Physical layout of wire-wrapped boards

[ChristofferB]

So, in my quest to make DIY computers, using primarily tools and techniques available at the time period of the machine I'm working on (usually '70 to '85), I'm in the midst of wire-wrapping a small
single board computer.

 I've noticed, however, that there isn't much resources on the actual layout of the IC's (that's probably a lie; more likely it is hiding very well).
Even today, finding rule of thumbs, and references of what is/was considered good practice, things like
"No more than 3 wrap's per pin"; "Avoid changing layers", "keep wire lengths as equal as possible", etc.are fairly obtainable,
but were there any guidelines or standards for placement of IC sockets, and components in general?

When browsing images of wrap-boards as case studies,
I see that some literally puts them as close as possible, others uses the area available.
 

thanks!

[SeanB]

Normally wire wrap puts the IC's in a line, with all pin 1 aligned to a grid location ( so you essentially have an IC every 10 holes on the board) with the decoupling capacitors being one per IC, or one for every 4, in the 0.1in space between the 14 or 16 pin DIL packages. On the bottom of the board the wire is run generally either shortest route for non bussed wiring, or in a channel formed from the wire wrap pins, so that they are neatly bussed together in similar lengths. Power is typically daisy chained down the row of IC's, with the top side ( pin 1 end) designated for the 5V busbar and the other end as the ground busbar.

Passive components like transistors and such are grouped together, using a IC socket to hold them in a form that plugs into a wire wrap socket, so you can have up to 8 resistors/diodes/capacitors in as single package on the board. Decoupling caps are either put in a single pin socket or are provided with sockets with built in caps between pins 7/14 or 8/16.

[JoeN]

Check out this guy's project.  It's large enough there are a lot of lessons you can learn.

http://www.homebrewcpu.com/construction.htm

[ChristofferB]

SeanB: Thanks! That's what I'll go for, then!
I think I'll stick with soldering power and ground to my chips (Don't have those wrap-specific boards with integrated power and groundplane).
Also; you can get IC sockets with decoupling capacitor built in, between two corner legs, could probably add that to the wrap sockets myself.

JoeN:

Yeah, that's one of my main sources of reference imagery.
Only he uses these double-eurocard boards with groundplane and powerplane built in, and hole arrangement pretty much dictates the layout.

--thanks

[Pjotr]

"No more than 3 wrap's per pin"; "Avoid changing layers", "keep wire lengths as equal as possible", etc.are fairly obtainable,
but were there any guidelines or standards for placement of IC sockets, and components in general?

thanks!

Phoehee, it must be over 30 tears ago I wrapped my last boards and backplanes (for a PDP11). DEC had a couple of pretty good pocket books about proper wire wrapping I remember. If you lay 2 or 3 wraps on a pin depends on the length of the pins. But 3 wraps are much more difficult to debug (repair/re-wrap) Main rule about placing is simply keep the mean of all the wires as short as possible. Further on, making a "wrapping list" is very important. It is kind of a net list as for PCB design. A wrapping list consist of two lists: First a signal list that lists all the signals and the pin to pin description of all the components. From that you derive a second list after placement of the components on the board. This list contains the board coordinates of each en every wire you will wrap. From this second list you will wrap the actual board. Using this method you have the least chance of mistakes. A fully wrapped board is a nightmare to debug because of wrapping mistakes. After the board is wrapped you use the second list again to check it with a "buzz box" (continuity checker). Do this twice, once forwards through the list and once backwards through the list. Why? With this kind of work people have the habit to make the same error again without noticing. Checking backwards catches this

With this list it is important to work in layers or levels. So first make a list for level 1, then for level 2 and eventually level 3. Do not jump or chain between different layers/levels. When you jump wraps between layers it can become a lot of work if you want to change connections later. With jumping you have often to remove all the wraps of that signal. Same for reworking wrapping mistakes.

[georges80]

I wire wrapped quite a few boards back in the time.

3 wraps per pin are fine, of course when it's time to rework it's always the bottom wrap that needs to be removed

The big thing with wirewrap is to NOT try to make it pretty. The more you run bundles of wires together, the more crosstalk you'll have and that becomes a problem as clock speeds go up.

The fastest processor board I attempted (was a combined processor and laser printer controller) was 15MHz and it took some time to get things stable and quite - mostly by spreading wires out to not crosstalk especially on the dram area.

I used (still have one somewhere) an OK electric wrap gun with the modified wrap tip. Nice took for getting consistent/quality wraps done and saves wear and tear on your fingers versus a manual tool. A decent stripper for the kynar wire was also a good thing to own.

I even have a few wirewrap sockets left in one of my parts drawers - though I can't imagine EVER wrapping up a board these days - it's just too easy/cheap to get a real PCB done and pretty well everything is SMT now and clock speeds preclude using wirewrap construction.

I remember when I had a stash of Rogers 'flat capacitors' that you slipped into the socket before pushing a chip in. Improved performance of some of the high speed stuff that had more critical decoupling issues. Rogers even made them for some of the weirder pinouts - i.e. units made for specific chips (CPU's etc).

You could also buy wirewrap sockets with 0.1uF capacitors going diagonally from pin 7/14 or 8/16 or 10/20 etc to ease/improve the decoupling path.

Have fun.

cheers,
george.

[flynwill]

I did my share of wire-wrap projects back in the day as well.

Power needs to be bussed in some way,  There were PCB boards designed for wirewrap that had rows of holes on 0.3" spacing and had interleaved power busses so that you could easily bridge the power pins were common, I'd suggest doing some searching.  As other have said bypass cap per chip.  (usually 0.1 uF ML ceramics).

A wire list is essential.  It should be straight forward to generated from the schematic if you  are using modern CAD tools.  Back in the day we would enter the list by hand sorted by IC connection (easy to type from the hand-drawn schematics) and then sort by signal name, to get the wire list.

Don't daisy chain connections, if you need to connect A to B to C to D first wrap A to B then C to D and finally B to C.  That limits how much you have to unwrap to fix an error.  Use 3 wraps on a pin when you need physically branch a signal to keep the total length short (eg your clock).

Install Labels on the back side of the board.  It's easy to get lost in the forest of pins.

I remember occasionally using twisted pair for really critical signals, Wrap two wires at one end (to signal plus and minus or signal and ground), twist them together to about 1 twist per inch, then wrap the other end.

Power wrap tool is also pretty essential.  It's really hard to get consistent wraps with the hand tools.  You want one that does a "modified wrap" (two turns of insulation around the pin) and where the wrapping tip is spring-loaded inside the tool.

Good luck!  Be sure to post pictures of the finished project.

[cosmicray]

I don't know what wire type (i.e. insulation) you are planning on using, but if you choose PTFE Teflon, be careful about pulling it tight when turning sharp corners. Teflon insulation has known behavior of cold flow under stress, and a 90-deg turn against a wire-wrap peg qualifies.

http://nepp.nasa.gov/npsl/wire/insulation_guide.htm

[Pjotr]

Pure PTFE isolation is surely NOT recommended. It is very difficult to strip by hand and and cannot be used at all with automatic wire wrapping bits. It was used in the early days of wire wrapping when pre-stripped wires were used with bits that did not strip the isolation off. But also with the modern Tefzel/Kynar isolation you have to watch out for "cold flowing": Do not lay wires with tight corners around pins. Using a "45 degrees" pattern helps with this.

[German_EE]

A long time ago in a workshop far away I built an Apple //e using wirewrap as I couldn't afford a real one. I would not like to repeat the project 

[SeanB]

I worked on stuff which used wirewrap in a **Production** unit, though seeing as the total production run was under 100 just doing 100 prototypes was not too onerous. The SENECA build team did do thier best to make it more reliable, by soldering every wire wrap joint after assembly. Was not fun when it came to changing anything like a socket or replacing a wire. On the test unit, which was also a rats nest of wire wrap, I had a few sockets that went intermittent.  Solution was to get some turned pin sockets ( not wire wrap style, but regular pcb pins) and insert them in place of the IC's involved ( a few TTL fusible link proms and some MSI counters) then solder the 2 sockets together, then plug the IC back in the new socket. No way I was going to spend all the extra time doing all the rework on the back, especially as the board was only vaguely documented, with a circuit that was correct, but where they often used gates from another chip and had not updated the drawing with it.

Plus I had a pile of black boxes that needed to go out. that whole AOG pressure of only having one serviceable unit, and it running from plane to plane as they went out to the flight line. Fixed the test bench, then did some unplanned overtime till 10PM getting a pile ready for the next day for them to have serviceable units. Nearly got shot by the roving guards, somebody forgot to tell them I was working late , though the Baboons were also disconcerting as they came through foraging in the dustbins for food. I had the blast door open for the fresh air.

[johnh]

Some photos of wirewrap board from the late 70's


Photo board2 show some of the balanced transmission lines

[wkb]

I did my share of wire-wrap projects back in the day as well.

Power needs to be bussed in some way,  There were PCB boards designed for wirewrap that had rows of holes on 0.3" spacing and had interleaved power busses so that you could easily bridge the power pins were common, I'd suggest doing some searching.  As other have said bypass cap per chip.  (usually 0.1 uF ML ceramics).

A wire list is essential.  It should be straight forward to generated from the schematic if you  are using modern CAD tools.  Back in the day we would enter the list by hand sorted by IC connection (easy to type from the hand-drawn schematics) and then sort by signal name, to get the wire list.

Don't daisy chain connections, if you need to connect A to B to C to D first wrap A to B then C to D and finally B to C.  That limits how much you have to unwrap to fix an error.  Use 3 wraps on a pin when you need physically branch a signal to keep the total length short (eg your clock).

Install Labels on the back side of the board.  It's easy to get lost in the forest of pins.

I remember occasionally using twisted pair for really critical signals, Wrap two wires at one end (to signal plus and minus or signal and ground), twist them together to about 1 twist per inch, then wrap the other end.

Power wrap tool is also pretty essential.  It's really hard to get consistent wraps with the hand tools.  You want one that does a "modified wrap" (two turns of insulation around the pin) and where the wrapping tip is spring-loaded inside the tool.

Good luck!  Be sure to post pictures of the finished project.

Like this board.  Photo taken from an EISA dual SCSI HBA I did in 1990 (...) at Philips Information Systems.

[German_EE]

Johnh, that's the first time I have seen wire wrap boards where the pins are the same side as the chips. It's logical when you think about it because it reduces the assembly height by about 50% as the bottom of the board will be nearly flat.

[Dave Turner]

If you look closely I think it's double sided wire wrap

[JoeN]

A long time ago in a workshop far away I built an Apple //e using wirewrap as I couldn't afford a real one. I would not like to repeat the project 

This implies you pirated the ROM images I suppose.  I guess that's not the worst sin ever committed.  How well did it work?  What peripherals did you have for it, were they genuine or hand-made?  Back then I had an Atari 800 which would have been impossible to build by hand because it had at least 3 proprietary ICs that I can think of.

[codeboy2k]

Johnh, that's the first time I have seen wire wrap boards where the pins are the same side as the chips. It's logical when you think about it because it reduces the assembly height by about 50% as the bottom of the board will be nearly flat.

Yeah me too.. first time I have seen that.  I worked with wire-wrap 30 years ago or so.. I can't add much to what's already been said, but definitely get an OK wrap gun if you can, you absolutely cannot do anything big without one.

[German_EE]

The ROM chips along with a couple of custom devices were purchased from Apple as 'Service Spares'. I wanted to build my own //e because the stock machine also required an 80-column card and a dual floppy drive interface before you got anything decent, so I decided to build everything on one board in an effort to make something smaller.

The computer booted and I could run any of the text based programs such as Appleworks but as soon as I tried to run a program that had graphics the screen got VERY noisy. I never did find the fault and eventually as the 80286 came along I switched to PCs and never looked back.

[SeanB]

You had a race with long wiring and timing, the original also had very critical timing issues, which is why there were those RC delays in the video circuitry. If you had played around with them a little it would have been clear.

[johnh]

If you look closely I think it's double sided wire wrap

No its not double sided.  When I get time I will post a photo of the other side

[Dave Turner]

Fair enough. It would have been unusual if it was.