Making PCB without a computer

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Since I do not plan to mass produce this PCB, I would like to avoid the hassle and great inconvenience of having to use a computer to make my amplifier PCB. Thirty years ago, around the end of 1980's and beginning of 1990's, making a PCB for a project was easy, fast and accurate enough for most projects. In those days, I used PCB transfers to protect the copper during the etching process. The transfers consisted of various lines of different widths, circles with a small hole in the middle and IC pin layouts. I used these to make PCBs in a matter of 2 hours.

When I searched the internet with google, I was shocked to learn these transfers may not be in production. Most google results are links to questionable toner transfer methods. My aim is to avoid using a computer at all costs as it hinders my work: I do not want to end up struggling with software and printer drivers just to make one PCB.
 
I have done it with a Sharpie (fine ink pen).

Air-brush artists use friskets: films and brush-on liquids which they cut with an X-acto knife and peel-away where they wish to spray.
Frisket - Wikipedia
Using a frisket is a very good idea. The frisket's material can be chosen so that it allows precise cutting. I have also in mind of using fine correcting pen.

So, I need more patience, but it can be done which is what counts.

A well deserved 'thanks' go to all those who contributed to this thread.
 
When I was a teenager back in the 80's, I had developed a similar technique. I was placing a sticker on the copper side, then drawing the traces and then removing the blanc sections with the knife. The pcb in the photo was made that way... and that age.
 

Attachments

  • P1020326.jpg
    P1020326.jpg
    630.5 KB · Views: 586
I used to draw PCB layouts with black felt tip pen on paper with a blue 1 mm grid printed on it, then copied it on overhead projector sheets and used a flat piece of glass and a TL lamp to transfer it to a PCB with photoresist layer.

I even managed to get a PCB with a 100-pin TQFP package with 0.5 mm pitch to work that way, using a 0.5 mm fineliner to draw the wires and shrinking everything by 50 % when I copied it on the overhead projector sheet. (I messed up two PCBs before I had one that was OK, but at least it worked eventually.)
 
Moderator
Joined 2011
When I searched the internet with google, I was shocked to learn these transfers may not be in production.

The manual stuff has been gone for many years, and good riddance. No one etches boards any more, you'd have to do it in your sink.

Some use a pcb milling machine, but these require a computer board file, the same as the board house requires. And no plated through holes.
 
Last edited:
Member
Joined 2016
Paid Member
I can’t be bother with etching myself - have not had great successs (operator malfunction?)

So for small boards I’ve been milling them manually. I use a 15*degree engraving bit in my rotary tool with the drill press set to a depth that puts the tip just through the copper.
 

Attachments

  • 9B029DF4-7A9C-4CE9-BA51-19D22995D1AC.jpeg
    9B029DF4-7A9C-4CE9-BA51-19D22995D1AC.jpeg
    703.3 KB · Views: 570
I remember in the 1980's our draughtsman designing pcb's that way.

The world has moved on greatly since then. No kidding, there are hundreds of pcb design packages for a pc. From very simple to highly complex packages.

I remember buying EasyPC in around 1990 and getting tripped up by all the bugs in it !

The cost of pcb's is now rock bottom. I just bought 10 off 100mm by 100m pcb's for $2 with jlcpcb. OK the p+p was a bit but I wanted them quickly. Slow post is just a few $.

I would recommend biting the bullet and getting stuck into a PC PCBCAD package. I often find I can pinch bits from different schematics and merge them for a new pcb saving me hours of work.

Autorouters are a big no no. Although, I run mine and then go in and fix star grounding etc.
 
Using software implies:
a) forced operating system updates that take very long to complete
b) probably having to use a heavyweight antivirus daemon eating memory and CPU resources
c) having to endure software implementation errors that result in a CAD program dying unexpectedly losing information in the process.
d) having to install a priter driver suitable for PCB printing. This means, being at the mercy of a printer's manufacturer, as most people are incapable of writing a printer driver themselves, I assume. Besides, writing a printer driver requires hardware knowledge, that a manufacturer is probably very unwilling to divulge for competitiveness reasons.
d) climbing a very slippery and steep learning curve that probably most people of my age are unable to endure.

I, humbly think, choice is not intrinsically bad. Manual methods should have their place even in a modern world.
 
Disabled Account
Joined 2010
Reminds me of the old days when I explored my first text maker (wordstar on C/PM).
There was a steep learning curve, but I never went back to the mechanical typewriter since.
Same with PCBs - I drew and glued many PCBs manually until the end of the 80ties.
Today I design these on my laptop using a Linux-System that gives much less hassle than windows.
No need for a printer driver or even a printer - I send the gerber files to chinese fabs and receive ready pcbs within 10days at a ridiculous price.

Admittendly all this had its learning curve.

Manually work will get harder as there is no support from industry any more.
So it's on you to choose your poison - this is DIY
 
Last edited:
Member
Joined 2002
Paid Member
Using software implies:
a) forced operating system updates that take very long to complete
b) probably having to use a heavyweight antivirus daemon eating memory and CPU resources
c) having to endure software implementation errors that result in a CAD program dying unexpectedly losing information in the process.
d) having to install a priter driver suitable for PCB printing. This means, being at the mercy of a printer's manufacturer, as most people are incapable of writing a printer driver themselves, I assume. Besides, writing a printer driver requires hardware knowledge, that a manufacturer is probably very unwilling to divulge for competitiveness reasons.
d) climbing a very slippery and steep learning curve that probably most people of my age are unable to endure.

I, humbly think, choice is not intrinsically bad. Manual methods should have their place even in a modern world.

hi edbarx,

I do agree with d).

Your choice is your choice so it can't be bad, it suits your requirements. Unfortunately, when you decide to be Robinson Crusoe there are not many people left to help. :D

I think you could also just do the circuit on strip board. That's pretty normal for one offs and prototypes.

There are also some guys on this forum that can do fantastic circuit boards in a couple hours. If you show your schematic one of them may take the bait.

regards

Reminds me of the old days when I explored my first text maker (wordstar on C/PM).
There was a steep learning curve, but I never went back to the mechanical typewriter since.
Same with PCBs - I drew and glued many PCBs manually until the end of the 80ties.
Today I design these on my laptop using a Linux-System that gives much less hassle than windows.
No need for a printer driver or even a printer - I send the gerber files to chinese fabs and receive ready pcbs within 10days at a ridiculous price.

Admittendly all this had its learning curve.

Manually work will get harder as there is no support from industry any more.
So it's on you to choose your poison - this is DIY

hi voltwide,

And Multiplan on CP/M. :)

Started off as a Unix fanboy, now I am OS neutral. They all work fine for me!

Agree, learning EAGLE or KiCAD for a one off PCB, is not feasible for mere mortals.

regards
 
Reminds me of the old days when I explored my first text maker (wordstar on C/PM). <snip>
I too use Linux, but Linux, is a wild untamed beast, as there are many distributions around ,and PCB making requires at least some form of graphics support. I assume, you create your PCBs using a graphical interface, although I may be wrong.

Which Linux application do you use and under which distribution? I am normally using Debian, or its new spin, Devuan, and install using debootstrap, as installers seem to put too many rods into my wheels.
 
Using software implies:
a) forced operating system updates that take very long to complete
b) probably having to use a heavyweight antivirus daemon eating memory and CPU resources
c) having to endure software implementation errors that result in a CAD program dying unexpectedly losing information in the process.
d) having to install a priter driver suitable for PCB printing. This means, being at the mercy of a printer's manufacturer, as most people are incapable of writing a printer driver themselves, I assume. Besides, writing a printer driver requires hardware knowledge, that a manufacturer is probably very unwilling to divulge for competitiveness reasons.
d) climbing a very slippery and steep learning curve that probably most people of my age are unable to endure.

I, humbly think, choice is not intrinsically bad. Manual methods should have their place even in a modern world.

a/ Win10 now updates in the background.
b/ Using a C++ cad package gives you very fast graphics and functions.
c/ My software has no bug list, if there is a bug I fix it right away.
d/ I just use a standard printer driver and have had no problems.
e/ Depends which package you buy, some are very simple to use.

Your just putting up a brick wall to new ideas.
I have sold 3,000 copies of my software and had very few problems even with older people.

Error checking of a good cad package is vital.
Lets face it one error and the pcb will have to be butchered to fix it.
 
Disabled Account
Joined 2010
I too use Linux, but Linux, is a wild untamed beast, as there are many distributions around ,and PCB making requires at least some form of graphics support. I assume, you create your PCBs using a graphical interface, although I may be wrong.

Which Linux application do you use and under which distribution? I am normally using Debian, or its new spin, Devuan, and install using debootstrap, as installers seem to put too many rods into my wheels.
I am using linux mint, a mainstream debian based distro for some years now on Lenovo X220 at the moment.

Using the wine windows emulator I made my pcbs with Eagle and meanwhile migrated to KiCAD.
Spice Emulation with LTSpice also works that way since ages.
FreeCAD, GerbView work as well.

There are no "special tricks" in my system, so I never cared about special graphic drivers or programs.
All in all I am a great fan of apt-get;)
 
Last edited:
Using software implies:
.............
I, humbly think, choice is not intrinsically bad. Manual methods should have their place even in a modern world.
Ok, now I´m drawing a picture of your problem.

You have not one but two problems:

1) creating the artwork itself.
You will have to draw it, black on white, one way or the other.
You may use a computer, not necessarily PCB software, you might even use any Graphic package (I know people using AutoCad or Corel Draw, go figure) and there is a very clever adaptation of Windows Paint, go figure.

Or you might draw it by hand, China Ink on regular or translucid paper, the real old way.

But then:
2) you have to transfer it to an etch resist layer on your copperclad board.

Won´t mention etching and drilling because you seem comitted to do that, whatever method you choose.

Direct transfer-lettering tracks and pads to copper surface seems to be your preferred choice, and takes care of 1) and 2) .... problem is that it does not seem to be availabke any more.

2 questions:

3) would you consider transferring an ink and paper (computer or hand drawn) design to light sensitive copper surface?
Yes/No

4) how complex is your project?
Please post the schematic, so far we have no clue about your project complexity.

IF not too complex/lots of long narrow tracks all over the place, you may even hand draw it on copper using an etch resistant felt marker, I have done it countless times for relatively simple one-offs.

But we can´t answer with precision not knowing the circuit you want.

5) and of course the above suggestion of using Veroboard or some other Stripboard is very valid and avoids etching and (most) drilling, go figure.

Just board size is not as compact as it could using other methods.

Please post your schematic :)
 
in the late 70s and early 80s (i.e. before "I" could afford a PC :) ), I used Statdtler Lumocolor #317 black pens instead of Sharpie markers.
Never had to retrace anything - those Lumocolor pens put down some thick, smooth ink.
I could crank out simple boards in under an hour.
Sometime I miss those days ...
mlloyd1
 
1) creating the artwork itself.
You will have to draw it, black on white, one way or the other.
You may use a computer, not necessarily PCB software, you might even use any Graphic package (I know people using AutoCad or Corel Draw, go figure) and there is a very clever adaptation of Windows Paint, go figure.
I am drawing it by hand on a piece of white paper.

2) you have to transfer it to an etch resist layer on your copperclad board.

Won´t mention etching and drilling because you seem comitted to do that, whatever method you choose.

Direct transfer-lettering tracks and pads to copper surface seems to be your preferred choice, and takes care of 1) and 2) .... problem is that it does not seem to be availabke any more.

2 questions:

3) would you consider transferring an ink and paper (computer or hand drawn) design to light sensitive copper surface?
Yes/No
I will not use photography. I will use primitive brute force drawing on the copper surface itself.

4) how complex is your project?
Please post the schematic, so far we have no clue about your project complexity.
I am planning to use modules for the separate 3 stages of the amplifier. This will allow me more flexibility. Every module will have its own PCB that will be mounted adjacent to each other on an aluminium angle bar.

The circuit will be based on the following:
http://circuitscheme.com/wp-content/uploads/2015/04/2000W-Power-Amplifier-Circuit-Diagram.jpg
2000W Class AB Power Amplifier - Circuit Schematic

I will make several changes to this circuit. These are:
a) a lower voltage for biasing the current sources instead of 10V
b) emitter degenerating resistors of 100 Ohm and 20 Ohm for the input differential pair and current mirror respectively.
c) The interfacing of the pre-driver stage with the driver stage. As it is, it is too dependent on resistor values. I will use an amplified diode with a voltage range of 200mV, that is, the amplified diode will have an adjustable voltage of 1.1V to 1.3V.
d) The output stage is ready as it was taken from a large public address amplifier that failed for some weird reason, most probably, its grounding scheme is at blame. So, I have the chassis, the power supply, the connectors and the output stages already available. However, I decided I had better redesign the pre-driver stages. You may deem this mad, but instead of throwing the amplifier away in a recycling facility, I decided to turn it into a challenging but learning experience at the ripe age of 51.
 
Since I cannot edit the last post before this, and I am now seriously trying to learn using KiCad, the previous post can be ignored, except for the part about which circuits I will keep as a reference. I will make the mentioned circuit edits, but first, I have to succeed in producing my first PCB through KiCad. Needless to state, the first attempt is not a complex circuit, but a humble power supply.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.