Issue link: https://iconnect007.uberflip.com/i/1126810
JUNE 2019 I DESIGN007 MAGAZINE 13 with the same relief sizes. This time, the per- son previewing the job says this won't fly and asks the customer to increase the relief sizes. Then, the customer is NOT happy, because they do not have the time to lay out the en- tire job again, saying, "Well, you did it on the four-layer last week." Now, you have an upset customer solely because you were trying to do them a favor on the four-layer, which you were not when you suggested that they could live with the undersized reliefs. The customer has a 14-layer and has used the same insufficient relief sizes. This is why we have capability minimums that should not be busted, even on the four-layer, unless the customer is made aware that they are risking internal misregistration issues and that you can do this one time. Do not make it the norm. These types of conversations should take place between a pre-engineering person and the designer, not a manufacturing person or a salesperson. You are doing the customer a dis- service by "pulling something off" when the risks of doing so should be clearly described to the customer. Don't be that fabricator! Another issue we should discuss is drill files. It seems that fairly often, the NC drill file comes in with a totally different numeri- cal format than the artwork. Sometimes, blind vias, buried vias, and through-holes all arrive with different numerical formats, leaving the preview person to figure out what the format should be. It's true that looking at the NC drill file co- ordinates can help determine whether the for- mat should be leading zero suppression, trail- ing zero suppression, or no zero suppression. Some come in as a mix of metric for tools and ing the trace width size from five different sce- narios to five different trace widths by one- tenth or one-hundredth of a mil. This was with the knowledge that the slight variations could not be resolved in fabrication, but this would allow a CAM operator to resize the unique trace widths without affecting the other sce- narios. Once everything was said and done, I presented the material type and dielectrics to the customer, who promptly said, "No, that material won't work for us." The best-laid plans do not always work in fabrication. That part was eventually rede- signed to reduce the layer count, opening up the possibilities for dielectric changes, high- speed materials, and reduced impedance sce- narios. It also had five trace widths from the same size to one-tenth or one-hundredth of a mil for us to pick them out for resizing to meet the desired impedances. After reading all of this, you're probably thinking, "What is your point?" I cannot overemphasize the need to check with your fabricator before starting layout, from the proper Er of the material you are us- ing to modeling the impedances with a true field solver and questions about the mask di- electrics and Er data associated with the mask. Just 10 years ago with 0.008"/0.008" spac- es and traces, it was no big deal to have a mil or two of mask over the trace. But with to- day's geometries of 0.003"/0.003", that extra mil of mask will definitely affect the imped- ances. Before you lay out a part that may have to change (which nobody likes), consult your chosen fabricator. "Helping" Your Customer This is also known as "How did we get into this mess?" Let me give you an example. Let's say the customer provides internal plane relief pads that are undersized based on your capabilities. But you say, "It is only a four- layer. Starting with 0.006" clearances from the edge of the non-drill-compensated plated hole will leave us with approximately 0.004"." But you are able to pull it off for that job. The customer thinks all the designs can use that same relief size, so they send a 14-layer It seems that fairly often, the NC drill file comes in with a totally different numerical format than the artwork.