Newbie questions

[Graham]

I've been charged with learning all about a firm's RV1s now they've decided to use it again after losing all the expertise as people left. There's the manual and some tweeking notes left but that's about it.

This forum has answered a lot of questions, filled in holes and shown me paths to be explored, and Mike has already been most helpful off-forum (many thanks Mike).

I'm working my way through RV Gerber with a new PCB (someone else's design that needs a fair bit of fixing) and I've come accross something I need some help on as the manual and Help are silent.

One component (a SOT89) isn't placed properly by RV Gerber. It's offset towards the lead end and I can't find a way of manually positioning it correctly.

Am I missing something or do I have to figure out how to edit the position info the TFR file?

 

[phonoplug]

Hi Graham

You'll come across this with various other components that are not symmetrical in both axis. For example, D-Pak, D2Pak, FFC connectors etc.

They are drawn offset on RVgerber but thats not really a problem. When you start to place the job in RVplace you then need to add an offset on these components to put it back into the right position. This offset should always be the same for a given component, though the rotation of the part will determine whether its an x or y offset, positive or negative. You only need to do this once and the offset will be saved in the job file from then on

I have a list of these offsets for each of the components I use that I know needs them, and I think you're probably looking for an offset in the order of 35 for that component (but don't quote me - and your particular CDF file will be different from mine too which would affect it).

There are other ways of adding offsets to components, and adding them to specific packages, but I've not had success with this before as it didn't seem to take into consideration the parts rotation on the board being built (which is not very helpful!). Perhaps I missed something though as I didn't spend much time on it once I realised it didn't seem to work!

[Gopher]

Never placed one of those, but looking at the package, I would expect it to place and come out of the oven perfectly safe and secure without any fiddling at all. Of course it depends somewhat on your footprint, but I don't even find I need any placement offset for D-Pak either, yes they are not perfectly central to the pads but neither are they outside them and they will move anyway when the surface tension kicks in at melting point.

[Graham]

@phonoplug: The puzzling thing is it gets placed with its centre of rotation nowhere near the centre of a square drawn around the footprint. It's not even at the centre of a triangle. The device's centre of rotation is exactly where you'd expect it to be.

Initially I thought it was somehow seeing a nearby via as part of the footprint but with it moved away it's placed in the same position. I'm wondering if the assymetrical footprint is what's confusing it.

@Gopher: With the leads off their pads it may centre on reflow, but it'll have to move some and may need a prod.

@both: The footprint isn't what I'd have used so I'll modifiy it and see if that has any influence on where RV Gerber thinks its centre is.

[Gopher]

Well I suppose when setting up using RV Gerber it is using the pad shapes to work out centre co-ordinates which throws an extra thing in the mix. But the offset you are describing is interesting, that package shouldn't place offset any more than  SOT23 should (i.e not at all), when placing you could try watching the offset described at the top of the screen, manually add those offsets into the pick locations for that  lane and lock X & Y for that CDF, I do this for SOT23 and it is a huge improvement on how it used to be, poor pick position does affect placement even when the RV is supposedly correcting for it, with some devices this auto-correction makes things worse because the pick position hasn't moved its just the jiggly nature of plastic tape.
Lots of what RV Gerver draws on screen is wonky cobblers so if that's the only place your seeing the issue so far, proceed to build. Plenty of my TFR files show things as not being perfectly aligned when they are quite clearly lined up like soldiers.

Of course is all else fails you can just draw a box round it in RV place and shift it there, I do this a lot these days just to compensate for our machines dire need of some calibratory love.

[phonoplug]

I *think* the way it works out the component centre is as follows, which might help explain why some are seeing different results:

When considering the centre of the component, it looks at the centres of the pads in the CDF symbol and works out the component centre from there, but when considering the centre of the placement site on the PCB, it looks at the outer edges of the pads on the board and works out the placement centre from those.

This works fine with a SOT-23, as the 3 legs are all the same size and shape, but with a SOT89 or D-PAK for example, there is a large pad on one side which upsets things. Ok I know what I mean but my explanation?? I'll do a sketch and post it in a minute. Should make a bit more sense then!

[Gopher]

I'd say that was about right, which in turn tells you why I don't need to displace my D-Paks, the CDF parameters mean it doesn't "see" the bottom edge of the large pad just the top one, thereby putting the centre, nearly in the centre.
I find with my machine at least, that the machine will see leads outside a package before those inside which is why I get that result. However QFN's were far easier than I ever expected so its not like seeing such pads is difficult.

[Graham]

@phonoplug: That's it! Well done that man.

I was assuming RV Gerber used the extremes of the pads, which would have dealt with asymmetric pads in most cases, not their centres.

The answer is to only highlight the centre pad of the SOT89 footprint used, which is a bit like this: .|.

The result is perfect placement in RV Gerber, and hopefully no playing with offsets.

I'm still going to re-do the footprint as it doesn't deal with the tab width.

Edit: That device was the least of my worries as it turned out. Unrouted nets including one across the supply, layers wrongly set up etc, inadequate clearances and more. A complete re-hash is planned after this run.

[Jason]

I *think* the way it works out the component centre is as follows, which might help explain why some are seeing different results:

Not sure that's quite right.
RVgerber seems to completely ignore the pin size and locations in the CDF and go purely on the centre of the body (call this the library part centre).
It then finds the extremities of the selected pads and finds the X and Y half way points (call this the board part centre).

It then puts the part on the board with the two centres aligned.

I tested this by editing all the pins off one side of a SOIC CDF and then highlighting
the full set of pads. Rather than ending up with the remaining pins in the wrong
place, because it uses the body centre (which hasn't changed) the remaining pins
were still spot on.

The means that for something like a D-PAK where the tab is much smaller than
the legs, RVgerber will always locate the body towards the leg pads more than
it should.

I'm not sure how that pans out when placing starts.

I'm not using RVgerber anyway as I'm creating .tfr files as direct exports
from Eagle using my ULP.