2022 09 14 SMO1 EMC
To find the root cause or
Mostly, how many more more components on the board, but it seems a bit fruitless after the.
So, I went to change the method and made the new feed proper analysis.
And this is shown on the next slide. Okay?
So this is the upside and here, we see, of course, around the quads area and a bit at the supply of the core voltage.
In with the numbers, the numbers are on the values and the spectrum analyzer.
So it's always reflecting digit of or different of one. Here means 10 dB difference, okay?
In the spectrum as it's relative because it depends on the pre-amplifier.
So this means we have one emission here that was known before and with a let's say with a larger magnetic field probe we find a high emission
around the oscillator area, but for this method,
I used nearby probe with a very narrow window so I can have resolution of maybe two millimeters to to. Yeah, pinpoint.
The mission more precisely and interesting is that in the lower area, the bot, we have a much higher emission dialogues.
So around this area.
And I also see a very high emission on the bottom side.
Have you traced here in this area is where you have 3.5 and 4. So basically a roughly 40 decibels per minute, have it traced back to see what signals signals.
We have to have in any high frequency, signals, switching noise.
That's usually not no switching at all. We have this lower left area. This part of our exercise analog part.
So we have ADCs and ADC supply, the parents voltage and supply. And this whole area is the ADC area.
The switching part is here, this is the supply of the car voltage, but especially in this area, there's no main peak of 100 megahertz.
So we see a bit here.
It's difficult to pinpoint on which capacitors and the seat here.
I did not yet measure by oscilloscope, with FFT to see which network it would be. It could be a next step.
This one question, the supply of this micron controller is done over the linear or with a. Yeah.
Buck converter. We have one computer here for the for the call voltage and so we will be interesting to make any 15 on the output of the, of the spark controller. Yeah, we can do that here. Yep. Across the survival district.
3.3 now the input supplies five volts and up from the payment and the output is 1.3 now I wanted three.
Okay. Yeah, I think it would definitely make sense, right?
To especially the supply voltage 1.3 volts, make an F50 to see what, what how the spectrum looks like.
And what's interesting for me, if you say on the right side we have the oscillator.
Yeah, let's say that's the supply lines are quite close to the oscillator area.
I don't know if we have any coupling effects due to layout or due to trace routing and so on.
But I think we should wait for the A50, analyzes to see what how the destroyer spectrum looks like. And then we can think about how we can implement some measures by battery working up.
Yeah, we have, especially one.
One dot three volt supply line towards oscillator just so it's here to the micro because the oscillator needs a 5 volts and a 1.3 volt supply.
Did you did you check all all these supplies?
Like the analog supplies, the digital supplies in the, in the frequency range already.
No, not yet. This is exactly one of the next steps to do. I don't think I have it next page. Just next steps.
I don't think I have even written it down, but it's quite logical to check the supply lines that this, this topic I put in, but yeah, not yeah.
So if it take a look on this, or to be want to discuss a bit further here on this.
That's just one question of from my side on the right side. We can see these four capacitors.
I think Valentine when the layout changed in C1. Exactly. At this point. Yes we changed. So something. Here we added additional capacitances.
Here to provide me a buffering here for the core, the power supply. So what I see here on the slide is 0 plus solution.
D robust it is, yeah, it is zero plus division D. So in the C1, you're right, coming up, there are some further capacitors on the bottom side, for the core voltage, but we still need to measure this sample.
We have the first samples available so this can be done.
Maybe even imperiled to the KV test of FFT, once not only C0 plus, but he also see one sample, okay?
But do you expect with this layout change? Also better, you see results or is it just layer change? Not referring to EMC.
I'm careful with forecast and it was doing improve it. This was the intention of the design change, but we have learned before not just in this project that the wishes do not always come true.
Yeah, right. So but you're right. We have already changed in this area for the C1 so it will be very interesting to compare the samples. It is difficult.
We need to device this set up because this is bottom side of the logic and I it's difficult to access it because you need to direct access if you want to measure, 100, megahertz, so, so because the PCBs are stacked, it's not accessible by a probe.
PW, we find a way to put some wires in between the PCBs.
Otherwise, it's yeah, not good to measure.
Okay, so so topside is easy that. So this will be a next step.
We have different five volts. Especially here.
Across the 3033. Yep. And sorry, Robert. Let me please correct. The value. We have the, on the output of one dot. 25 volt. Not one dot.
Three volt.
Just correction the supply voltage of the micronutrient mean. Yeah, I'm not 25 okay.
Yeah, this is but this is controlled by the micro itself within its a lot tolerances. So it has a feedback into measures the supply and adapts.
The buck new recycle. I'm here, but below here and the bottom layer of microcontroller, we have there can aligns, right?
Which I go out and go in the PCB now.
Layers.
Right for the can I have checked all the supply lines? But of course you're right.
We have two cans attached to the micro. I would need to check where we have the cans.
Exactly. But I would first focus on the supply lines and in a second step we can also check the can I've also made the near field probe close to the can transceivers and that's a possible emission path. But nearby probe signal is not that high.
So so it's so let's say one or megahertz can be detected 90% around the microcontroller so it's the source and the sure it's a source it has internal megahertz. Exactly.
I think that this peripheral croc, there's also backup clock into internal. It also 100 megahertz. Yeah, yeah.
The main internal system clock also 100 megahertz. And to so all the main clocks are at 100 megahertz. A question is, can we change this clock to another frequency? Because maybe this system have a resonance near to 100 megahertz.
And therefore we see this high peak, the missions, the questions that we have the possibility to change this clocking For example, into the software, I guess. Yeah for example, to shift it to 15 minutes or to 200 that is far away. I think from there is not point, It's just 50.
I wouldn't, I wouldn't choose because you will have the second harmony harmonica that, yes.
Yes.
Right.
75 or something like that. What would be interesting if there is a possibility to to change one of the internal clocks somehow just to see if how the effect is yeah. For analysis purposes? Yeah. Yeah. So please check this possibility.
Robot.
I, I guess to suffer critics will be extremely happy but how it's like thank you. And so this, I will need to add to the next steps because it's also.
So what I have in mind is also, if you poop on the C1, the influence of the power PCB, still to be checked that we can by cross swap of find out if there's any influence because the pole PCB has an emission path through clamped 40 and 41.
And then power supply, concept, defined tests.
We already talked about FFT and then of course it will also be interesting to see the pre-TV test results.
With the C1 should C1 here with this.
You mean have you had in the past this issue or already bear within the 3D videos?
So long it says, this needs to be prepared. We have seen in the, the TV test with this.
Is your, plus we have seen the one other megahertz and the previous tests with a B2.
We have not seen it at least not in that way. But the first test, where it occurred, mainly was a CP that was very clear, the dominating peak.
And I think we did not make CP will be two so we can only compare the test like strip line. That was not that visible before.
So it happened and that occurred the first time during the, the PV test campaign.
And maybe also because we have reduced the other emissions would be what billionaire interesting to know is, once again to recheck, all the changes, how the changes in the node changes, you've done. Yeah, which we might affect us, this issue, or lead to amplify amplifying of this issue.
And so, from between the between the, I don't be too simple and the current sample stage make a double check once again.
What exactly has has changed there?
Yeah.
I mean maybe you can try to isolate the failure even if it is in the layout or we on due to some more other schematic changes schematics.
We have already checked quite a lot into this, especially on the supply lands.
There was no real schematics change but of course, in the layout there are always some or several changes and sometimes they seem minor in my time effecting. So this is the yeah, we have already checked but your right.
This is where I put this here review. Power supply concept, exactly.
We need to double check if there is something. And yeah, some significant change towards the B2 from my point of view. No. Generally speaking I think it's totally makes sense to make a 50 analyzes on each power supply line.
Yeah, I think you have different power supplies within your schematics so check check all of them.
And what we're interesting to know, based on the results where you have to highest, resonance closer to 100 megahertz, and then maybe check this within the layout where you can get some, some coupling effect here.
If you have some, some coupling twice and onto some analogous signals or to some digital signals. But as far as I'm still from Karina the emissions here on the, on the kale, for this is the highest. Yeah.
And basically yeah this means this should definitely come from a power supply or from the power supply one of the parts of place which is connected to the kale 40, you know, or it happens to be a ground or if there is a couples here, the there are a lot of capacitors between care, 40 and care 41.
So it can couple back towards the carefully. Yeah, I think one of the yeah, well good good, good topic. Ground concept would be interesting to know how the ground concept looks like and the stood from you last time that we have any way.
The issue that we have not. We don't really have a ground for the kale. 40 after the 48. Volt supplies. We have just a. Yeah, b. Plus. Yeah.
And of course this is this isn't really the the nicest way to have the in dismiss that all the currents must be led back through the kill 31.
And I think one of the it's a bit delicious way. We have the kale 41 but we do not have the kale so yeah. Sorry, yeah.
This is what I wanted to say, thank you. Yeah. Before you have till 41, but you don't have the kill 31. Okay answer. But actually the supply or the power supplies. You have our own kale 30 in the feedback lines are or the.
Yeah. The current is going back through the kill for the one. Exactly.
Yeah, yeah. Okay. So this means basically the yeah, one of the interesting things to to double check would be also the ground concept both within the layout as well as. Yeah.
Within this schematic side.
I would start with the layout because if you say you've done some similar changes would be would be interesting to to see if the ground concept has been changed or moved.
I don't know the ground planes, or if you have additional ground traces, which have been added or moved somehow the best area.
The main ground concept on the logic. PCB has not changed to have one in a layer completely as a ground where for four layer board and the second layer from the top is completely ground.
And then, of course, on the other layers, we have various ground chaps, and because when we change routing, we will have to, we have to change the ground layers between there.
We have some changes that's for sure.
Okay. Now general question, can we get an update from your side maybe next week? Also, in a special hardware meeting regarding the F50 analyzes, you've done. I think, from my point of view, we should speed up a little bit.
The yeah the analyzes your the work here on this topic because we're getting closer and closer to the actual partition house.
And to be honest with you, we it looks like we haven't really founded the the root cause in so far we could not define any countermeasures.
So I would really like kind of ask you to speed up the activities here.
Okay, I think we can make an FFT first even before Nifty probe or cross swaps.
We became put their 50 at the first of first steps. Yeah. So we can change the order, make an F50 in the first and then have an update next week, then maybe the other dates might be shifted a little.
Yeah.
But that's possible. Okay. Should I don't Karina? Should we schedule already?
I'm meeting for Thursday or Friday next week or something like this. So, it's over next to somebody next week. All week can also discuss it in the heart. We're meeting maybe to have an order. Yeah, slot in the hardware meeting.
Yeah, I agree.
Sounds good. Okay, thank you for my side.
I have no for the questions.
Welcome. I've also no fair question.
Just one point can you share this slides with us in Panama? How many I can simply use?
So this is, as you can see, derived from the management meeting. So you the others, there are more slides again shortly if you wish, but you can ignore the other side, then you have already seen them. So, how many people if you can put it to Panama?
Then it's great. Like, if the internally don't? Yeah, no problem. Thank you.
Welcome. All right. That there are no other questions.
So comments that nothing will close the meeting. Yes, thank you also. Thanks.
Bye. Bye. Have a nice day. Have a nice day. Thank everyone. Goodbye. Thank you. Bye.
Bye together.
