Thanks, Zach! One point worth mentioning is that sometimes the variation is not between different calculation methods, but between the ideal design parameters and the real-world fabricated results. For example, I think Altium assumes that one ounce copper is 35 um thick (which is close; IPC uses 34.3 um). But actual copper comes in at the low side of the accepted tolerance (to save money on copper) and then becomes even thinner after processing. IPC-6012 Table 3-13 allows a minimum copper thickness of 24.9 um after processing for inner layers. Another factor is that Altium's Layer Stack Manager only shows the starting copper thickness of the material used on the outer layers, and ignores the extra plated copper on top of it that results from plating the hole barrels. (Altium needs to added PLATED COPPER layer to the Layer Stack Manager!!!). Final trace widths are commonly measured to be thinner than in your design, etc. The point is, the bare board fabricator results may differ from Altium's estimate, not because Altium's calculator is wrong, but because their final result has to match the desired Impedance AFTER processing.
You're right there does need to be a copper plating option included in the layer stack manager specifically to address nickel-based platings. The problem is it is based on a Simbeor model and I do not know how they queue that up, if they do it with a pre-built numerical model built with multivariate regression or if there is actually a method of lines field solver running inside the layer stack manager.
I don't want to comment on specific impedance calculator products, the intention here is only to point out why some calculators give different results and to help users understand what different calculators are doing.
Hello Zach, I have a question about Impedance in Altium. You know that we can calculate the Impedance in Layer Stack Manager and then from the rules we can assign the this Impedance value to some signals as 100ohm. However, when we make the Signal integrity Simulation in Altium, we always see the signals are 200-250 ohm but we define 100 ohm. It is not possible. I have read in diffedent forma that others have encountered this error as well. What is the reason of this? Is this a bug?
I don't know exactly why you would see this without looking at your project and your PCB layout. It sounds like you have a thick board (62 mils) with 2 layers and you are trying to route thin traces, in this case those thin traces will have very high impedance. If you have a multilayer board, then you probably removed all the ground beneath those traces. In either case you would get very high impedance of 200-250 Ohms. Just for fun, I tried to reproduce your situation, When I take a 50 Ohm trace for a 6 mil thick dielectric, but I increase the dielectric thickness to 60 mils (Dk = 4.2), the SI simulator calculates an average impedance of 150 mils, just as you would expect. Note that the "define 100 Ohm" condition is only a target impedance for unrouted nets and I think it is only used in the crosstalk/reflection simulation, it is not something that is used as a design target in the simulator. Note that you have to specifically size your traces and route them over ground to get a specific target impedance. If you enter "100 Ohm" into that box it will not use that for any nets that you routed. If you enter some other value into that box when you are preparing the simulation, then the simulator will only model signal behavior using that number for unrouted nets.
Would like to see some maxwells calculators calculations. I have seen many open source 3D field solvers like: “openEMS” and “MEEP”. What are your thoughts on this? Maybe someday can show how to use them? Thanks
We showed something awhile ago with Simbeor, that one uses method of lines and some other methods to build solutions. I downloaded openEMS but have not played with it yet, but as soon as I learn I'll have to show how to use it.
I can't believe it's a free content with so many good explanations. Thank you!
Glad you enjoy it!
Thanks, Zach! One point worth mentioning is that sometimes the variation is not between different calculation methods, but between the ideal design parameters and the real-world fabricated results. For example, I think Altium assumes that one ounce copper is 35 um thick (which is close; IPC uses 34.3 um). But actual copper comes in at the low side of the accepted tolerance (to save money on copper) and then becomes even thinner after processing. IPC-6012 Table 3-13 allows a minimum copper thickness of 24.9 um after processing for inner layers.
Another factor is that Altium's Layer Stack Manager only shows the starting copper thickness of the material used on the outer layers, and ignores the extra plated copper on top of it that results from plating the hole barrels. (Altium needs to added PLATED COPPER layer to the Layer Stack Manager!!!). Final trace widths are commonly measured to be thinner than in your design, etc.
The point is, the bare board fabricator results may differ from Altium's estimate, not because Altium's calculator is wrong, but because their final result has to match the desired Impedance AFTER processing.
You're right there does need to be a copper plating option included in the layer stack manager specifically to address nickel-based platings. The problem is it is based on a Simbeor model and I do not know how they queue that up, if they do it with a pre-built numerical model built with multivariate regression or if there is actually a method of lines field solver running inside the layer stack manager.
What about Saturn PCB calculator?
That's a great question, I use it a lot for some things.
Rogers impedance calculator is good too
I don't want to comment on specific impedance calculator products, the intention here is only to point out why some calculators give different results and to help users understand what different calculators are doing.
Hello Zach, I have a question about Impedance in Altium. You know that we can calculate the Impedance in Layer Stack Manager and then from the rules we can assign the this Impedance value to some signals as 100ohm. However, when we make the Signal integrity Simulation in Altium, we always see the signals are 200-250 ohm but we define 100 ohm. It is not possible. I have read in diffedent forma that others have encountered this error as well. What is the reason of this? Is this a bug?
I don't know exactly why you would see this without looking at your project and your PCB layout. It sounds like you have a thick board (62 mils) with 2 layers and you are trying to route thin traces, in this case those thin traces will have very high impedance. If you have a multilayer board, then you probably removed all the ground beneath those traces. In either case you would get very high impedance of 200-250 Ohms. Just for fun, I tried to reproduce your situation, When I take a 50 Ohm trace for a 6 mil thick dielectric, but I increase the dielectric thickness to 60 mils (Dk = 4.2), the SI simulator calculates an average impedance of 150 mils, just as you would expect.
Note that the "define 100 Ohm" condition is only a target impedance for unrouted nets and I think it is only used in the crosstalk/reflection simulation, it is not something that is used as a design target in the simulator. Note that you have to specifically size your traces and route them over ground to get a specific target impedance. If you enter "100 Ohm" into that box it will not use that for any nets that you routed. If you enter some other value into that box when you are preparing the simulation, then the simulator will only model signal behavior using that number for unrouted nets.
Would like to see some maxwells calculators calculations.
I have seen many open source 3D field solvers like: “openEMS” and “MEEP”. What are your thoughts on this? Maybe someday can show how to use them? Thanks
We showed something awhile ago with Simbeor, that one uses method of lines and some other methods to build solutions. I downloaded openEMS but have not played with it yet, but as soon as I learn I'll have to show how to use it.
Imagine different "calculators" that gave you different amounts on your paycheck.
😔 ｐｒｏｍｏｓｍ
please do the speak slowly session