PCB review

[cosam]

I've been messing around in KiCad, trying to get up to speed with actually making a PCB as opposed to simply drawing schematics. I was wondering if any of our resident experts might be inclined to have a quick look at the results to see if I'm on the right track. The routing was done on freerouting.net and I've added a ground plane on each side. The project isn't entirely off-topic - it's a simple signal distribution board for a DEC RQDX3 disk controller. If there's enough interest I may run off a small batch.

The KiCad project can be found here (Subversion users can check out the entire project using that URL). There's also a PDF of each side for non-Kicad users. I can run off other formats on request.

Many thanks in advance for looking.

 

[glitch]

It looks to be laid out fine, but it's impossible to tell if it's correct without a schematic. What I usually do, and what we do at work as well, is get two people, one with a copy of the schematic, and one with a printout of the top+bottom traces with component numbers/values. You manually check each point against your master schematic. It saves a lot of time in troubleshooting the boards once you get them.

I don't use KiCad...does it not have a router built in? If not, you may be interested in trying CadSoft EAGLE, which is free for smaller boards (3x4" I think). It includes an autorouter, though I usually route everything manually.

 

[Lou - N2MIY]

I will gladly check this for you. I have wire wrapped my own version of the BA23 signal distribution board and made replacements for the BA23 button panel (as it affects the the drive signals). I have also used M9058s and been through those prints. So, I have been through the signals a few times now.

I am no modern cad expert. I will simply plot out your work and check the hardcopy. At work I am known for my red pencil rather than marking up prints electronically. This will give me something to do on the plane when I travel for buisness in the next few days.

I might even be interested in a board or two because I am down to my last M9058 and making duplicates of something I already wrapped once gets annoying.

Is board making really that cheap / easy? I have been thinking of wrapping an RX8E (M8357). They are hard to find, but I almost have all the ICs (except an SP314) and could make my own. With all the interest in omnibus machines, there might be interest in people building (but not engineering) their own options. I enjoyed building the SBC6120 and would certainly buy bare etched/drilled omnibus options, like say the EAE.

Lou

 

[glitch]

Is board making really that cheap / easy? I have been thinking of wrapping an RX8E (M8357). They are hard to find, but I almost have all the ICs (except an SP314) and could make my own. With all the interest in omnibus machines, there might be interest in people building (but not engineering) their own options. I enjoyed building the SBC6120 and would certainly buy bare etched/drilled omnibus options, like say the EAE.
Lou

Yes, and if the board is single-sided you can even etch it yourself using the toner transfer method. I regularly produce one-off single sided prototypes this way, including surface mount prototypes. Getting double sided boards is especially cheap for me as I can give the Gerber file outputs for my boards to my boss, to be included in the large panelized prototypes we order. BatchPCB offers basically the same service, except you wait a bit longer since they depend on submissions from various people. You end up paying a $10 setup fee, and $2.50 per square inch for double-sided boards with BatchPCB.

The service we use at work is offered by Advanced Circuits, and is called their "Barebones special" -- you get double-sided boards with plated through holes and tin dip, nothing else. It's fairly cheap, and if you end up getting a production run through them, you get some (all?) of the barebones cost back. They're also an American company (we try to keep our orders local, as long as it's cost effective). They're also really fast.

 

[Lou - N2MIY]

For a small board, that's a pretty good price. A quad height extended length dec board (8.5"x11") would be about $100. No doubt, if I did this for a living, the PCB route would be the way to go.

The single sided route could be much cheaper. I actually bought some of those toner transfer sheets about 20 years ago, but never followed through on the rest. I remember that Radio Shack used to sell resist pens, copper clad board, and etchant - I'd have to order from Jameco or DigiKey now... I assume the hard part about a home made double sided board is getting the artwork aligned on both sides, and then I'd have to solder wires through the vias and solder all leads on both sides.

Steve,

Can you make a .PDF of the schematic? At first glance, looking at the traces, I think I understand. I see a SIP pull-up resistor pack next to the 50 pin connector (J12 on an M9058 ), the 34 and 20 pin connectors for the two RD disks, and the one 34 pin connector for the RX. Then some ready/write protect signals for the drives go to the 2x4 and the 14 pin connectors.

It might be worth considering adding the WP and RDY status LEDs. The ready LEDs would also need PNP driver transistors.

Lou

 

[glitch]

I do the single-sided transfers using inkjet photo paper and a HP LaserJet 4 -- you don't need the expensive laser transfer sheets. They're easier to start out with, but after a little fooling around with photo paper, you can get equivalent results first time, every time.

Aligning the two sides is part of the challenge. The other is that you can only etch one side at a time. Since you have to iron the transfer to melt the toner to the copper clad, trying to do both sides would be very difficult. I've seen it done by coating one side of the board with resist after applying the toner transfer to the other side, etching, then stripping the side that wasn't etched and coating the previously etched side with resist. You then iron-transfer the other side and etch.

If you need copper clad FR4 boards, let me know. I can get you single or double sided at really great rates, as I'm in the process of liquidating the contents of our warehouse at work. The tenant disappeared and stopped paying his rent for a few years, leaving the warehouse full of electronics surplus. I've got enough copper clad FR4 to probably cover the floor of the warehouse (literally several hundred square feet).

You can also make your own etchant from Hydrochloric Acid and Hydrogen Peroxide, from what I've heard. I haven't tried it, but several DIY friends have used it as an etchant with better results than the traditional Ferric Chloride. There are directions on the Internet for how to make and use it.

 

[Chuck(G)]

I've heard that using clay-coated (shiny) paper from old magazines provides one of the best transfer mediums for laser PCB creation. You may have to tape it to a regular sheet of paper to print it, but the transfer is apparently much cleaner than any other method. Besides, what are you going to do with those old issues of BYTE?

 

[cosam]

It looks to be laid out fine, but it's impossible to tell if it's correct without a schematic.

I didn't even thoroughly check the actual "wiring" myself yet. I was actually more worried about the fabrication side of things: track width, pad size, clearance, ground planes, etc. but if you guys are up for some trace tracking that's of course great I've added a PDF of the schematic. Not the tidiest ever (I just wanted to get the connections made and get onto routing) but you get the idea.

I don't use KiCad...does it not have a router built in? If not, you may be interested in trying CadSoft EAGLE, which is free for smaller boards (3x4" I think). It includes an autorouter, though I usually route everything manually.

KiCad does have a simple autorouter, but it doesn't appear to be very smart. I'm impressed with freerouting.net though; it's great even just to watch it do its thing. I can't fault the export/import between the two applications. I did look at Eagle but 3"x4" is a bit limiting and I'm a bit cheap

I will gladly check this for you. I have wire wrapped my own version of the BA23 signal distribution board and made replacements for the BA23 button panel (as it affects the the drive signals). I have also used M9058s and been through those prints. So, I have been through the signals a few times now.

Thanks! I don't have the M9058, which is partly why I wanted to try this. The only proper distribution board I have is built into a BA23 so not really something you can easily swap out ;-) I also built my own little board, but on a piece of stripboard. It looks even more of a mess now since I added a floppy connector. I figured a PCB version would be a nice simple, yet useful project to figure out the software.

At first glance, looking at the traces, I think I understand. I see a SIP pull-up resistor pack next to the 50 pin connector (J12 on an M9058 ), the 34 and 20 pin connectors for the two RD disks, and the one 34 pin connector for the RX. Then some ready/write protect signals for the drives go to the 2x4 and the 14 pin connectors.

Yup, that's pretty much it. The connector next to the RX is for a front panel but the 2x4 are there so you can jumper the WP/READY signals to pull-ups instead. The silkscreen is in the KiCad project, but I ran off a PDF of that too.

It might be worth considering adding the WP and RDY status LEDs. The ready LEDs would also need PNP driver transistors.

Could do, but I'm not sure they'd be all that useful on the board itself unless you're running it on the bench. Maybe a little front panel board would be more useful? There's a simple circuit on this RQDX3 reference page (which is incidentally also where I got the pinouts from) but it specifies 5mA LEDs, so I think a transistor would indeed be better.

 

[glitch]

If you need design rule check (DRC), you can use Advanced Circuits' FreeDFM:

http://www.freedfm.com/

I use it for boards even if they aren't going out with Advanced Circuits. I think they give you a discount for running your files through it before you place an order so that they don't have to deal with any errors on their end.