BUG in SON Generator: DFN2x2-6L 3D model fails with specific L and b parameters

[javier.rodas]

When creating a SON (DFN2x2-6L) package using the parameters L = 0.3mm and b = 0.3mm, the generated 3D model renders incorrectly (as shown in the attached images).

Currently, the only workaround is setting the b parameter to a slightly smaller value to get it to generate properly; using b = 0.29mm or smaller works as a temporary fix.

Please review the SON Generator logic for this package type to resolve this geometry conflict.

[jorge_garcia]

Hello @javier.rodas ,

Thank you for the images. So if you look at the datasheet b is specified as min= 0.25 and max = 0.35 however in neither of the two captures do you show that combination. When Min and Max are set the same you limit how flexibility built into the IPC-7351 standard. I would suggest that when using the generator the datasheet be followed exactly.

Right now I can't take this to a developer because the first thing they are going to call out is that you are not following the datasheet specification. If you follow the datasheet numbers does the generator work?

Let me know if there's anything else I can do for you.

Best Regards,

[javier.rodas · Accepted solution]

If you use different L and b parameters, it generates the 3D package using the larger volume (the max values) and it usually works. However, in this specific example, when L = b, the current SON generator may fail. So, setting aside this specific datasheet, please fix the issue when L = b = 0.3 mm, which is a perfectly valid possibility in any datasheet (a perfect square SON or QFN pad).

[jorge_garcia]

Hello @javier.rodas,

A square pad would be allowed. The issue isn't the fact that L=b in this case, the issue is that there is no min and max for either value. In the data sheet L and b both have min and max values. The problem comes from setting min and max to be the same not that L=b.

I'm going to share this with our librarian team but I still don't think there is a bug here.

Let me know if there's anything else I can do for you.

Best Regards,

[javier.rodas]

I'm sorry, Jorge, but I'll have to disagree here. I have literally hundreds of 3D packages in my libraries that use identical min-max values. That isn't the problem.

The issue only occurs with L = b = min = max = 0.3mm in the SON package generator, which is why I reported it as a bug. The 3D representation of the pads below the package ends up in a round shape, while the exterior pad edge is flat, and that internal pad radius edge calculation has a bug when the pad length is equal to the pad width, that is, when the SON package pad is perfectly square. The QFN package generator probably has the exact same issue, but I haven't tried it yet.

You can enter min = max = mean_value on any other package generator and it works 100% fine all the time. Using the mean value for package dimensions is completely valid, and almost nobody wants to waste time entering min-max tolerances for every package, which makes the 3D package creation task too tedious.

Again, literally 99.9% of my 3D components have min=max=mean_value, as I always calculate them from the min-max values specified in the datasheets. That way, I don't get what is, in my humble opinion, a wrong "max" 3D representation of the package for every leg and body of the component. Fusion 360 uses only the max value to calculate the volumes, right? At least the embedded "3D package generator webview" in EAGLE worked like that, and I never liked it. The mean values represent the actual leg and body volumes much better.

[jorge_garcia]

Hi @javier.rodas 

According to the various footprint standards IPC-7351, JEDEC MO-229, etc. L should always be greater than b even if slightly so. Every package follows slightly different formulas which is why some of the other generators don't misbehave.  Just because the mean value has always worked for you till now doesn't mean that's how the standards are defined.

The Max values are not the only value used to calculate the pad shapes. The IPC-7351 defines three different density levels LEAST, NOMINAL, and MOST. By default the NOMINAL density level is used  and it has different factors based on the min and max values you provide. If you find that the pads are too large for your liking you can go to LEAST density level and it will be the tightest "reliable" pad size for the given component.

In any case, I did pass it on to the developers and this is the feedback they gave back.

Let me know if there's anything else I can do for you.

Best Regards,

[javier.rodas]

I get your point that L should generally be larger than b, but in reality, I already gave you a counterexample with the HUSB332B package. L can be 0.2 mm while b_min is 0.25 mm, so b can definitely be larger than L. This means the SON pad could be perfectly square, or even wider than it is long. The SON (and QFN) package generator in Fusion 360 should support b >= L, and having min equal to max in any field shouldn't cause an issue; it should generate the 3D package correctly in all of these cases.

Thanks again for your time replying to our issues.

[javier.rodas]

Here is another example of a package where L has a min=max=0.4 mm, with no tolerance range.
This proves that min and max can perfectly be defined as equal values for any parameter of the package.

[jorge_garcia · Accepted solution]

Hi @javier.rodas ,

I've passed this along to our developers and they'll make the final call. In the meantime, here's the official recommendation.

Use the datasheet values as specified whenever possible. The mean value may work but if you run into a situation where it doesn't work, then you must use the values as specified in the datasheet.

Let me know if there's anything else I can do for you.

Best Regards,