MSPM0G1105: Regarding the load capacity of HFXT

Hi Hiroshi，

Correct me if my understanding to your question is wrong, if you want to choose the right capacitors on both sides of HFXT, you can refer to the following calculation formula.

Where CL is the load capacitance of the crystal, which you can find on the datasheet of the crystal you choose. Cs is the stray capacitance of the PCB, which is related to the layout of your board, the general empirical value is 3-5pF. Cg and Cd are the two external capacitors on both terminals, generally, we choose Cg and Cd to be the same.

Best Regards,
Peter

Hi Peter,

Thank you for your prompt comment.

I apologize for the poor way I asked the question.

I understand the above formula.

Simply put, just like the above formula, what is the CL value required by the MSPM0G1105?

I don't think it's 1pF.

Am I wrong?

Best regards,

Hiroshi Yamada

Hi Hiroshi,

I believe CL is the total load capacitance required by the crystal, not MCU, like it is 16pF when you choose to use MURATA’s “CSTNE8M00G550000R0”. And Integrated effective load capacitance(Cl,eff​) would be included in the Cs which is the stray capacitance may also include other parasitic capacitance like PCB trace capacitance or some, I think normally we choose Cs as 3-5 pF to calculate in the formula because we also take that Cl,eff​ into consideration.

Best Regards,
Peter

Hi Peter,

Thank you for your quick comment.

I have been in charge of the design so far, but the load capacitance (CL) of the crystal unit and the connection side (MSPM0G1105) needs to match.

I think it is necessary to adjust the load on the crystal unit side with a capacitor so that it matches the load capacitance required by the connection side, using the formula Peter-san showed us previously.

Am I wrong?

Best regards,

Hiroshi Yamada

Hi Hiroshi,

Yes, I think that is exactly what the formula want to tell us. It indicates that how to combine the board parasitic capacitance and add external start-up capacitors on both side of the oscillator to achieve the capacitance matching so that complete normal oscillation.

Like the schematic we provide in the LP-MSPM0G3507_UserGuide, the C12 and C10 which is corresponding to the Cg and Cd in the formula, it's better to calculate their value based on the formula at first and have a test to make sure the crystal can stably oscillate.

Best Regards,
Peter

Hi Peter,

Thank you for your prompt comment.

I understand what you are saying.

Please tell me the load capacitance value (CL) of the MSPM0G1105.

Again, this is generally written on the device you connect, right?

Best regards,

Hiroshi Yamada

Hi Hiroshi,

Sorry for misunderstanding you, forgive me but I had never heard of the MCU's load capacitor before, and all relevant specifications are listed on the datasheet, if not, means it is not currently available so far. Below I have attached the input capacitance of the IO port, is it the load capacitance that you are concerned about? But even so I do not think that the input capacitance of the normal Digital IO pin is the same as the specific crystal input pin, maybe still need to refer to the Integrated effective load capacitance(Cl,eff​) 1pF according to the datasheet.

Best Regards,
Peter

Hi Peter,

Thank you for your quick comment.

Even if the AM5708 is used in another project, you can specify the load capacity of the crystal resonator to be connected.

I think we'll try to connect an external capacitor to the crystal resonator to match this capacity.

For final matching, IC matching test is performed by the quartz resonator manufacturer.

This is a story that is carried out and determined the exact load capacity.

Best regards,

Hiroshi Yamada

Hi Hiroshi,

Yes, thanks for providing the Info, and I think that is exactly what I mean as I mentioned previously, it is the capacitor for Crystal which mostly determined by the Crystal's load capacitor, not for MCU, it has the same meaning as Cg and Cd in the previous formula. and I think it just give a rough range according to the datasheet, and it is also necessary to select the appropriate capacitance through experiments, because it may still be affected by the parasitic capacitance of the board, input capacitor of the pin, etc.

Best Regards,
Peter

Hi Peter,

Thank you for your comments many times.

I'm not sure if my thinking is wrong or if I'm not getting my point across in English, but it seems we're not on the same page.

What I'm trying to say is that the load required for the crystal oscillator (Cf1 and Cf2) is 12-24pF.

The data sheet for the crystal oscillator also lists the load for proper operation.

In short, you need to select a crystal oscillator that operates properly at 12-24pF. For example, if you connect an appropriate crystal oscillator at 8pF, I think you'll run into problems such as no oscillation.

Conversely, I think you'll run into the same problems with a load of about 8pF, which is the crystal oscillator's specification.

In that case, I think 12-24pF is the required specification for the connected device (in this example, the AM5708).

Am I wrong?

Sorry for asking so many times.

Best regards,

Hiroshi Yamada

Hi Hiroshi,

Yes, if there are strict requirements(Cf1 and Cf2 is 12-24pF) for AM5708 due to some special structure, it indicate that the crystal's load capacitor should be limited within 6pF-12pF due to the formula, it needs to be strictly follow the datasheet and choose the crystal carefully. But since the M0 does not have any info mentioned in the datasheet for this, I believe it is sufficient to just meet the formula and fine-tune the capacitance value according to the actual situation to achieve the best oscillation effect.

Best Regards,
Peter

Hi Peter,

Thank you for your multiple comments.

So M0 is a rough standard?

I originally posted this to investigate the cause of the low amplitude of the crystal oscillator I found in my prototype.

So the cause is unknown.

For reference, how many volts of amplitude is required when connecting a crystal oscillator to HFXT?

In my prototype it is about 400mV.

Best regards,

Hiroshi Yamada

Hi Hiroshi,

What is the value of the capacitance at both side of the crystal oscillator you use, or you can directly provide the schematic of this crystal part for me.

I also would like to ask that since the amplitude was not your expected value, does it have any impact on the system, for example, does any problem occur in your normal application that makes you suspect that it is a problem with the crystal oscillator?

And I will test the amplitude of the crystal from my side and will update to you later.

Best Regards,
Peter

Hi Peter,

Thank you for your repeated comments.

I have attached the circuit diagram.

The URL for the Murata Manufacturing ceramic resonator I am using is below.

https://www.murata.com/products/productdata/8801161740318/SPEC-CSTNE8M00G550000R0.pdf?1559743212000

The MSPM0G1105 has frozen because the amplitude is small.

Thank you for your cooperation.

Best regards,

Hiroshi Yamada

Hi Hiroshi,

I have a test in the launchpad from my side, and amplitude between HFXOUT and GND is ~410mV, all functions perform normal. So I think if your prototype is about 400mV, it should be reasonable.

Additionally, what if we try disconnecting this resistor as shown below and test again to check whether the MSPM0G1105 has frozen? and by the way, what does frozen mean? Does it mean that the MCU doesn't perform the function as you expected, like toggle a led or some, or any other strange phenomenon?

Moreover, what are the value of these two capacitor as shown below?

Best Regards,
Peter

Hi Peter,

Thank you for your comments.

Thank you for checking.

The resistor in the red frame is not mounted. (N.M)=Not Mount

Regarding the freezing, this issue is in a debug environment with a debug probe XSD110 connected, but the XDS110 is getting an error saying that it can't find the MSPM0G1105? Can't connect?

It's freezing because the MSPM0G1105 doesn't work at all.

The other question is about the capacitor in Murata's ceramic resonator. It refers to the part in the data sheet I contacted you about earlier, "Built-in L or dCapacitance (C1, C2 at 1MHz): 33pF ±20%/max."

Best regards,

Hiroshi Yamada

Hi Hiroshi,

If my understand is correct, this is the maximum value by the datasheet, have you ever tried decreased the cap value and do some test? Can the problem be solved? And it's best to also measure the waveform of the crystal to check whether it is normal.

In addition, are there any screenshots of the corresponding debug error information? 

Best Regards,
Peter

Hi Peter,

Thanks for your reply.

Murata Manufacturing has built-in ceramic resonator loads (C1, C2), so changing them would require changing components, so I haven't done so.

I previously inquired about the crystal used in an evaluation board using the MSPM3507 and was told that it uses the "Seiko Epson TSX-3225."

It has a 40MHz, CL=12pF specification.

It was difficult to obtain, so I considered using the current Murata ceramic resonator and inquired on the forum, which seemed to suggest there would be no problem, so I used it.

In retrospect, C1=C2=33pF is probably a little large since CL=16.5pF.

I don't have a screenshot. I can set up the environment and check, but it won't be possible in the near future.

Best regards,

Hiroshi Yamada

Hi Hiroshi,

hiroshi.yamada said:

In retrospect, C1=C2=33pF is probably a little large since CL=16.5pF.

Yes, I do think so, perhaps the capacitance value can be reduced to around 28pF to have a test.

Best Regards,
Peter

Hi Peter,

Thank you for your comment.

It will be difficult to confirm the changes in the near future, so I will consider it later.

Peter-san commented that it would be reasonable if the prototype's amplitude was around 400mV.

If so, is there a possibility that there are other factors at play? If so, please let us know what they are.

Best regards,

Hiroshi Yamada

Hi Hiroshi,

You can do a comparative experiment, use the internal SYSOSC crystal instead of the external crystal to see if all functions are normal, so as to further determine whether it is just the external crystal oscillator problem.

In addition to hardware setting, there may be software issues. We provide a related HFXT code project in the SDK(Your MSPM0_SDK Download Folder\mspm0_sdk_xxxx\examples\nortos\LP_MSPM0G3507\driverlib) and you can have a test.

Best Regards,
Peter

 Hi Peter,

Thank you for your comment.

I have already switched to the internal clock and checked it.

The results were OK.

However, there is one unit that cannot connect to the XDS110 and I cannot switch.

Therefore, I think the problem is related to the external oscillator.

I will need to check with the person in charge about the software.

I have one question.

I think you said that the external oscillator amplitude of around 400mV was not abnormal, but is that correct?

In that case, I think it might be a problem with the software settings (registers?).

Best regards,

Hiroshi Yamada

Hi Hiroshi, 

hiroshi.yamada said:

I think you said that the external oscillator amplitude of around 400mV was not abnormal, but is that correct?

I tested different Launchpads boards again, the amplitudes are all around 400mV and the functions are normal.

hiroshi.yamada said:

In that case, I think it might be a problem with the software settings (registers?).

Maybe can check the clock tree setting part if you configure the clock by Sysconfig.

Best Regards,
Peter