Altium vs Orcad

messagenews: snipped-for-privacy@e9g2000prf.googlegroups.com...

I meant what kind of future designs are you looking at doing. $10K package like Altium is overkill for simple 2 layer stuff.

When you generate Gerber files for manufacturing, you will get top and bottom layers as seperate gerber files, or you can tell the software to only generate the bottom gerber layer. Just send the bottom layer file to the manufacturer and bingo, you have a single sided PCB.

You shouldn't be using an autorouter at all! Especially for that incredibly simple circuit.

Dave.

Do they automatically route them or something? or convert the vias into jumpers? I don't understand what th gerber file contains? If its info for a routed board and if I routed it with vias then I can't see how they can fix that?

huh? Why can't I? And the circuit is not so simple? Remember, its 8x each mosfet and not just one. Its got almost 100 nets involved. Doesn't seem all that simple to me(It might be simple in the grand view of things but if its so simple then autoroute shouldn't ahve any problem at all with it).

When all you are doing is a single layer board, how hard can it be to hand route it?

It would seem that auto routing it would even be a poor choice.

So does little brother Easy-PC.

[ stuff about autorouting, autoplacing, auto-pinswapping ]

Hi Jon,

I don't know your motives for using an autorouter, but the general consensus among pros and amateurs (but especially pros) is that autorouting isn't worth the trouble.

I, personally, find the results aesthetically abominable, but then the only autorouter I have experience with is Eagle's, and that one is really crappy. I've heard good things about the Bartels router.

In general, autorouting is OK for simple boards with mainly non-critical signals, but those you might route by hand just as quickly. Complicated and critical boards need to be routed by someone who understands the circuit, so an autorouter is out. Upshot: Why bother with autorouting at all?

robert

Something as primitive as that can be hand-placed and hand-routed in 15 minutes and it'll even look good. With repetitive elements it's important to figure out once how to lay out the power buses, from then on it's just fun.

Here's a 5x5 relay cross switch I'm working on. Routing took a single afternoon because essentially it's N times the same thing. (I've done boards less than half that size that took me a week.) An autorouter would have made a complete mess of it because it would have tackled the routes one-by-one, completely oblivious to the big picture. It also wouldn't have done the guard traces for the MCX connector's shields, and it wouldn't have done such a nice job of routing the DB25 connector.

Knowing me I'll spend another day tweaking the layout and especially the silk screen layer.

What you should do is hand-route it as a two-layer board, taking care to make the traces on the bottom layer (=jumpers) straight. To manufacture the PCB, you just use the copper data (=gerber file) of the top layer and do the jumpers by hand.

Depending on your design rules you may or may not be able to route between 0603 pads, which will greatly influence the number of jumpers needed.

--Daniel

Well, the boards I'm working are simple and I figure it would be much easier just have it do it. Instead of a few hours routing it(potentially) it takes a few seconds. Of course thats if I was working with two layers. Because its one layer its impossible to route without jumpers(and I'd have the same issue manually routing but at least I could manually add jumpers quite easily)

I guess I'm going to have to do it by hand or modify an autoroute since I can't seem to figure out how to efficiently use jumpers(having to use 0-ohm resistors seems like a hack and is probably just as much trouble)

Your are using more than one layer. Try do that with a single layer and no vias... then is it easy?

I'll probably need to change it, right now I can but I'd rather have larger trace widths. Its a mess the way it is and I doubt anyone could route it manually on a single sided board without using jumpers(cause many of the components can be easily trapped and not be able to connect to to any other components(That is the problem with autoroute and no jumpers), at least in

15 mins. That said, I doubt one would need any more than 8 jumpers.

I'll try and route it manually and see what happens. I guess I just expected more of auto routing.

Jon

messagenews: snipped-for-privacy@i13g2000prf.googlegroups.com...

A gerber file is the output file you send to the manufacturer (you can also just send the Altium/Orcad file etc). Google will tell you all about gerbers I'm sure.

*ANY* PCB program can be used for single layer design by using the top layer for the links. Then all you do is manufacture the *bottom layer* only, simple. Nothing needs to be "fixed" or "converted", your vias (or pads) become the holes for the links.

You can, but you shouldn't! You have fallen into the classic beginners trap. Autorouters are not a tool for beginners or simple circuits, they generally produce

*horrible* results, and the simpler the board, the more horrible a result you tend to get. Especially for "single sided" boards!

Autorouters are designed to be used by very experienced PCB designers who know how to set up the rules correctly and can use it in a controlled manner. A beginner using an autorouter is asking for a disaster.

Do yourself a *big* favour, take some pride in your PCB layout and learn to route everything manually. Autorouters are a very *bad* idea except in specific cases, especially for you and your particular circuit.

Wow, 100 nets! Any half decent PCB designer can manually route 100 nets in probably

10 minutes. A beginner might take maybe 30 minutes?, an hour?, two hours? 100 nets is nothing.

What's more, with a "multi-channel" design like you have it's even easier. You just route one channel and then copy and paste the other channels. Then you connect the I/O and power, it's really simple.

No it won't, but it will produce a horrible result. Try it and post your result here for comment, and you'll see what I mean.

As for your original question of Altium vs Orcad. These are high end packages that cost a lot of money and appear to be gross overkill for what you need. Unless you want to do big complex boards and designs, any cheap basic PCB package will do what you want. Is this for work or home? What type and complexity of boards will you be working on the future?

Dave.

I know what a gerber(well, basically) is but I don't see how leaving off the bottom layer helps in preventing vias without causing the whole pcb to be worthless. I have no experience with sending in a board to be manufactured though and have only etched my own single sided boards.

You mean that I have to connect the "links" myself or they jump them at manufactoring? Obviously if there are throughholes with the vias you can link them using whatever method you want but htey have to be linked because that is the routing for the circuit. Running a wire or jumper instead of a copper trace is the exact same thing as far as its still bottom layer work and requires holes. I'm tired of drilling holes on single layer and that is why I'm trying to smt work but I thought I could get away with no holes with it and just a few jumpers.

The results are fine if its got more than one layer. The result is impossible for single sided. The problem is, is that the auto router isn't intelligent enough to insert jumpers to make it simple enough.

I figure the autorouter is smarter than me but maybe not? Most of the pcb's I've done are very small and actually could be done with just a few jumpers but were all TH. (so jumping was pretty easy)

Ok, can you route it in 10 mins on single sided(no vias)? I don't see how you can do it very well without adding a lot of jumpers. Sure I can route it in probably 20 mins(as I just did) but I still cannot route some nets because they require jumping and jumpers won't fit near the components because they are tightly compacted. Sure its an easy think of every component has 1in^2 to itself.

I'm not saying its necessarily hard but you guys seem to make it sound like its a walk in the park. I agree with vias it is and if you convert those vias to jumpers then its the same thing as using a bottom layer(essentially because it requires holes.. in that case I might was well just get a 2 sided board and etc both sides and drill the holes).

lol. I don't think so. I think you need to actually try the circuit I did. Show me if its so easy ;) (remember, single sided) When I route it every component gets in the way of some other component. If you route around it then it blocks 3 more. I don't think its as easy as you think.

Basically on single sided, when you supply power to each channel it cuts off routes for every output of the device(or input) so you have to jump for each one. Then supplying the ground it cuts off the input(or output) and you have to have around 16 jumpers. Maybe it can be optimized for less but I don't think it can be done well in 10 mins unless you through jumpers all over the place. I'm trying to minimize jumpers because its smt and I'll probably have trouble soldering the components specially when traces go through pins. I also want it to look decent without 100 jumpers all over the place.

So if you mean that "It can be routed in 10 mins" then I'm sure it can... but how well can it be done? Can you do it in 10 mins without any jumpers and minimize space? I'd like to see it. Can you do it with just 3 jumpers and no TH's/vias?

The result isn't horrible but incomplete(well, its messy). With 2 layers its not so bad.

Its for home but I eventually would like to make more complicated stuff. I've used the PCBExpress software and routed my own powersupply a while back. It wasn't difficult but there was no need for so many jumpers as in this case.

Extremely simple: Just route them as straight tracks on the other side of the board. When you get the finished (one-sided) board, it'll have the pads, copper, drill and all, into which you can solder your jumper. No need to include jumpers or 0-ohm-Rs in the schematic.

Is it possible you've never done PCB layout before?

robert

What? There is no other side. ITS SINGLE LAYER!! sheesh. How many times do I have to say this. So what happens? When you do it for two layers to get the "virtual jumpers", guess what? The end up going from one side of the board to the other. So do you expect me to run wires like a spider web all over the board?

Yes I have but its always been single layer, manual routing, and smaller components. I did a power supply that had 4 linear regulators, two variable and two fixed on a 2inx2in board. It was easy because all "channels" had only 3 "ports". I had like 3 jumpers. In this project every channel has 5 ports and about 15 components gets boxed in. Sure I can use vias but then I have to jump from one side of the board to the other cause the autorouter doesn't minimize the length of the traces on the bottom side(mayber there is a rule for that though).

Now I can easily route the board and put in jumpers manually but its not efficient and not the best way. I want to learn to the best way(or a better way) for future projects too. Your method isn't very good because the board its end up a lot of jumpers that are not needed and to long.

I have to admit that the routing has to use jumpers because of the design. That is not the problem. The problem is to minimize the jumpers. Its quite easy just to use jumpers haphazardly and get a route. I can do that in about 10 mins. (even minimize the jumper length in the process) But that doesn't mean its the best way to do it. (of course maybe I should just do it anyways which I actually am because that seems to be the only way. I end up using 0-ohm resistors because it looks nicer than a much of jumper wires cluttering up the board)

It won't be worthless. The vias are what you solder your jumpers into.

That's going to be difficult if you can't do 8/8 rules because you'll want to route between SOIC and 0603 pins.

You could jack up the "cost" for vias and for routing and bends on the opposite side, then you'll end up with few, short, straight jumpers.

What is indeed a walk in the park is the decision of whether you can do truly via-(jumper-)less design based on whatever your design rules are. When your design rules allow 2 traces underneath 0603 parts or between SOIC pins it might be doable. If you can't do either, it's impossible.

Yes, you might as well. One-sided layout is a lot harder (if not impossible) using SMD. Your SMD board might end up having lots less holes than the corresponding through-hole design, but it'll still have plenty.

Welcome to the reality of PCB routing. The problem is not a lack of intelligence of you or the autorouter, the problem is that what you're trying to do is simply impossible single-sided.

That's just topology, and no amount of computing power will get around that. To get to an island you'll have to use a boat (or an airplane, or swim, or dig a tunnel) and that's that. 16 jumpers would have been my estimate as well.

It won't take as much as 10 minutes to figure out that you can't reduce the number of jumpers.

Unless you can get plenty of traces between 0603 and SOT-23 pads you'll have to live with the jumpers.

No. Who could?

Better get into double-sided PCBs.

robert

No, then it'd be simply impossible. I just posted the layout to demonatrate that a seemingly complex PCB is routed quite easily if there are plenty of repetitive elements.

--Daniel

I put a rule that said minimize vias and I guess it did but it still made the length minimal so a simple small jumper would work(even when it could).

No, its not doable in those cases either, in general, It becomes a maze essentially and its easy to get into a dead end(so maybe there is even a mathematical solution but chances are the autorouter will find it will be

0).

My point is not to find any jumperless solutions cause there isn't any in this case. But to minimize them as I have said repeatedly. I think that is the difference between what you guys think is easy and what in reality is a hard problem.

I suppose you mean jumperless routing? Yes, I know. Minimize jumpers? Not impossible at all. Difficult? Maybe.

Well, I don't have any real experience with it and for the smaller pcb's I just used jumper where I had to but I did minimize them to some degree. I didn't just put a jumper when I needed it but I thought about the best place to put it.

I don't agree. It depends on the placing and orientation of the components. Changing them can have a drastic effect. Now maybe there is some type of theorem which states the minimial number of jumpers need but I doubt it.

I guess the problem is that I shouldn't be worrying about perfection to much here and just get it done. Although the difference between 16 and 8 jumpers a big difference to me.

hehe, yes yes yes... I know... but how many? Maybe tons of jumpers are normal on single sided layouts of any complexity?

Don't know. Thats what I got from you and David. Acting like its easy to do it. (normally when you do something you don't try and do it half ass so I take it when David said do it in 10 mins it means he could do it very efficiently(very few jumpers, maximal density, etc...)).

It seems like a lot easier. But for such a small project here I don't see any need. I don't have any double sided boards at the moment so one I use up my single sided I'll probably start doing that or I'll start sending them off for manufacturing.

But this project is pretty small and its a good learning experience and I need to do some more etching myself since I haven't done any in a while.

I guess I might just be epecting to much. I thought you guys were implying that it was very easy to do an optimal route in a few mins and I was having trouble believing that. I guess though its my nature to want to make something the best I can but I'm going to go ahead and just try to route it and worry about optimization some other time when it actually matters.

Jon

yes, but you used two layers. I never said that using two or more layers was difficult. What I am finding difficult is to use single layer, maximize density, and minimize jumper and jumper length(well, actually I guess a fixed length would be ok). I was under the impression that you guys were saying this is easy. Of course now I get the impression that it is not and so I don't need to worry about it to much(this project isn't all that critical so no big deal). Of course this started from the autorouting approach and I thought that maybe it could find the optimal solution.... but seems auto routing is not as good as I thought it would be.

I'm just going to go ahead and route it and put jumpers in where they need to go and not worry about optimizing it. Shouldn't take to long to do that.

[...]

There is a useful compromise that I use when I do a "home-made" SMT board.

Route the board "single sided" but with a ground plane on the opposite side (with vias for ground points). Then make the board with double sided material, but only etch the "tracks" side. Drill the ground connections and use through-pins soldered both sides (or bits of resistor legs etc) to connect the ground vias to the ground plane.

- no alignment issues, drilling not critical

- very good quality signals and power bypassing due to ground plane

- the "ground" signal usually has the most nodes, so removing it from the tracking layer frees up lots of routes, often enough to be able to complete the rest of the design single sided.

- You *can* use through hole parts too, but you then need to isolate the "non-ground" pins. This can be done by countersinking on the groundplane side with a suitable drill bit, and you need to solder on the tracking side.

--

John Devereux

You shouldn't be using an autorouter for single-sided, period.