FEDEVEL
Platform forum

Imad_RF , 07-04-2024, 08:19 PM

Also tried on AC, No spikes

Imad_RF , 07-04-2024, 08:19 PM

One min!

Imad_RF , 07-04-2024, 08:25 PM

Imad_RF , 07-04-2024, 08:26 PM

Don't look at the numbers on the right this scope is #%&* 🤨

QDrives , 07-04-2024, 08:31 PM

It is on 2V per div, and 20MHz bandwidth?
What is the lowest your scope can handle? I would think you could get to 10mV/div or lower.
If you have a SMPS, you should see the switching ripple.

Imad_RF , 07-04-2024, 08:32 PM

lemme see

Imad_RF , 07-04-2024, 08:35 PM

Imad_RF , 07-04-2024, 08:36 PM

this is across the IC VDD & GND

QDrives , 07-04-2024, 08:45 PM

So something like 60 kHz switching frequency. Those a quite high spikes.
Do the spikes on the supply match with the spikes you see on Vref and Vout? As in do they happen 'synchronous' or if you are in luck, at the same time?
You might want to see what the voltage is when triggered by oa spike on one of Vref or Vout.

Imad_RF , 07-04-2024, 08:46 PM

yes this smps has a switching frequency of ~65KHz according to its datasheet

Imad_RF , 07-04-2024, 08:48 PM

Imad_RF , 07-04-2024, 08:48 PM

This is Vref

Imad_RF , 07-04-2024, 08:49 PM

Imad_RF , 07-04-2024, 08:49 PM

and this is Vout

Imad_RF , 07-04-2024, 08:54 PM

Out of curiosity, this is VDD to GND of the IC when the amplifier is loaded on 4 ohms dummy load

Imad_RF , 07-04-2024, 08:55 PM

Sooooo, these voltage spikes are getting picked by? Idk... the inductor on the top side?

QDrives , 07-04-2024, 08:56 PM

Do you only have 1 channel on your scope?

Imad_RF , 07-04-2024, 08:58 PM

Imad_RF , 07-04-2024, 08:58 PM

THIS IS VREF! When the amplifier is loaded

Imad_RF , 07-04-2024, 08:58 PM

2

QDrives , 07-04-2024, 09:10 PM

What if you:
Channel 1 - Vref (or Vout)
Channel 2 - 5V (at sensor)
Both AC and 20mV/div
20us / div
Trigger on channel 1 , either -40mV or +60mV
Vertical shift the 2 channels so they do NOT overlap as much as possible. Something like 1 to 2 divisions.
To see peaks you may want to set the vertical scale to 50mV/div.

2 channel measurement can be a lot harder than 1 channel, so take your time and do not cause a short (or even worse, damage)


I am thinking that the noise on the supply may be causing the noise on the outputs. If so, then we first need to make the power better.
Second is Vref as that should be a stable voltage too (no AC component). The last is Vout.

QDrives , 07-04-2024, 09:10 PM

Does loading make a (big) difference?

Imad_RF , 07-04-2024, 09:14 PM

yes!

Imad_RF , 07-04-2024, 09:14 PM

before loading:

Imad_RF , 07-04-2024, 09:14 PM

Imad_RF , 07-04-2024, 09:14 PM

after loading:

Imad_RF , 07-04-2024, 09:14 PM

QDrives , 07-04-2024, 09:17 PM

Using the ground spring tip right?
Can you do the same for the 5V supply?

Imad_RF , 07-04-2024, 09:18 PM

yes using spring tip

Imad_RF , 07-04-2024, 09:23 PM

so that was the 5V with the amp ***unloaded***

Imad_RF , 07-04-2024, 09:24 PM

5V with the amp ***loaded***

Imad_RF , 07-04-2024, 09:25 PM

I can't even tell where is the SMPS conducted ripple !!

QDrives , 07-04-2024, 09:37 PM

Your timescale is different in the two supply measurements. Does the 20 in the bottom with the red circle mean that the scope is in 20MHz bandwidth limit?
The Pk-Pk is not so different for the supply. The peak voltages varied also in the 20us/div version. I would have expected a more constant peak voltage.

I would do the measurement of the 2 channels (Vref or Vout and 5V) to see if the output noise is synchronous with the switching as I stated before.

User , 07-05-2024, 06:45 PM

Do you have a ressource where I can learn about that phenomenon please?

QDrives , 07-05-2024, 07:09 PM

No really. I got the question from a video that Robert made. In that video it also showed the return currents on multiple Gnd layers (https://youtu.be/nPx2iqmVAHY?t=374)
Heidi Barnes was in Altium Live, of I think 2 years ago. After her presentation I asked her about this phenomena.
Another message through LinkedIn with the link to the video and the question if this is skin effect (or more the lack of it).
I received a power point with 2 slides. The first page was the setup (as in the video). The second is as shown above here.

Her text: "*I did some quick simulations, and I have to admit that my intuition is better at the high frequencies... it is fascinating to see how the lower frequencies penetrate the lower ground layers. To answer your specific question it will take some more simulations.... and I need to think about how this relates to inductance :)*"

So I would say very few people are aware of this.

User , 07-05-2024, 07:28 PM

That's crazy man 😅

User , 07-05-2024, 07:29 PM

I guess I'll ignore that phenomenon for now, I surely have more important phenomenon to study right now like common mode currents

User , 07-05-2024, 07:29 PM

Thank you for your answer

User , 07-05-2024, 07:38 PM

Btw, Very often I have difficult questions to ask experts or knowledgable people let's say. I don't know any in my real life environment, so I have to contact people online. You mentioned Heidi, do you know other experts available on linkedin to answer technical questions?

QDrives , 07-05-2024, 07:48 PM

There are plenty of experts on LinkedIn that can answer questions. The 'difficulty' is finding the right moment for the right question.
In this case Heidi was at Altium Live where I asked her already if she would be willing.
Just asking out of the blue will not give as good as a result as when you ask the question on a post about such subject.

Take your common mode current for instance. There are plenty of EMC experts who talk EMC subjects. When they post, you could ask your question in a comment and they, or others, can answers.
However, do keep in mind that most answers need to be able to be given in less than 5 minutes 'work'.


The simplest 'answer' to you common mode current question is: it flows where you do not want it.
However, once you 'know' the main causes, it is easy to see the paths. A lot more difficult can be to change the path!

Gagan , 07-05-2024, 08:08 PM

Well, if you just want to know what skin effect is, I could shed a light upon that

Gagan , 07-05-2024, 08:09 PM

I mean it is a common phenomena what is observed to be in transmission lines also

Gagan , 07-05-2024, 08:09 PM

corona effect and skin effect

User , 07-06-2024, 08:25 AM

Please read the conversation further to know what exactly we were talking about.

Gagan , 07-06-2024, 08:48 AM

The lack of skin effect and because of lack of skin effect the lower frequencies are allowed in the ground plane ? Is that right

QDrives , 07-06-2024, 03:05 PM

Almost right. At the lower frequencies a single ground plane does not 'shield' the signal (or return current). In fact, 3 ground planes still do not shield a 1MHz signal.
Given that most SMPS switch at frequencies lower than 1 MHz, it (almost) does not matter how many ground planes you have between it and sensitive circuits, it can get there.

Gagan , 07-06-2024, 03:33 PM

Thank you for the insight @QDrives , now as talking about transmission lines the stranded conductor are made of number of twisted wires and the centre conductor has the maximum flux linkage because of all other conductors wrapped around it, therefore it has the maximum inductance and therefore more reactance, which leads to less current flow in the centre and hence called skin effect, is the same case with the grounds also, does the lack of skin effect means there is less flux linkage to the first ground plane and hence less inductance, and therefore it does not restrict the lower frequency currents and only restrict the higher frequency currents to flow through it.. please correct me if I am wrong

QDrives , 07-06-2024, 06:54 PM

I am way less of an expert than Heidi Barnes and her answer had this "*To answer your specific question it will take some more simulations.... and I need to think about how this relates to inductance :)*"
I did not ask (many) follow-up questions as I was hoping she would come with some more. Or she was hoping that I would buy the ADS tool too.

Question is: Is it inductance? If we should think in electric and magnetic fields, is it perhaps more that the fields are not able to penetrate the copper fast enough?
With lower frequencies the fields have more time to do so.

User , 07-06-2024, 07:07 PM

Return currents using the 2nd and 3rd return layer means that the 3rd and 2nd layer offers an input impedance (for the return current) not that higher compared to the first one. Of course, I talk about loop Impedance. As we all know, the higher the frequency, the more the loop inductance dictates which are the paths of lower impedance. Since GND2 and GND3 are used by the return currents, it may implied that the loop inductance isn't dominating the loop impedance yet as it does at 10MHz or 100MHz (where only GND1 is used); and therefore; the lowest loop inductance path SIG-GND1 is not that lower impedance compared the loop formed by SIG-GND2 and SIG-GND3. Which means that the return current see 3 path of comparable input impedance, and will flow in all of them. It's my understanding of the phenomenon at least.

User , 07-06-2024, 07:09 PM

Of course, Inductance is related to magnetic Field, I talk about inductance but you can also see it as Magnetic Field, it is the same, I found it easier with inductance