Electronics > PCB/EDA/CAD

Length matching + impedance matching questions

<< < (5/6) > >>

VEGETA:
I adjusted the routing from the BGA pins as seen in pictures, also made skew + length matched as before. Should be perfect now.

I asked the manufacturer if I can break 2 transmission lines (2 pairs), each pair into an IC which duplicates it... so that one pair goes as default here and the other pair goes elsewhere unrelated (for making analog video output)... they suggested making this for all 8 pairs to avoid timing issues. it will be very pricey though and totally useless since I only need 2 pairs not all 8. I am trying to see if I can just getaway with just 2 ICs for 2 pairs.

if that is the case, how can I route them in order to avoid any signal timing problem?

Fronberry:

--- Quote from: VEGETA on October 04, 2021, 07:21:42 am ---
what are reflections to begin with?

Also, how to know their amount on my current design? vias are necessary to use in this design, and even in the reference board so I cannot avoid them.

--- End quote ---

Think about what a signal *really is*, a wave of energy traveling through electrons in a medium.  Waves share a lot of physical properties between material/media/modes of transfer.  Whether it's energy traveling through water (waves on an ocean). waves traveling through electrons (electricity), or waves traveling through air (sound).

Now, think about what happens when a wave hits an area of uneven resistance to its motion.  Like a wave of sound hitting a wall, it will reflect.  Electricity/signals share this property.  When you have an area of non-constant impedance, some of the energy will reflect backwards along the transmission line, causing all sorts of effects.

It's hard to measure the exact values of reflection and distortion due to it, even with high end simulation.  The most common approach is to follow design rules (such as the placement of AC caps) based on past experimentation.

https://en.wikipedia.org/wiki/Signal_reflection

VEGETA:

--- Quote from: Fronberry on October 07, 2021, 03:28:27 am ---
--- Quote from: VEGETA on October 04, 2021, 07:21:42 am ---
what are reflections to begin with?

Also, how to know their amount on my current design? vias are necessary to use in this design, and even in the reference board so I cannot avoid them.

--- End quote ---

Think about what a signal *really is*, a wave of energy traveling through electrons in a medium.  Waves share a lot of physical properties between material/media/modes of transfer.  Whether it's energy traveling through water (waves on an ocean). waves traveling through electrons (electricity), or waves traveling through air (sound).

Now, think about what happens when a wave hits an area of uneven resistance to its motion.  Like a wave of sound hitting a wall, it will reflect.  Electricity/signals share this property.  When you have an area of non-constant impedance, some of the energy will reflect backwards along the transmission line, causing all sorts of effects.

It's hard to measure the exact values of reflection and distortion due to it, even with high end simulation.  The most common approach is to follow design rules (such as the placement of AC caps) based on past experimentation.

https://en.wikipedia.org/wiki/Signal_reflection

--- End quote ---

Thanks for explanation.

So the summary is that I should place the caps as close to source IC as possible right? I think I have put them correctly, I've done as reference board which actually made and works. Caps are not far from source IC, they are physically about 15mm from it. traces are 20mm length matched.

Do you know what should I do about the question in last post? meaning adding the distributor IC for 2 transmission lines (pairs) along the way after the caps to TX IC.

this distributor IC requires its own AC coupling cap on its output too.

Fronberry:
Is this the standard you are dealing with?  https://www.thine.co.jp/files/user/img/corporate/VBOSTD-V1P52-0000_Abridged%2BEdition.pdf

It seems like you should refer to section 3.3, which doesn't seem to be available in my brief Google search...  :-//

VEGETA:

--- Quote from: Fronberry on October 07, 2021, 06:48:24 am ---Is this the standard you are dealing with?  https://www.thine.co.jp/files/user/img/corporate/VBOSTD-V1P52-0000_Abridged%2BEdition.pdf

It seems like you should refer to section 3.3, which doesn't seem to be available in my brief Google search...  :-//

--- End quote ---

Yes this is the one, how did you know about it? are you Skynet?  :P

It seems like section 3 is not in the document you linked.

However, in the HDMI TX IC which takes VBO-HS inputs, it says this:

it is recommended to separate at least 3 times the dielectric thickness between the
signal layer and the reference layer to any other adjacent signal or GND plane in order to reduce noise
inference and jitter. (or 25 mils is enough space in almost PCB stack)

the dialect is 122um, so distance between any of the 2 signals of the pair vs any other signal is about 0.5mm in my layout which is very good as they recommend. About the pair itself, it is 100 ohms differential with 0.1mm width and 0.2mm distance between the pair traces exactly as the reference board.

what do you think?

Navigation

[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Thanking...
Go to full version