Whats’ the point of running the Design Rule Checker when those options are turned off???
Apologies…a typo. The caption should have been "Note how the blue layer, GND, is tied to pin 21.
It does not appear there is a way to make a custom footprint to tie the two ground areas together under the MAX17633 only using pin 21 and not net ties. I am going back to AD and proposing just one GND plane for all ground connections, unless they have some really good reasons why that will not work. I think the board layout is clean enough and keeps related paths in their own spaces.
Power flows into the board from the bottom left to charge the batteries (MAX17633 Land), flows out of the batteries on the middle left from MAX17320 Land, into the relays and igniter in the upper left in Relay/Power Land (3-6 A for 100 - 200 msec). The Pi is connected to the 40 pin socket and the DAC/ADC and other digital stuff is in Pi Land. The LEDs on the lower right are on or off, so no switching issues. Fastest signals are the slow SPI bus and slow I2C bus in Pi Land.
Thanks @paulvdh and @mousey and @Piotr for your very helpful comments!
Mark
I am sure there was no pun intended in your comment…
Consider that User can set/tweak the Rule’s - a handful of parameters. Thus, can avoid having to deal with them when DRC’ing. I don’t usually bother with DRC/ERC as I know what I’m doing in the design process.
Occasionally I use DRC to check something but, to me DRC rules are a bother, especially if I don’t want to be tweaking them per project or accept defaults… That said, most folks probably use it/need it.
The Zone-Fill DRC rule has a Checkbox so, Kicad developers must have realized a reason/justification for having it’s On/Off setting and, my logic is that sometimes a design doesn’t want certain items involved in Refilling Zones. Items such as THT, Vias… as we’ve been discussing.
User can Ignore the DRC Errors/Warnings, too…
People who don’t use safety belts in cars have the same attitude. They are “safe drivers” and don’t get into accidents.
Safety comes in layers, and is build into the design. Not using DRC is just plain bonkers. It’s one of the most important features in KiCad. In aeronautics they use the “Swiss cheese model”. The Idea is to recognize that any safety measure has flaws, and some error can slip though one of the holes. But by having several layers of safety (cheese), it becomes increasingly unlikely that an error slips through all the holes.
In my rocket launcher design I use mechanical relays instead of solid state relays. Why? because a mechanical relay fails in the open position if the circuit is designed correctly. Solid state relays typically fail in the closed position.
I do not want to be attaching the launcher’s igniter leads to a rocket motor igniter when the relay has failed and the motor immediately ignites. 3,000+ degrees F rocket motor exhaust, even for a brief amount of time, will do a lot of damage to hands and eyes. Not to mention the potential injury from grabbing two leads designed to deliver 6 - 10 A when they are hot.
BTW, rocket launches, from model rockets to NASA launches, have a countdown for a reason, and it is not to build Hollywood type suspense.
Mechanical relays do not always fail “open” It also happens that damaged contacts get heated and get welded together. In “safety relays” there is a mechanical linkage between the “power contacts” and a “feedback contact”. If the relays fails in either the open or the closed position, the feedback circuit will be able to give feedback on this, so some action can be taken.
@paulvdh I didn’t say always. And I qualified failure in the open position when the circuit is designed correctly. Meaning, the relay is NO and the current through the contacts is far below the max current for those contacts.
And yes, I have visual feedback status on the state of the relay through the LEDs on the board and through the Pi and its control software.
That has nothing to do with a “safety relay”. Those have “forced contacts”. Look up how they work when you are interested in designing safety into electronics. For things like rocket motors, I would also think about adding an extra relay with NC contacts (or simple mechanical switch), that short out (after a fuse) the ignition circuit to make sure the thing can not get ignited unintentionally. Adding a “crow bar” circuit to intentionally blow the fuse on detected faults is also a consideration. In car racing (except from 5 point seat belts, safety cages etc) there is also a big red switch that cuts off the battery when turned. It’s nearly always the first thing that gets done when such a car is in an accident.
https://duckduckgo.com/?q=ralley+master+battery+isolator&iax=images&ia=images
That is there, too. I call it a Master Arm switch.
@pmi I couldn’t resist…
I love Rockets! I designed TVCS (Thrust Vector Control Systems) and LOFS (Laser Ordnance Firing Systems) for Nuclear Weapons to Space Exploration. We didn’t have Software with DRC heck, we didn’t have computers in the early days so we Engineers did our own DRC… And, yes, we had Countdown’s not for Hollywood…
Most all of my work was Secret/Top-Secret and the only photos taken were Controlled Document’s so, I haven’t any to post except as shown but, you get the idea… Top left is MX-Missile launching from Vendenberg, I’m standing next to back-end of MX-Third Stage (blocking the view of my TVCS )
10, 9, 8,7,6,5,4,3,2,1 Fire… Wait, wait! Did somebody do DRC?
So has every Engineer before you and will long after you. The reason you didn’t have DRC ‘back in the day’ is because the technology wasn’t there, it is now and its pre installed and free Had it it been available to you and you didn’t use it in such a dangerous industry you would have been out on your ear
3,2,1, BOOM 20 dead…DRC would have caught that…Have Mr Coffee come to my office.
From these thread I suppose that BlackCoffee designs PCB as I was doing it (in 70s) when I then draw tracks directly at PCB and then etch it in trichloride. When you use technology that accept clearance of 0.2mm and small distance you have only in footprints and then you draw all your tracks with 0.7mm (or more) distance from one to another than you can do it safely, I think.
The problem begins when you don’t control everything. Zones are practically out of your direct control as they are filled by KiCad. But from how BlackCoffee suggested to connect these vias I supposed he assumes that he don’t allow KiCad to refill zones. So if you have several net zones one under the other and you place via there it will be connected with all of them until they will be refilled.
You can imagine design order when zones have to be filled first and then you add several vias where you want them but then you not allow zones to be refilled but I would never used such design order.
Very interesting point you make and quite fascinating when a topic uncovers what could be a historical book as there is so much content The difference between the tools, materials and techniques between the 70’s and today is extraordinary and as a consequence of this we understand just how much we have given up to automation or not as the case maybe
We really do have a lot of ways to ‘skin a cat’
True. And, of course what to do depends on the Circuit, Fab-House…etc.
NetTie’s are a good solution if space permits. The ‘Tie’ is a Filled-Zone on the Top or Btm layer that bridges and touches the Via’s in the NetTie part.
Zones have Priority settings that affect what gets filled first and what outlines get deleted…
To me, THT’s are simlier to facilitate connections but, perhaps PCB Fab houses prefer Vias for tying Nets together… THT’s enable ‘me’ to have control.
A stock NetTie having 3-Vias 3-THT’s Red loop’s show Contact to Via-Cylinder. Arrow is the Bridge.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.