Electronics > Projects, Designs, and Technical Stuff

FLIR Lepton board

(1/2) > >>

LukeW:
The attached schematic shows a FLIR Lepton 2.5 sensor, in the appropriate SMD Molex socket, connected to a Raspberry Pi.
Both I2C and SPI are used, along with the clock, and appropriate voltage rails.

I have a PureEngineering breakout board, and this works fine - same Lepton sensor, same Raspberry Pi, same pinout, same OS, same software, same 5V power supply which powers the downstream regulators and the RP. It's the same thing basically. I essentially want to start by replicating a minimum working setup on my board, but my version doesn't work.

The PureEngineering board runs the I/O ring at +2.8V but I think this is a bit dodgy when we are connecting it to a Raspberry Pi which expects 3.3V levels, and 3.1V (without additional level translation) is still a bit of a dodgy hack but it's a little less dodgy. Well, supposedly they are using a 2.8V regulator but upon actually measuring the rail voltage on the board, it's actually closer to 3.3 volts. (The other separate 2.8V regulator that provides the 2.8V core voltage on the PureEngineering board is about 2.8V as expected.)

The green scope traces (attached) are from the working PureEngineering board, showing the 3 voltage rails (2.8V, 2.8V and 1.2V) and the 25MHz clock.

The purple traces are from my board, showing the 2.8V, 1.2V, 3.1V rails and the clock.

This is the schematic for the commercial, working board.
https://files.groupgets.com/lepton/LeptonAdaptorSchematic.PDF

The 3.1V rail is supplied from a TLV70231DBVR which can supply up to 300mA.
I've run the I/O ring at 3.1V which is the maximum officially supported by the Lepton.

The MCP1700T-1202E/TT that supplies the +1.2V rail can supply up to 200mA and the MCP1700T-2802E/TT that supplies the +2.8V rail can supply up to 250mA.

I've tried removing the C45 1uF from the PWRDWN rail (it shouldn't really be there) and I added the 1uF capacitor to ground that is supposed to be there on the reset line, in place of the reset button.

The 25 MHz clock is supplied by a ASFLMB-25.000MHZ-LC-T MEMS oscillator. This appears to be supplied fine to the correct pin in the Lepton socket. Doesn't look very square, but this could be a bandwidth limitation on the 100MHz scope.


Any ideas or suggestions as to what's failing here? I'd appreciate a fresh set of eyes on it. Cheers.

LukeW:
Attachment: 25 MHz clock on the PureEngineering board (green traces) and on my board (purple trace)

LukeW:
Attachment: Voltage rail comparisons

frogg:

--- Quote from: LukeW on May 15, 2019, 11:10:45 am ---The PureEngineering board runs the I/O ring at +2.8V but I think this is a bit dodgy when we are connecting it to a Raspberry Pi which expects 3.3V levels, and 3.1V (without additional level translation) is still a bit of a dodgy hack but it's a little less dodgy.
--- End quote ---

+2.8V is well within the limits of driving 3.3V logic and so should be suffiicient.

Level shifters would probably be less dodgy than breaking / adding caps :)

Could you also be more clear about what is "failing" here? I can't find anything in your post about what's actually failing.

LukeW:
I haven’t sniffed the SPI/I2C transactions in detail, but the example video capture code which talks to the camera fails and just gives a red box. (Works perfectly on the other board.) And i2cdetect fails to see the I2C device on the bus, but it does on the other board.

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod