PCB layout - JHarvey

[Fred]

At the top it says moved and in the source it says :

Code: Select all

<script language="javascript">
		//copyright circuitcalculator.com
		//please do not copy without permission

[AbeFM]

That's just the one I was using.

My own arguments the first time I went through this were:
1) Other boards seem to use small traces and are just fine
2) The current isn't on very high most of the time
3) There's only 100 mils or so between pins anyway, a trace can only be so large!

Due to the asymptotic nature of cooling, you'd think that a 50% duty cycle wouldn't run 50% cooler... And since this is the ground for EVERY return path of the whole car... Well, at least the accessories as mentioned earlier, and the injectors (Which I'm still nominally running one pair at hold and one pair at peak)... Which is what led me to the one suplimentary ground trace - something for the 300-800 mA here and there of the solenoids, and then one from the MS case (which houses the injector drivers although the argument could be made that's a bad idea) which must just get out the other end of my card. I may just use a 40 mil trace and run one more fatty (You'd think with twice (ok, 1.25x) the current capacity of the two injector lines) from the MS to the connector.

Should I stay away from making the sensor signal lines 10 mils?

[AbeFM]

At the top it says moved and in the source it says :

Code: Select all

<script language="javascript">
		//copyright circuitcalculator.com
		//please do not copy without permission
                //This is just a simple test to see if you can read my email address off the bottom of the page

[AbeFM]

130 Degrees C.
266 Degrees F.

Details:
The 130 C is a sustained temperature. That is the temperature that they are UL rated for V0 (which is the highest fire rating UL has). The laminate actually starts to melt between 285 and 300 C. This is called Tg or the temperature at which the resin begins to gel.

___________

Ok, well, based on that, let's say "ambient" is 50 C (120f) so, 80 degree rise? Ok, gotta leave the country - ttyl

[jharvey]

I seem to recall most PCB houses claim 5/5 or 6/6 as the low cost tolerance. Sparkfun has a tutorial that cautions that 5/5 is X,Y and that at an angle. Many fab houses can't hold 5/5 on an angle, even though they claim then can, so Sparkfun recommend at least 8/8 as a rule of thumb. I think the 10mils is a nice round number, so I'd avoid going any smaller than that if you can. 10mil traces with low frequencies, won't likely have significant capacitive or inductive issues.

I forget what the built in fuse widths are that I put in my spin of 1.0. I seem to recall I pushed the limits, but I think it should be fine as there are no traces near the fuse trace.

Also don't forget your thermals may limit your current capabilities for a pin as well.

Do you have a draft posted, I'd be willing to review and toss back a more detailed explanation of things. I'm still a fan of multiple grounds if you can. Give the injectors and coils their own wire directly back to the battery.

[jharvey]

At the top it says moved and in the source it says :

Code: Select all

<script language="javascript">
		//copyright circuitcalculator.com
		//please do not copy without permission

I seem to recall this fellow didn't want copies just floating around the Inet begging for bit rot. If he made an update, he wanted that update to stick. I don't think he objects to it being used else where. I suspect the copy note is an attempt to let people know when it has been updated. I'd bet permission isn't a big deal with him.

[Fred]

Fair enough, but it's also not on geocities anymore either so no big deal. Either way I cached it so it's here if required.

Fred.

[jharvey]

In the past, I've mentioned spinning a brain board with FPGA capabilities. I'm thinking of spinning an AVR32 board with future FPGA growth capabilities. Even better, I see Atmel has an eval board for $80 and they release the schematic, orcad design files, and gerber files for it. I can't help but think half of the respin is already done.

Some of my frustrations with that TA card include the higher costs, less active dev for it's GCC tools, PIT timer is abnormal, and it's less common to find open minded folks using that FreeScale chip.

Here are some things I like about the AVR32
64 or 48 pin design
temp range -30c to +85c
one 7 channel PWM
one 3 channel timer/counter
http://www.newae.com/loonboard/lub.html
community that is very friendly to the opensource enviroment
onboard USB, UART, ect.
eval board ATEVK1101 found here http://www.atmel.com/dyn/products/tools ... ol_id=4175
includes USB, SD and MMC Card Reader, ambient temperature sensor

I think it can be done fairly easily. I'm curious if others might find this as an interesting approach. I see many features that are already partly developed, like the integration into an FPGA via the lub boot loader variant. I also AVR code as something I can develope with. Biggest issue I see at first glance is CAN. But we aren't focusing on the for round 1 any how. So adding it to a future card shouldn't be that big of a deal.

Thoughts?

[jharvey]

Perhaps an even better AVR32 platform.

NGW100 found here. http://www.atmel.com/dyn/products/tools ... ol_id=4102

Includes 2 eth ports, cost $89.

[gearhead]

The hardware is only part of the equation, as Fred could corroborate. You also need to get all the background stuff working like burns and comms and tuning etc. I am not saying it is either good or bad, but I'd hate to start over. From what I read, the chip selected, S12XDP512, certainly looks like a very solid solution which has more capability than I think we will need initially and is part of a developing product line. 32 bit math could possibly help with speed, but I think we are pretty good with the one we have selected.

I have been quite busy lately and have not had much time to devote to the development of the hardware. I will get back to it, soon. I also think the TA board is a bit pricey for what it is and am advocating developing a dedicated board and soldering a S12XDP512MAL to it as a development board for engine management uses. From a rough calculation, I think we can have a 4 layer board made and solder the chip down with the majority being through hole or combo through/SMD pads for a true DIY board for some people to put in their projects for a similar price to the T/A board. I have advocated for a DB37 pin compatible to MS to facilitate the swappable nature of this initial development phase and am not advocating this as 'the way forward' but suggesting it as a reasonable approach. I just look at my stuff and think that it would be beneficial if I could plug a jimstim to a board for bench development and then when I think it is ready to start an engine or even drive it, I can plug it on my wiring harness and go. I think I am not alone. My goal is for this to be 4 cylinder COP (low level drivers), sequential injection with appropriate relay controls and some high current controls on board (boost, idle, etc). This board would also run a V8 with W/S and ganged injection... More channels could be developed on a riser board to utilize the many outputs of the chip.

If we can get this to the level of the current MS2 (which I think we can with a lot of work), I feel we would really have a nice product that others may be interested in. The objective must be to get this project to a reasonable point and, I think, changing processors at this stage may be a bit off topic and a bit derailing. Once we have the basics down (which we do not, yet), someone may be interested in porting to a new processor. I think we should leave that to later.