Replacing 16MHz Crystal with MEMS Oscillator

Learn MoreFAQsTutorials

15 posts / 0 new
Last post
ahiggs
Offline
Last seen: 1 month 3 weeks ago
Joined: 2019-12-17 07:10
Replacing 16MHz Crystal with MEMS Oscillator

Hello,

I have an application using the DA14585 that is subject to extreme EMI. I believe that this is disturbing the high frequency oscillator and causing the microcontroller to not run or constantly reset in situations. Would it be possible to replace the 16MHz crystal with a MEMS oscillator to improve this? 

The data sheet only lists a minimum and typical external clock voltage of 1 and 1.2V, respectively, but no maximum voltage. Would this oscillator work

Thank you!

Keywords: 
Device: 
PM_Dialog
Offline
Last seen: 27 min 23 sec ago
Staff
Joined: 2018-02-08 11:03
Hi ahiggs,

Hi ahiggs,

Thanks for your question and for your interest in our BLE solutions. Let me check it and I’ll get back to you.

Thanks, PM_Dialog

PM_Dialog
Offline
Last seen: 27 min 23 sec ago
Staff
Joined: 2018-02-08 11:03
Hi ahiggs,

Hi ahiggs,

Although there are a some drawbacks using an external oscillator (e.g. power consumption, phase noise, cost, ..), the DA14585 can handle an external 16MHz clock. The output voltage of the oscillator should be divided to 1.2V.

On the other hand, if EMI is the issue here, wouldn't some kind of shielding do a better job?

Thanks, PM_Dialog

ahiggs
Offline
Last seen: 1 month 3 weeks ago
Joined: 2019-12-17 07:10
Hi, thanks for the response.

Hi, thanks for the response. We are already using PCB mount shielding, but the environment is harsh enough for this to not be sufficient. I can upload an image of the PCB where you can see the mounts for the shielding and general layout if anything looks off if this would help.

 

 

PM_Dialog
Offline
Last seen: 27 min 23 sec ago
Staff
Joined: 2018-02-08 11:03
Hi ahiggs,

Hi ahiggs,

Yes, please upload an image of your PCB. It would be helpful.

Thanks, PM_Dialog

ahiggs
Offline
Last seen: 1 month 3 weeks ago
Joined: 2019-12-17 07:10
Hi,

Hi,

Here is a PDF of the layout. The device is powered by a small 3.6V rechargable battery for measuring a thermocouple in a high voltage environment with extreme EMI.

Attachment: 
PM_Dialog
Offline
Last seen: 27 min 23 sec ago
Staff
Joined: 2018-02-08 11:03
Hi ahiggs,

Hi ahiggs,

We have the layout reviewed and the grounding of the xtal is looking unusual. We combine the ground ring around the xtal with the top layer ground. So we don't have the space marked with the green arrows – see attached.

Thanks, PM_Dialog

Attachment: 
ahiggs
Offline
Last seen: 1 month 3 weeks ago
Joined: 2019-12-17 07:10
Hello,

Hello,

I have tested boards with both ground designs as I have encountered advice both for and against the grounding ring. We did not notice a difference in performance in the harsh environment between the two. 

Do you have suggestions of what to do with unused GPIO and LFxtal pins? 

 

Thanks,

Ansel

PM_Dialog
Offline
Last seen: 27 min 23 sec ago
Staff
Joined: 2018-02-08 11:03
Hi ahiggs,

Hi ahiggs,

Could you please answer on our questions below?

  1. How did you proof that the 16MHz oscillator is suffering from the EMI environment?
  2. Which xtal (datasheet) is used?
  3. Why is there no EMI protection on the trace to the antenna?
  4. No antenna matching is needed?!
  5. Does the board work in a non-EMI environment? How is possible to be tested it this without a UART?
  6. What EMI frequencies and field strength are we talking about?

The default setting for all GPIOs is input pull down. Only P0_0 is used in this application. There are no traces at the other GPIOs or the LF xtal. We would say it is quite unlikely, that those pins contribute to the issue.

Is this your final layout? If not, then please upload your final layout with the exact information when it is ready.

Thanks, PM_Dialog

ahiggs
Offline
Last seen: 1 month 3 weeks ago
Joined: 2019-12-17 07:10
1. I do not know for certain

1. I do not know for certain that the oscillator is causing issues. The code running on the 14585 is based off the DSPS example and periodically samples an ADC (via the easy_timer) and notifies a receiver with the value, it has been well tested on this project and others and has never encountered issues. (answered further in 5.)

2. This crystal. 16MHz, 18pF, 80ohm, 30ppm. Has been used on other 14585 designs without issue (standard industrial environment)

3. That the shielding does not cover the trace to the antenna? I did not think that this was necessary as it is closely protected by the ground plane. 

4. This desgn is used for laboratory testing with close proximity of transmitter and receiver and does not require exceptional performance to warrant the cost of impedance control or measurement.

5. This board works in a non-EMI environment and with moderate EMI present. The amplifier connected to P0_0 measures a differential thermocouple signal with a rapidly changing common-mode voltage. When the CMV changes between 0 and 10kV, the 14585 functions as expected and is connectable and outputs correct data. If the CMV changes above ~10kV the 14585 will either stop advertising, advertise but not send any notifications, or advertise but immediately disconnect-reconnect-disconnect-etc from the receiver until the EMI has cleared. The CMV increases and decreases rapidly, rate of multiple kV per microsecond, and occurs at about 15Hz.

6. I do not know the exact frequency or strength. I believe it is well in the sub-GHz range but we do not have the equipment to measure it.

This is the most recent layout. As this is not a design for series production, we are able to relatively quickly make changes and test new layouts.

 

Thanks!

Ansel

 

PM_Dialog
Offline
Last seen: 27 min 23 sec ago
Staff
Joined: 2018-02-08 11:03
Hi ahiggs,

Hi ahiggs,

The 16MHz xtal does not fit to our specification.

The load capacitance should be 10pF and the 80Ohm ESR is only valid for xtals with a frequency between 20MHz and 29.9MHz. The 16MHz xtal has an ESR of 150Ohm, which is too high for our oscillator.

Before trying the MEMS oscillator, changing to an xtal, that fulfills our specification would be an option.

Were you able to trim the 18pF xtal to 16MHz?

Please also check the AN-B-054: DA14585/586 Application Hardware Design Guidelines.

Thanks, PM_Dialog

ahiggs
Offline
Last seen: 1 month 3 weeks ago
Joined: 2019-12-17 07:10
Ok, wow, I completely

Ok, wow, I completely overlooked that and the incorrect resistance value on Digikey also didn't help. This crystal appears to be more suitable. I will look at changing it when we next revise the boards.

I did not attempt to trim the crystal as the 14585 worked fine with it in previous designs. 

Thanks!

PM_Dialog
Offline
Last seen: 27 min 23 sec ago
Staff
Joined: 2018-02-08 11:03
Hi ahiggs,

Hi ahiggs,

I’ve checked the datasheet and the load capacitance is 8pF. This should be 10pF.

Thanks, PM_Dialog

ahiggs
Offline
Last seen: 1 month 3 weeks ago
Joined: 2019-12-17 07:10
Hello,

Hello,

Unless I am mistaken, the datasheet says the part number ECS-160-10-36Q-AES-TR indicates the capacitance which should be 10pF for this version. Otherwise the resistance, shunt capacitance, and tolerance are all within the reccomended range?

Thanks!

PM_Dialog
Offline
Last seen: 27 min 23 sec ago
Staff
Joined: 2018-02-08 11:03
Hi ahiggs,

Hi ahiggs,

My apologies -  yes the load capacitance is coded in the part number. A ECS-160-10-36Q xtal has a load capacitance of 10pF indeed.

The frequency tolerance is little big, but if you are not using it at the full temperature range and you will trimm the 16MHz oscillator. In this case the xtal will fit.

Thanks, PM_Dialog