In this post, I will write about the lessons I learnt when designing the MRR ESPA. Specifically, lessons related to the high currents used for the heated bed, which I learnt during the early phase of the design.
1. Separate power connectors for heated bed. Heated beds can draw a lot of current. So using the same power supply for the heated bed and the rest of the 3D printer (control board, motors, hotend, etc.) can result in a very high current being carried from the power supply unit (PSU) and the board. If the power connector on the board is not rated for such high currents, they can melt, burn, and become a fire hazard. This is why some boards are designed with a separate set of power connectors for the heated bed. This allows the same PSU (which usually have a few sets of output leads) to be connected to the board, with one set of output leads for the board and such, while another set of leads for the heated bed. This lowers the current being drawn on each set of leads, keeping temperatures lower.
2. Separate power supply for heated bed. While boards may have separate power connectors for the heated bed, this does not mean they are designed for separate PSUs. To safely use separate PSUs on the same board, the different power rails should be isolated from each other. When an external MOSFET is used for the bed, this MOSFET (like the Little-Driver) usually has an optocoupler that separates the bed PWM signal from the bed supply. In designing the MRR ESPA's heated bed circuit, I included this optocoupler into the design. This separates the bed power supply from the rest of the board, which means separate PSUs can be used on the same board. (Note: There are people who say you can use separate PSUs on the same board as long as the GND of the separate PSUs are tied together to get a common reference. I am not a certified electrician and therefore cannot say how safe this is. I prefer not to take the risk, and use the optocoupler instead since that is a much safer design.)
3. Trace width. High currents need thicker trace widths in order to have sufficient cross-sectional area to carry those currents. Simply put, the larger the current, the larger the cross-sectional area. Otherwise, the resistance in the traces will result in greater heat, and can use the PCB to melt and burn.
4. MOSFET selection. MOSFETs generate heat. While it is more complex, an easy way to calculate the amount of heat generated is to multiply the square of the current being driven by the on-resistance of MOSFET. (This is only one part of the equation, though.) This means that a low on-resistance is preferred for the heated bed MOSFET. But another factor is the thermal resistance, which is basically the temperature rise per watt of power generated. Some packages have 62 degC/W, while some are 50 degC/W. Obviously, the 50 degC/W MOSFET will get less hot when driving the same current. (And I learnt a lesson: Chinese clones of MOSFETs may not always work as expected.)
5. MOSFET cooling. Since MOSFETs generate heat, cooling is required. MOSFETs can usually be cooled by the PCB if there is a large enough "heat sink" area. The thermal resistance on data sheets usually state what that "heat sink" area is when they tested the MOSFET. If your board does not have a sufficient "heat sink" area to help cool the MOSFET, the thermal resistance will be significantly less than what is stated on the data sheet. This is why the bed MOSFET on the MRR ESPA has such a large copper area exposed on the reverse side of the board.
6. MOSFET gate voltage. MOSFETs usually have a range on-resistance that gets lower as the gate voltage increases. This is why it I designed the MRR ESPA with a jumper that controls the voltage used to drive the optocoupler that is connected to the MOSFET's gate. This optocoupler should use 12V if VBED is 12V so that the bed MOSFET can be driven with a high gate voltage. But if VBED is 24V, the optocoupler needs to be driven at a lower voltage since the maximum gate voltage is 20V. The jumper is based a connection to a voltage divider to divide the voltage by half; on a 24V VBED, this means the optocoupler and gate voltage is 12V. Other options used in board designs are MOSFET drivers, but these may or may not be isolated gate drivers.
7. Fuses. Since 3D printers can draw high currents for the heated bed, it can be a fire hazard if too much current is being drawn. Boards have components designed for a maximum current rating; above that, components may melt and burn. Fuses are therefore absolutely necessary to ensure that boards do not draw more current than what they are designed for.
Nowadays, when I see a new board on the market, I always look at the following (external appearance, and schematic if available):
- Does it have a separate connector for bed power?
- Does it have fuses?
- What is the MOSFET used for the heated bed?
- Does it have an optocoupler to isolate the bed supply from the supply used for the rest of the board?
- Does it have sufficient "heat sink" area under the bed MOSFET?
- What type of connectors does it use from/to bed supply, and from/to heated bed?
Hopefully, this helps people when they are selecting/design a 3D printer control board. And help prevent unnecessary fires.
1. Separate power connectors for heated bed. Heated beds can draw a lot of current. So using the same power supply for the heated bed and the rest of the 3D printer (control board, motors, hotend, etc.) can result in a very high current being carried from the power supply unit (PSU) and the board. If the power connector on the board is not rated for such high currents, they can melt, burn, and become a fire hazard. This is why some boards are designed with a separate set of power connectors for the heated bed. This allows the same PSU (which usually have a few sets of output leads) to be connected to the board, with one set of output leads for the board and such, while another set of leads for the heated bed. This lowers the current being drawn on each set of leads, keeping temperatures lower.
2. Separate power supply for heated bed. While boards may have separate power connectors for the heated bed, this does not mean they are designed for separate PSUs. To safely use separate PSUs on the same board, the different power rails should be isolated from each other. When an external MOSFET is used for the bed, this MOSFET (like the Little-Driver) usually has an optocoupler that separates the bed PWM signal from the bed supply. In designing the MRR ESPA's heated bed circuit, I included this optocoupler into the design. This separates the bed power supply from the rest of the board, which means separate PSUs can be used on the same board. (Note: There are people who say you can use separate PSUs on the same board as long as the GND of the separate PSUs are tied together to get a common reference. I am not a certified electrician and therefore cannot say how safe this is. I prefer not to take the risk, and use the optocoupler instead since that is a much safer design.)
3. Trace width. High currents need thicker trace widths in order to have sufficient cross-sectional area to carry those currents. Simply put, the larger the current, the larger the cross-sectional area. Otherwise, the resistance in the traces will result in greater heat, and can use the PCB to melt and burn.
4. MOSFET selection. MOSFETs generate heat. While it is more complex, an easy way to calculate the amount of heat generated is to multiply the square of the current being driven by the on-resistance of MOSFET. (This is only one part of the equation, though.) This means that a low on-resistance is preferred for the heated bed MOSFET. But another factor is the thermal resistance, which is basically the temperature rise per watt of power generated. Some packages have 62 degC/W, while some are 50 degC/W. Obviously, the 50 degC/W MOSFET will get less hot when driving the same current. (And I learnt a lesson: Chinese clones of MOSFETs may not always work as expected.)
5. MOSFET cooling. Since MOSFETs generate heat, cooling is required. MOSFETs can usually be cooled by the PCB if there is a large enough "heat sink" area. The thermal resistance on data sheets usually state what that "heat sink" area is when they tested the MOSFET. If your board does not have a sufficient "heat sink" area to help cool the MOSFET, the thermal resistance will be significantly less than what is stated on the data sheet. This is why the bed MOSFET on the MRR ESPA has such a large copper area exposed on the reverse side of the board.
6. MOSFET gate voltage. MOSFETs usually have a range on-resistance that gets lower as the gate voltage increases. This is why it I designed the MRR ESPA with a jumper that controls the voltage used to drive the optocoupler that is connected to the MOSFET's gate. This optocoupler should use 12V if VBED is 12V so that the bed MOSFET can be driven with a high gate voltage. But if VBED is 24V, the optocoupler needs to be driven at a lower voltage since the maximum gate voltage is 20V. The jumper is based a connection to a voltage divider to divide the voltage by half; on a 24V VBED, this means the optocoupler and gate voltage is 12V. Other options used in board designs are MOSFET drivers, but these may or may not be isolated gate drivers.
7. Fuses. Since 3D printers can draw high currents for the heated bed, it can be a fire hazard if too much current is being drawn. Boards have components designed for a maximum current rating; above that, components may melt and burn. Fuses are therefore absolutely necessary to ensure that boards do not draw more current than what they are designed for.
Nowadays, when I see a new board on the market, I always look at the following (external appearance, and schematic if available):
- Does it have a separate connector for bed power?
- Does it have fuses?
- What is the MOSFET used for the heated bed?
- Does it have an optocoupler to isolate the bed supply from the supply used for the rest of the board?
- Does it have sufficient "heat sink" area under the bed MOSFET?
- What type of connectors does it use from/to bed supply, and from/to heated bed?
Hopefully, this helps people when they are selecting/design a 3D printer control board. And help prevent unnecessary fires.
No comments:
Post a Comment