Is it possible to use M.2 Hat plate without 40 connector?

Hello! Few days ago I asked, how to connect to Raspberry more external devices, if all 40 GPIO pins are occupied by connector under M.2 HAT plate. Mr. KK informed me that Hailo module does not use GPIO at all and advised to connect directly to RPi GPIO pins. Now I have disassembled HAT plate from connector and removed connector. But I am not sure, that HAT plate will work without connector. Please, advise, what to do. Thanks.

If you want to strictly follow the HAT+ specification you should connect the GPIO header. There is an EEPROM on the HAT to identify the hardware. This way Raspberry OS could change the device tree or set some settings automatically. There is also an additional power connection.

Raspberry Pi - HAT+ Specification

Did you try?

Not yet. Because I am afraid to spoil the M2.HAT plate, it costs much money as for me. Are you sure, that this plate does not use connectors’ pins? I could not find information or scheme of connection of M2.HAT to Rpi. . I really cannot understand, why the holes for 40 pins are situated on the HAT plate. Please, confirm once again about safe usage of M2.HAT without connector. Thank you.

If you want to be safe, you should use the GPIO header.

Hi Guys, The GPIOs are used also to read the HAT+ eeprom to detect which HAT is connected (Using I2C) and also to supply extra power and GND connections. You don’t want to count on the PCIe flat ribbon to power your HAT…

In the AI Kit (m.2 HAT) the I2C is not a must however in the AI HAT the I2C is used to detect it is the AI HAT and set the PCIe to Gen3 automatically. If not connected the PCIe will use gen2 as default resulting in very poor performance. (In the AI kit you should configure it manually).

So to sum it up it can work w/o the GPIO header but you shouldn’t do it… :wink:

Hello, Giladn! Thank you so much for your explanation. OK, I understand, that 40-pins connector is useful and even necessary for M2.HAT plate. But I cannot understand, why so clever guys from Hailo occupied all 40 pins of RPi and did not leave any free pins for connection of some additional units. It looks, that I have bought expensive HAT+ AI with Hailo 8 in vain… I hope, that HAT designers will explain, how people could get to some pins of Rpi with installed HAT module. For example, I need an access to serial port and couple digital ports for my idea.

Hi, You can definitely use all the pins.
I understand that the GPIO header supplied by Raspberry Pi in the kit is physically blocking access to the pins but this is just a physical limitation. The HAT is a Raspberry Pi product and designed according to RPi HAT+ specification.
It is RPi decision which GPIO header is included.

So what can you do?
You can buy an extra long GPIO header or something like Raspberry Pi 40 Pin GPIO Extension Header

Quick and dirty solution:
Remove the standoffs in the GPIO header side and push the HAT a bit more to gain access to the pins.

Note
The I2C pins used as part of the HAT+ standard should be used with more caution as they are used on boot to read the EEPROM, check Raspberry Pi documentation about that.

Hi, Giladn! Thank you very much for your answer. Now everything is clear, no problems. All the best!