Trouble with CR95HF and crystal oscillator not starting

[lmarckx]

Hello, 

I am working on implementing the CR95HF RFID reader into a project, and I'm having some trouble with the external crystal starting; I'm wondering if anyone has had a similar issue. 

On my design I have the CR95HF connected to an NRF52840 MCU through UART. When I send the ECHO command (0x55) over the UART bus, I do not get any response from the chip. I have verified the following:

• The chip is properly receiving 3.3V at all of its power supply pins.
• The SSI_0 and SSI_1 pins are properly tied to ground to select UART communication
• Baudrate for UART is set to 57600, the default for this chip
• I pulse the UART_RX/IRQ_IN pin low for ~150us on startup to properly select the protocol

On an oscilloscope, I can see that the XOUT pin that goes to the crystal rises to just above 0.5V on powerup, but the crystal never begins oscillation (see attached image). I am using the XRCGB27M120F3M00R0 from Murata, which is 27.12MHz with ~150 ohm ESR and 6pF load capacitance. I will provide the schematic for the entire circuit for reference.

The especially troubling part about this is that we have used this exact circuit and layout on different boards for different projects, and the CR95HF has worked just fine in those cases. It's just now on this new board that we can't get the crystal to start oscillating.

Any help would be appreciated, and if necessary I can provide more information (more scope traces, layout, etc.)

Thank you!

[Brian TIDAL]

Hi,

During the startup sequence of the device at power up, the HF oscillator (HFO) normally starts after the initial low pulse on IRQ_IN for a setup duration of t3.

• could you check on a second board: do you face the same issue?
• could you check whether the crystal has been soldered in the proper position (i.e. not swapped)?
• could you check whether the crystal has been properly soldered ?
• could you replace the crystal with a functional known one?
• could you probe the crystal on another project board to have a comparison?

Rgds

BT

[lmarckx]

Hi Brian,

Thanks for your response. To address your questions:

• I do not have a second board to test with just yet. I have confirmed via PCB x-ray that all the traces between the CR95HF and the oscillator look good and are going to the right pins, etc.
• The crystal has been soldered in the correct orientation; this specific part is symmetrical, and I have confirmed the pinout to the crystal is correct. I did try rotating the crystal 180 degrees just as a check and I get the same behavior
• The crystal is properly soldered to the board and making good contact with the pads
• I swapped the crystals between this board and another one from one of the other working projects. The crystal from this board that is not starting did work properly on the other project's board, and the known working crystal from the other project still failed to start on this board.
• I will attach some scope traces to show the startup behavior and functional oscillation on the working project to compare with the non-working project. The exact procedure I followed is: I apply power to the board (5V, which gets converted to 3.3V through a linear voltage regulator which then powers the CR95HF), then I manually short the IRW_IN pin to GND to simulate the initial low pulse to start the oscillator. On the working project, you'll see that as soon as the IRQ_IN low pulse is received, the excitation voltage spikes up to ~1V, then settles to ~0.6V, and then a few miliseconds later the oscillation begins. 

 startup on the working project's board

 transient startup response on working board

 oscillation on working board

 startup on not-working project's board with the same crystal from the working project.

 

 

[lmarckx]

Update: I also tried swapping the CR95HF chips between the two boards, and this did not make any difference. The working project still works and the non-working project still does not start the crystal, despite the XOUT voltage being present.

[Brian TIDAL]

Hi,

On the last scope trace (non-working board), the time constant appears to be too large. Could you check the C55 and C56 capacitors for the correct 6 pF value, proper soldering, and other potential issues?

Regards,
BT

[lmarckx · Accepted solution]

Thank you everyone for the attention on this. Unfortunately the solution was not very interesting (and rather frustrating) -- the footprint of the crystal was made backwards, so the pinout was swapped. Schematic is good, but the layout was wrong. So, what I was seeing was just the excitation voltage passing straight through to the other pad because it was going through the ground plane of the crystal...

Thanks again for the help, I'm closing going to close this.

[Brian TIDAL]

Thanks for the update. Please note that this device is Not Recommended for New Design (NRND status).