Hello Tim,

Could you please check and confirm the loop stability and the calculation of the output capacitor based on the information provided above?

We are currently implementing brightness control for LED displays via LED dimming. The output voltage is 5V and displays are working, but we’re observing oscillation at 5V rail and audible noise, which sounds like buzzing. Similar to the sound at 1 min of this video (www.ti.com/.../6100466795001)

I suspect this might be coming from the inductor. The switching waveform at the switching node appears to have the same switching frequency.

When I disconnect the control cable, the converter works fine. However, once the control cable is reconnected, the noise returns.

Additionally, is it normal to experience this behavior during a short circuit condition? the sound level of this case is very low compared to above.

Many thanks,
Malek

Hi Malek,

Is the cable adding inductance after the output caps? If so, this second LC stage may result in instability.

--

Tim

Hello Tim,

Thank you for your feedback.

I’m not sure if the inductance is causing the issue, but the setup is the same as the initial test with the same cables. The only change has been the firmware update, which now controls the content on the LED displays. We’ve added brightness control and text movement features.

Additionally, the noise doesn’t appear with all colors. Sometimes, red with specific text works fine without noise. However, when the same text is displayed with white (turning on all RGB LEDs), the audible sound becomes louder.

It seems like there are a lot of step loads due to the switching of the LEDs, as LEDs turn on and off rapidly. There’s no overheating or damage, but the only issue is the audible noise.

Do you think I need to modify the compensator to make it faster, or should I increase the output capacitance? By the way, I am running in FPWM mode, not diode emulation.

I am measuring using 10x oscilloscope probe with ground spring.

For reference, here is the output at the 5V power supply (AC coupled) with no load, and there’s no audible sound. I’m able to clearly capture the switching frequency.

the input of the 5V power supply (AC coupled) with no load, and there’s no audible sound.

the output at the 5V power supply (AC coupled) with  load, and there’s audible sound.

the input of the 5V power supply (AC coupled) with load, and there’s audible sound.

Looking forward to your thoughts.

Best regards,

Malek

Hello again, Tim,

I’m sharing more waveforms to provide further clarity.

This shows the output voltage under full load and full brightness, with no audible noise present.

Input voltage when Full load and full brightness. No audible noise.

Output voltage when Full load and low brightness. Audible noise.

Input voltage when Full load and low brightness. Audible noise.

Input voltage when Full load and low brightness. Audible noise. but bypass input EMI filter.

Malek,

At least some of these look like load transient waveforms - please check the load current waveform is steady DC or if its varying.

--

Tim

Hello Tim,

Thank you for your help! I really appreciate your support.

Yes, you are correct – the load current is varying periodically, as I can observe from both the clamp meter and the input current of the main power supply. Unfortunately, I'm unable to capture the output current waveform.

Is the audible noise considered normal behavior? Are there any hardware changes that could help minimize it? The converter seems stable for me, so I assume this isn't an instability issue, correct?

What options do I have moving forward?

Hello Tim,

Thank you very much for your continued support. I truly appreciate it.

I recall being uncertain about which value to input for the total output capacitance in the Excel calculator, as I have two types of capacitance at the output: 4 x 47uF/10V and 220uF/20m. However, after reviewing several application notes on how to calculate the equivalent capacitance, I moved to simulation but did not achieve better results than those suggested by the calculator. I made several attempts, including importing the derated models and studying the response, in order to determine the correct compensator values to input into the calculator.

Unfortunately, I do not currently have access to a network analyzer to measure the closed-loop response of the designed buck converter. However, I did try to solder the compensator suggested by the calculator (with an output capacitor of 75uF/2m) to generate a 25mV ripple current, as shown in the images above. The response of the converter is shown below. Based on this, I suspect that the phase margin is lower than desired do you agree?

The current response is shown below, and the buck converter with this compensator has been tested under various conditions. It meets the application requirements, but I would appreciate your feedback.

Regarding the audible noise, I agree with your observations. It seems that the repetition rate of the load transient is causing low-frequency noise around 400Hz. The inductor I’m using is the MDA1365-3R3M from KOHERelec, which is a shielded construction. I did not select a ferrite inductor, as I wanted to save cost and space.

There are no issues with the converter itself—only the audible noise when the brightness is lowered. We have left the device running for several days without any adverse effects or overheating problems.

However, we are open to suggestions on how to minimize this issue through hardware changes.

Best regards,
Malek

Hello Tim,

Thank you for sharing your thoughts and the file.

I agree with your comments.

I used similar equations that I found online, but I'm not entirely sure if the effective capacitance at the switching frequency should be used in this field, since I am studying the response versus frequency. Finally, I decided to use the total capacitance at low frequency (with DC biasing for the ceramic capacitor) and the equivalent ESR at the switching frequency after several attempts with SPICE simulators.

These are the results when use effective values at switching frequency.

The good news is that I got results close to yours. When I used your suggested values in the calculator with the current compensator, it produced good results:

C = 300 µF
ESR = 5 mΩ

Additionally, here are the results from my extracted values from the SPICE models:

C = 330 µF
ESR = 2 mΩ

If there is a resource or detailed explanation that can help clarify this, I would appreciate it and am eager to learn more!

In the past, I have designed compensators using MATLAB in a similar way, as shown below.

www.ti.com/.../slva301.pdf

Best regards,

Malek

Hi Malek, I'll study this and reply tmrw.