Slashdot Mirror

Here is one more, I did several boards with them when I was in college, their customer service was very helpful with my first board.

http://www.4pcb.com/

http://4pcb.com/ great manufacturer located in Denver. They have great deals, willing to work with you at all levels of production (prototypes vs. actual production releases). If you are new to design they are very helpful and willing to test your designs before printing, to warn of any mistakes you may have.

Yep... that was the golden age for a lot of reasons. I still like writing software for embedded systems, not least because the microcontrollers you get to use are often at about the same complexity as a computer from the early 80s. Sure, they're faster and have more RAM, but you generally don't have complex memory controllers and other peripherals that get in the way. Plus, you can buy a PIC or other similar little jellybean micro for about a dollar and slap it on a $33 custom PCB (if you're a student you can buy just one and get a free pizza, otherwise you have to buy three and no pizza, but you still get the standard bag of microwave popcorn). If you can fit it on a two-layer board, for about $50 total you can put together your own custom embedded device and understand exactly what's going on.

Good review! One question: what method does the author recommend for getting the motherboard PCB made? I would imagine hand-drawn layouts and home etch kits would scare off some potential robot newbies, so I hope he makes some mention of the semi-pro hobbyist alternatives: software like gEDA and Eagle, and board houses like Advanced Circuits for cheap, small quantity fab runs.

For PCBs, be sure to check out Barebones. I've done a couple of runs through them, including one involving a chip-scale BGA package, and they did a great job. They're really a front for one of the previously-mentioned PCB operations, Advanced Circuits. For FPGA and CPLD prototyping, definitely go to Digilent. I've bought a few products from them and have been consistently impressed. Finally, it's probably an obvious choice, but eBay is always a good resource too. Many deals can be had in the "Electronic Components" category of their "Business & Industrial" section.

As far as I see, it is now a lot easier to get into game programming and hardware design today than it was 10 or 20 years ago. There is a device, you know, called a PC. If you can write good games for that, who needs mad 6502 hacking skillz?

Also, the only package that's impossible to solder with a $5 soldering iron is BGA, which are not that popular. I have successfully soldered 0.5mm spacing SMD chips (about the finest they make) with a cheap soldering iron, a homemade PCB, regular solder, and a roll of Radioshack brand solder wick. If you need a complex PCB with lots of tiny plated vias and such, there are companies that will do it for a very reasonable price.

In short, I really fail to see the point of this. It's not very good for learning to program -- why would you want to learn to hack obsolete hardware? It's not good for learning hardware -- someone already did all the work for you! It's not a usable game console. What's the point?

Yeah, acid-etching is great for simple designs, but anything requiring really small traces is out of the question. Double-sided boards and SMDs can be a bit of a trick, too, especially when you do both double-sided and surface mount.

There are tons of great, cheap PCB manufacturers out there, though. This Pad2Pad service is really only good (IMHO) if you have a lot (i.e. hundreds or thousands) of boards being made, and you can get them to assemble the majority of each board.

Olimex is a decent place to order from. 4PCB is also good, even more so if you're a student.

Also, if the board is small, you can possibly panel it to save some money (depends on how many you're making, etc.).

If you're a student, try Advanced Circuits next time you need a board fabbed. They do small runs (can order 1 or 2 boards - don't need to order in the 10's or 100's) at reasonable prices. Their boards are high quality - you don't just get copper on fr4, you also get a solder mask and silkscreen. Give them a try.

I agree, a professional electronics technician is no cheap fellow, nor should he be. And i dont understand why this is even relevant, pc board manufacturers have worked this way as long as i can remember. You design your own board in any number of packages, email/upload/whatever a standardized version to the factory and voila, it appears a week later. These guys will even give you a free bag of popcorn with your board. :)

AVR is indeed good. My first MCU was 8080, then some flavors of 8051; before that was just TTL, lots of it. Got to practice with them when I was repairing IBM 360/370 at school. Today I like AVR a lot, they are simple and easy to use, and AVR-GCC is fairly reliable (haven't failed on me yet, at least.) The C code translates into machine commands quite well - AVR was marketed as RISC optimized for high level languages. I can believe that.

PIC ... never worked with them; I treat them as dinosaurs of MCU world. But at previous job we had a portable device where a PIC (24-pin or something) was used. Why? Because the engineer knew PICs and had the tools. That's the only reason.

And with regard to Ford Pinto ... I am not *that* old, thank you :-) I just read a lot. But I don't own anything Ford anyway, just in case :-) Prefer to stay with something more reliable (being dirt cheap helps too :-)

I use PADS PCB ver 7.0 for DOS to generate the Gerber photoplot, aperture, and excellon drill files. Once I have the fileset intact, I zip them and email my rep ( in my case, Anthony Estes ) over at Advanced Circuits with my zipfile package as an attachment, and about a week later I get a large padded envelope in my mailbox with my boards in it. Every one has been exactly what was ordered. Excellent workmanship. And on time.

Getting started is the hardest part. They know that too, and thats why they have the reps. There is a little hand-holding to be done to make sure you are sending the right stuff, as you do not want a botched job, and they do not want unhappy customers either. When I started off, I first talked via email to my rep, sent him my filesets, and had him review them to make sure they made sense to him too, and were complete and manufacturable. I was impressed with my rep, as he knew just by looking at my files in his viewer if I had screwed up the aperture files or had a bad excellon file. ( The gerber photoplot files shows how the lines are routed, the aperture files show how wide the traces or pads are, and the excellon files show where and what size drill for the holes. You get one photoplot file and one aperture file per layer, and one excellon file per board. ) Once I had confirmed I had a producible fileset, I called him on the phone to set up an account and get the boards made. Since then, every time I need a board, I email him my fileset, and he sends my boards back.

They have quite a sophisticated system to track the progress of manufacturing the boards. You note that little window on their opening page. When you place your order, they will give you a key number to find your order in their database. They know minute to minute where your order is in the manufacturing process - and you can verify at any time where it is. It looks as if they have some very sophisticated PCB manufacturing automation going on over there. My guess is that its like a giant one-hour photo machine. Files go in, boards come out.

No, I do not work there. I am just a pleased customer. I feel I have spewed enough venom at those who did not come through that I feel I must give equal time commending those who did.

I use PADS PCB ver 7.0 for DOS to generate the Gerber photoplot, aperture, and excellon drill files. Once I have the fileset intact, I zip them and email my rep ( in my case, Anthony Estes ) over at Advanced Circuits with my zipfile package as an attachment, and about a week later I get a large padded envelope in my mailbox with my boards in it. Every one has been exactly what was ordered. Excellent workmanship. And on time.

Getting started is the hardest part. They know that too, and thats why they have the reps. There is a little hand-holding to be done to make sure you are sending the right stuff, as you do not want a botched job, and they do not want unhappy customers either. When I started off, I first talked via email to my rep, sent him my filesets, and had him review them to make sure they made sense to him too, and were complete and manufacturable. I was impressed with my rep, as he knew just by looking at my files in his viewer if I had screwed up the aperture files or had a bad excellon file. ( The gerber photoplot files shows how the lines are routed, the aperture files show how wide the traces or pads are, and the excellon files show where and what size drill for the holes. You get one photoplot file and one aperture file per layer, and one excellon file per board. ) Once I had confirmed I had a producible fileset, I called him on the phone to set up an account and get the boards made. Since then, every time I need a board, I email him my fileset, and he sends my boards back.

They have quite a sophisticated system to track the progress of manufacturing the boards. You note that little window on their opening page. When you place your order, they will give you a key number to find your order in their database. They know minute to minute where your order is in the manufacturing process - and you can verify at any time where it is. It looks as if they have some very sophisticated PCB manufacturing automation going on over there. My guess is that its like a giant one-hour photo machine. Files go in, boards come out.

No, I do not work there. I am just a pleased customer. I feel I have spewed enough venom at those who did not come through that I feel I must give equal time commending those who did.

ExpressPCB- Has an offer that will let you make 3 3.8"X2.5" 2-sided boards for about $60, as well as a more general off that's not too much more pricey. They have their own board design software you have to use, which is a bit primitive but adequate for hobbyist use, though it's a problem if you were to ever want someone else to make your boards.

Advanced Circuits- Has a deal to make 2-sided boards for $33/ea, min qty 3. These boards have a solder mask, which is required for dealing with fine-pitch SMT parts, and makes your board all pretty and professional-looking ;-)

Sierra Proto Express- Has a similar deal to Advanced Circuits, but also has a good price to make four-layer boards.

Some general notes- I've used the first two, and it worked alright, but I haven't tried the third one. The latter two require Gerber and Excellon data- this is the standard format for PCB plotting and drilling information information. Essentially any board layout software should be able to generate them. However, it is not trivial to figure out what precisely to send the board manufacturer- you can't just blindly trust your layout software to do the Right Thing. I keep meaning to write a little tutorial on my hard-won knowledge about this, but I've never gotten around to it :-/

By the way, the hardware used may have been expensive, but the hardware PRODUCED to do it was valued by the author at about $50. So a device could be created to spit out the codes easily and cheaply.

I just wanted to interject a quick reality check. Sure, it looks cheap and easy when quickly reading the paper (or just reading comments on slashdot, most written by people who themselves skimmed or did even read it). It looks so simple and easy...

The bare circuit board was made by Advanced Circuits using their $33 each service (that I've used a few times for my own projects). At the time they had a minimum of 2 boards, now it's three. $99 (plus shipping) is still a GREAT price for prototype circuit boards with 6 mil spacing. The norm for the industry is in the $300 neighborhood.

But that $100 only gets you a tiny bare circuit board with a LVDS to TTL buffer chip and 6 mil traces at the same spacing as the traces on the xbox circuit board (nice of them to route the signals on the outer layer instead of an inner layer with the vias burried under the BGA package).

Another component he used as a Xilinx development board, which probably sells for several hundred dollars, and featured a nice Virtex series FPGA chip (expensive). Even if you get the chip as a free sample, you'll need a 4 to 6 layer board (which is way outside of the $33 double sided service), and the ones with flexible choices of I/O signalling only come in BGA packages... which requires very expensive equipment or hiring an board assembly company to solder it. Those chips can only be programmed using proprietary software. Xilinx does provide some limited free software, but the full version sells between $700 to $2500 depending on which chips is supports.

Now I suppose if you're working in your basement, your labor might be free... but consider the difficultly of soldering those 6 mil traces to the matching 6 mil tracks on the xbox PCB. Also consider that he hand-routed the signals inside the FPGA chip for 200 MHz performance... a very difficult and time consuming task, and he manually tweaked the propagation delay of the clock to get his sampling into the center of the stable bit times of the waveforms on the xbox board.

I've spent quite a bit of time designing with FPGAs (eg, the mp3 player on my website), and I can tell you that this hand optimizing the internal layout of the FPGA, custom tweaked for the other delays in his system, is some very serious voodoo magic that takes an incredible amount of time and patience.

Anyway, my point is that the cost is much more than $50... as a student or engineer with access to much of the equipment, you can discount those other costs. Even if the hardware and software were free, the skill required is absolutely astounding. I know it's easy to read a paper like that and lump it into the collective memory of blubs that "appeared on slashdot" without any (or much) appreciation for what an incredible feat it was.

That's why I'm writing this long-winded message... to remind and armchair would-be hardware hackers out there that reading a paper like that prepares one for mastery in hardware hacking about as well as watching the olympic on television prepares one to be a champion figure skater.

So a device could be created to spit out the codes easily and cheaply. It also would not have to be attached for a long period of time, just long enough to retrieve the key. As such you could, theoretically take your xbox to a shop, and be handed the key 2 minutes later. Wouldn't have to solder anything either.

It would be trivial for Microsoft to make all those signals in inner layers of the circuit board in future revisions. Many other more sophisticated counter measures are also possible. Technically unsophisticated laws, like say, the DCMA also serve as a pretty good deterant (at least against a shop doing the work for profit).

But even with the xbox, as it was 1/2 a year ago, the key extraction is a very tough job. The bug in the secret bootloader that allowed the lookup tables for hardware config to bypass the entire process has almost certainly been fixed by now (reportedly, Nvidia recently reported a significant loss on an inventory of xbox specific chips that had to be scrapped... one can only assume they had the original bootloader code and the chips they're making now have a different key and that bug fixed).

So next time you watch figure skating, and they make it look so easy... the same is true with this sort of hardware hacking. Anyone who really does design and play with hardware can tell you that the process described in that paper was absolutely astounding. And while it was relatively cheap, it certainly costs MUCH more than $50.