New MKS SBASE Smoothieware-compatible board from China
Posted by vreihen
|
Re: New MKS SBASE Smoothieware-compatible board from China June 23, 2015 10:20AM |
Has anyone tested this board yet? I've ordered it now as a simple to use / cheap electronics kit. I'm also curious if these boards have mosfet protection with a freewheeling diode to avoid the high voltage spikes from the parasitic inductances of the heat bed / heater cartridge.
I hope by the time I get it someone will have already ported DC42's method for quick delta autocalibration to smoothieware
Edited 1 time(s). Last edit at 06/23/2015 10:26AM by Dejay.
I hope by the time I get it someone will have already ported DC42's method for quick delta autocalibration to smoothieware
Edited 1 time(s). Last edit at 06/23/2015 10:26AM by Dejay.
|
Re: New MKS SBASE Smoothieware-compatible board from China June 27, 2015 12:08PM |
I received the MKS SBASE V1.0 Motherboard yesterday and spent the night checking it out.
The small MOSFETS for E0 E1 and FAN are LR7843 International Rectifier 30V 3.3mOhm 161A/140W. The one large MOSFET for the BED is VS40200AT Vangaurd Semiconductor 40V 3mOhm 200A/300W. The MOSFETs as well as the DRV8825s are heatsinked with vias through the board to the back, where they meet a second copper pad that's then coated in solder. All the MOSFETs are driven by TC4427 MOSFET drivers which are supplied with 12V for gate voltage by a L7812C2T linear regulator on board. Current for the steppers is set using Microchip MCP4451-503 digital pots, and there are headers for external steppers or for linking E1 to Z for driving two motors from one control. The caps are 105C rated ChengX capacitors.
The board is powered by a STmicro L5970 step-down switching regulator or via USB rated for 1A. Power draw is 100mA at 12V, down to 70mA at 24V without the USB plugged in. Power drops to 47mA at 12V, 40mA at 24V with the USB plugged in. Because the regulator is rated for 1A, I figued what the heck and plugged a Pi into it. The voltage dropped to 4.65V but the Pi booted and ran along with an Edimax wifi dongle. So theoretically you could run OctoPrint right from this, but I'm not sure the designers had this in mind when the build the SBASE so the inductor is probably not rated for this load.
There's 5 LEDs that indicate the status of the board (like the 6 smoothie has except without the VBB?) they blink and all that good stuff to help troubleshoot. There's also an LED by each MOSFET just like smoothie. The ethernet works, that's all I have to say about that.
There's a 6 pin programming header, a well as an IBOOT jumper (not sure what that does). The connectors are:
EXP1
1 - P1.31 - Buzzer
2 - P1.30 - Click button Pin
3 - P0.18 - LCDE (MOSI0)
4 - P0.16 - LCDRS (SSEL0)
5 - P0.17 - LCD4 (MISO0)
6,7,8 - NC
9 - GND
10 - 5V
EXP2
1 - P0.8 - MISO1
2 - P0.7 - SCK1
3 - P3.25 - Encoder Pin A
4 - P0.28 - SD select (SCL0)
5 - P3.36 - Encoder Pin B
6 - P0.9 - MOSI1
7 - P0.27 - SD detect (SDA0)
8 - Reset
9 - GND
10 - 5V
AUX1 is 5V/GND. The RX, TX, RTS, CTS (labeled as MKS TFT in the seller image) do not appear to be connected directly to the CPU, however using a TTL serial cable I get the smoothie console if I connect to RX/TX/GND. EXP1 and EXP2 are the pinout to connect to a 12864LCD with encoder and SD card. A "2004" LCD will not work as the other LCD data lines are not connected to the CPU. There is a block of 3 jumpers labeled J16. These connect the EXT2 MISO, MOSI, and SCK pins to the controller. If you pull the jumpers off I guess you could use the SPI bus for something else?
I tested a 1.4 ohm heated bed at 12V (8.5A) on the large FET. 20 minutes later it was still nice and cool. The area around the fuse (slightly below and to the left of the FET) is actually hotter.
For fun I put the heated bed on one of the extruder FETs. Hey they're rated for it, why not? It seem to work ok as well
Kysan 42BYGH4803 steppers were attached and current set to 1.5A. After one hour of running they were still chugging away. They get up to temperature pretty quickly but stay there.
For comparison I dropped the stepper current to 1A and they chips cool as expected.
So my impression of this board is that it is quite good. For the price I really expected it to be a lot lower quality and poorly built. I sort of hoped it would be a piece of junk so I could confidently say that the official smoothieboard was better in every way, but I can't, this is just a good cheap knockoff. The downside for me is that I would like to control the printer's LED lights, as well as a second fan and this just doen't have the headers to be able to wire mosfets to for that purpose. If that's what you need, then smootheboard all the way, but this is a solid alternative for running a dual extruder printer with the power of smoothieware on the cheap.
Edited 3 time(s). Last edit at 06/27/2015 02:04PM by CapnBry.
The small MOSFETS for E0 E1 and FAN are LR7843 International Rectifier 30V 3.3mOhm 161A/140W. The one large MOSFET for the BED is VS40200AT Vangaurd Semiconductor 40V 3mOhm 200A/300W. The MOSFETs as well as the DRV8825s are heatsinked with vias through the board to the back, where they meet a second copper pad that's then coated in solder. All the MOSFETs are driven by TC4427 MOSFET drivers which are supplied with 12V for gate voltage by a L7812C2T linear regulator on board. Current for the steppers is set using Microchip MCP4451-503 digital pots, and there are headers for external steppers or for linking E1 to Z for driving two motors from one control. The caps are 105C rated ChengX capacitors.
The board is powered by a STmicro L5970 step-down switching regulator or via USB rated for 1A. Power draw is 100mA at 12V, down to 70mA at 24V without the USB plugged in. Power drops to 47mA at 12V, 40mA at 24V with the USB plugged in. Because the regulator is rated for 1A, I figued what the heck and plugged a Pi into it. The voltage dropped to 4.65V but the Pi booted and ran along with an Edimax wifi dongle. So theoretically you could run OctoPrint right from this, but I'm not sure the designers had this in mind when the build the SBASE so the inductor is probably not rated for this load.
There's 5 LEDs that indicate the status of the board (like the 6 smoothie has except without the VBB?) they blink and all that good stuff to help troubleshoot. There's also an LED by each MOSFET just like smoothie. The ethernet works, that's all I have to say about that.
There's a 6 pin programming header, a well as an IBOOT jumper (not sure what that does). The connectors are:
EXP1
1 - P1.31 - Buzzer
2 - P1.30 - Click button Pin
3 - P0.18 - LCDE (MOSI0)
4 - P0.16 - LCDRS (SSEL0)
5 - P0.17 - LCD4 (MISO0)
6,7,8 - NC
9 - GND
10 - 5V
EXP2
1 - P0.8 - MISO1
2 - P0.7 - SCK1
3 - P3.25 - Encoder Pin A
4 - P0.28 - SD select (SCL0)
5 - P3.36 - Encoder Pin B
6 - P0.9 - MOSI1
7 - P0.27 - SD detect (SDA0)
8 - Reset
9 - GND
10 - 5V
AUX1 is 5V/GND. The RX, TX, RTS, CTS (labeled as MKS TFT in the seller image) do not appear to be connected directly to the CPU, however using a TTL serial cable I get the smoothie console if I connect to RX/TX/GND. EXP1 and EXP2 are the pinout to connect to a 12864LCD with encoder and SD card. A "2004" LCD will not work as the other LCD data lines are not connected to the CPU. There is a block of 3 jumpers labeled J16. These connect the EXT2 MISO, MOSI, and SCK pins to the controller. If you pull the jumpers off I guess you could use the SPI bus for something else?
I tested a 1.4 ohm heated bed at 12V (8.5A) on the large FET. 20 minutes later it was still nice and cool. The area around the fuse (slightly below and to the left of the FET) is actually hotter.
For fun I put the heated bed on one of the extruder FETs. Hey they're rated for it, why not? It seem to work ok as well
Kysan 42BYGH4803 steppers were attached and current set to 1.5A. After one hour of running they were still chugging away. They get up to temperature pretty quickly but stay there.
For comparison I dropped the stepper current to 1A and they chips cool as expected.
So my impression of this board is that it is quite good. For the price I really expected it to be a lot lower quality and poorly built. I sort of hoped it would be a piece of junk so I could confidently say that the official smoothieboard was better in every way, but I can't, this is just a good cheap knockoff. The downside for me is that I would like to control the printer's LED lights, as well as a second fan and this just doen't have the headers to be able to wire mosfets to for that purpose. If that's what you need, then smootheboard all the way, but this is a solid alternative for running a dual extruder printer with the power of smoothieware on the cheap.
Edited 3 time(s). Last edit at 06/27/2015 02:04PM by CapnBry.
|
Re: New MKS SBASE Smoothieware-compatible board from China June 27, 2015 08:22PM |
Thank you CapnBry for testing it out! Now I feel even better for ordering it
Depending on how much I like smoothieware I'll just make a direct donation to it.
I didn't order any LCD for the board. I want to just use it over the web interface and without USB / SD card. Is that a bad idea? Can I even copy a gcode file over ethernet onto the board?
Depending on how much I like smoothieware I'll just make a direct donation to it.
I didn't order any LCD for the board. I want to just use it over the web interface and without USB / SD card. Is that a bad idea? Can I even copy a gcode file over ethernet onto the board?
|
Re: New MKS SBASE Smoothieware-compatible board from China June 30, 2015 08:34AM |
Oh I totally agree. It is a shame it is so slow though because I'm sure a lot of code went into making it work. I think half of the problem is the UIP library they use for the networking. I used this on an Arduino many years back to add networking to my project and it was pretty A) amazing that a whole network stack can be implemented in like 15kB 
awful because you could only support one connection, and only send frames up to like 1kB and only have one "on the wire" at any time. The other half of it is Windows' delayed ack which waits 200ms before sending an ack to a TCP connection, expecting more data to come. If you can only have one frame on the wire at any time that adds 200ms additional latency to every block of data (which were already small to begin with). UIP is fine for sending small chunks of data at high latency but not at all for moderate throughput. A microcontroller needs, like a few hundred KB of RAM to be able to "fill the pipe" to get throughput up.
I have a Raspberry Pi hooked to my current printer, and this one will get one too, for running OctoPrint. As someone pointed out, this adds $50-55 to the cost of the board, but I like the functionality of OctoPrint, upload an STL, it slices (through the cloud) and prints when it is ready add a webcam and you can check on it from the other room. The network performance still tops out around 6 MBytes/s but that's enough for my workflow, I'm not moving hundreds of megs of data there.
awful because you could only support one connection, and only send frames up to like 1kB and only have one "on the wire" at any time. The other half of it is Windows' delayed ack which waits 200ms before sending an ack to a TCP connection, expecting more data to come. If you can only have one frame on the wire at any time that adds 200ms additional latency to every block of data (which were already small to begin with). UIP is fine for sending small chunks of data at high latency but not at all for moderate throughput. A microcontroller needs, like a few hundred KB of RAM to be able to "fill the pipe" to get throughput up.I have a Raspberry Pi hooked to my current printer, and this one will get one too, for running OctoPrint. As someone pointed out, this adds $50-55 to the cost of the board, but I like the functionality of OctoPrint, upload an STL, it slices (through the cloud) and prints when it is ready add a webcam and you can check on it from the other room. The network performance still tops out around 6 MBytes/s but that's enough for my workflow, I'm not moving hundreds of megs of data there.
|
Re: New MKS SBASE Smoothieware-compatible board from China July 05, 2015 04:24PM |
Well I found something to complain about now that the rest of the printer is up and running. I could not get it to maintain a hotend temperature on my official E3D v6 hotend. It would fluctuate 3-5 degrees either way, and the PID controller was going nuts trying to maintain it. I tried the PID autotune but it didn't help, so what I did was just set the PWM heater output manually to a fixed value and waited for the temperature to stabilize:
So yeah, there's no way any PID algorithm is going to be able to control anything if the input is bonkers. Watching the temperature on the LCD, it will be at a temperature then jump up 2 degrees in one second then jump back down. There's just way too much noise for the software to work properly.
I'm not sure if the real Smoothieboard just does a better job of suppressing the noise, but the thermistor chosen by E3D seems ill-suited for this application. At 230C, there is just 4.03 ohms per degree , which would be over 1 degree C per 10 bit ADC tick, and you'd need less than 3mV of noise to be able to measure that which we can see here the noise is at least 3 times that:
So yeah, there's no way any PID algorithm is going to be able to control anything if the input is bonkers. Watching the temperature on the LCD, it will be at a temperature then jump up 2 degrees in one second then jump back down. There's just way too much noise for the software to work properly.
I'm not sure if the real Smoothieboard just does a better job of suppressing the noise, but the thermistor chosen by E3D seems ill-suited for this application. At 230C, there is just 4.03 ohms per degree , which would be over 1 degree C per 10 bit ADC tick, and you'd need less than 3mV of noise to be able to measure that which we can see here the noise is at least 3 times that:
Sorry, only registered users may post in this forum.