eez-open / modular-psu Goto Github PK

Since adding of LSE xtal (#35) consumed two additional MCU pins, SER_OUT and PAR_OUT cannot be driven directly from MCU anymore. An I/O expander (I2C) has to be introduced on the BP3C to provide SERial and PARalell coupling control.

Modify fan filtering and driver to improve EMI by decreasing current peaks caused by charging capacitor connected in parallel with fan power lines (added to clean up EMI generated by fan itself).

UART interface is required for programming firmware of peripheral module MCUs (see also #41). Reassign DIN, DOUT and SD_DETECT pins that new UART_TX and UART_RX can be used. Use two currently unassigned pins on the 40-pin DIB backplane connector.

Add on the MCU board selection of input power for audio amplifier (IC12). Default is +3V3.

Both coupling relays on the BP3C PCB have to be rotated since module in slot #2 cannot be inserted

The OVP crowbar is connected to OUTF+ instead of POST_OUTF+

For chosen TFT display (RFE43BH-AIW-DNS) JP4 default position should be changed from RESET to GND.

I have seen selected MCU is an STM32F769IGT (LQFP176) which is very good anyway I think for future and to manage even more things it will be a must have to switch to pin compatible STM32H7 CPU with the big advantage to have Dual Core 1x CortexM4F @ 240MHz and 1x CortexM7@480MHz tons of FLASH/RAM peripherals ... that will allow to add special hard real-time monitoring / tasks... in one core and doing other things in an other core ...
It seems STM32H757IITx LQFP176 is full pin compatible (more check to be done to be sure) marketing status is active but it seems hard to find so far (maybe contacting ST directly will help)

USB B-type connector cannot be used for OTG that required 5-pins (with ID). It has to be replaced with Mini AB. Currently ID instead of GND is connected to the pin 4.

IC12 power inputs are wrongly connected (i.e. V+ to pin 11 instead of 4 and vice versa).

It seems that PLLI2SR cannot be used in MUX for generating 25 MHz on MCO2 output. Instead, HSE output can be used that require replacement of 10 MHz (Y1) XTAL with 25 MHz.

DCM220's UART lines are swapped before instead of after digital isolator.

Currently is not possible for MCU to find out if HW OVP is tripped or not. Therefore bring existing OVP_FAULT signal to one of available I/O Exp input.

The MAX31760 I2C address is set to 0xA0 that is already assigned to the EEPROM on the MCU board. Change it to the 0xAE to be over all EEPROM addresses (on MCU and peripheral modules).

Add place on the AUX-PS PCB for 2K2 pull-up on fan controller's THERM pin.

Make existing X4 20-pin JTAG connector optional or remove it completely and add 0.1" header for SWD/SWIM connection. Rearrange usage of IC6 and IC7 inputs.

• Adjust X1 pin mappings to follow "backplane" PCB.
• Move up X1 for 0.5 mm.

It seems that encoder's pinout is mirrored (i.e. A and C pins are swapped). Has to be checked and eventually rectified in Eagle library.

Hey, 5093d6b breaks the Kitspace pages because we currently need the Gerber files in Git (not in zips). The pages are still up but won't update any more. The options to fix it:

1. Unzip the zip files
2. Wait for kitspace/kitspace#216 and Kitspace will plot the Gerbers for you from the KiCad files

Which option would you prefer?

Replace PCB jumper with real one for BOOT0 input for easier selection of MCU bootloader when USB DFU is used for firmware upgrade.

Provide dedicated BOOT output for each BP3C slot that can be used for firmware programming of on-board MCU. That will require the following modifications:

• Expand 26-pin DIB periperal module connector to 28-pin to carry UART lines (RX, TX)
• Assign BOOT input on one of the existing GND pin
• Replace 4-pin I2C I/O expander with 8-bit version for dedicated BOOT outputs and programmable RESET output
• "OR-wire" I/O expander peripheral reset output with master RESET (with diodes)

Loudspeaker is wrongly connect. Swap polarity.

Currently used AR1021 is recently marked as "Not recommended for new designs". Therefore it has to be replaced with something else that comes in convenient package (not QFN, BGA, etc.), e.g. TSC2007 or TSC2003.

Add 470u elco over fan terminals to improve EMI. Change fan supply input filter capacitor from 10u to 1u and remove 100n.

Rotate X3 16-pin IDC connector for 180 deg and swap its position with TR1 if possible.

The L3763 power dissipation is considerable on 2-layer PCB with 48 Vdc applied on its Vin input. Reduce Vin by introducing simple Vout tracking with MOSFET that Vin is always about +7 V above Vout

Additional xtal is required for RTC otherwise it is going to stop counting while power is switched off and battery is used.

LSE xtal pins are already occupied with DOUT1 and DOUT2. Move that functions to other unassigned pins if available. If no pins are available unassign OUT_SER and OUT_PAR and introduce an I2C I/O expander on BP3C board.

Currently used audio amp and PCB loudspeaker is too costly and cannot offer audible enough sound. Therefore, move back to simple buzzer with driver as in case of H24005. For buzzer combine PCB holes for both 9 mm and 12 mm buzzer (e.g. AI-1223-TWT-5V-5-R).

I2C address keys on the BP3C is wrong and has to be fixed. Use PCB jumpers to offer more flexibility for defining module IC2 address.

Use +5V instead of +3V3as power input for LCD backlight LED boost converter.

• IC2 lines for MCU EEPROM (i.e. SDA and SCL) are wrongly connected.
• Replace with 2K SOIC8 (e.g. AT24C02D-SSHM-T)
• WP (Write protect) pin move from 3V3 to Gnd.

Split power and signal GND on the AUX PS board. Also trace dedicated GND for fan.

the following changes have to be implemented to improve switching between two ranges:

1. Connect D15 cathode between 0R010 current sense resistors
2. Add 100n capacitor in parallel with R59
3. Add hysteresis resistors (1 Meg) over IC12B and IC12C

All mentioned new added parts values have to be further optimized for as fast as possible switching (without introducing stability issues).

Use remaining IC12 input to protect VBUS line.

Replace X2 16-pin IDC connector from right angled to vertical and move it to the PCB's left edge.

Suggested by JBB: https://www.eevblog.com/forum/projects/eez-h25005-a-possible-successor-of-eez-h24005-programmable-power-supply/msg2854548/#msg2854548

Move BOOT0 switch from AUX-PS to MCU module

Replace existing digital I/O drivers/buffers with 74HC4050 (use 2 for input and 2 x 2 for outputs).

Add third relay that is also controlled via I2C I/O Expander to connect together Out- terminals of all three modules

Include additional relay that can under firmware control connect Ch#1 Out- and Ch#2 Out+ for making split rail output without external wiring.

Add PCB jumper (open) that MCU battery on the MCU board could be used to provide +VAUX for the DIB backplane.

Add ferrite beads on +5 V power input and USB power (IC9).

EEPROM VP (Write protect) pin move from 3V3 to Gnd.

https://github.com/eez-open/modular-psu/tree/master/mcu/KiCad are obsolete they use EEZ DIB MCU r1B2 when official latest version (Eagle files) is EEZ DIB MCU r1B6

It will be great to add details in the mcu/README.md that official design are done on Eagle to avoid confusion and mistake if someone want to use/work with latest validated design files.

A must for a fully open source hardware (especially certified oshwa https://certification.oshwa.org/hr000002.html) will be to have all latest design files converted (or even natively done on KiCad maybe for next iterations to completely avoid Eagle ...) to free/open source software and here especially the amazing KiCad

The current fan connector pinout is not in line with PC standard (power lines are swapped). Also use 4-pin connector (has to be Molex 47053-1000) that can accept both 3- and 4-wire fan connectors.

The length of the bus for the SDRam chip should have all the same length.
The bus operates at 143MHz, so the total trace length should be all equal.

Also, the length of each trace at each layer should have the same length!

Replace existing fan controller with one that has two tacho inputs (e.g. MAX31760).

OVP circuit as currently wired cannot provide 4 V or more as OVP_FAULT signal for the I/O expander. Therefore that signal has to be inverted with adding one MOSFET with optionally pull-up on its drain (if internal I/O expander pull-in is not going to be used).