320pin BGA Pin Layout suggestions

one.

is 220MHz but the

Absolutely use

chewing up

1. Even

probably

simple: it ends

experience

design.

like these

here...

reason

power ball

You ARE Dimbulb!

I misunderstood your question. It's unnecessary. Microstrip is plenty good at 60MHz. The fields are contained quite well in the dielectric.

If you really want to get whacky put the power planes on the outside.

Don't be silly.

I know people who, at least until recently, used OrCad on DOS because they thought the 'doze version was so awful. I never had a lot of problems with 9.x, but...

The chances of all that being true at the same time is MUCH less likely than scenarios I envisioned. Hence, at the very least, suggesting a 4-layer board would be a good idea is a much less likely shot in the dark. You have to admit, you are stretching. When you have to try to manipulate unknown quantities or qualities to such an extent to make your arguement, it generally means it is likely invalid. You are basically saying in 1000 cases, 999 of them go to my view, one to yours. Not exactly a victory. Same with Kevie.

You really didn't say microstrip was a better idea, did you? And put the planes on the outside and chew them up with the part pads? You are a moron. Its hard to argue someone who just keeps getting dumber with each response. Because you babble something assinine doesn't mean its correct, or even clever.

So we should buy 6 layer boards and just have one layer etched clean instead of using it to reduce EMI or improve performance? Yeah, no moronic stretch there to try to make some beyond stupid point.

Your assumption that it is a processor and nothing else is much more unlikelythat other signal exist on the board. No stretching there....

Hey Dumbass, the chip runs at 220Mhz max internally, hence being referred to as HP, something the designer must just be "wasting" blipping a few outputs back and forth, hey? How slow can you be to figure out it is almost definitely an LPC3180 we are talking about? And you think unspecified means this board has nothing else on it and thats a more valid assumption than there being a great deal more to it? Weak, weak stretch there.

And EMI does not equal "able to route it". You aren't very good at this, are you?

Wow, smash me with the PC board arguement. In fact, some STILL ARE 4 layers. Know how they do it? Not microstrip (duh), they use dual stripline techniques, lots of design reuse, lots of near identical versions, etc., to reduce the time to make the first one. How do I know? Well, back when I was employed at IBM, thats what we did and it hasn't changed a ton (of course, wasn't 30 years ago, I was there for the first Pentium boards, PC and AS400).....

The only thing more stupid than to follow your advice would for me to keep discussing all this nonsense you come up with. I am confident the OP will have a very good idea how to proceed.

By the way, please tell us how they make microstrip on 4 layer boards like YOU would do. Its real easy to control the impedance during normal manufacture, right? No cost penalty to make them accurate, right? Aye!

If I had to guess, I have surpassed 1000 pcb layouts. I have easily purchased MILLIONS of PCB's. You aren't fooling anyone.

I am not saying that, you are the one manipulating numbers. The OP asked what numbers of layers are needed and why. The minimum answer is 4 layers, and I could think of countless scenarios where you could get away with just 4 layers. But, yes, as I said in my first post, more than 4 would be more usual, but I don't think it's 1000 to

We don't know the details, so any answer could be the correct one for the OP.

Dave.

^^^^^^^ Which is what I suggested, Dimmie.

You are the one making all sorts of assumptions, none of which have been supported by the OP. If some of your assumptions are true, the solutions isn't automatically more planes. Let the OP describe his problem further before you try to show how smart you are. Your shit does stink, believe it or not.

By the way, MassivelyWrong, the name is Keith.

Now see, look at your manipulation again to try to be "superior". If, as is true, you came in with the less likely scenario as admitted, it seems most logical it was YOU creating an argument and have tried to "win" without first thinking. I don't care about winning, I just gave the OP the best answer. Sorry that upsets you so. If it makes your ego less sore, OK we'll say 100:1 more likely to not me a 220Mhz processor alone on a board using scant I/O.

When I shit, its only to make you smell better.

Look, it boils down to you want to believe your assumptions are more correct than mine, and your assinine excuses for advice are of value when they aren't. Your suggestions have been dumb, real dumb.

Orcad was ALWAYS awful. I used the DOS verson sometime in the early

Today, if I hated Windows that bad, I certainly would rather move to linux than keep DOS around.

P.S. - Eagle sucks :)

No, it was a proof that you're talking through your ass. A five- layer board will not be more expensive than an eight layer and cannot be more expensive than six. IOW, you lose that one.

I haven't any idea what the rest of the board is. Neither do you, so stop pretending to be so smart.

You haven't a clue, as proven here.

Back when I was employed at IBM... LOL! What happened, get tossed out?

Your choice. You are looking pretty stupid though. I do hope you treat your customers better than the OP.

Within 10% it's pretty simple, yes.

Wow! I'm impressed by your pomposity. Does your arm hurt?

Exactly. There simply wasn't enough information to make such decrees. With all the smoke and heat the OP may not come back with more information.

LOL! I am not the least bit upset about anything, you are the one who seems to have the problem here. I have not argued a single thing, apart from that it is possible to use a 4 layer board if the situation suits.

Get over it.

Dave.

It's always good to know the mentality of the people I'm dealing with.

No, ditz. I made no assumptions. You're the only one here making assumptions.

That seems to be a matter of differing opinion.

Ok, Some differ.

Never used it, though I'd try it if I had reason.

Hi All,

Thanks for your replies.

I note comments between 4~8 layers (may be more to assist routing).

I would like people to comment on routing the 0.5mm BGA balls...

I hear routing between 0.5mm balls will be difficult, will all the major ball signals have to pass through layers to give maxium ease to the PCB routing?

Here's the correct link to the BGA320 package.

Thanks in advance

Joe

Nice try, but you're not the OP. ...perhaps one of Brian's sock puppets.

On 3/20/2007 12:21 PM, The digits of krw's hands composed the following:

o.k. I'm convinced. 60 MGz, big BGA. 6 layes min. The Compliance folks would probably prefer 8. KRW has not a clue. Must be an academic.

Clue 1) No one makes odd layer board anymore. Clue 2) Physical ability and FCC and Signal Integrity issues do not go well together especially as speed goes over 25MHz

On 3/21/2007 11:02 AM, The digits of krw's hands composed the following:

I shouldn't feed the trolls. Really I try, but your comments on the topic and lack of knowledge of modern design and fab practices show you are not doing any current real world design. Perhaps some academic stuff at the most. You sound like the Academic folks I know with no real world experience or knowledge. Are you an academic or a hobbyist? When was the last board you designed a manufactured board with a clock over 50Mhz that actually passed FCC compliance testing? I have not a clue who Brian is, your comments and what they say stand on there own. Brian's show he has real world design knowledge. Simple as that.

Now I got to get back to work. I'm working on a high speed camera board for a client. I'm sure you could have done it in 3 layers, but I had a pile of .5mm pitch BGAs and feel pretty proud to have gotten it done in

1. Thankfully nothing had more then 54 pins other than the TQFP package. Do you even know what a TQFP is? Perhaps you can tell me the trace and space and drill used to break out a .8mm BGA. Your answers to this "might" give you some credibility but that is going to take quite a bit to fix.

Hawker