kicad-developers team mailing list archive

[Vesa Solonen <vsolonen@...>]

The following has almost nothing to do with fonts anymore, be warned ;)

> R200 is already a pretty recent GPU. A LOT of machines only use Intels (and BTW we have a lot of P3 and
> Celeron class boxes around... still at <1GHz speed).

There is a serious flaw in the system. I've seen so many pallets of P4 
> 2GHz machines going to "recycling" that it hurts when I hear that somene 
really would have needed them. It's a pity that there is no reuse 
directive in effect that would really force reuse before element level 
recycling for younger than five years machines. Mostly it is about 
companies which push new machines every three years and want to mill 
everything to dust in the name of data security. What a load of bs... If 
those companies say they _want_ safe reuse, they'll get it. Naturally it 
all boils down to is, is it economically viable and who is going to say 
what you do with your property.

> The 'dirty trick' I talked about is EXACTLY a frame buffer object (implemented before frame buffer objects
> :P)

:P I'm a bit lagging in my skills here... I used vertex buffers for 
speed testing though. BTW, have ever looked at Peter Clifton's branches of 
gschem and geda pcb? The problems being solved are essentially the same.

> Indeed all Apple machines have a later than Radeon 9600 class GPU :D

http://developer.apple.com/graphicsimaging/opengl/capabilities/
It seems there are also some older ones too...

> I'd say that without PCB making machines (I mean the fully automatic ones) the outsourcing give you 'perfect'
> results (if it isn't good they have to redo it for free :D) in at most 4 days, and it cost a little more than
> my hourly pay :P add the cost of the HM drilling bits, and you spend LESS outsourcing even the prototyping!

Very true.

> And, for hobby use, we can get a fully processed 2 side eurocard (160x100, 35um copper, 2 sides, 2 solder
> masks, component silk) for about 50 EUR.

Some friends have used Olimex, but it was damn difficult to get a board 
_without_ ENIG pads. If you can live with ENIG, they may be the cheapest 
around. With hobby users I mostly meant guys who do the stuff because they 
can - it has nothing to do with economics :P

> For me RF is >10 MHz, yet I had some problems one times with the coils positioning (mutual flux:() in a
> 1.6MHz power oscillator for ultrasonics (well, it's a 50W oscillator, flux IS NOT trascurable :D).

I've played with some ready made 1.2 MHz ultrasonic fog makers. They make 
a nice demo with plastic bucket full of water. When you turn the beam 
towards the plastic edge. Soon there is smoke coming off the lossy plastic 
:D

> The 300mil board is a somewhat simple appliance... it has a 650W heater running at 24V for safety :D It was a
> funny board to design, in 160x90 it has 230VAC, 24VAC, 30VDC, 5VDC (regulated by a SMPS) AND the millivolt
> signal from a termocouple :D just imagine the stuff riding the board when the heater switches (hint: is a
> wound wire heater -> HIGHLY inductive :D)

Sounds like a challenge. There must be a helluva snubber on the relay if 
that's what controls the heater? BTW, if you have any chance to affect the 
winding of the heater, get it with bifilar windings and wire those in 
opposite directions (series better) -> inductance cancelled :) Some 
synchronization to zero crossings may also help if relay switching times 
are well known and stable.

Best regards,

-Vesa