Single Layer Prop board layout?

Whelzorn

I'm looking for a single-layer prop D40 board layout.
I need one that:
-breaks out IO pins 0-27
-has a header for a prop-plug
-has a power supply (5v & 3.3v reg.)
-has an EEPROM socket

I tried designing one myself, but I always end up isolating pins that need to be grounded, or some signal wire gets crossed, etc. I'm trying to stay away from using jumpers if possible.

Thanks

rjo_

you came to the right place

deSilva

LOL
But maybe this will help? It was the simplest thing I could conceive of some months ago...
Mind the "wrong" orientation of the EEPROM.

It fits on 50mm x 160mm Perf Boards available @ 60 Eurocents in Germany; the rest is low-cost as well. All inclusive it is around 2 Euros...

Ample space for some additions, place the headers at your own convenience. Note it can become a little bit tight where EEPROM socket and PropPlug meet....

Post Edited (deSilva) : 10/6/2007 10:13:54 AM GMT

Leon

I can let you have a PDF of the artwork for a simple single-sided proto board I made at home. I've attached the schematic and PCB layout. It includes a sizeable prototyping area.

Leon

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle

Post Edited (Leon) : 10/6/2007 11:45:00 AM GMT

LoopyByteloose

Well it looks impossible to do without some jumpers.
You can use Eagle to make the one-sided board with automatic routing including the jumpers.

The trick is that you create art work for double sided board which includes vias with all wires on one surface routed verticle and all wires on the other surface routes horizontal.

Here, I would make the top surface vertical. In that way, it wouldn't interfer with your 40pin DIP.

Instead of using the vias, you just enlarge with pads [noparse][[/noparse]and maybe relocate for clearence]. And you print out only one side for actual artwork - the bottom. The top is just there for reference.

There are ways to improve on that routing approach in Eagle [noparse][[/noparse]it allows you to override if you don't like a certain route], but this will get you started thinking clearly.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"Everything in the world is purchased by labour; and our passions are the only causes of labor." -- David·Hume (1711-76)········

···················· Tropically,····· G. Herzog [noparse][[/noparse]·黃鶴 ]·in Taiwan

Lawson

well I haven't needed anything purely single layer. I DO like making layouts that have everything but the ground plane on one layer. So...

The layout I posted here => http://forums.parallax.com/showthread.php?p=582447 (my first post in the thread) Has a Prop layout on the right side that's almost one layer. (the image has also been attached.)

Hm... looking at it, closing the ground connection on one layer is going to be interesting. The 'X' that the power and ground connections under the Prop chip make is the biggest problem. Getting around that will require snaking the ground (or power) in and out around some pins. Otherwise plenty of space exists under the DIP-40 prop to complete the routing. Also, be careful of large loops formed by the power and ground wires. A loop with a large area makes a good magnetic pickup and will feed magnetic noise into the chip.

That should give you a good start,
Marty

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Lunch cures all problems! have you had lunch?

hippy

Lawson said...
The 'X' that the power and ground connections under the Prop chip make is the biggest problem.

Other circuits I've seen only use one or the other sets of Vss/Vdd and previous posts suggest it is not absolutely necessary to connect both. It may be possible to get away with just both Vss's connected across the DIL. It should be possible to put a wire link under the DIL socket if one is being used which ought to simplify creating a cross-over.

Ken Peterson

@deSilva: Thanks for posting your prototype board design. I was going to do something similar and you saved me some design time! [noparse]:)[/noparse]

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


The more I know, the more I know I don't know.· Is this what they call Wisdom?

deSilva

You are welcome! When I think back I tried 3 or 4 versions until I found something with few cuts and few wires

I have a 20 pin male edge connector at the left hand side, so I can e.g. plug the whole thing into a solderless breadboard, using I/O 0 to 15.

As it's really low-cost without any parts, I recklessly solder things on the small lower part

I used it also as a first approach to multi chip connections, just pluging some of them into another striped breadbord with simple 20 pin female connectors, thus forming a 16 bit bus

Leon

It's not a good idea to leave supply pins unconnected, they are there for a reason. They should both be decoupled, as well.

Leon


▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle

deSilva

As they are internally connected (but no one seems to know to which current rating...) I think one cap should do

Phil Pilgrim (PhiPi)

Whether nor not one cap will do depends on how the supply lines are routed. If they consist of short, direct traces running underneath the 40-pin DIP to connect both sides, one cap very close to a Vdd/Vss pin-pair will likely suffice, but two would still be preferred. If the power traces connecting both sides snake around the outside of the chip, though, two caps will very definitely be needed. It's almost never a good idea to use a chip's supposed or actual internal connections as a dodge for good power layout or adequate bypassing. Whenever a chip has two or more sets of power connect pins, you can bet they've been put there for a reason.

-Phil

deSilva

One reason might be: convenience for the user?

But maybe a statement from Parallax is appropriate... I never asked for one, but as this matter came up a few times even during the short time i read this forum ("Are the pins internally connected? Yes, thay are - somehow!") I think an explanation would be appreciated.

It is no use to have an obviously working feature without knowing the limits!

Another matter concerns cap decoupling. I gave this advice many times, and it seems it had some favourable impact, but the feedback was never unambigious..

Parallax does not use it in any of their designs..... Why?

Leon

I asked them, they just said they didn't have any problems with a single capacitor. I always decouple every supply pin, the cost is negligible.

Leon

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle

deSilva

Do you have a reference?

Phil Pilgrim (PhiPi)

deSilva said...
One reason might be: convenience for the user?

LOL. Not likely. By connecting and bypassing all the power/ground pins, you minimize the currents that run across the chip's own power/ground busses (assuming there's an internal connection to begin with), thus keeping voltage differentials and ground bounce to a minimum. If this wasn't a potential issue, other, more interesting uses would've been found for the extra pins.

As Leon points out, bypass caps are cheap. There's really no excuse not to use them abundantly — assuming there's room available.

-Phil

Fred Hawkins

Oldbitcollector,
do you have these bypass caps in your pdf? The PEKit users could use some guidance here, we don't have any in any of our lab schematics. (I think).
Fred

Whelzorn

Wow, thank everyone. I definitely have some things to look at now. I attached my single board layout, and this time it doesn't require any jumpers. Theres probably some massive flaw in the design, but since I asked you all for help, I figured I'd post it to see what the response is.

Phil Pilgrim (PhiPi)

I'm guessing that IC2 is a 5V regulator and that IC3 is 3.3V, correct? The polarized filter caps for these two regulators need to be positioned close to the regulators to be effective. Also, I see no bypass caps. You need at least one 0.1uF ceramic cap positioned near the Prop's Vdd pin. (See the multitude of postings above on this very topic.) Another positioned near the EEPROM is also required. Finally, even though I know you want to avoid jumpers, I would add one anyway to join the two distantly-connected pieces of groundplane underneath the Prop near the middle where they come close together.

I note the part designator for the EEPROM is 24LCS21. That's a really tiny EEPROM (128x8) and not nearly big enough for use with the Prop, unless you're not planning to save programs in it. An I²C EEPROM of at least 32Kx8 (e.g. 24LC256) is what you need. They're not expensive. You will need to ground pins 1-4 (A0-A2 and Vss), BTW.

This may not be an exhaustive list, but those are the things I caught at first glance. You've got a great start! With a little more effort, you'll have a nice board.

-Phil

Addendum: Oops, I did miss something, although one of my suggestions above would fix it — sort of. The ground plane that connects the Prop's pins 9 and 10, is isolated from the rest of your circuit. All your ground plane sections need to be strongly connected. One thing I always look for when laying out a PCB is the feng shui of current flow. Where are the bottlenecks, the roundabouts? How circuitous a path must the supply current take to reach the things it has to power? One of the best things you can do to ensure success is to keep supply routes short and fat.

Post Edited (Phil Pilgrim (PhiPi)) : 10/7/2007 2:05:55 AM GMT

Lawson

Phil Pilgrim (PhiPi) said...
The ground plane that connects the Prop's pins 9 and 10, is isolated from the rest of your circuit.

Not quite, It looks like he's using the reset switch as a jumper. (look at the reset line for the prop-plug header.) Darn clever if you ask me [noparse]:)[/noparse] Personally I'd put the regulators on the far end of the prop away from the eeprom, reset, and prop-plug header. I've got a few ideas on how to snake the power up under the Prop that I'll try out monday.

Later,
Marty

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Lunch cures all problems! have you had lunch?

Whelzorn

Phil: yes, you are correct about the voltage regulators. And the only reason I used such a small EEPROM in the layout was because it was the only one Eagle had in it's libraries that used the 8 pin DIP package with the same pinouts as the 24C512, which is the part I'm actually using. And I was too lazy to create a new component in eagle so that also explains why pins 1 and 2 aren't grounded.

Lawson: haha yeah, I wasn't even going to bother with a reset switch until I realized it would kill two birds with one stone.

thanks for the help!

Beau Schwabe

All,
The Dual set of Power and Ground terminals are not there for convenience. ALL Power and Ground should be connected ... Although a chip may "appear" to work correctly when you just use one set of Power and Ground terminals, it will suffer IR (Current due to Resistance) drop within the chip.

For an example, the resistance per square is 0.078 Ohms.
Suppose that your Power and Ground wires within the chip are 100um wide.
If the chip is 7.28mm on a side, that’s 7280um.

To determine the “number of squares” divide the length of your wire run by the width of your wire.

7280um / 100um = 72.8 squares

To determine the resistance over that length, multiply the number of squares by the resistance per square.

72.8 squares * 0.078 Ohms = 5.7 Ohms per side

....Now, if you only provide power at one point no matter how you slice it, you have to effectively travel along two sides to get Power and Ground to the opposite side of your chip.... that's 11.4 Ohms, but since the Power and Ground form a big loop, that 11.4 Ohms is in Parallel with another (<-- the other two sides) 11.4 Ohms so the resistance is the same as it would be for a side... 5.7 Ohms.
By using BOTH Power and Ground terminal pairs, the effective resistance becomes 2.8 Ohms for EACH. (<-- Power and Ground)

Realistically, the Power and Ground is strapped throughout the chip, so this value is lower ( I measured .9 Ohms on the current Propeller), but the effects are still present when you fail to connect all Power and Ground connections.· IR drop can cause an imbalance in the way a chip may perform, due to excessive heat buildup.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Post Edited (Beau Schwabe (Parallax)) : 10/7/2007 6:01:51 AM GMT

Phil Pilgrim (PhiPi)

Ah, I see the connection through the switch. Hmm. I'd still jumper the two sections together beneath the Prop, though. And I'd also jumper directly from the regulator ground to the center ground plane.

You see, simple DC continuity is not a good measure of even adequate design where power and ground traces are concerned. This is because a digital circuit is not a DC circuit; rather, it's one replete with switching transients, current surges ... all manner of nasties conspiring to defeat one's design. This point cannot be stressed enough! You have to consider where and how the current flows in your circuit, and you must include adequate bypassing. A large filter cap is no substitute for multiple, well-placed bypass caps. They are all necessary!

The attached image reveals the shortest ground return paths from the Propeller back to the regulator in the current design. The jumpering I've suggested will shorten them considerably. Oh, did I mention bypass caps?

-Phil

Phil Pilgrim (PhiPi)

Beau,

I assume (based on the QFP pinout) that the Propeller die has four pairs of bonding points for power and ground. The DIP40 has two pairs of Vdd/Vss pins. How is the bonding done internally in this package? Does each pin bond to two points on the die? If so, how do you keep the bond wires that cross over the die from shorting to something they shouldn't?

-Phil

Fred Hawkins

Whelzorn said...
·I figured I'd post it to see what the response is.

I'd try it with the upper jack·section made·as simple as possible, the prop plug shoved to the right, the eeprom rotated, and the power section and its jacks brought to the bottom. Consider rotating·JP4 to the left edge.
Doing this even as an exercise will help reveal what's worthwhile and what's too tricky by half.
Roughly:

deSilva

So it will become more alike to deSilva's breadboard

hippy

Beau Schwabe (Parallax) said...
The Dual set of Power and Ground terminals are not there for convenience. ALL Power and Ground should be connected

It may be a good idea to update the Hardware Connection diagrams published which don't have all sets of Vdd/Vss connected so they are -

http://www.parallax.com/propeller/media.asp
http://www.parallax.com/dl/docs/prod/prop/PropellerHardwareConnections.pdf
http://www.parallax.com/dl/docs/prod/prop/WebPM-v1.01.pdf

The datasheet ( PropellerDSv0.3.pdf ) does show both sets connected.

Not everyone using the propeller will understand the nuances and implications of "minimal circuit" and of "should". It would be best IMO to purvey the impression that the connections are required rather than not.

Lawson

As promised this is my take on a one layer layout of the core components of a Propeller system. All of the free I/O are easily accessible. The reset line could be routed better, but it'll work.

Marty

P.S. I just threw this together, If anyone spots wiring errors please pipe up.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Lunch cures all problems! have you had lunch?

Whelzorn

wow, that's tiny compared to mine...
I don't see any wiring errors, but you are feeding that 3.3v regulator whatever Vin is (in my case it's 9v) so I would probably damage it here. I always feed the 3.3v reg. the output from the 5v reg. just so I'm not limited to the smaller regulator's input.
Also, is there any need for that little triangle spike thingy on the gnd trace near the xtal?

Thank you though, I may base mine on this one if it's alright.

Harley

Some questions:
1. How do you plan to mount the Prop (socket, solder to pcb, etc.)?
2. Where are any 'header' pads to connect to the I/O pins? Hopefully you don't plan to solder wires to the Prop pins.
3. If the Prop is socketed, and you need to remove the Prop for any reason, looks like you are boxed in on both ends by components.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Harley Shanko

Lawson

Both ways of feeding the regulators has it's merits. Cascading spreads out the heat, while direct connection allows the supply to sag farther before the Prop will brown out. Heh, no the little "spike thingy" probably isn't needed. A little chunk of ground plane over there wouldn't hurt though. Keeping the power and ground wires closely paired is important to keep noise to a minimum.

Hm... If you're willing to hide some resistors under the Prop's socket, I made another layout. This one gives free access to ALL the I/O pins.

Feel free to use it, Kind'a hard for me to stop you once it's posted on a public forum

Marty

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Lunch cures all problems! have you had lunch?