OMG! The zero ohm jumpers!

 

and here I am making 4 layer boards just to hide all the traces on inside layers to not mess up the visuals....

 

That is pretty insane. I've taken boards with parts on both sides and laid them out with parts only on one side, but the bare boards were still multi-layer. This was to remove the extra step of soldering parts to the bottom of the board. I don't remember how much money we saved by redesigning it like that.

The engineering group that I worked with at the time was impressive that way. We modified a lottery ticket reader that was based on a fax machine scan head. The operator could insert the ticket in any direction and it would read the barcode on the ticket. The initial design used 4 PC boards, a power supply board, a board with an FPGA and a CPU and two entry/exit sensor boards. We ditched the exit sensor board, merged the power supply and controller sections onto a single board that also had a single ticket entry sensor on it. We mechanically tripped the entry sensor with a bar that was pushed by the ticket being inserted. Basically 4 boards to 1. We also were able to replace the costly FPGA with a PLD and two FIFO RAM chips. If I recall, the four board design was about $300 for all four boards. Our single board was just shy of $100. Over a 1000 piece order that savings added up.

 

At JLCPBC.com they charge the same price for 2 layer and 1 layer. It has been years since there was much difference.

 

I designed a board once for a 11KW SCR resonant-converter. The area for the capacitor section was about 5" x 5" and had to handle 75A average (150A peak) current, 700 volts DC, in a 50 degree C ambient temperature and pass UL. It took a 2-oz copper with 2-oz reflow solder (4-oz total conductor thickness) on each side of the board, with both sides connected in parallel. At the time, it was actually pushing circuit board technology a bit.

 

Someone once told me that it was possible to track ANY design on a double sided board by running the tracks on one side north-south and east-west on the other and linking with vias.
I've never designed anything I couldn't track on two layers, so it must be true - mustn't it?

 

Someone once told me that it was possible to track ANY design on a double sided board by running the tracks on one side north-south and east-west on the other and linking with vias.
I've never designed anything I couldn't track on two layers, so it must be true - mustn't it?

I use two layers whenever possible.

I have an engineer who insists on 4 layers, regardless of the complexity.

I let him debug and patch the prototypes.

 

Someone once told me that it was possible to track ANY design on a double sided board by running the tracks on one side north-south and east-west on the other and linking with vias.
I've never designed anything I couldn't track on two layers, so it must be true - mustn't it?

As long as you have plated though holes on the vias. Otherwise there could be a large number that would have to have wires soldered into them.

 

I've never designed anything I couldn't track on two layers, so it must be true - mustn't it?

I produced some boards that truly required 6 layers. It was basically a motherboard with half a dozen 96 pin "Euro" connectors. The board had to be darned small to fit a very specific package. Indeed, at one point, I wasn't quite sure I would even be able to get all the components onto the board, that before running a single trace; simply placing the components was a nightmare. And very little was "bussed", so rarely rarely did pin 1 on one connector run to pin 1 on the connector beside it. Usually it would run to pin 90 on a connector several inches away. Add in the requirement that the backplane had to supply power to all the cards, this back in the days when "an amp or two" was what a single chip used, not a whole card, and so two planes for power and ground were unquestionably required.

Now add the complication that it was a navigation system for search and rescue marine helicopters, meaning a horrendous physical environment, ridiculous amounts of vibration and the absolute requirement for high reliability. I did not trust buried vias in such an application, and I had in my own mind a minimum trace width for resistance against cracking, both these two criteria eating up even more real estate.

Thankfully, there was nothing too much over a megahertz and most signals much slower than that, so very few traces had to have an additional "high frequency" consideration. Even so, it was a tough job.

I literally had nightmares about it, with coloured traces running to and fro in my head before slamming into some brick wall at a dead end, kinda like the movie Tron.

When the EE came back a while later asking for a "small" number of changes, I almost broke down...

 

Someone once told me that it was possible to track ANY design on a double sided board by running the tracks on one side north-south and east-west on the other and linking with vias.
I've never designed anything I couldn't track on two layers, so it must be true - mustn't it?

Yes, it's true, but it might require the parts to be spread out excessively to accommodate the tracks. Also, from an EMI and timing standpoint it becomes a nightmare.

 

exactly... for example this may not hold when there is a bunch of high current circuits. then spacing between pads of a single component (relay, connector, etc.) becomes the bottleneck.

 

I have come across this situation many times. I worked for a company years ago. Sometimes I needed to make PCB for project prototypes and also for read products. The per-unit cost of a double-layer PCB was twice that of a single-layer PCB. So, I was told to accommodate all my designs in a single layer. The designs looked very clumsy. Traces needed to be at least 30th and all had to be in a single layer! As a result, I needed to use a lot of vias. Electroplated vias were not available where I worked. Vias were constructed manually with leftover resistor legs etc. That made the PCB look bad sometimes. Proper via implementation is something I’ve always found challenging, and this article provides a great overview of the via manufacturing process, types, and best practices: Understanding the PCB Via Process. If I had access to better via fabrication options back then, it would have saved a lot of manual work and improved PCB reliability.

 

Someone once told me that it was possible to track ANY design on a double sided board by running the tracks on one side north-south and east-west on the other and linking with vias.
I've never designed anything I couldn't track on two layers, so it must be true - mustn't it?

That is not a rule but a guideline which I tend to follow in my PCB layouts.
Simply speaking, it helps in resolving difficult layouts, and sometimes results in fewer vias.