6-layer stack confirmation

Log in or register to post comments
Go To Last Post
13 posts / 0 new
Author
Message
jars121
jars121's picture
Level: Wannabe
Joined: Thu. Jun 7, 2018
Posts: 65 View posts
#1
Posted by jars121: Sun. Jun 30, 2019 - 03:49 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Afternoon,

 

I've been back and forth with various PCB manufacturers this week regarding their 'standard' 6-layer configurations. The following diagram represents the configuration I'm after:

 

 

I only need 3 signal layers, and want each to have a closely coupled GND. From my discussions with the various manufacturers, it seems that the above stack isn't typically catered for in 'standard' configurations, as I've not had much luck determining a suitable configuration. The below is the closest I've come across (outside of going with a custom layer configuration):

 

 

My concern with the above, for which I'm seeking some input, is the spacing of the internal layers. In the top image, the separation between layers 3 and 4 is quite pronounced, to ensure the top GND plane (layer 2) is closely coupled to the second signal layer (layer 3). In the above stack, this separation is 14 mil, while the 'coupling' between layers 3 and 2 is 13 mil, which seems like much less of a pronounced separation than I would expect.

 

Is there any issue with the above stack? My concern is the power plane (layer 4) being used as the return path for layer 3, as the power plane is fragmented with multiple different voltage. I've used plenty of stitching vias for signal transitions between the various signal layers, as they share a reference GND plane; I don't have the space to use stitching capacitors, hence the need to ensure layer 3 used layer 2 as the return reference.

 

The top diagram specifies 'largest distance', which the above configuration satisfies; am I just being paranoid/pedantic? I'm not sure on the cost difference, but would it be worth specifying a custom configuration which increases the separation between layers 3 and 4, and reduces the dielectric height between layers 2 and 3, and layers 4 and 5?

 

Thanks!

 

 

Tags:

Learning and Information, General Electronics
Top
ka7ehk
ka7ehk's picture
Level: 10k+ Postman
Joined: Fri. Nov 22, 2002
Posts: 20227 View posts
Location: Tangent, OR, USA
Posted by ka7ehk: Sun. Jun 30, 2019 - 02:19 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

My considered opinion is that "you take what you can get". Personally, I have never had to go beyond 4 layers nor have I had to be concerned about coupling between layers. 

 

Are you absolutely certain that you need what you are asking for?

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

Top
avrcandies
avrcandies's picture
Level: Raving Lunatic
Joined: Wed. Jun 30, 2004
Posts: 3183 View posts
Location: Cincinnati Oh
Posted by avrcandies: Sun. Jun 30, 2019 - 04:35 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I only need 3 signal layers

Only?  Why would you need more than 1 or 2? Is this a super-sensitive amplifier circuit detecting nano-volt signals?  High density component  count?  Or perhaps some RF application?   Many, many circuits work fine with no ground plane layer at all (though much nicer to have one)...why do you need 3?

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Top
ka7ehk
ka7ehk's picture
Level: 10k+ Postman
Joined: Fri. Nov 22, 2002
Posts: 20227 View posts
Location: Tangent, OR, USA
Posted by ka7ehk: Sun. Jun 30, 2019 - 07:33 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I can see these sorts of requirements if you have buried strip lines or such. Or, VERY high density. But, what is being requested is fairly extraordinary.

 

The response that is really needed on this one is from Ross who is in Australia. He ought to be coming along fairly soon if he is going to log in today.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

Top
jars121
jars121's picture
Level: Wannabe
Joined: Thu. Jun 7, 2018
Posts: 65 View posts
Posted by jars121: Sun. Jun 30, 2019 - 07:51 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for the responses guys. The need for 3 signal layers is due to density; the previous version of the board was on a 4-layer, but I've since increased the component count and decreased the board footprint, so the move from 4- to 6-layer was somewhat of a necessity.

 

I don't think the stack I've posted will cause any issues, just figured I'd raise it in case there were any obvious concerns.

Top
ka7ehk
ka7ehk's picture
Level: 10k+ Postman
Joined: Fri. Nov 22, 2002
Posts: 20227 View posts
Location: Tangent, OR, USA
Posted by ka7ehk: Sun. Jun 30, 2019 - 10:13 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Well, increasing the number of layers because of smaller size or higher component count is no surprise.

 

Its that fat layer in the middle that seems so odd. Maybe you need to order 8 layer with nothing on that middle dielectric. That is probably what a custom board fabricator would do anyway.

 

I am sure that you know that increasing the distance between layer 3 and layer 4 does not eliminate coupling; it only reduces it relative to standard spacing. 

 

If reduced coupling is SO important, I would go with 8 layers and add a ground layer between 3 and 4.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

Last Edited: Sun. Jun 30, 2019 - 10:15 PM
Top
avrcandies
avrcandies's picture
Level: Raving Lunatic
Joined: Wed. Jun 30, 2004
Posts: 3183 View posts
Location: Cincinnati Oh
Posted by avrcandies: Mon. Jul 1, 2019 - 12:24 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Are these sensitive or high speed signals?  What are the coupling constraints or concerns? Are there high currents involved (maybe you don't need a dedicated power plane, especially if you have 3 layers for signals).  Don't fall into the trap of throwing layers at the problem.

Without a schematic, it is hard to say.

 

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Top
jars121
jars121's picture
Level: Wannabe
Joined: Thu. Jun 7, 2018
Posts: 65 View posts
Posted by jars121: Mon. Jul 1, 2019 - 12:37 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

All very valid questions and points, thanks again for your input.

 

I'm not looking to eliminate coupling altogether, as you've mentioned that's not the aim of increasing the spacing between layers 3 and 4. I am however looking to implement a sound/considered EMI approach, which extends to other aspects of the design (e.g. enclosure). As for the power plane, I do have some reasonable current demands across the various voltages; if I absolutely had to I could look at using traces rather than polygons, but I'd rather keep the plane for the time being.

 

The board itself is mixed signal, ranging from analog in and out, through low-speed digital (e.g. I2C, UART, etc.) and higher speed digital (LVDS, USB 2.0, USB 3.0).

 

I think I may end up going with the initially posted stack, assemble the board and see how it goes. 

Top
meslomp
meslomp's picture
Level: Raving Lunatic
Joined: Wed. May 2, 2007
Posts: 5758 View posts
Location: Nieuwegein, Netherlands
Posted by meslomp: Mon. Jul 1, 2019 - 05:43 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

With multi layered boards ( specially over 4 layers) you need to select a manufacturer and go with them. You can then after you have made a layer plan just go to others and ask if they can make it, but each manufacturer will have different production approaches that will have different layer thickness requirements to suite their needs of easy and cost effective production.

 

I find it strange that they wanted the largest distance in the middle section. But I guess that is to do with how they make the board and what your special requirements are, specially on blind and burried vias.

Top
clawson
clawson's picture
Level: Moderator
Joined: Mon. Jul 18, 2005
Posts: 96699 View posts
Location: (using avr-gcc in) Finchingfield, Essex, England
Posted by clawson: Mon. Jul 1, 2019 - 07:51 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I know little about electronics but I assume we're not talking about a 16/20/32MHz AVR here? Is this some kind of 100'sMHz ARM or something? Perhaps up in the GHz arena? Obviously the precautions you take in such circuits could be radically different to the low speed AVR experience here? I doubt if many people have put a Tiny2313 on a 6 layer board??

Last Edited: Mon. Jul 1, 2019 - 07:52 AM
Top
jars121
jars121's picture
Level: Wannabe
Joined: Thu. Jun 7, 2018
Posts: 65 View posts
Posted by jars121: Mon. Jul 1, 2019 - 08:14 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

meslomp wrote:

With multi layered boards ( specially over 4 layers) you need to select a manufacturer and go with them. You can then after you have made a layer plan just go to others and ask if they can make it, but each manufacturer will have different production approaches that will have different layer thickness requirements to suite their needs of easy and cost effective production.

 

I find it strange that they wanted the largest distance in the middle section. But I guess that is to do with how they make the board and what your special requirements are, specially on blind and burried vias.

 

This is more or less what I've done :) I've spoken to a range of manufacturers, including manufacturers I've used in the past, as well as some others I've found through research. The stack I posted earlier is their 'standard' configuration with the larger middle dielectric, which at this point in time I'm inclined to try.

 

clawson wrote:

I know little about electronics but I assume we're not talking about a 16/20/32MHz AVR here? Is this some kind of 100'sMHz ARM or something? Perhaps up in the GHz arena? Obviously the precautions you take in such circuits could be radically different to the low speed AVR experience here? I doubt if many people have put a Tiny2313 on a 6 layer board??

 

Correct :) This particularly board has a 300MHz ARM MCU (among others).

Top
avrcandies
avrcandies's picture
Level: Raving Lunatic
Joined: Wed. Jun 30, 2004
Posts: 3183 View posts
Location: Cincinnati Oh
Posted by avrcandies: Mon. Jul 1, 2019 - 08:16 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Been to too many design reviews where someone shows up with a tangled 4 or 6 layer board & they went back reduced the layers needed by 2.

The original component positioning was usually as though someone threw the parts against the board, spatter painting style.   A slight rearrangement can have a large impact (beneficial or detrimental).

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Top
N.Winterbottom
N.Winterbottom's picture
Level: Resident
Joined: Tue. Oct 1, 2002
Posts: 904 View posts
Location: Redbourn, Herts, UK
Posted by N.Winterbottom: Mon. Jul 1, 2019 - 08:46 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

PCB suppliers will each  have a prefered "Standard" Multi-Layer buildup. A typical specification I have seen specifies that a standard 6 layer build comprises 2 inner CORES of 13thou each.  These 2 cores carry the 4 inner copper layers which be be freely selected as Signal or Plane. The inner pre-preg layer is 8thou, the outer pre-preg layers are both 11thou.

 

  COPPER 1

  prepreg  11thou

  COPPER 2

  core     13thou

  COPPER 3

  prepreg  8thou

  COPPER 4

  core     13thou

  COPPER 5

  prepreg  11thou

  COPPER 6

 

 

Top