Welding problem or SI problem? You decide!

yvonneGan

Welding problem or SI problem? You decide! [Copy link]

I believe many fans know that Mr. Gaosu has moved into a new office. When we were busy tidying up the laboratory, we received a board sent by a customer. After testing, the customer reported that the impedance at the SMA head position was too low, so we put down the tidying up work and started the first debug test of the new headquarters in the messy laboratory!

Since the customer designed it himself and we processed and welded it, the customer suspected that we processed or welded it. Let's take a look at the customer's PCB file first. We can see that the customer designed a test board and the SMA head used is a coaxial connector welded on the edge of the board.

Since the customer sent us the soldered board, the first step was of course to test and verify whether the impedance was really low. So we tested several SMA head positions and found that the impedance was obviously low. The position where the SMA head contacts the board was only 42 to 43 ohms.

From the PCB design, we can see that from the SMA head to the routing, the routing line width does not change, so the impedance deviation caused by processing is unlikely, because the subsequent routing can be controlled within the allowed range of 50 ohms. So we locked the position of the low impedance at the contact point between the SMA head and the board. In fact, not only do we suspect this position, but the customer also feels the same way.


In fact, this customer has good design and simulation capabilities. You heard it right, with simulation capabilities! ! ! The customer first performed a 3D simulation of the contact position between the SMA and the board, so the customer had reason to suspect that it was a problem with our processing or welding. Mr. Gaosuo usually puts his posture in a relatively low position to listen to the feedback and opinions of customers, and the customer also showed us their 3D modeling and simulation results. Mr. Gaosuo met a colleague and immediately felt that he had found a fellow traveler. Anyway, give him a thumbs up!


We saw that the customer's modeling and simulation results were like this. It can be seen that the customer was very careful in design and simulation, and the simulated impedance was accurately around 50 ohms. Friends who have done 3D modeling of SMA heads will also know that this operation is not an easy task.

After reading the customer's simulation model and results, the whole debug became more confusing, because at first glance, Mr. Gaosuo felt that there seemed to be no problem with the customer's SMA modeling and simulation results.

So we took the physical board, looked left and right at this position, and suddenly found a very important point, that is...
it turns out that the SMA head is welded! ! !


But what Mr. Gaosuo wanted to express was not so superficial, but from the perspective of SI, based on our experience, the welding process will more or less affect the signal quality, so we will evaluate the difference between the customer's simulation results and actual measurements from this perspective. First, let's take a look at the details of the board before and after welding with the naked eye. You can clearly see that after welding, there will be a string of solder residue at the place where the wiring window is opened, so we have a bold idea...

We simulated the customer's original model based on the relevant parameters of solder and the influence of solder to see what the result would be.

The simulation results verified Mr. Gaosuo's conjecture. After simulating the impact of welding, the impedance at this location dropped to a level very close to the test results.

Judging from the simulation results, the difference between adding welding influence and not adding welding influence is so huge.

This is the main content of the article. Is it the welding problem or the SI problem that causes the low impedance? Mr. Gaosuo will not comment too much here, and you guys can judge for yourself.