- Joined
- Sep 20, 2016
- Messages
- 7
- Points
- 3
Almost two years ago, I posted a PCB and details of a boost regulator design using the MP3412 Synchronous Boost White LED Driver by Monolithic Power Systems. I posted it to LPF because I saw the regulation was more ideal in my application than another regulator I had started out with. (see "MP3412 Boost Regulator Board" through an internet search linking to LPF).
I've been making small revisions and design changes to improve the design for my application as a lower-powered LED driver for a personal flashlight. Through this process, I've found I wanted to test more and more to fully understand what's happening during different driver states and when using different loads.
While steady-state regulation has been as good as ever, transients became a concern. I started measuring startup and shutdown transients as I had changed my flashlight switch design to become FET-based rather than mechanical. I remembered learning that faster switching and inductive elements could mean the generation of large spikes in voltage, and I wanted to ensure that this wasn't occurring (if such spikes could damage components or LEDs).
In addition, the MP3412 was designed for white LEDs, so my use of colour LEDs with various forward-voltages became a concern, with regards to how the difference in load would affect the regulator performance.
Two groups of results I will mention below:
Regarding point 2, I observed some startup transients with various loads including white and colour LEDs (green, red in series, or with resistors or normal diodes). I found white LEDs give a smooth startup (as per the datasheet), green had some slight differences, but red had significant differences depending on variation. I need to test more and fully categorize what happened where.
The MP3412 regulator is supposed to ramp current up to setpoint over a span of about 150 us. I'm suspecting that the regulator is expecting a constant-voltage load (ie. a single diode). A white LED is fine, a green may have a different but similar forward-voltage, possibly accounting for the slight difference in current ramp.
By using a red LED and then adding voltage drop with a resistor means that the total forward-voltage will change during the ramp, so the driver won't see current feedback like it's expecting, resulting in transients. I saw relatively large ripples (like sawteeth) in the LED current that stopped after 5 or 6 cycles.
(I need to test this soon) I suspect two red LEDs in series should give enough forward-voltage, but may respond differently to the ramp control because two diode junctions probably don't act like a single diode junction to the applied ramp.
My point: one strategy for testing regulators is to use a dummy load, or a load that will not cost too much in case it is broken from testing. This may not show all the characteristics being tested before use with the actual load. I tested with white LED's initially, but when questioning transients, switched to the above-mentioned combinations and found issues that could affect laser diodes (which I understand are much more sensitive!).
So, if you decide to use this regulator, or my previously posted design, please be careful and test appropriately.
When I find more time, I can edit this post and answer questions for clarity, as well as gather and post scope captures to show what I'm seeing.
I've been making small revisions and design changes to improve the design for my application as a lower-powered LED driver for a personal flashlight. Through this process, I've found I wanted to test more and more to fully understand what's happening during different driver states and when using different loads.
While steady-state regulation has been as good as ever, transients became a concern. I started measuring startup and shutdown transients as I had changed my flashlight switch design to become FET-based rather than mechanical. I remembered learning that faster switching and inductive elements could mean the generation of large spikes in voltage, and I wanted to ensure that this wasn't occurring (if such spikes could damage components or LEDs).
In addition, the MP3412 was designed for white LEDs, so my use of colour LEDs with various forward-voltages became a concern, with regards to how the difference in load would affect the regulator performance.
Two groups of results I will mention below:
- Low forward-voltage loads affect the MP3412 steady-state regulation; follow the minimum per the datasheet.
- Different loads have different effects on the startup behaviour of the MP3412.
Regarding point 2, I observed some startup transients with various loads including white and colour LEDs (green, red in series, or with resistors or normal diodes). I found white LEDs give a smooth startup (as per the datasheet), green had some slight differences, but red had significant differences depending on variation. I need to test more and fully categorize what happened where.
The MP3412 regulator is supposed to ramp current up to setpoint over a span of about 150 us. I'm suspecting that the regulator is expecting a constant-voltage load (ie. a single diode). A white LED is fine, a green may have a different but similar forward-voltage, possibly accounting for the slight difference in current ramp.
By using a red LED and then adding voltage drop with a resistor means that the total forward-voltage will change during the ramp, so the driver won't see current feedback like it's expecting, resulting in transients. I saw relatively large ripples (like sawteeth) in the LED current that stopped after 5 or 6 cycles.
(I need to test this soon) I suspect two red LEDs in series should give enough forward-voltage, but may respond differently to the ramp control because two diode junctions probably don't act like a single diode junction to the applied ramp.
My point: one strategy for testing regulators is to use a dummy load, or a load that will not cost too much in case it is broken from testing. This may not show all the characteristics being tested before use with the actual load. I tested with white LED's initially, but when questioning transients, switched to the above-mentioned combinations and found issues that could affect laser diodes (which I understand are much more sensitive!).
So, if you decide to use this regulator, or my previously posted design, please be careful and test appropriately.
When I find more time, I can edit this post and answer questions for clarity, as well as gather and post scope captures to show what I'm seeing.
