Recore A3

Recore Rev A3

This document is for Revision A3, please see Recore_A2 for the previous version.

Hardware overview

Software for Recore

Recore should come pre-flashed with ReFactor image. If you want to update that, here are the steps:

ReFactor

Flashing ReFactor from a USB drive

Get the latest ReFactor image from here: https://github.com/goeland86/ReFactor/releases for example:

wget https://github.com/goeland86/ReFactor/releases/download/v3.0.0-RC4/ReFactor-recore-v3.0.0-RC4.img.xz

Assuming you have managed to boot Recore from a USB flash drive and you have the ReFactor image available in the same folder, you can use dd to do a direct transfer of the image from the USB drive to eMMC.

xz -v -d -c Refactor-recore-v3.0.0-RC3.img.xz  | dd of=/dev/mmcblk0

After that, make sure you flash the right u-boot version.

From Scratch

Starting with Armbian Buster Server for Pine 64 (kernel 5.4) https://redirect.armbian.com/pine64/Buster_current Flash the software to a USB drive.

Once flashed, it's necessary to add the Recore device tree to the right folder.

cp sun50i-a64-recore.dtb /media/xxx/boot/dtb/allwinner/

With a clean board, it's necessary to start the board from a host computer using FEL mode.

sunxi-fel -v -p spl sunxi-spl.bin write 0x44000 bl31.bin write 0x4a000000 u-boot.bin reset64 0x44000

The output will appear in teminal UART0. You can use minicom on a host computer to watch the boot process and get a login screen

minicom -D /dev/ttyUSB0 -w

If a dongle with Armbian is present in one of the USB host slots, it should boot from the dongle.

Booting from USB if an OS is present on the eMMC

If the eMMC is parititioned and has an OS, u-boot will launch right into that. In order to override that and boot from a USB dongle, stop u-boot by pressing a key and then write:

setenv devnum 0
run usb_boot

Install distro to eMMC

Create Partition:

export card=/dev/mmcblk0
blockdev --rereadpt ${card}
cat <<EOT | sfdisk ${card}
1M,,L
EOT
mkfs.ext4 ${card}p1

Copy Kernel and rootfs to eMMC

Use the flash script in Redist

mount /dev/mmcblk0p1  /mnt/
rsync -aAx --human-readable --info=name0,progress2 --exclude={/dev/*,/proc/*,/sys/*,/tmp/*,/run/*,/media/*,/mnt/*,lost+found/} /* /mnt
echo "rootdev=/dev/mmcblk0p1" >> /mnt/boot/armbianEnv.txt

Update the UUID in fstab to the UUID of the eMMC part 1

export UUID=`blkid | grep mmcblk0p1 | awk -F'"' '{print $2}'`
sed -i "s/UUID=.*\//UUID=$UUID \//g" /mnt/etc/fstab

Make eMMC bootable

Place U-boot-spl in sector 15 (8KB offset) Clean card:

dd if=/dev/zero of=/dev/mmcblk0 bs=1k count=1023 seek=1

Flash the binary

dd if=u-boot-sunxi-with-spl.bin of=/dev/mmcblk0 bs=1024 seek=8 conv=notrunc

Note: U-boot needs to be patched in order to use the eMMC as the first (and only) boot device. The changes can be done in "spl_mmc_get_device_index" where dev_num should be 0 instead of 1.

Install Klipper

It's good to add a separate user for Klipper:

useradd klipper -s /bin/bash -G sudo,dialout
passwd klipper
mkdir /home/klipper
chown klipper:klipper /home/klipper

su klipper
cd ~
git clone https://github.com/intelligent-agent/klipper
cp klipper/config/generic-recore.cfg ./printer.cfg
./klipper/scripts/install-recore-a3.sh

You can check the log by writing

tail -f /tmp/klippy.log 

Install STM32 firmware

This flash scripts requires gpiod

sudo apt install gpiod

cd /home/klipper/klipper
cp .config_stm32 .config
make
sudo ./scripts/flash-stm32.sh out/klipper.bin (might have to be run several times)

Install AR100 firmware

See Refactor#AR100_firmware for instructions

Running Klipper

After a reboot, the ar100 firmware needs to be reflashed. We will get this into an automatic routine eventually, bit for now:

sudo /home/klipper/klipper/scripts/flash-ar100.py /home/klipper/klipper/out/ar100.bin

While details are ironed out there is a script that can be used to restart Klipper that will handle resetting things and setting up the steppers etc.

/home/klipper/klipper/scripts/restart-recore.py

Giving right privileges for AR100 restart

The AR100 flasher and reset script needs to be run as root. We create a group called flashers, add klipper to that and gie the group sudo access to the script

sudo addgroup flashers
sudo adduser klipper flashers
echo '%flashers ALL=(ALL) NOPASSWD: /home/klipper/klipper/scripts/flash-ar100.py *, /usr/bin/gpioset *' > /etc/sudoers.d/flashers

Install Octoprint

su klipper
cd ~
sudo apt update
sudo apt install python-pip python-dev python-setuptools python-virtualenv git libyaml-dev build-essential
mkdir OctoPrint && cd OctoPrint
virtualenv venv
source venv/bin/activate
pip install pip --upgrade
pip install octoprint
./venv/bin/octoprint serve

Add a systemd start-up script

nano /lib/systemd/system/octoprint.service

Add the following content

[Unit]
Description=The snappy web interface for your 3D printer
After=network-online.target
Wants=network-online.target

[Service]
Type=simple
User=klipper
ExecStart=/home/klipper/OctoPrint/venv/bin/octoprint

[Install]
WantedBy=multi-user.target

Enable and start script

sudo systemctl enable octoprint
sudo systemctl start octoprint

Manual control and testing

To use the examples in this section, first install gpiod

apt install gpiod

Power domains

There are 9 power domains around the board controlling voltage on different pins:

• Input power controls power to 4 high power outputs, the 4 fans and the 6 stepper drivers. This input has current monitoring, fast acting over current protection, voltage monitoring, temperature monitoring and reverse polarity protection.
• Thermo couple power Controls +5V/1A output to the ADCs. This can be used for turning on power to the analog pins.
• End stop power Controls +5V/1A output on the six endstops. Ganged. ES5 can be switched to have 12V output. The 12V output has a

100mA internal current limit.

• 4 USB host power domains Controls 5V/1A current output to the usb host connectors. These are turned on by u-boot.
• HDMI 5V output Controls the 5V/1A HDMI output. Turned on by u-boot.

Input stage

Reset over current protection and set it in "one-shot" mode.

gpioset 1 193=0
gpioset 1 193=1

The over current protection can also be set in "transparent" mode, where the current alarm will be reset automatically. Note that this is a bad idea for general operation, only use for testing.

gpioset 1 193=0

Enable 24V input

gpioset 1 194=0

End stop 5V /12V

End stop 0 to 4 has a programmable +5V output voltage. End stop 5 has +5V or +12V selectable.

To enable +5V on ES 0...4

gpioset 1 203=1

To disable again

gpioset 1 203=0

To switch to 12V output on ES5

gpioset 1 199=1

To disable again

gpioset 1 199=0

End stops values

To see what value the end stops have

gpioget 1 228 # ES 5
gpioget 1 229
gpioget 1 230
gpioget 1 231
gpioget 1 232
gpioget 1 233 # ES 0

HDMI power

echo 201 > /sys/class/gpio/export
echo out > /sys/class/gpio/gpio201/direction
echo 1 > /sys/class/gpio/gpio201/value

TMC2209

In window/shell 1:

stty -F /dev/ttyS2 raw -echoe -echo
xxd -c 4 /dev/ttyS2

In window/shell 2:

echo -n -e '\x5\x0\x6\x6F' > /dev/ttyS2

The return should be along the lines of

00000000: 0500 066f  ...o
00000004: 05ff 0620  ... 
00000008: 0001 4058  ..@X

To read the OTP registers

echo -n -e "\x5\x0\x5\x21" > /dev/ttyS2
echo -n -e "\x5\x0\x5\x21" > /dev/ttyS3
echo -n -e "\x5\x1\x5\x97" > /dev/ttyS2
echo -n -e "\x5\x1\x5\x97" > /dev/ttyS3
echo -n -e "\x5\x2\x5\x7a" > /dev/ttyS2
echo -n -e "\x5\x2\x5\x7a" > /dev/ttyS3
echo -n -e "\x5\x3\x5\xcc" > /dev/ttyS2
echo -n -e "\x5\x3\x5\xcc" > /dev/ttyS3

To set the OTP register to use RDSon:

echo -n -e "\x5\x0\x84\x0\x0\xbd\x6\xac" > /dev/ttyS2

To check that no errors are reported by the steppers write this on the command line. If a "1" shows up, that is an indication that the motor is reporting an error.

gpioget 1 128
gpioget 1 129
gpioget 1 130
gpioget 1 131
gpioget 1 132
gpioget 1 133

STM32F031

There is a small firmware to test all functionality that is controlled by the STM32F031 chip on the board. Once uploaded to the board using the flash script, the firmware will turn on all LEDs it controls and send back ADC readings.

To flash this firmware to the STM32:

gpioset 1 197=1
gpioset 1 196=0
stm32flash -i -196,196 -w Sumato-f031.bin -v -g 0x00 /dev/ttyS1
gpioset 1 197=0
stm32flash -i -196,196 /dev/ttyS1

To see the ADC readings:

stty -F /dev/ttyS3 38400 raw
cat /dev/ttyS3

You should see something like

ADC T0: 0
ADC T1: 0
ADC T2: 0
ADC T3: 0
ADC U: 2338
ADC I: 0
ADC TB: 3417

The following one-liners assumes gawk is installed.

Reading Current

The current shunt resistor is 1 mOhm, the amplifier is 20 times, the vref is 3.3V, so we get: (adc_val/4096*3.3)*1000/20 = (adc_val/4096)*165 You can convert the current (as in amps) reading to something useful like this:

cat /dev/ttyS3 | awk '/I:/ { printf "scale=4; (%i/4096)*165\n", $3; fflush(); }' | bc -l

Reading Board Temperature

To see the temperature readings, try this :

export beta=4327
cat /dev/ttyS1 | awk '/TB:/ { printf "scale=4; 4327/l( (4700/((3.3/(3.3*%i/4096))-1.0))/0.05306820342563400000) -273.15\n", $3; fflush(); }' | bc -l | awk '{print strftime("%k:%M:%S"), $0; fflush();}'

Reading Input Voltage

To see voltage on the input, it can be converted like this:

cat /dev/ttyS3 | awk '/T:/ { printf "scale=4; (%i/4096)*3.3\n", $3; fflush(); }' | bc -l

Thermistor inputs

To enable pull-up on input for using thermistors for temperature measurements:

gpioset 1 102=1
gpioset 1 120=1
gpioset 1 160=1
gpioset 1 161=1

Setting Op-amp gain to 1:

gpioget 1 100
gpioget 1 235
gpioget 1 145
gpioget 1 34

In the device tree file, the gpio/ldos must be set to gpio input:

&gpio0_ldo {
  function = "gpio_in";
};

Thermocouple input

Either a thermocouple or a thermistor can be used as input on the T0-T3 inputs.

Bias

The thermocouple is set up to have a Bias of 0.7V/100 on the input which enables negative values (lower than the board temperature) as input to the ADCs. With the input short circuited, the bias can be recorded. Typical values are 750 This needs to be subtracted from the result.

Cold junction compensation

Another important aspect is cold junction compensation. There is a thermistor connected on the board close to the thermocouple inputs which can be used to measure the temperature close to the input junction. This temperature must be added in software.

For Rev A3, the cold junction compensation is actually located a bit away from the junction, so it might not be accurate. The connected thermistor is a TDK NTCG104EF104FT1X The connected Thermocouple has a (25/100) beta value of 4327.

In order to use the inputs for thermocouple, the pull-up resistors must be deactivated:

gpioget 1 102
gpioget 1 120
gpioget 1 160
gpioget 1 161

Also, the op-amp needs a gain of 100:

gpioset 1 100=0
gpioset 1 235=0
gpioset 1 145=0
gpioset 1 34=0

Finally, the input bias needs to be enabled. The following LDOs needs to be enabled:

reg_ldo_io0
reg_ldo_io1
reg_dldo3
reg_dldo4

AR100

See also: AR100

Loading a binary into SRAM A2

Get Crust in order to load the AR100 bin file into memory

git clone https://github.com/intelligent-agent/ar100-fw
cd ar100-fw
make defconfig
HOST_COMPILE= make tools

Once the load program is compiled, an AR100 binary blob can be loaded with

./ar100-fw/build/tools/load scp.bin

Please note that the AR100 binary file bust be 8 byte aligned, meaning that the size must be divisible by 8.

Performance

Start-up time

With a maximum clock for MMC2 of 200 MHz, here is the start-up time:

root@recore-3:~# systemd-analyze 
Startup finished in 5.128s (kernel) + 11.567s (userspace) = 16.695s 
graphical.target reached after 11.421s in userspace

Measured voltages

Instruments used:

• Siglent SDM3065X, 6.5 digit multimeter
• HP 3458A, 8.5 digit multimeter
• HP 3245A, Universal Source

Microcontroller voltages

VDDA uC: 3.312 V
VDD uC: 3.312 V

Voltage from the PMIC supplied as bias:
0.706 V

Offset and Gain Measurements

Input voltage measured with SDM3065X, output voltage measured with 3458A. Programmable Voltage reference used was the DIY DAC cape made by Elias. Measurements should be repeated with HP 3245A.

Op-amp gain measurements
Vin (mV)	Vout (V)	Gain
1.0103	0.754	95.73
2.0038	0.8492	95.78
3.013	0.9457	95.72
4.0032	1.04	95.60
5.0108	1.1372	95.78
6.0016	1.232	95.76
7.0094	1.3285	95.76
8.0015	1.4235	95.76
9.0106	1.5201	95.76
10.0005	1.615	95.77
11.0102	1.7116	95.76
12.0042	1.8067	95.75
13.0142	1.9035	95.76
14.0056	1.9985	95.76
15.0145	2.095	95.76
16.0055	2.1899	95.76
17.0145	2.2865	95.75
18.0055	2.3813	95.75
19.0163	2.4783	95.76
20.0067	2.5723	95.72

An experiment was done using an HP 3245A and a HP 3458A measuring the input and output voltages and calculating offset and gain. This will become a part of a calibration routine during final testing and flashing, values should be stored with the board.

Input 1	Vin	Vout	Offset	
	0.0003	0.6685	0.6685	
				
Vin set	Vin	Vout	Gain	Deviation
-5 mV	-4.9983	0.1840	96.93	0.16
0 mV	0.0003	0.6685	96.76	-0.01
5 mV	5.0000	1.1523	96.77	0.00
10 mV	10.0013	1.6355	96.69	-0.08
15 mV	15.0005	2.1198	96.75	-0.01
20 mV	20.0005	2.6025	96.70	-0.07
	Average gain		96.77	
				
				
				
Input 2	Vin	Vout	Offset	
	0.0017	0.6485	0.6483	
				
				
Vin set	Vin	Vout	Gain	Deviation
-5 mV	-4.9961	0.1722	95.30	-0.07
0 mV	0.0017	0.6485	95.37	-0.01
5 mV	5.0014	1.1253	95.37	-0.01
10 mV	10.003	1.6025	95.39	0.01
15 mV	15.0021	2.0793	95.38	0.01
20 mV	20.0023	2.5559	95.37	-0.01
	Average gain		95.38	
				
				
Input 3	Vin	Vout	Offset	
	0.0013	0.6560	0.6559	
				
				
Vin set	Vin	Vout	Gain	Deviation
-5 mV	-4.9969	0.1794	95.35	-0.02
0 mV	0.0013	0.6560	95.35	-0.03
5 mV	5.0012	1.1329	95.38	0.01
10 mV	10.0025	1.6098	95.37	-0.01
15 mV	15.0015	2.0864	95.36	-0.02
20 mV	20.0013	2.5631	95.35	-0.02
	Average gain		95.36	
				
				
Input 4	Vin	Vout	Offset	
	0.0024	0.6498	0.6496	
				
				
Vin set	Vin	Vout	Gain	Deviation
-5 mV	-4.9954	0.1714	96.98	1.18
0 mV	0.0024	0.6498	94.50	-1.31
5 mV	5.0015	1.1283	95.96	0.16
10 mV	10.0028	1.6069	95.83	0.02
15 mV	15.0024	2.0853	95.78	-0.02
20 mV	20.0022	2.5642	95.78	-0.02
	Average gain		95.81	

5V Buck converter ripple and noise

The 5V step down/Buck converter is a AOZ2151PQI-10. It can provide 4A and has an input voltage range of 12-28V. It is hard wired to be in PWM mode. PFM mode results in a loud whine.

Below is a screenshot of the ripple and noise with a 4 A load across the 5V rail. An experiment was done with adding more capacitance. The three examples below show 80 uF, 120 uF and 160 uF capacitor bank. File:80uF.png File:120uF.png File:160uF-ripple-RevA2.png

Input stage

An experiment with a constant current load drawing 20 A from the bed connector with and without a 92 mm fan on top of the board for forced convection. The temperature measured was with the on board thermistor. Using a thermal camera, the temperature displayed matches well with the temperature reported by the camera. However, in Rev A2, this is not the hottest point on the board. The MOSFET controlling the bed seems to consistently get the highest temperature. The reason for this is probably due to a 3.3 V VGS.

Inrush current

During turn on of the high power stage, there will be an inrush current to charge up the bank of capacitors present on the board. If no loads are turned on, the inrush current reaches 31 A at peak. This is important to consider when setting the current limit with device tree.

Temperature low pass filter

A Low pass filter has been added on each of the temperature input channels. The cut-off frequency is below what my oscilloscope can handle, but a plot in the time domain verifies a 7 Hz -3dB point.

ADC offset and gain measurements

To measure the ADC offset error, measure the voltage on the uC input pin and record when the transition from 0 to 1 occurs. It should occur at 0.5 LSB = 3.3/4096/2 = 0.4028 mV. Offset: 55.35 mV - 0.4028 mV = 54.95 mV

Known Issues

Linux

The default config option for Pine64 seems to be a maximum of 4 uarts. In order to increase that, it is necessary to compile a new kernel.

Rev A4

High current draw on USB lines can cause instability in some devices. Specifically having a webcam or a wifi dongle and a Manga Screen 2 can cause flicker on the screen.

Rev A3

Ethernet slow at appearing, limited to 100 MB/s

There is an issue with the ethernet connection, that both makes ethernet slow to come up, and only comes up as 100 MB/s. It looks like it is due to an error in the schematic where some of the connections are crossed. Here is a typical dump of dmesg on a particularly long wait:

root@recore-3:~# dmesg | grep eth
[    4.586525] dwmac-sun8i 1c30000.ethernet: IRQ eth_wake_irq not found
[    4.592983] dwmac-sun8i 1c30000.ethernet: IRQ eth_lpi not found
[    4.599124] dwmac-sun8i 1c30000.ethernet: PTP uses main clock
[    4.605035] dwmac-sun8i 1c30000.ethernet: supply phy not found, using dummy regulator
[    4.613054] dwmac-sun8i 1c30000.ethernet: supply phy-io not found, using dummy regulator
[    4.623598] dwmac-sun8i 1c30000.ethernet: Current syscon value is not the default 6 (expect 0)
[    4.634957] dwmac-sun8i 1c30000.ethernet: No HW DMA feature register supported
[    4.649102] dwmac-sun8i 1c30000.ethernet: RX Checksum Offload Engine supported
[    4.656372] dwmac-sun8i 1c30000.ethernet: COE Type 2
[    4.661378] dwmac-sun8i 1c30000.ethernet: TX Checksum insertion supported
[    4.668226] dwmac-sun8i 1c30000.ethernet: Normal descriptors
[    4.673920] dwmac-sun8i 1c30000.ethernet: Chain mode enabled
[   16.682181] dwmac-sun8i 1c30000.ethernet eth0: PHY [stmmac-0:07] driver [RTL8211E Gigabit Ethernet] (irq=POLL)
[   16.705860] dwmac-sun8i 1c30000.ethernet eth0: No Safety Features support found
[   16.713354] dwmac-sun8i 1c30000.ethernet eth0: No MAC Management Counters available
[   16.713366] dwmac-sun8i 1c30000.ethernet eth0: PTP not supported by HW
[   16.713387] dwmac-sun8i 1c30000.ethernet eth0: configuring for phy/rgmii link mode
[  249.157988] dwmac-sun8i 1c30000.ethernet eth0: Link is Up - 100Mbps/Full - flow control rx/tx
[  249.166587] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready

root@recore-3:~#  sudo ethtool eth0
Settings for eth0:
	Supported ports: [ TP MII ]
	Supported link modes:   10baseT/Half 10baseT/Full 
	                        100baseT/Half 100baseT/Full 
	                        1000baseT/Half 1000baseT/Full 
	Supported pause frame use: Symmetric Receive-only
	Supports auto-negotiation: Yes
	Supported FEC modes: Not reported
	Advertised link modes:  10baseT/Half 10baseT/Full 
	                        100baseT/Half 100baseT/Full 
	                        1000baseT/Half 1000baseT/Full 
	Advertised pause frame use: Symmetric Receive-only
	Advertised auto-negotiation: Yes
	Advertised FEC modes: Not reported
	Link partner advertised link modes:  10baseT/Half 10baseT/Full 
	                                     100baseT/Half 100baseT/Full 
	Link partner advertised pause frame use: Symmetric Receive-only
	Link partner advertised auto-negotiation: Yes
	Link partner advertised FEC modes: Not reported
	Speed: 100Mb/s
	Duplex: Full
	Port: MII
	PHYAD: 7
	Transceiver: internal
	Auto-negotiation: on
	Supports Wake-on: d
	Wake-on: d
	Current message level: 0x0000003f (63)
			       drv probe link timer ifdown ifup
	Link detected: yes

root@recore-3:~# mii-tool eth0 -v
eth0: negotiated 100baseTx-FD flow-control, link ok
  product info: vendor 00:07:32, model 17 rev 5
  basic mode:   autonegotiation enabled
  basic status: autonegotiation complete, link ok
  capabilities: 1000baseT-HD 1000baseT-FD 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD
  advertising:  1000baseT-HD 1000baseT-FD 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD flow-control
  link partner: 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD flow-control

An experiment was done with patching the wires on a board level and Gigabit ethernet seems to work fine in that instance.

Reset button not responding

Simply pressing the reset button does not always cause a reset of the board.

- Does not seem to be an issue for Rev A3, needs to be a part of the testing routine.

On-board thermistor source impedance

The on-board thermistor chosen for A2 is 100 K, which in conjunction with the 4.7 K pull-up causes the source impedance to be above the recommended 50 K external input impedance. (100 K sinking) When sampling the value from the on-board thermistor, the ADC is loaded too much causing a dip in the value. The result is that a lower value sampled than what is actually present on the ADC pin. If the sample time is kept at minimum (1.5 cycles) the offset will be large. Increasing the sample time fixes that problem. File:SDS5034X PNG 68.png