USB-to-Serial Module based coders cable

**pboese** · March 5th, 2005, 01:37

This is a copy of a message I just sent to my list of interested buyers. Just an FYI to those who might build their own cable.

-----Start of message-----
I have found a problem with the current design. It appears that the USB interface does not operate reliably when attached to "some" computers. I believe the problem is caused by the choice of the master clock frequency. The FTDI FT232BM chip is designed to operate with a 6.000Mhz clock. To make the RS232 serial side work at the maximum baud of 1.56M baud for the DC I needed to change the clock to 6.144Mhz.

It works great on my old computer. My new computer does not detect the cable properly. I sent a cable to Propeller (of DCFreeDev fame) and he has not been able get it to work. I also sent a prototype to Dan Potter to test on a MAC. He was able to use the cable, but not at the maximum baud due to driver issues under OSX. Of the four computer involved the cable worked on only half of them to date. That's a 50/50 chance the cable will or won't work for any of you.

If I go back to the 6.000 Crystal the maximum serial speed that can be used is 500K Baud. Mind you that is better than the Lik-Sang coders cable which is limited to 115200 baud.

I think it may be possible to modify the pcb layout to add a second crystal and a jumper so the the master clock speed can be switched depending upon how finicky your PC's USB controller is. You would have to open the case and move the jumper manually. (with the cable not connected to anything of course, and you will use good anti-static handling procedures, won't you

Getting 1.56M baud out of your cable may still be possible if your USB ports choke on the USB Coders Cable. You can try adding a ISA/PCI USB add-on card. But don't ask me which ones "might" work

Adding a second crystal and jumper will increase costs by a dollar or two. Your chances of being able to operate it at max baud is 50/50. But you would still have 500K as a fallback.

Send me your comments and let me know if you are still interested.

Paul Boese a.k.a Axlen
-----End of message-----

**pboese** · March 7th, 2005, 04:07

I do not like the "band-aid" solution to the USB
problem. It's is a poor solution to the problem.
Therefore I am going to redesign the USB Coders Cable.

The design goals this time a mostly the same with the
following addition. No surface mount components. Only
through hole parts will be used. This will make it
possible to "kit" the device so "You" will have the
option to build it yourself.

I am currently looking at using a MicroChip PIC
18Fxxxx series USB capable microprocessor. MicroChip
provides sample code to creating the USB-Serial type
solution. I like this PIC series chip because the USB
core and processor core can use seperate clock
sources. That means the USB side will be solid and I
can still get the device to talk to the Dreamcast at
1.56M baud.

This is all preliminary so if you have any ideas please post them here.

Regards,

Paul Boese. a.k.a Axlen

**SiZiOUS** · March 10th, 2005, 00:16

I am going to redesign the USB Coders Cable.

Take your time. When you succeed, please notify me.

Good luck.

**pboese** · March 16th, 2005, 00:23

Originally Posted by neoak

Don't worry, it seems that Yahoo sent your email to the Junk folder. Don't know why. Anyway, do i send the email to coderscable@etc?

And for the redesign, that means you are opening preorders again?

If I can get this into an EASY to build format without any surface mount components then I'd considers opening things up again. And of course the device has to work for more than me and a handfull of other users

I've already ordered some PIC development tools and have a sample part in my hands already.

By the way, for those who might be interested I put up the CAD files for current USB Coders Cable: Rev-F Schematic and Layout. This does not include any crystal switching jumpers or sockets. So if you feel inclined to get your hands dirty with Eagle and improve the current design yourself; have at it. Just be sure to observe the GPL and copyright info in the ZIPed package: USBCC-F.ZIP

**semicolo** · March 16th, 2005, 05:08

Here's a hacked version with 2 crystals and jumpers to configure it : ftp://semicolo.kicks-ass.org/dreamcast/USBCC-Fbis.zip (online from about 14h to 4h GMT, sorry, my sister sleeps in my computer room and she doesn't like fan noise)

**pboese** · March 21st, 2005, 00:20

Here's a picture of the basic development setup for the USB coders cable redesign. No USB cable yet since I'm just testing the development environment and getting used to the IDE. These PIC chips are really neat to play with.

Sourcing all through-hole parts for this version has been going great so far. Just discovered Spark Fun Electronics: a place that does prototype double sided PCB's with soldermask and silkscreen for $2.50 per square inch. As you might well have guessed I'll be making every effort to make this board as small as possible to keep costs to a minimum.

**pboese** · March 21st, 2005, 00:36

Originally Posted by semicolo

Here's a hacked version with 2 crystals and jumpers to configure it : ftp://semicolo.kicks-ass.org/dreamcast/USBCC-Fbis.zip (online from about 14h to 4h GMT, sorry, my sister sleeps in my computer room and she doesn't like fan noise)

Thanks for seeing to this semicolo; I won't be doing anymore work on this revision. I've learned my lesson with surface mount designs. They are great for commerical products, but they can be a real challenge for hobbyists. Some kit producers are soldering all the smt parts onto their boards because too many buyers were unable to do it properly.

The FTDI based design has been a great learning experience

...expensive, but fun... Onward!

**GPF** · March 21st, 2005, 20:29

just a quick question about the USB-Serial interface, I was looking at existing USB/Serial adaptors and saw one on outpost http://shop4.outpost.com/product/3811787 that states that it Support over 1 Mbps data transfer rate. Would something like this hooked up to a standard coders cable and your software changes to dc-tool-serial enable it to reach that speed .

Thanks,
Troy

**pboese** · March 22nd, 2005, 02:05

Originally Posted by GPF

just a quick question about the USB-Serial interface, I was looking at existing USB/Serial adaptors and saw one on outpost http://shop4.outpost.com/product/3811787 that states that it Support over 1 Mbps data transfer rate. Would something like this hooked up to a standard coders cable and your software changes to dc-tool-serial enable it to reach that speed .

Thanks,
Troy

Your basic USB-Serial cable typically uses a single clock source to drive the USB module, cpu and serial USART. Depending on the USB version, 1.1, 2.0, the speeds the cable supports, the USB portion of the circuit will operate at some multiple of the applied clock frequency, typically 4Mhz or 6Mhz by using a PLL (phased locked look) to generate a 48Mhz internal clock (USB 2.0). The Dreamcast scif ports are driven by a 50Mhz clock. The problem is that the baud generators for each device (USB-Serial cable / Dreamcast) when programmed for 1.56M baud will not be within +-3% of each other and will not be able to communicate. The key here is the difference in actual serial speeds between the two devices beacause they operate at different clock speeds. It might work at 500K baud though depending on the drivers for your USB-Serial cable. Some drivers will limit you to 480K baud.

I tried to fudge this with the previous design by using a 6.144Mhz crystal where a 6.000Mhz is called for, and it worked okay with my PC hardware. It was a compromise that looked promising, but the more systems I was able to test it on revealed that is was not a realiable solution,

The new design uses two seperate clocks. One for the USB module running at 4.000Mhz externally and 48Mhz internallly, and a second 25Mhz clock source to drive the CPU and serial hardware. That way I can keep the USB side of the equation in spec and still be able to talk to the Dreamcast at 1.56M baud.

Aside from the baud issues you will still have to translate the RS232 signals from the USB-Serial cable to the Dreamcasts 3v TTL levels either by hacking the USB-Serial cable or adding additional circuitry to translate the signal levels.

That said, you might be able to get the USB-Serial cable to work but it will take some experimentation. Try it and let us all know how it goes. Maybe you will find a cheaper way to accomplish the goal.

**GPF** · March 22nd, 2005, 16:04

thanks for information. Just more currious about that cable. Its been awhile since i look at usb/serial adaptor last time i looked the cheap one were only up to 115kb, so i was surprised to see one that said it supported over 1 Mbps

I actually got a Broadband adaptor yesterday, so good luck with your new design, I look forward to reading about it.

Troy