Using a Xtal with 20pF load capacitance

Tip / Sign in to post questions, reply, level up, and achieve exciting badges. Know more

cross mob
AmSa_4551891
Level 2
Level 2
5 replies posted 5 questions asked First reply posted

Hi ,

I am using the cy8c4128lqi-bl543 and the cy8c4128lqi-bl553 product. We have placed an Xtal of 24MHz with  20pF  load capacitance (Cypress requires 8pF) in  ~1000 PCBA.

Can we use the tuning suggested in the https://www.cypress.com/file/139476/download    (Xtal tuning reference design) for tuning of that load to the correct one?

0 Likes
1 Solution
PradiptaB_11
Moderator
Moderator
Moderator
500 replies posted 250 solutions authored 250 replies posted

Hi,

The oscillator of PSoC4 BLE devices is designed to work with an 8-pF load capacitance, which requires that the chosen crystal have an 8-pF load capacitance. A high capacitive load will result in a lower oscillation frequency. This frequency offset will be directly translated to an offset in the RF carrier frequency and symbol timing accuracy of the device.

It is highly recommended that the load capacitance of the selected crystal is 8 pF. 20 pF is a very high value and tuning may not work in this case. If the change was say around 2pF meaning load cap was 10 pF then tuning would have worked.

Thanks,

Pradipta.

View solution in original post

0 Likes
5 Replies
PradiptaB_11
Moderator
Moderator
Moderator
500 replies posted 250 solutions authored 250 replies posted

Hi,

The oscillator of PSoC4 BLE devices is designed to work with an 8-pF load capacitance, which requires that the chosen crystal have an 8-pF load capacitance. A high capacitive load will result in a lower oscillation frequency. This frequency offset will be directly translated to an offset in the RF carrier frequency and symbol timing accuracy of the device.

It is highly recommended that the load capacitance of the selected crystal is 8 pF. 20 pF is a very high value and tuning may not work in this case. If the change was say around 2pF meaning load cap was 10 pF then tuning would have worked.

Thanks,

Pradipta.

0 Likes

Hi,

Thanks for the answer. Assuming that we try and modify the frequency according to the document:

https://www.cypress.com/documentation/application-notes/an95089-psoc-4proc-ble-crystal-oscillator-se...

Do you see an issues? or you think it is not feasible to reach this range of tuning with the current layout ?

Thanks in advance.

0 Likes

Hi,

The crystal you are using is designed to provide 24 MHz oscillations with 0 ppm error when 20 pF load Capacitance is applied. If you change the CL value you will not observe 24 MHz value, there will be error in the output frequency. Using the tuning strategy you can tune C1 and C2 (PSoC 4 device internal tuning Caps) between 3.7 pF to 16.428 pF meaning you can tune the CL upto 8.214 pF(16.428/2) without adding external Caps.

To operate the crystal properly you will need to add external caps to amke the CL 20 pF. PSoC driver will not be able to guarantee safe and reliable operation at such high capacitive load.

So just by using the tuning the internal caps as mentioned in the app note what you are trying to achieve is not feasible.

Thanks,

Pradipta.

0 Likes

Hi Pradipta ,

Thanks for info. we are trying to tune it only with PSOC SW, we don't have provision for caps on our PCB. I have some comment/questions about what you have replied:

  • Q: We are using PSOC 4.2 for tuning. the values there varies between 7.5 pF to 24.5 pF, and not from 3.7pF to 16.428pF - can you explain that? 
  • We do use 20pF load Xtal , but it has a very low ppm (10), and therefore we some "spare" for meeting the SIG requirements (50 ppm). when we use the default values of the PSOC4.2 version caps (15.2 pF and 17 pF) , we are getting 54ppm which is not enough.
  • Then we tried to raise the Xi and Xo caps, so for 24.5 and 24.5 pF we go only 17.5 ppm, which is very good, BUT- we see that we have issues with the connectivity (advertise is OK , but we have problems with the connection initiation with the unit), and we do see that uP sometimes halts (and eventually Watchdog reset the unit) when it needs  to start transmit.
  • We have reduced the values for now to Xi=21pF and Xo=23 which looks OK from ppm perspective (35 ppm)  and seems for now that it doesn't halts (we check with 10 different units).  but we need to produce 1000 unit using this Xtal (due to Covid-19 we cannot get a lower load +replace one now...), for a product that needs to last for 2.5 years.
  • My question is (other than the question in above )  :
    • is there anything else which we need to look for other than the going below 50 ppm
    • is the 24.5pf/24.5pf issue I see is related to this sentence you wrote:

pastedImage_0.png

Appreciate your help

Amir

0 Likes

Hi Amir,

Kindly find my comments below.

Q) We are using PSOC 4.2 for tuning. the values there varies between 7.5 pF to 24.5 pF, and not from 3.7pF to 16.428pF - can you explain that?

Ans) I took these values from the app note only from section 4.1.2. These are general values and there can be variations within the PSoC versions and family. The only thing to note here will be that the load capacitance value should be close to 8pF for the on chip ECO circuit to operate correctly.

Q)Is there anything else which we need to look for other than the going below 50 ppm

Ans) As specified in the app note section 7 you need to check all the parameters listed there.

Q)Is the 24.5pf/24.5pf issue I see is related to this sentence you wrote

Ans) It may be the case, but i cannot confirm on this until i observe the issue. I am assuming you will be using CYBLE_API_RESULT_T CyBle_GapcConnectDevice (const CYBLE_GAP_BD_ADDR_T *address) to initiate the connection. Can you once check the  Return value indicates if the function succeeded or failed.

I would suggest you to refer to section 3, 6 and 7 of the app note. There can be a lot of factors like temperature, load capacitance, drive level etc that can cause you to exceed the 50 ppm limit. The ECO frequency should be within 20 ppm for reliable operation. The product will last 2.5 years as you expect but you might experience poor RF performance.

Thanks,

Pradipta.

0 Likes