AMX Schematics & Layout

[KingJL]

Carl, Can you check the footprint you have for IC4 ( LTS2338 ). When I overlay the footprints I have for the MSOP16, they don't match up. The pitch is off and the width is WAY off. I even went and used "library loader" to retrieve the symbol and footprint. The results were the same... they don't match up.

[KingJL]

Carl, Can you check the footprint you have for IC4 ( LTS2338 ). When I overlay the footprints I have for the MSOP16, they don't match up. The pitch is off and the width is WAY off. I even went and used "library loader" to retrieve the symbol and footprint. The results were the same... they don't match up.

on edit: I think I see what the problem is... the footprint on the PCB is a DFN 16 package. The LTC2338 does not appear to be aavilable in the DFN package... all of the alternate parts (LTC23376, LTC2377, LTC2378, and LTC2380) are available in both the DFN and MSOP package, but the LTC2338 only lists the MSOP 16 package.

[Dean Sarelius]

Top
Ground
Power
Bottom

Hi Carl,
So gather this is a double sided PCB only is that correct, just wantto make sure you are following ground return paths.

[Carl-NC]

on edit: I think I see what the problem is... the footprint on the PCB is a DFN 16 package.

You are correct. I copied the '2338 from another part, added the MSOP, and forgot to delete the DFN, and then it defaulted to DFN.

So gather this is a double sided PCB only is that correct, just wantto make sure you are following ground return paths.

No, it's 4 layers.

[Carl-NC]

Here is V1.1. Changes are:

ADC footprint corrected
Polarity of C51 corrected
Added R56 (0Ω) to the output of IC7b for DAC option

Thanks to everyone for the feedback!

AMX_SchRev1.1.pdf
AMX_GerberRev1.1.zip

[Dean Sarelius]

Here is V1.1. Changes are:

ADC footprint corrected
Polarity of C51 corrected
Added R56 (0Ω) to the output of IC7b for DAC option

Thanks to everyone for the feedback!

[ATTACH]n411747[/ATTACH]
[ATTACH]n411746[/ATTACH]

Hi Carl
I have taken a look at your gerbers and would need to make some recomendations.
Bear in mind firstly since you are using a 4 layer stackup, inner layers 2 and 3 will be more tightly coupled to there adjacent top and bottom layers respectively since the inner core typically is much thicker.
What this means is that the top and bottom traces will look for a ground return path respectively along layers 2 and 3 and any vias between the top and bottom connecting these traces would need to be provided with a continous ground path between the layers.
When a trace is unable to find a common ground return path for example when crossing a layer with a cutout the ground return will spread out as it seeks to find an alternate path around the slot.
It used to be common practice that isolating adjacent ground returns from a more sensitive part of a circuit that a slot would be used however this has proven problematic is managing the ground returns.
Nowadauys it is recoemended not to use slots but to look at how the ground return paths are managed.
When laying out a 4 layer PCB rather than use a power plane I simply route power and signal on the top and bottom and dedicate layers 2 and 3 as ground layers which provides the neccessary ground return coupling for the top and bottom traces.
Whenever I have to run a signal or power from the top to the bottom I place a small ground via next to the trace to provide a continous ground path for the trace and the ground return via also connects the top and bottom ground pours.
There are several good videos covering these concepts here i have provided a couple of the best links.
Rick Hartley https://www.youtube.com/watch?v=ySuUZEjARPY&t=109s
Zach Peterson https://www.youtube.com/watch?v=BsOETPSszJQ
Using these techniques I have been able to isolate ground returns from adjacent signals in very high gain circuits yielding exceptionally low noise and low EMI performance.
There is no need to try to isolate the adjacent high gain circuits from noisy power or switching traces by placing cutouts since the traces are tightly coupled to the ground plane there is very little to no spreading.
Once you start impimenting these concepts into your own design I guarentee you that you will see much improved perfromance as I have.
I have just completed a PCB based on moodz Magpie which I have included the gerbers for your reference, just waiting for the components before I can assemble and test.
Regards,
Dean

[Tinkerer]

Hi Carl
I have taken a look at your gerbers and would need to make some recomendations.
Bear in mind firstly since you are using a 4 layer stackup, inner layers 2 and 3 will be more tightly coupled to there adjacent top and bottom layers respectively since the inner core typically is much thicker.
What this means is that the top and bottom traces will look for a ground return path respectively along layers 2 and 3 and any vias between the top and bottom connecting these traces would need to be provided with a continous ground path between the layers.
When a trace is unable to find a common ground return path for example when crossing a layer with a cutout the ground return will spread out as it seeks to find an alternate path around the slot.
It used to be common practice that isolating adjacent ground returns from a more sensitive part of a circuit that a slot would be used however this has proven problematic is managing the ground returns.
Nowadauys it is recoemended not to use slots but to look at how the ground return paths are managed.
When laying out a 4 layer PCB rather than use a power plane I simply route power and signal on the top and bottom and dedicate layers 2 and 3 as ground layers which provides the neccessary ground return coupling for the top and bottom traces.
Whenever I have to run a signal or power from the top to the bottom I place a small ground via next to the trace to provide a continous ground path for the trace and the ground return via also connects the top and bottom ground pours.
There are several good videos covering these concepts here i have provided a couple of the best links.
Rick Hartley https://www.youtube.com/watch?v=ySuUZEjARPY&t=109s
Zach Peterson https://www.youtube.com/watch?v=BsOETPSszJQ
Using these techniques I have been able to isolate ground returns from adjacent signals in very high gain circuits yielding exceptionally low noise and low EMI performance.
There is no need to try to isolate the adjacent high gain circuits from noisy power or switching traces by placing cutouts since the traces are tightly coupled to the ground plane there is very little to no spreading.
Once you start impimenting these concepts into your own design I guarentee you that you will see much improved perfromance as I have.
I have just completed a PCB based on moodz Magpie which I have included the gerbers for your reference, just waiting for the components before I can assemble and test.
Regards,
Dean

Thank you for the excellent links.

[ivconic]

Hi Carl
I have taken a look at your gerbers and would need to make some recomendations.
Bear in mind firstly since you are using a 4 layer stackup, inner layers 2 and 3 will be more tightly coupled to there adjacent top and bottom layers respectively since the inner core typically is much thicker.
What this means is that the top and bottom traces will look for a ground return path respectively along layers 2 and 3 and any vias between the top and bottom connecting these traces would need to be provided with a continous ground path between the layers.
When a trace is unable to find a common ground return path for example when crossing a layer with a cutout the ground return will spread out as it seeks to find an alternate path around the slot.
It used to be common practice that isolating adjacent ground returns from a more sensitive part of a circuit that a slot would be used however this has proven problematic is managing the ground returns.
Nowadauys it is recoemended not to use slots but to look at how the ground return paths are managed.
When laying out a 4 layer PCB rather than use a power plane I simply route power and signal on the top and bottom and dedicate layers 2 and 3 as ground layers which provides the neccessary ground return coupling for the top and bottom traces.
Whenever I have to run a signal or power from the top to the bottom I place a small ground via next to the trace to provide a continous ground path for the trace and the ground return via also connects the top and bottom ground pours.
There are several good videos covering these concepts here i have provided a couple of the best links.
Rick Hartley https://www.youtube.com/watch?v=ySuUZEjARPY&t=109s
Zach Peterson https://www.youtube.com/watch?v=BsOETPSszJQ
Using these techniques I have been able to isolate ground returns from adjacent signals in very high gain circuits yielding exceptionally low noise and low EMI performance.
There is no need to try to isolate the adjacent high gain circuits from noisy power or switching traces by placing cutouts since the traces are tightly coupled to the ground plane there is very little to no spreading.
Once you start impimenting these concepts into your own design I guarentee you that you will see much improved perfromance as I have.
I have just completed a PCB based on moodz Magpie which I have included the gerbers for your reference, just waiting for the components before I can assemble and test.
Regards,
Dean

I wanted to look at your work, but not one of the several online services I use, nor the Sprint Layout 6 software; can't open your gerber files?
Whereas, say, I can open Carl's gerber files in any of those.

[Dean Sarelius]

I wanted to look at your work, but not one of the several online services I use, nor the Sprint Layout 6 software; can't open your gerber files?
Whereas, say, I can open Carl's gerber files in any of those.

Try these CAM files, let me know if you have any issues.
https://www.dropbox.com/scl/fo/6ggat...psv0ogq08o7a1m

[Dean Sarelius]

I see the gerbers do not show the ground vias very well so have added a 3D pdf to the dropbox link and when you zoom into the PCB near the MCU for example you can see the GND retrun vias little blue dots next to the signal vias.

[ivconic]

No, again:
Sprint Layout 6



JLPCB:

[Dean Sarelius]

I just dragged the CAM file onto the google browser desktop and it opened the file.
Suggest to give this a go and if you have any issues I will try to convert them to PDF.

[ivconic]

Rick Hartley https://www.youtube.com/watch?v=ySuUZEjARPY&t=109s
Zach Peterson https://www.youtube.com/watch?v=BsOETPSszJQ

Excellent lectures! Especially from the first link.
But the man explains a very wide area. While we are interested in problems with signals at low frequencies.
However, a good part of the lecture also applies to what we are doing here.
So the video is very educational and illustrative.
It would be too ambitious to say that I already knew most of this, of course I didn't.
But before I started designing the pcb myself; years ago I reviewed a "million" pcbs drawn by someone else.
I was particularly interested in pcbs drawn by people at metal detector companies.
Because that is directly the issue that interests me.
And looking at all those pcbs over the years; without prior specialist education in pcb drawing; I slowly started to understand some rules.
There is a neat analogy for this. When you go to a foreign country and you don't even know the language spoken there.
You move among people for several months and listen carefully to what and how they say in different situations.
Meanwhile, you begin to understand that language more and more. Then you start saying it yourself.
At first it was bad, but with time it got better and better. After a few years, you already started to speak a very correct language.
If you move only in a certain group of people who are engaged in one activity; your language will be excellent in that field.
But if you go to another group of people, of a different social position and with different activities; you will notice that your language
is not good enough and that you will have to learn some things again, in order to reach the level of that group in speaking.
No formal education; this way of learning in practice is only somewhat good, but not good enough.
If we only draw pcbs for analog detectors, eventually we'll get really good at it one day.
But if one day we move to a more modern technology that includes digital signals; things are changing drastically and we will have to learn
some new rules and some things even from the beginning.
The point is, with practice and dedicated work; a person can learn a lot and achieve a lot.
But without formal education; there will remain many things that man will not know.
That's why lectures like this are a real refresher and a source of new knowledge. Of course, who wants to study further.
I'm not ashamed to say that I don't know something.
Because from the first lecture, you can hear in several places that even several university professors did not know the answers to some things.
Technology develops so fast and man learns some new things so fast; that it is necessary to keep up with that progress.
Long story short, this doesn't belong in this thread anyway, it would be productive to move all of this to a new dedicated thread.
As a hobbyist, the task is easy for me. Because I do works in very limited ranges.
I very rarely decide on SMD, and mostly for the largest components, which does not deviate too much from "through-hole" components.
I mostly look to do single layer, lately more often two layer designs.
Since thanks to the advent of cheap services like JLPCB, I can give myself so much comfort to work that way.
But I will probably never make a pcb with more than 2 layers.
And the amount of ignorance is measured by the amount of failure in work. in this case in pcb drawing.
I remember struggling the most with the TGSL pcb. I did a "zillion" revisions and barely ended up with one pcb design where the detector worked acceptably well.
My next biggest failure was trying to draw a pcb for the Minelab Musketeer but in my own way, completely different from the original and with through-hole components,
while the original is with SMD components.
How important, even crucial, is pcb design in fact; I learned from those two examples.
And of course, that's when I definitely realized that I actually know very little.
I have to admit that the later transition to digital technology, or mix analog-digital; simplified things in many ways. Just the opposite of what I feared.
Because the increasingly frequent and massive use of integrated circuits; it makes the job a lot easier.
Of course, by this I mean mostly battery-powered devices, with very low voltages and currents.
And at relatively low frequencies.
And I was trained for the exact opposite, for very high frequencies, high currents and voltages.
But after schooling, I did any of that in practice for a very short time, so it can be said that that schooling is of little use to me today.
Because things are diametrically opposite here dealing mostly with low freqencies, low currents and low voltages.
And finally, I will say something here that few of you will accept without surprise.
PCB design for high frequencies, high currents and high voltages... is much easier and simpler than what we're doing here, believe it or not.
All the more credit to Carl and his attempted pcb for a PI detector that has very complex requirements.
Because AMX combines all possible complications from all mentioned fields of work so far.
All possible "trouble" that may occur in pcb drawing, for any of the mentioned technologies; are unfortunately present in this project!
That's why the task of drawing the pcb for this project is much more difficult.

[ivconic]

I just dragged the CAM file onto the google browser desktop and it opened the file.
Suggest to give this a go and if you have any issues I will try to convert them to PDF.

Pictures like that may help, to be imported as background in Sprint Layout and than redrawn.
Usual method I practice when have not any other way to do it.
"...I just dragged the CAM file onto the google browser desktop and it opened the file..."
How You done it?
I tried do drag the cam file to Chrome icon and happens nothing.
Than I tried to drag it again into opened Chrome browser and again nothing?
We take back roads instead of taking the right path.
What pcb drawing software do you use?
Does that software have an export option for gerber files in a standard format, interchangeable one?
The easiest way to check it is to pack all the gerber files into a folder, "zip" them or "rar" them and then upload them to the JLPCB server.
It will instantly unzip them and open them in "pcb viewer". If it doesn't open them, it means that it doesn't recognize that format.

[ivconic]

I guess this is one of "those" days when nothing works for me.
In meantime I tried few other online gerber viewers and also cam viewers... NONE can open the zipped folder or any file alone??