Occilator Location

Armored Cars

I'm working on an adaptor for the SX58 to make it thru-hole for prototyping stuff.· The only problem is I just read that the occilator should be as close to the chip as possible.· Could someone look at my drawing and see if the occilator location will be a problem?· The traces from the occilator are in blue (top layer) and pink (bottom layer).· It will be placed in the three holes to the far left.

Oh yeah, and if you see any other problems I missed please let me know.

SteveW

If nothing else, I'd thin the tracks on the pads. There's no need for the extra copper there, and it just encourages solder bridges between pads, especially if your solder mask isn't perfect.
(Also, are you really doing this on a 4-layer board? That seems a bit generous - I suspect it would fit on 2 layers if you went with thinner tracks, and it's not as if 4-layer boards are easy to make at home...)

Steve

Paul Baker

Steve is correct, you shouldn't have to resort to 4 layers for signals (4 layer typically reserves the inner two layers for ground and Vdd planes). I see two issues that are a problem with your oscillator, In my opinion the oscillator is too far from the SX and both traces have two vias. Vias cause an electrical discontinuity in the path serving as a source for reflections (something to be avoided if possible). The oscillator should be the second component layed onto your board (after the SX) and should be placed as close as possible while still allowing adjacent traces on the SX to be routed without interference. Instead of trying to reivent the wheel, why don't you have a fab house make some SX52 con carne for you (through hole prototype board for the SX52 in a DIP package). The gerber file was recently released since the product has been discontinued and is availible here.

KenM

Or ou can wait about 10 days for one already made. Parallax should release the SX48 and SX52, mounted on a small protoboard for about $10.00, yes, ten bucks.

June 28th it should be available.

Armored Cars

I'm not going to be using 4 layers, (cost too much) I just turned the layers on to help me think better when drawing the traces. I will be ordering these through Express PCB (or something like that). It is small enough that I will be able to get 18 of them for $51, about $3 each which is dirt cheap for a custom board.
When I am done they should act as stand-alone chips, like a Stamp except with the power of an SX and built in reset button.

I wanted to keep the lines thick because I don't know how much power I will be using and would rather have the traces too big than too small. I might just make 12 with small traces and 6 the way it is now.

Well, I'll redraw this and post a pic when I get done. Oh yeah, Paul Baker I couldn't figure out how to use that file.

Paul Baker

Gerber files are the end files sent to a fab house, it is considered the industry standard. There are freeware programs such as Viewmate to view them, but it isn't nessesary to do so, they are completed and all set to be sent for fab exactly in the state that they are. ExpressPCB uses a proprietary format, meaning they will not accept gerber files and other fab houses will not accept thier file format. If turn around time isn't important, www.sparkfun.com offers fab services for a $5/sq in (no minimum, no hidden costs) and accept gerber files. The SX52 con carne is under 2 sq in (and might be under 1 sq in, Ill have to check) so the cost for a single board will be $5 or $10 at most. You could do a side by side comparision (your board and the SX52 con carne) if you want without much additional cost.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔

Armored Cars

That sparkfun site is awesome!· Heres a pic of the updated layout.· It actaully worked out much better that way.
I'll have to wait until school starts to get the sparkfun programs downloaded and redraw it then.· (We have 28.8 Kbps dial-up)· When I redraw it I will run all the traces exactly as they are in this pic.

Paul Baker

Yeah its a fun site, glad you like it. The oscillator looks much better now.

A few other things about the layout, your concern for trace width is unessesary, an SX pin can only conduct 25 mA safely, a 1 inch 8 mil width trace can carry 1/2 amp and only raise in temperature 1 C. The only traces you should have wide is the power traces.

And if your layout software can make flood filled areas you should make the entire back flood filled ground (it will route around your signal traces).

Pin #1 should come off parallel, not at an angle, the odd layout you have can create electrical "echo chambers", it is so short that its resonancy is probably in the gigahertz but its better to be safe.

If you reroute the trace going to the top pin of the oscillator so that it loops under and around the second pin your need for more than two layers disappears.

On your I/O lines try to avoid 90 degree turns, this can create reflections in the signal.

KenM

Maybe I am missing something.....it looks like many of the pads for the SX are not lined up???

It appears some leads will solder to pads, while ohther leads will solder to traces, was that your intent?

Paul Baker

Oh yeah Ken I missed that, that is caused by the trace starting in the center of the pad and since he takes traces inward and outward which are wider than the·pads that effect is produced.

Here is a couple excerpts of my last board using an SX52, I don't claim its a perfect layout but I did spend considerable effort researching and thinking about the layout. The first is a photo without flood fills shown, its easier to pick out the components. #1 is the SX, #2 is the oscillator, I placed this further away because it is an actively driven oscillator which can be placed further from the processor, you are using a resonator which relies upon the SX to drive it so it must be placed closer. #3 is an SPDT switch to switch between the oscillator and the SX-Key header, an actively driven oscillator and the SX-Key must not drive the OSC line simultaneously because the oscillator will damage the SX-Key. #4 is the SX-Key header to program the SX. #5 is the reset button and #6 is one of the 4 bypass capacitors of the SX to reduce ripple in the power supply.

The second picture shows the flood filled areas, the blue is the backside ground plane, and the red underneath the SX is a Vdd flood fill on the top side of the board. The ZOhm resistors are zero ohm resistors (a jumper) to connect areas of the ground plane which are broken up by the Vdd traces, you want to provide as close of a direct loop for power as possible, the Vdd and Gnd going to each component should be as straight of a shot to and from your power source as much as possible. (My power jack is just off-screen to the bottom right)

Perhaps these examples can help illustrate some of the points I was making.

Post Edited (Paul Baker) : 6/23/2005 3:13:04 PM GMT

Armored Cars

I'll make the changes when I redraw it in the new program.
No 90 deg. turns
No short little pieces like on pin 1
Smaller traces (for 25mA max)

Thanks!

kjennejohn

H, Armored Cars.
Two things strike me about your layout:
1. It won't work on a typical breadboard. How were you going to use this? I ask because...
2. The pins in the three border pins in the upper left are considerably larger then the rest of the border pins. If the majority of these holes are designed for the typical 0.025" wirewrap pin, then those three larger holes are the "odd men out", size-wise.

It's been a while since I did wirewrapping or much work on tenth inch prototyping boards. I'm about due for a revelation.

Later!
kenjj

Miner_with_a_PIC

Good stuff, vias and bends in traces will create reflections but fortunately the reflected amplitudes will be considerably less than transmitted amplitudes. Provided that the lengths between these points of reflectance are less than the first POSSIBLE harmonic of 1/4 wavelength (this actually requires very special conditions to be present, the 1/2 wavelength is more common) you should be just fine. For a 50 Mhz circuit the shortest length between discontinuies for resonance to occur would be a on the order of 5 feet. Inductance created by loops and the ever present parasitic capacitance will become an issue far before issues with reflections will come into play. 50 Mhz is not too large a frequency and the oscillator layout design for the SX is going to be very forgiving.

JT

Post Edited (Miner_with_a_PIC) : 6/26/2005 2:00:00 AM GMT