Gerbv

FWIW, I only ever send the drill ident.gbr map and never checked the .drl file.

What I see is similar to you: that gerbv seems to be interpreting values at 1/10 of what they should be. For example:

In the original .drl file a coordinate such as X07904Y02588 is being interpreted as X = 790.4mil Y = 258.8mil instead of 7904mil & 2588mil resp.

If I manually edit the drl file to remove the leading zeros and add a trailing one then it comes good.

This is with Gerbv v2.6A. Like I said: FWIW.

w drill locations in a drill file look out of location. I seem to have nar rowed the problem to the use of trailing zero suppression. I can verify th e drill file (which contains only line feeds it appears, so not so easy to view) has a few entries with the trailing zero suppressed. In Gerbv these are all being interpreted as if zero suppression were off. In fact, it tel ls me when it translates automatically, it is using no zero suppression. I f I manually turn on trailing zero suppression it so munges the drill file that no drill show up on the screen at all and if I click the zoom to exten ts button it takes me far off the board and shows me a small area with no v isible features.

ro suppression and see if that fixes the problem at my end.

resp.

What program is generating your file?

I'm also using v2.6a. Gerbv seems to have manual controls but defaults to automatic. I tried setting trailing zero suppresion and it didn't work. Y ou seem to be using both trailing and leading zero suppresion which won't w ork anywhere. If you lop off a zero at the beginning and one at the end, w ho could you figure out where the zeros were lopped off? In your case I wo uld have the generating software not trailing zero suppress (it doesn't sav e so much in the file anyway). Or if there is nothing in the 0.1 mil digit , you can tell Gerbv there are only 3 bits past the decimal point. Don't k now if this will work since this feature seems to be less than tested.

In my case I found I needed a command, INCH,LZ which says work in inches an d leading zeros are included which means missing digits are assumed to be t railing zeros.

Rick C.

I would try the newer version: 2.6.2

Bye Jack

I'm using DesignSpark PCB 8.1.1 to generate the Gerbers.

This sort of thing isn't bread and butter stuff for me so that's one reason I haven't paid much (any!) attention to the .drl drawing particularly. I just used a cheapo PCB fab found at random on the 'net and they didn't require/ask for .drl files.

Sorry I can't be of more help.

Yes. I use it to laser cut mechanical templates before fab.

I've used (I think) the default settings in Kicad and drill files and gerbers match, so I haven't tried fiddling with settings in gerbv. The board houses (including entirely-automated-nobody-speaks-English board houses in China) have been fine with them. Also Altium, but I don't remember whether I changed the settings.

It may be that Kicad is outputting a different format by default to not trip into this. Can you change your output settings? Theo

"Mechanical template" is a space model?

And who do you use for the laser cutting?

Tim

out

if

kes

rip

I'm not generating the file. The assembly house has redone the array goin g from 4 up to 5 up to get better utilization of the full PWB panel size. In the process they rotated the gerbers which makes it a *LOT* harder to ve rify the coordinates of any given feature. Instead of a simple translation by X,Y now you have to swap the coordinate, add to the offset X and subtra ct from the offset Y. I can't do that in my head so easily. Couple that w ith their use of LF rather than LF/CR and that is why it was rather hard to figure out what was wrong.

The problem with the current state of PCB design documentation is that it i s very similar to SPI ports. The documentation covers one original impleme ntation and there are many permutations on that which are not very compatib le. Excellon files in particular are a real PITA. IPC had a good idea, bu t the industry was too lazy and complacent to make a change that would requ ire work. They'd rather continue with the confusion because it is only a c onstant low level of irritation.

Rick C.

Yes. A piece of cardboard of the same thickness, shape and with the same holes as the PCB, to 1:1 scale.

Ourselves. We have a GCC Laserpro Spirit 40W laser cutter.

The flow is: Kicad export to gerber Gerber opened in gerbv only display the layers to be cut. The GCC allows one raster layer and multiple vector layers - my usual flow is raster for images to etch like tracks, one layer for holes, another layer for edge cuts. Give each layer a distinct colour (black=raster, red=holes, green=edge) Export as EPS, load into Adobe Illustrator CS3 (the laser cutter only plays nicely with a few programs) Tidy up the EPS to keep the laser cutter happy (line thickness = 0.01pt, make sure objects have the right line colours) 'Print' to the laser cutter, setting appropriate powers for the different colours. The red (holes) layer is cut before the green (edge) layer.

It works quite well, and is handy for debugging mechanical issues before going to fab.

Theo

Sounds great. And only "from $21,995" :) Until I'm a millionaire I'll keep getting semi-routed double sided boards from China...

Tim

Well, unless your boards are 1m x 60cm x 1 inch thick, I think our laser cutter might be a /little/ overkill :)

There are small lower power laser cutters that do a few watts for a few hundred dollars, which might be suitable for cardboard. This sort of thing, though it might need a higher power version, and different software:

Theo

Somewhat remotely related but while on a mechanical wave perhaps someone would weigh in as to how to weld PT100 leads to teflon wire. Soldering is not an option, they work around 300C and if they lose feedback etc. they could get to 400. Well at 400 the joint will not be the only victim but still. Any experiences?

Dimiter

====================================================== Dimiter Popoff, TGI

Do search on "pulse arc welder". Used for very thin wire welding, thermocouples etc...

Chris

Thanks, I think I get it how it is done now. Not sure I'll buy such a welder though, I only need a few welded for now.

I had a similar problem a year or so ago when rebuilding an old HP double oven frequency standard, where the heater element was probably nichrome, and which needed to be bonded to ptfe tinned copper wire. Thought the pulse arc welders very good, but as you imply, too expensive for a one of job. Eventually used a power supply to charge a large electrolytic and a tungsten probe in series with an iron cored choke to generate an arc on break. A few tries at various voltage levels got the job done as per a proper pulse arc welder...

Chris

Oh thanks, this is very encouraging. I think I can manage it somehow. The best part of it is that you did not use any inert gas? I have a handful of 4700 uF caps - some fast, some not - in a drawer here, might just do the job... I'll experiment around next few days.

Please share any details you might remember - voltage, capacitance etc., I mean ballpark range.

Dimiter

It was all a bit of a hack, but it was using the stored energy in the choke, back emf, to generate the arc on break. The choke was an old C core item, about 4" square, 5 amp rating, fwir. Wires stripped off and tightly twisted together, trimmed level at the ends, then copper clip to the twists and a tungsten scriber for the electrode at the tip.

Tried various approaches, but thinking back, used a psu in current limit mode to charge the inductor, then arc on break. It's possible to get a neat rounded weld without gas, but it is a bit hit and miss. As this was to be refoamed up, used a magnifier to make sure there was a good weld between the two metals.

If you have one of the old 5v 100amp or more computer psus, you could try that with home made copper electrodes to crush spot weld the twisted wires at the ends, as an alternative method. Not everyone has a junk box with iron cored chokes these days, though a large 50Hz low voltage transformer secondary winding might get the job done. Just got to experiment :-)...

Chris

Thanks Chris, I'll try various hacks once I know it can be done. I'll figure out something. I don't have a choke similar enough to yours but I'll find something suitable somehow. The alternative method you suggest I could try using the transformer of an old microwave, replacing the HV secondary with 1-2 windings of thick cable, I know people do it - not sure how it will work on copper wires but well, it should melt them allright.

Dimiter

That might work well. Have a spotwelder in the workshop here, but a bit to clunky for fine wire work. I'm still considering buying one of the commercial pulse arc welders, but would look at those made for jewelery work, rather than those for thermocouples, where you put the twisted ends into a hole in the front panel. Afaics, the jewellery use a tungsten electrode, where the tip is withdrawn from the workpiece on contact. Then a hv supply starts an arc across the subsequent small gap, conducting and forms a plasma from the discharge of a capacitor. Seems a clever idea to me and could be useful in the lab here for other purposes as well. More fun to build one though, given the time :-).

I did tried soldering, as the oven temp is only ~60 c, but failed miserably, even using acid flux...

Chris