Is it just simple impedance control? Answer with real examples!

yvonneGan

Is it just simple impedance control? Answer with real examples! [Copy link]

This content is originally created by yvonneGan , a user of EEWORLD forum . If you want to reprint or use it for commercial purposes, you must obtain the author's consent and indicate the source

Original article by Mr. Gaosu, the self-media of YiBo Technology | Wu Jun and Huang Gang

"I just want to control the impedance, but you suggest doing simulation. That's really unfair."

This is a conversation between Captain Gaosuo and a marketing staff of our company last week. The customer required that the impedance of the entire channel from the tin finger to the gold finger be controlled at 100Ω±10%. However, the customer had made two versions before, and the impedance of the final test did not meet the requirements, so he contacted our marketing staff to redesign.

(, downloads: 0)

Our market is also "battle-hardened", thinking that it is just 100Ω impedance control, which is easy to do, so we started designing. When it was almost time to submit the board, I always felt something was wrong, so I consulted Mr. Gaosuo. After understanding the detailed needs of the customer, Mr. Gaosuo's team determined the simulation plan. Then our marketing staff "went crazy" on behalf of the customer: I only asked to control the impedance, but you suggested doing simulation, which is really unfair...

Mr. Gaosuo has no choice but to return to his home court to discuss what the requirement of “100Ω impedance control” is…

Back to our design, the adapter board design of the 10G optical module looks like a very simple board, consisting of three parts: solder fingers, routing and gold fingers.

(, downloads: 0)

The stackup is also very simple, 4-layer board, ordinary FR4 board.

(, downloads: 0)

Let us briefly describe the customer's version 2 test background. We use the familiar SMA to enter (test fixture), then connect the optical module and adapter board, enter the adapter board through the gold finger, and then connect the cable after the finger comes out.

The customer tested two versions, and the impedance of the first version was very low. This was due to design problems. The lower layer of the gold finger and tin finger pads was laid with ground, resulting in low impedance of the pads on both sides, thus lowering the overall impedance.

(, downloads: 0)

In the second version, after learning from the previous experience, I used impedance calculation software such as polar to accurately calculate the impedance of the gold finger pad, and then modified the size of the gold finger pad (pad was changed to 0.55mm, spacing 0.25mm), and confidently calculated to 100 ohms. In fact, it is very good to be able to calculate this kind of coplanar impedance, and not all of my friends around me can do it...

(, downloads: 0)

As a result, when the board was produced (the second version in the picture above), I found that the impedance of the gold finger was still obviously low, only about 93 ohms. Then I had all kinds of doubts, doubting the processing ability of the board factory, doubting that the software calculation was wrong, and then I even doubted... Life is here.

Then this difficult job is handed over to us. Now it’s my show time. I will focus on analyzing the gold finger part. How can we optimize it so that the impedance is not much different from the actual measured impedance?

First of all, is our software correct? How can there be such a large error? Let's take the gold finger part and simulate it accurately to see.

(, downloads: 0)

In order to distinguish the positions of solder fingers, traces and gold fingers, so that everyone can see more clearly, we deliberately adjusted the impedance of the traces to 95 ohms, so that the changes at both ends can be distinguished. Our simulation results are as follows:

(, downloads: 0)

The impedance of the gold finger pad we simulated is still 100 ohms, which is the same as calculated by the software.

That means the impedance calculation software is fine, so why are the actual measured results so different? ?

We must know that when we test, the gold finger must be inserted into the slot to connect, so when it is actually working, the gold finger part is not just a pad, it is actually a combination of two structural keys, which we generally call the launch structure. So at this time, the whole structure becomes the following model:

(, downloads: 0)

Then if we combine the two structures and simulate them again, we can see from the simulation results that the impedance of the gold finger becomes about 93 ohms, which is very close to our measured data.

(, downloads: 0)

It seems that our launch simulation is relatively accurate, but the question is, how can we obtain a gold finger design with a measured resistance close to 100 ohms?

Looking at the previous design, there is no reference under the gold finger at the bottom of the table, so we can only increase the impedance after launch by modifying the size of the gold finger pad. We directly scan the simulation results of different pad sizes and finally find that when our pad is modified to 0.47mm, the simulation results show that the gold finger pad can basically reach 100 ohms.

(, downloads: 0)

Are you curious about another question? If you calculate the impedance in polar or other similar impedance calculation software according to the modified pad size, what will the result be? ? ? 115 ohms!!!

(, downloads: 0)

Does it overturn your imagination? The fact is that after the launch, the stub behind the pad launch will lower the impedance of this part, and it will change according to the length of the pad itself, so the simple pad impedance calculation is almost meaningless and may only be misleading. Only when we have the simulation conditions can we characterize its impedance through relatively accurate three-dimensional simulation.

Finally, I would like to end with this sentence: "I only asked to control the impedance, but you suggested to do simulation, which is really unfair..." Do you suddenly feel that if you don't do simulation, the result of the board will be really scary...

bobde163

It's too advanced. I can only worship it. It's awesome!

Is it just simple impedance control? Answer with real examples! [Copy link]

Latest reply