Puma board for FreeEMS

[nitrousnrg]

Finally I have my camera back:

http://img340.imageshack.us/img340/3006/dsci0036c.jpg

Jesus Christ man! You're rough :-)

Instead of assembling mine, perhaps you could put all the components in a plastic bag and post them to me. Then I could sprinkle them on and drive back and forward over it a few times to attach them. Am I being too harsh? LOL. You might want to customise the init.c file to disable all the output pins and avoid frying your CPU straight off... At least some of the outputs are pulsing on and off all the time. You better hope its not a bridged one.

Nah, that was when it came off the oven, I had put too much solder paste hoping to see most of it evaporating. That didn't happen.

The final result was a working, but quite overheated IC due to the copper strip used to remove the extra solder (can't remember the name, it was 4 letter and had a 'w' or 'k')

About the reg, why the change? Is it making heat or is it noisy?

Why remove the fusing? I was tempted to suggest we add some overvoltage protection. A common approach is to use an SCR fired by a chip like MC3423P, such that if you have an OV condition, it shorts and blows the upstream fuse.

It turns out the reg doesn't stand 125°C. 125 is the max junction temperature. Also, my ultra cheapy chinese multimeter measured 4.9v and I didn't like that.

I'd like to put a simple varistor, and move the fuses with the rest of the car fuses. Or put a different kind of fuse, what we have there looks odd to me (not automotiv-ish).

does BDM offer a similar test layer?

I hardly doubt it.

What solder paste are you using?

I'm using rework solder paste, but the first thing I have to do is
1. Put less paste
2. Repair the temperature sensing circuit, I'm working blind with the little oven. That circuit isn't working because I connected backwards the supply cables (yay!)

[Fred]

Fair enough re the paste thing :-)

Why remove the fusing?

Because it belongs outside the box.

Marcos, junction temp is how the cpu is specced too. Environmental temp is impossible to spec because it depends on heat, which depends on clock speed, etc. They give you junction, and case to junction, and you do the rest.

"I hardly doubt it" >> "I highly doubt it"

Fred.

[jharvey]

I can recommend Kester 236. Or was it 256? I think it was 236, I let a friend use it, so I can't go confirm the container. It's the same stuff Cash Olsen sells, that's good stuff, however not RHOS. Needs no flux off when done, which means no build up either.

I know a regulator design that's already drafted in KICAD. You can use that one to get an automotive grade regulator. I seem to recall it's 125C compliant, and it's a switching supply, so it makes less heat. We can chat more about it when we move into spin 2. On a side note, why sequential names? I like spin doctor, it's more prestigious than the number 2. Ah, my humor is so obscure, and bad, it will probably be the death of me.

I see what you are saying about the fusing. I know several after market devices like automotive amps come with fusing at the device. However, that practice invites people to rely on this fusing instead of using fusing in the proper place at the source. I drew up DFH such that the PCB trace was thinner between the two fuse pads. The theory was that when laying down 10 amp PCB traces, they would typically be designed for 20 amp draw. This would prevent issues with voltage drops ect. Then at the fuse, the trace between it would be rated for the 10 amps. It's short length would minimize voltage drop, and would minimize heating. It was intended that the fuse was not typically populated. Because it's DIY, we all know that some poor blok will direct connect it or will make a mistake and use a 40 amp up stream fuse. This would mean that when they exceeded the 10 amp design spec, it would burn the board in a repairable controlled location. Repairable by adding the fuse holders, and inserting a fuse.

That was the theory. However, there is some concern that a PCB trace won't burn as expected. I seem to recall that some folks recommended to use a real fuse all the time, such that a 10 amp draw could be guaranteed.

[Fred]

On a side note, why sequential names? I like spin doctor, it's more prestigious than the number 2. Ah, my humor is so obscure, and bad, it will probably be the death of me.

Maybe save the release names for releases rather than pre releases? This is the equivalent of me typing make. I don't even number those :-)

Because it's DIY, we all know that some poor blok will direct connect it or will make a mistake and use a 40 amp up stream fuse. This would mean that when they exceeded the 10 amp design spec, it would burn the board in a repairable controlled location. Repairable by adding the fuse holders, and inserting a fuse.

I disagree with trying to cover every mistake from every idiot. It's a waste of space that he's already short of. More pins or more protection could be created/used without them, too. Basically, if someone is dumb, they deserve to suffer. It's called natural selection, unfortunately our western societies thinks it's good to keep idiots alive. The Spanish don't, unsure about Argies.

Fred.

[jharvey]

I'm not particularly tied to either fuse approach. I see pros and cons of either way. I don't think the pros or cons of either approach are really all the distinguished, or important. I say pick an approach and go with it. However, that's for the next spin.

[nitrousnrg]

Marcos, junction temp is how the cpu is specced too. Environmental temp is impossible to spec because it depends on heat, which depends on clock speed, etc. They give you junction, and case to junction, and you do the rest.

No, max CPU junction temp is 145°C or something like that. The cpu can stand 125° ambient as long as it doesn't draw more than 80mA or so.

I know a regulator design that's already drafted in KICAD. You can use that one to get an automotive grade regulator. I seem to recall it's 125C compliant, and it's a switching supply, so it makes less heat.

Next spin (doctor?:p), could have switching supplies, like the IT34063, or something similar. Also, we need a proper reference voltage source. For the secondary 5v supply, its probably better to keep the regulator, I don't want a switching supply for the analog circuitry.

[jharvey]

Why remove the fusing?

Because it belongs outside the box.

Please use content when you debate like this. I will completely ignore comments like this one.

Marcos, junction temp is how the cpu is specced too. Environmental temp is impossible to spec because it depends on heat, which depends on clock speed, etc. They give you junction, and case to junction, and you do the rest.

How much current does the CPU draw typically? I believe it's expected to be under that 80ma Macros noted. 80mA dropping 3.3v with a fairly large package, vs a 10v drop of the regulator with a fairly small package. I see a hot regulator as an issue long before the CPU is an issue.

[nitrousnrg]

I see a hot regulator as an issue long before the CPU is an issue.

True, the regulator has to be changed by something better.

[Fred]

Why remove the fusing?

Because it belongs outside the box.

Please use content when you debate like this. I will completely ignore comments like this one.

37 boards have been made and are being distributed soley because *I* spread the word, and because *I* recommended them to my friends and acquaintances. Marcos was going to order 4. Don't think I have any loyalty to anything but the best outcome here. If a device isn't up to spec, I will not publicise or recommend it. That will reduce the quantity of buyers and testers significantly. It belongs outside the box for many many reasons. I'm not going to list them. I don't have time.

I knew nothing about this board when I put out the word, except that I believed Marcos had probably done a fairly good job. I didn't look over any of the architecture stuff at all. Mark my words, though, I will, and shlt that is wrong will get fixed before the word gets spread again. This is part of the dev process, incremental improvements. These requirements will get added to the FreeEMS compliance document as I go through this process with Marcos and anyone else willing to spend time on it.

I believe it's expected to be under that 80ma Macros noted.

I measured more. XGATE is enabled now, so it'll be higher again. I may have cocked up the measurement, but I don't think so. I've also measured my cpu a significant amount above ambient due to that power dissipation. This info is somewhere in my hw thread.

Fred.

[wrcbfmr]

.