SMT MMU & GLUE dilemma

[exxos]

PLCC vs SMT:

The mongrel board was basically going to have the SMT MMU & GLUE rather than the PLCC chips... I have about 80 of each of those SMT chips so it be a shame not to use them, but this basically forces people to buy the chips from my store rather than reusing PLCC chips people may have anyway. SMT is likely going to cause some people some problems soldering as well. Though the mongrel is moving in that direction anyway.

However with the PLCC being in short supply, or basically non-existent in terms of spares, I thought I would mention this and see what people generally think about this ?


Another possible option:

It could be possible that could send the boards of do have the SMT chips assembled, Maybe even some other basic parts such as sockets and even the bulk of the caps ( I am not thinking about fully assembling these, or supplying bags of parts either, as my sanity cannot cope with that again!) , though this of course is going to greatly increase costs for people over just buying the motherboard on its own if I start sending boards off to have bits soldered.

Incidentally, I have been trying to optimise these boards to use less different parts, and even a lot less chips.. But then I started to add stuff like the video switch etc which is also SMT stuff.


More food for thought:

We have a lot of decoupling capacitors on this board, for which are through-hole parts.. I did mention this a while ago, but I don't think anyone actually replied to it, in that I am thinking of changing them for SMT caps, and may even mount these on the back of the PCB out of the way, and underneath each chip which would actually give slightly better decoupling... The thought of cutting 100's of legs on caps again makes me shudder!


Future thoughts:

I was thinking about adding both types on the board, but with not much space for the tracks, I don't think this is just going to be possible any more. As people know, there was the SMT adapter boards to fit the SMT chips into the PLCC sockets, but having different brands of sockets, make these difficult to fit and I generally do not think they are reliable solution after the STE booster chaos, so I don't want to go down that route now.

There is of course the FPGA cores, Which basically will not be possible to fit to the mongrel, but they can still be tested out on the Alpha or regular motherboards. Ultimately, in theory, the next board after the mongrel would likely be a lot of SMT FPGA stuff anyway.

[PhilC]

I guess you'll have to take a poll of the people who have expressed their interest in the new board.

As for me, after doing the TF534 and the SMT MMU adaptors I'm more than happy to have the Mongrel with the SMT components.

The thing we need to remember is that the Mongrel is the next step in development, not the last step or the finished item, so as such, it's a case of making do with what is practical from the R&D side of things and lets face it, SMT GLUE and MMU are not going to be anywhere near as expensive as the PLCC chips.

Are the FPGA's going to be SMT or PLCC? Again, people may need to purchase those from you too as not everyone will be able to flash new cores onto them.

You're not going to make everybody happy but if you can make things cheaper too, even if it is at the slight expense of making it more complicated, then I don't think people will mind that either.

Today money is more important.

[exxos]

I guess you'll have to take a poll of the people who have expressed their interest in the new board.

Then it be 50:50 and be no further forward :P

As for me, after doing the TF534 and the SMT MMU adaptors I'm more than happy to have the Mongrel with the SMT components.

The FPGA of course be SMT, so it is really the shape of things to come anyway..

The thing we need to remember is that the Mongrel is the next step in development, not the last step or the finished item, so as such, it's a case of making do with what is practical from the R&D side of things and lets face it, SMT GLUE and MMU are not going to be anywhere near as expensive as the PLCC chips.

That is true.

Are the FPGA's going to be SMT or PLCC? Again, people may need to purchase those from you too as not everyone will be able to flash new cores onto them.

FPGA's are pretty much all SMT... But my initial thoughts will be to have this as a plug-in board, then if people set fire to the plug-in board, is not so much of a problem as setting fire to the main motherboard. It will be possible I could get a small batch of them made up for people.. But these are problems for much further down the road anyway....

You're not going to make everybody happy but if you can make things cheaper too, even if it is at the slight expense of making it more complicated, then I don't think people will mind that either.

Today money is more important.

It's cost versus complexity really.. The SMT chips are not expensive, but harder to solder... But I have moved over to the SMT video DAC anyway now... so I actually just realised it probably is a little irrelevant as the difficulty and soldering is going to be there anyway now..

[PhilC]

so I actually just realised it probably is a little irrelevant as the difficulty and soldering is going to be there anyway now.

And there ya go, answered your own question, lol :)

Its your board, make it what you will.

[JezC]

For me it's more important for the development to continue as fast as you're happy to push on with it, so if the 'Mongrel' is quicker to produce with SMT rather than PLCC parts (or that's how you want to spend your time), then that counts for a lot.

I'm really excited about the prospects of the developments beyond the 'Mongrel' even if they are many months if not years away, so anything to shorten the development timescales is worth considering.

I can understand how the option for a self-build board might be complicated by lots of surface mount parts but as you mention, the options for through-hole parts are fast diminishing (and/or prices rising rapidly) so i feel this is a bullet we will have to bite at some point.

From a personal point, I have many old STFM boards which could supply some of the parts for any new designs but I'd expect that the relatively small cost saving of re-using the PLCC parts from them could easily be covered by making the overall board smaller & so cheaper.

As with all these developments, it's your call & I'm sure we are all immensely grateful for your time & effort to develop these updated designs.

[exxos]

For me it's more important for the development to continue as fast as you're happy to push on with it, so if the 'Mongrel' is quicker to produce with SMT rather than PLCC parts (or that's how you want to spend your time), then that counts for a lot.

It probably doesn't make any odds if it goes with SMT or PLCC from a design point of a view. Its more if people are likely going to kill the board trying to solder these things on..If people are not happy about this, then either it will have to be assembled for them, or we stick with PLCC stuff as long as possible.

I'm really excited about the prospects of the developments beyond the 'Mongrel' even if they are many months if not years away,

Yes, FPGA is the way to go...Basically because the blitter's are impossible to buy these days, so we have no choice but to go down the FPGA route.. Of course we can also add in the MMU & GLUE in there as well..

Icky is working on the blitter board currently.. We need to thoroughly test out each core and make sure they are all compatible before we officially move over to this stuff, as said previously, the cores do have some slight issues and are not currently 100% percent compatible, but Wolfgang will update them should any problems be found anyway.

I can understand how the option for a self-build board might be complicated by lots of surface mount parts but as you mention, the options for through-hole parts are fast diminishing (and/or prices rising rapidly) so i feel this is a bullet we will have to bite at some point.

They are getting hard to find these days, generally the resistors and capacitors are easy to find still, but there are a lot more items available in SMT. SMT generally not difficult to solder, a simple gas soldering iron to solder these parts on is all thats needed really, but mostly I just want to avoid cutting hundreds of legs of things again, that is incredibly time-consuming for some reason.

From a personal point, I have many old STFM boards which could supply some of the parts for any new designs but I'd expect that the relatively small cost saving of re-using the PLCC parts from them could easily be covered by making the overall board smaller & so cheaper.

As with all these developments, it's your call & I'm sure we are all immensely grateful for your time & effort to develop these updated designs.

Thanks. It's never easy to make these design decisions, as there is always a lot of pros and cons for each.

One thing I did forget to mention in this thread, is that the reason for the SMT chips mostly, was because they were one of the last production runs of those chips. So they should stand a better chance of overclocking than some of the older PLCC chips.. Of course only a idea and they may not overclock any better or worse than the with the original chips anyway.

[exxos]

So what I thought of trying to do was had both versions (SMT & PLCC) Basically just seeing if the autorouter can wire these things up...

The MMU was a bit tricky because of the series resistors, so I just bodged in a bus wire to save me doing the labels individually.. It is only currently a test anyway... So once I finished I will see if the autorouter can mash it all together..

Capture.JPG

EDIT:

It routes.. just not very well... Will try some other variations when eagle stops hanging...

Capture.JPG