• Please review our updated Terms and Rules here

A general purpose Flip Chip adapter board - Worth doing?

vrs42

Veteran Member
Joined
Nov 24, 2009
Messages
655
Location
Beaverton, Oregon
I got 0.0955" by taking the offset to the middle of the first pin (0.158") minus half the cell width (0.125"), or 0.158 - (0.125/2) = 0.0955". This represents the right edge of the cell for the first pin.

Of course, you are completely right, half a thousandth is meaningless in this context. So 0.095" is the better value to use.

I took a 3D printing class tought by a mechanical engineer. One of the most important things he tought us was that if one part has to fit inside the other, there better be at least a 0.01" difference in size between the inside one and the outside one, or they'll never actually go together, let alone come apart.

My edge connectors have the leftmost dimension line at 0.105". The rightmost dimension line for that tab is at 2.345". (Yes, that puts the gold quite close to the board edge.)

It's important to be able to "rock" the boards slightly in their sockets, to get them out. (This is exactly the sort of thing that I was referring to when I suggested a couple of revisions might be needed to get things working.)

Vince
 

DDS

Veteran Member
Joined
Sep 5, 2013
Messages
588
I took a 3D printing class tought by a mechanical engineer. One of the most important things he tought us was that if one part has to fit inside the other, there better be at least a 0.01" difference in size between the inside one and the outside one, or they'll never actually go together, let alone come apart.

My edge connectors have the leftmost dimension line at 0.105". The rightmost dimension line for that tab is at 2.345". (Yes, that puts the gold quite close to the board edge.)

It's important to be able to "rock" the boards slightly in their sockets, to get them out. (This is exactly the sort of thing that I was referring to when I suggested a couple of revisions might be needed to get things working.)

Vince

Sliding fit -- when the parts need to move while in operation.

Clearance fit -- when the parts need to be easily assembled and not move.

Interference fit -- when the parts are press fit one time and "never" come apart.
 

Dare

Experienced Member
Joined
Dec 28, 2015
Messages
70
I took a 3D printing class tought by a mechanical engineer. One of the most important things he tought us was that if one part has to fit inside the other, there better be at least a 0.01" difference in size between the inside one and the outside one, or they'll never actually go together, let alone come apart.

My edge connectors have the leftmost dimension line at 0.105". The rightmost dimension line for that tab is at 2.345". (Yes, that puts the gold quite close to the board edge.)

It's important to be able to "rock" the boards slightly in their sockets, to get them out. (This is exactly the sort of thing that I was referring to when I suggested a couple of revisions might be needed to get things working.)

Vince

Agreed. However I think that's allowed for in the spec'd dimensions. For example, I have three hex boards in front of me now that all have varying dimensions for the A connector (.105/2.352, .116/2.358, .119/2.359) and all of them fit fine. I ended up with .104/2.34 on my M9042 clone board and the fit is perfect.
 

intabits

Experienced Member
Joined
Jan 22, 2019
Messages
145
Location
Melbourne, Australia
My edge connectors have the leftmost dimension line at 0.105". The rightmost dimension line for that tab is at 2.345"...
That makes the tabs 2.240" - perzactly the size specified by DEC,
So
...I think that's allowed for in the spec'd dimensions.
seems to be correct.

It's important to be able to "rock" the boards slightly in their sockets, to get them out.
maybe they were designed at 2-1/4", less 0.01" for "room to rock"?...
 

intabits

Experienced Member
Joined
Jan 22, 2019
Messages
145
Location
Melbourne, Australia
Some parts arrived from Mouser and Digi-Key, so I can make my "full dress rehearsal" physical mock-up of the adapter modules.
Despite some low quality workmanship, and the lack of suitable tools to do a better job, I'm quite happy with the result.
Professionally made modules should give me the capabilities that I'm hoping for.

Two adapters attached to a large PCB in the quad height configuration.
The small 1.5mm aluminium straps/plates on both sides of the two pcbs join them solidly together.
This PCB is even longer than an extended length module. But I would never be making project cards anywhere near this size (certainly not homemade ones)

SJlWQhm.jpg



The Adapter module/Project board combination is easy to insert and remove.
Shorter length projects might need some nearby boards to be removed to gain working room for this.

Zq4TdoX.jpg



The combination of the two PCBs has no rigidity problems.

iE9V9Rq.jpg



Even just the adapter module alone is useful for gaining quick and easy access to bus signals.

rfL8iH1.jpg
 

Dare

Experienced Member
Joined
Dec 28, 2015
Messages
70
> Even just the adapter module alone is useful for gaining quick and easy access to bus signals.

Was thinking you should place a big ground pad in the center span of the PCB to serve as a convenient place to attach the ground clip for your scope.

--Jay
 

cbscpe

Experienced Member
Joined
Apr 11, 2017
Messages
369
Location
Switzerland
Nice builds.

I did not see it mentioned, but what I'm using is the drawings found in EK-KA680-TM-001_KA680_TechMan Appendix F-41. As already mentioned prices for PCB have been falling in the last decade quite substantially and therefore I tend to build full sized PCB cards at least for dual-width designs. I don't bother about gold plating, the stuff I build will not be in operations for an extended period. In most cases I'm updating my designs and replace them with newer enhanced versions. And champfered edges can be done easily with the correct tools. For signals to be accessible I prefer a backplane with wire-wrap pins. However I need to mention I'm limiting myself to Q-Bus machines, which require less sophisticated debugging tools.

Peter
 

intabits

Experienced Member
Joined
Jan 22, 2019
Messages
145
Location
Melbourne, Australia
>
Was thinking you should place a big ground pad in the center span of the PCB to serve as a convenient place to attach the ground clip for your scope.
I still haven't checked whether ground (and power) pins are consistent across the various DEC bus systems, so the adapter does not assume anything in that regard.
It may also be used in non-bus systems, such as specialized connectors inside a peripheral, that may have other ground pin assignments.
One could always knock up "scope friendly adapter" suited to a particular test scenario, that plugs into the basic adapter. (based on just an IDC header)
I'll be trying to make a DIY 10x probe cable with the "dupont" pins shown in the pic, that is suited to this usage. We're not talking GHz here, so hopefully a simple compensation network (at the scope end?) can be made.
 

intabits

Experienced Member
Joined
Jan 22, 2019
Messages
145
Location
Melbourne, Australia
I did not see it mentioned, but what I'm using is the drawings found in EK-KA680-TM-001_KA680_TechMan Appendix F-41.

Thanks. That's a much higher quality diagram than available elsewhere, I wish I'd thought of looking at QBus manuals.

They've also fixed some of the errors, but I think some remain. Eg: the 10.457" quad height should be 10.438".
From the handle holes, 8 +2 + 2x0.219 = 10.438" (also, no decimal point shown on "219")
You get 10.458 only when the max tolerance of 0.01 on the 0.219 as applied. (The much earlier diagrams showed the quad height as 10.437"). The single and double heights are correct.
 

ftcnet

Experienced Member
Joined
May 4, 2019
Messages
125
Location
BC Canada
I still haven't checked whether ground (and power) pins are consistent across the various DEC bus systems, so the adapter does not assume anything in that regard.
Good thinking. Ground and power pins are not consistent between Omnibus and Unibus backplanes. Not sure about Qbus, but I'd be surprised if it was.
 

intabits

Experienced Member
Joined
Jan 22, 2019
Messages
145
Location
Melbourne, Australia
A couple of weeks ago, I had some of these adapters made by JLCPCB.
As suggested by Roland, I didn't bother with the ENIG Gold fingers (and it would have near doubled the price).
The boards are too short for the fab to do the beveling, so that must be done manually on some sandpaper.

FCAB-1.jpg

For anyone interested, these are the gerbers: View attachment FlipChipAdapter-2-1.125.zip


I've also thought of another use for them - they could be used to make a "poor man's extender board", which might be desirable given the cost and difficulty of getting DEC style edge connector blocks.
It uses one set of adapters to connect the Omnibus (or whatever) to ribbon cables that reach to above the surrounding boards, then pairs of adapters at the top end of the cables are used to sandwich the edge connector tabs of the DEC board under test (laying on some insulating material over the surrounding boards)

Some sort of spacer would be needed between the two adapters to match the DEC board thickness, along with some fiddling to remove the black plastic strip on the headers that that join the adapters together, and receive the ribbon cable IDC connector.

Perhaps bolts in the holes would be enough to make good contact between all the fingers, but some sort of clamp (wood or 3D printed?) over the fingers, with other bolts between the tabs would be better. This clamp would also keep the adapters aligned properly with the DEC board.

Something like this:-

FCAB-2.jpg

It may even be better than an extender board in one respect: the board under test can be laying on something solid, rather than flapping around in the breeze, making it easier to probe and attach clip leads to.
(Needs a lot of adapter boards though - 6 for a quad board, 9 for hex)
Something to think about...
 

Attachments

  • FCAB-2.jpg
    FCAB-2.jpg
    132.5 KB · Views: 1

intabits

Experienced Member
Joined
Jan 22, 2019
Messages
145
Location
Melbourne, Australia
Here is my first attempt at a home-made Omnibus interface card using these adapters.

It's a mash-up of the boot-loader designs by Malcolm Macleod and Roland Huisman, plus some extra stuff (power supply voltage monitoring, and provision to read the memory address and memory data buses)

Yes, I do know about professional fab houses that are some or all of: cheaper, faster, easier, better. (I just had 5 different designs done that way a couple of weeks ago).
But I want to explore the limits of what I can do at home, especially when my design may be total rubbish, so I don't want 5 expensive copies of a failure.
It may also be easier to "bodge" on fixes with a DIY board (wider tracks & spacing, no solder mask or copper fills), which is desirable for development.

Single sided PCBs are more practical for DIY, so that's what I did, using wire links on the "solder" side.
That was the worst part of this project - next time I would try to do double sided.

Being a DIY board, and single sided, I decided on surface mount because:-
* Less holes to drill
* No pins on the "solder" side, so wire links can run under components without interference.
* Smaller components leave more space for traces.
Even the through-hole style Arduino was mounted as an SMD.

I used the "toner transfer" method, which went quite well:-

OIV01-1.jpg

As did the etching (I don't know why the forum software insists on rotating this one to vertical) :-

OIV01-2.jpg

Then came the drilling, this took about an hour:-
OIV01-3.jpg

The bottom rows are 1mm, most of the others are 0.8mm. I would have preferred 0.6mm, but all my bits of that size are now broken. Only one 0.8mm bit broke doing this, but probably more 0.6mm bits would have broken if I had any.

OIV01-4.jpg

Then the many wire links:-

OIV01-5.jpg

This took several hours, and I would not do this again. Next time I'll try double sided, and if that doesn't work, it's off to the fab...
The 0.8mm holes are a bit too big for the via sizes used, so I'm expecting some problems with dodgy joints on some of these.
OIV01-6.jpg

Finally, all the parts soldered on:-
OIV01-7.jpg
Because it's single sided, the 40-pin adapter connectors are on the other side, so that the pins can be soldered on this side.
I still need to make a couple of small 4-hole plates (probably from FR4) to bolt the board and adapters together.


Glue used to keep the wire links from shorting:-

OIV01-8.jpg

Over the next weeks, I'll try to get this thing to do something useful...
 

Attachments

  • OIV01-5.jpg
    OIV01-5.jpg
    103.7 KB · Views: 1
  • OIV01-5.jpg
    OIV01-5.jpg
    103.7 KB · Views: 1
  • OIV01-5.jpg
    OIV01-5.jpg
    103.7 KB · Views: 1
  • OIV01-3.jpg
    OIV01-3.jpg
    115.5 KB · Views: 1
  • OIV01-4.jpg
    OIV01-4.jpg
    101.6 KB · Views: 1
  • OIV01-1.jpg
    OIV01-1.jpg
    113.5 KB · Views: 1

BitWiz

Experienced Member
Joined
Sep 7, 2021
Messages
235
Location
Glen Ellyn, Iil
What is the likelihood of removing the board and having the dual in line connectors separate leaving the edge connectors in the back plane?
 

intabits

Experienced Member
Joined
Jan 22, 2019
Messages
145
Location
Melbourne, Australia
What is the likelihood of removing the board and having the dual in line connectors separate leaving the edge connectors in the back plane?

Zero!

Not shown in the last photo, I made small plates with 4 holes to join the board and adapters firmly together.
An earlier photo in this thread shows my "dress rehearsal" mock-up using aluminium plates.

I've added the plates (made of FR4) to the board above. It's shown in this thread:-
https://www.vcfed.org/forum/forum/ge...rogress-so-far

It inserts and removes to the backplane as easily as any other board.
 
Last edited:
Top