FesZ, outstanding video, thank you. Remote sensing is a very complex and important topic. Wire harness design and fabrication is the critical design area.
Wow .. Finally Power supply Series started pending from long time .. My humble request to you to cover all topic related to power supply in detail... For example LDO and SMPS stability
I guess you could have a high-pass filter from the output cap to the feedback network, and a low-pass filter from the load to the feedback network, in order to have good DC accuracy while maintaining stability. I came into an unwanted computer PSU recently, making a high-power CC/CV buck-boost benchtop supply from that sounds like fun, and I’ll be sure to have a seperate sense terminal.
The conceptual idea of testing power supplies starts with a know load characteristic like a constant current (CC) load or a load resistor (which is not CC as it heats up, or the supply output voltage changes). So, if you want an ideal static load then use an electronic load. However, the real issue with ideal loads is that the power supply is designed to satisfy the requirements of the load, and the load is usually very dynamic in how it uses power, which ultimately leads to looking at the load as being fed by a Power Distribution Network (PDN). Which is especially important at meeting microcontroller supply decoupling requirements. So, it is good to get into the habit to look at power supply design as powering a specific circuit or board subsystem. You may ask okay if the power supply design is dependent on the load how can I design a basic generic power supply? the answer is you cannot. You can however, design the PDN to allow a generic supply to meet the load requirements. A good example of this is your basic generic personal computer power supply. It is about as generic as you can get with virtually hundreds of options available to you off-the-shelf. However, the power supply does not directly feed the load(s) (except perhaps a cooling fan which is not a complex load), it is what is called a pre-regulator, so there is a motherboard regulator that feeds the microprocessor, which is very specific in meeting the needs of the microprocessor and also is characterized by a local PDN. So, the trend in design for addressing the load requirements is, in almost all complex load designs, you will see the use of a generic pre-regulator supply feeding a PCB that will have a local regulator and PDN to feed the load. Remote sensing is not really needed if the power supply is at the load (in very close proximity like 5mm). In a preliminary static test with a constant current load (CC electronic load) at rated power, you are mainly looking at power dissipation (heat flow) at room temperature. To get an understanding of the thermal performance of your design you need to let the power supply stabilize, which will take some time, and is useful to monitor with a thermal camera to see what is happening, to make sure it has stabilized, and to understand any hot spots. If the supply works as designed at room temperature. Then you can decide how you want to proceed. Off-line AC to DC supplies are interesting to test because you now have to also include AC line transients, not just low line or high line which is used for static testing. Also, a universal supply with power factor correction and relatively high current requirements (like 500W) can do interesting things during line transients that are not easily modelled. You will need a programmable AC source to simulate the regulatory required transients (they will be specified depending on where you want to sell your product) and also one's you make for yourself to better get an overall worse case test for your design. The starting point for a regulated DC power supply is usually to compare it to a large DC battery of the same terminal voltage, which is essentially the idea of understanding the output impedance of the supply and its rather impressive capabilities with the understanding that it will have some dynamic behavior because it uses feedback. On you board layout where the sensing feedback trace inters the large plane have the trace enter at 90 degrees to the plane and not off angle where one side of the trace enters at an oblique angle. This will create an unwanted large trace stress riser which is where you well get a crack in the field after operational heat cycling. The advantage of having the power supply controlling switch on-board a regulator IC is you can monitor its temperature and shut down the supply when there is a fault which causes overheating. If you have an off-chip regulating pass transistor you should monitor its temperature and shut down the supply if it gets too hot. You can get a star ground effect using unbroken planes if you place (group) the components correctly for their operation, which usually means at least a four layer board. You are basically laying out the board to eliminate parasitic effects. The two internal planes can form a high frequency capacitor which is helpful in designing the PDN.
Hurray ! More power electronics !
FesZ, outstanding video, thank you. Remote sensing is a very complex and important topic. Wire harness design and fabrication is the critical design area.
Wow .. Finally Power supply Series started pending from long time ..
My humble request to you to cover all topic related to power supply in detail...
For example LDO and SMPS stability
wuhuu PSU series ;)
Thanks for posting
I guess you could have a high-pass filter from the output cap to the feedback network, and a low-pass filter from the load to the feedback network, in order to have good DC accuracy while maintaining stability. I came into an unwanted computer PSU recently, making a high-power CC/CV buck-boost benchtop supply from that sounds like fun, and I’ll be sure to have a seperate sense terminal.
I look forward to a current limiting smps.
Very good video! When choosing switching converter, also test for various interference emissions 🙂
What is the song you are using, is it from the yt audio library, or is it something you or a subscriber put together ?
Its from the yt library - I think its called "Vibe tracks - Alternate"
The conceptual idea of testing power supplies starts with a know load characteristic like a constant current (CC) load or a load resistor (which is not CC as it heats up, or the supply output voltage changes). So, if you want an ideal static load then use an electronic load. However, the real issue with ideal loads is that the power supply is designed to satisfy the requirements of the load, and the load is usually very dynamic in how it uses power, which ultimately leads to looking at the load as being fed by a Power Distribution Network (PDN). Which is especially important at meeting microcontroller supply decoupling requirements.
So, it is good to get into the habit to look at power supply design as powering a specific circuit or board subsystem. You may ask okay if the power supply design is dependent on the load how can I design a basic generic power supply? the answer is you cannot. You can however, design the PDN to allow a generic supply to meet the load requirements. A good example of this is your basic generic personal computer power supply. It is about as generic as you can get with virtually hundreds of options available to you off-the-shelf. However, the power supply does not directly feed the load(s) (except perhaps a cooling fan which is not a complex load), it is what is called a pre-regulator, so there is a motherboard regulator that feeds the microprocessor, which is very specific in meeting the needs of the microprocessor and also is characterized by a local PDN. So, the trend in design for addressing the load requirements is, in almost all complex load designs, you will see the use of a generic pre-regulator supply feeding a PCB that will have a local regulator and PDN to feed the load. Remote sensing is not really needed if the power supply is at the load (in very close proximity like 5mm).
In a preliminary static test with a constant current load (CC electronic load) at rated power, you are mainly looking at power dissipation (heat flow) at room temperature. To get an understanding of the thermal performance of your design you need to let the power supply stabilize, which will take some time, and is useful to monitor with a thermal camera to see what is happening, to make sure it has stabilized, and to understand any hot spots. If the supply works as designed at room temperature. Then you can decide how you want to proceed.
Off-line AC to DC supplies are interesting to test because you now have to also include AC line transients, not just low line or high line which is used for static testing. Also, a universal supply with power factor correction and relatively high current requirements (like 500W) can do interesting things during line transients that are not easily modelled. You will need a programmable AC source to simulate the regulatory required transients (they will be specified depending on where you want to sell your product) and also one's you make for yourself to better get an overall worse case test for your design.
The starting point for a regulated DC power supply is usually to compare it to a large DC battery of the same terminal voltage, which is essentially the idea of understanding the output impedance of the supply and its rather impressive capabilities with the understanding that it will have some dynamic behavior because it uses feedback.
On you board layout where the sensing feedback trace inters the large plane have the trace enter at 90 degrees to the plane and not off angle where one side of the trace enters at an oblique angle. This will create an unwanted large trace stress riser which is where you well get a crack in the field after operational heat cycling.
The advantage of having the power supply controlling switch on-board a regulator IC is you can monitor its temperature and shut down the supply when there is a fault which causes overheating. If you have an off-chip regulating pass transistor you should monitor its temperature and shut down the supply if it gets too hot.
You can get a star ground effect using unbroken planes if you place (group) the components correctly for their operation, which usually means at least a four layer board. You are basically laying out the board to eliminate parasitic effects. The two internal planes can form a high frequency capacitor which is helpful in designing the PDN.
ذ