285 replies to this topic

[vampirolatino2]

Can't wait for this to happen

[mjr]

 

Yes 3 chimes.  I'm also adding a 5" bell from a Golden Arrow that is essentially a knocker inside a big bell and would like the same on that.  not sure how many other toys this might apply to, or how many people are making em based cabs that have this stuff, but flexibility is always a good thing.

 

Okay, here's my current thinking.  I'm working on a *third* 10x10 board to go with the others, this one designated as the "Chime Board".  I'm using a 74HC595 to set up 8 new digital outs - these chips are 8-bit shift registers that are basically the digital out equivalent of the TLC5940 chips used on the MOSFET board for PWM outs.  It looks like 5 time-limiter circuits will fit.  Since we get 8 outputs from the 74HC595,..

 

Well, it looks like full 8 timer output circuits will fit after all.  That should be enough for just about any virtual cab, I'd think, but multiple chime boards can still be daisy-chained if not.

 

Now I just have to breadboard the 74HC595 setup and make sure it actually works.

[mk47]

 

I searched for for a ribbon cable suitable for the 5 pin 1 row headers, but i didn't find anything. Isn't it better to change the layout to use 2 rows of 3 pins. 

 

Good catch - I'll make the change.

 

Great! But I just noticed that 6 connectors don't seem to be very common. Maybe we should consider to use 8 connectors (2 times 4 pins), even if we have 3 unused pins in that case.

[mjr]

Great! But I just noticed that 6 connectors don't seem to be very common. Maybe we should consider to use 8 connectors (2 times 4 pins), even if we have 3 unused pins in that case.

 

Do you think so?  From what I can see on the electronics dealer sites, it looks like the IDC housings are equally available for just about any pin count.  

[wshadow]

Hey great project - can't wait to build my cab (which only exists in my mind so far).

 

About the maximum current of the hi-power traces - why don't you just omit the solder mask on those traces?

This way the traces either get tinned at the PCB factory (if the customer selects tinning) or they can be tinned manually during assembly.

That's about the industrys cheapest trick ever to increase the maximum current for hi-power traces.

Although it's impossible to predict the real current limit for tinned traces, it's absolutely safe to say that the limit can be doubled (if the traces are tinned evenly).

[mjr]

Hey great project - can't wait to build my cab (which only exists in my mind so far).

 

About the maximum current of the hi-power traces - why don't you just omit the solder mask on those traces?

This way the traces either get tinned at the PCB factory (if the customer selects tinning) or they can be tinned manually during assembly.

That's about the industrys cheapest trick ever to increase the maximum current for hi-power traces.

Although it's impossible to predict the real current limit for tinned traces, it's absolutely safe to say that the limit can be doubled (if the traces are tinned evenly).

 

Hmm, interesting idea, although I don't think it's necessary - the trace widths should be wide enough to handle the kinds of loads we need easily.  Plus it seems a bit untidy to expect people to tin the traces - the boards are pretty dense.

[wshadow]

Yeah well, just an idea

I was only talking about the thick traces at the MOSFET outputs, where board density doesn't really matter for tinning.

The remaining traces would still be covered, so you can't short an output trace against anything else.

 

About the daisy-chain connectors... I'd suggest using 2x5 pin headers and make one row GND, the other row carry the signals.

If you use ribbon cable for the interconnection, you will end up having a signal/gnd/signal interleaving at the actual wires, which provides at least a bit of shielding.

 

Additionally, as we are talking about shielding, from the EMI point of view, you should convert the free space on the boards to large GND planes.

This can be done in EAGLE automatically.

[mjr]

About the daisy-chain connectors... I'd suggest using 2x5 pin headers and make one row GND, the other row carry the signals.

If you use ribbon cable for the interconnection, you will end up having a signal/gnd/signal interleaving at the actual wires, which provides at least a bit of shielding.

 

Interesting idea - I'll see if there's room for expanding the headers that way.

 

 

Additionally, as we are talking about shielding, from the EMI point of view, you should convert the free space on the boards to large GND planes.

This can be done in EAGLE automatically.

 

Already done.  If you hit RATSNET after loading one of the boards, you'll see it come up - for whatever reason, EAGLE doesn't show the polygon fill automatically at load.

[mk47]

 

About the daisy-chain connectors... I'd suggest using 2x5 pin headers and make one row GND, the other row carry the signals.

If you use ribbon cable for the interconnection, you will end up having a signal/gnd/signal interleaving at the actual wires, which provides at least a bit of shielding.

 

Interesting idea - I'll see if there's room for expanding the headers that way.

 

 

That would be great!

[wshadow]

I've also been thinking about two more things...

 

Your hi-power traces often go from layer to layer, using vias.

It's actually quite uncommon to route hi-power traces from layer to layer using only one via per connection.

You'll virtually never find that on industry designed PSU boards or similar, because routing high current through vias puts quite some stress on them.

If routing through vias is really required, the industry designers put multiple small vias there and overlap the traces on both layers to reduce resistance on the through-connections.

Anyway, it's still better to keep a hi-power trace on one layer only from end to end and adapt the signal traces instead (to "evade" the hi-power traces).

 

Another thing is the connectors for the hi-power outputs.

You're just about to remove the 1x5 pin header for the board2board interconnection because the plugs (where you actually put your wires in) are said to be uncommon.

The same actually applies to all pin headers on your boards.

For the button inputs you can actually use ribbon cable connectors, but ribbon cables aren't suitable for the midrange (darlington) and hi-power (MOSFET) outputs.

For the darlington outputs, AWG 21 or 20 wire (or 0.5mm² for us european guys) can be crimped into DuPont connectors, but this requires a lot of squeezing and stuffing (I did that in the past).

When it comes to hi-power output, I can't really think of a suitable connector for .1" pinheaders that can be put onto AWG 18 or 17 wire (~1mm²) without using specialized tools.

I guess that's one of the major reasons other controllers (like LEDWiz) use screw terminals for this purpose.

Pluggable connectors with screw terminals at plug side and solderable pins on PCB side (so you don't have to screw/unscrew every single wire) are available, but require at least a 3.81mm (3/20") pitch at PCB side.

 

PS: Yes, I am aware that the projected loads can be routed through wires thinner than what I described above, but in terms of voltage drop etc I really suggest using at least the said gauges.

 

Please don't get me wrong about all my "critcs" - I can't wait to have your project in my cab, I'm only trying to be constructive here

Edited by wshadow, 18 November 2015 - 09:23 AM.

[mjr]

It's actually quite uncommon to route hi-power traces from layer to layer using only one via per connection.

 

Yeah, I'm not wild about that either, but there's no way to route these in one layer with this density.  I'm just going to have to try the board with the vias and see how it works.  I actually think it's pretty overengineered for power in general, though - devices in a cab all operate intermittently and briefly.

 

 

For the button inputs you can actually use ribbon cable connectors, but ribbon cables aren't suitable for the midrange (darlington) and hi-power (MOSFET) outputs.

For the darlington outputs, AWG 21 or 20 wire (or 0.5mm² for us european guys) can be crimped into DuPont connectors, but this requires a lot of squeezing and stuffing (I did that in the past).

 

No, you don't want to use ribbon cable for the power outputs - I'm specifying crimp housings for those.  The crimp pins are good to 3A to 4A, and you can get pins for 20-24AWG.  I use 22AWG throughout my cab (22AWG is good to 7A).

 

It would probably be ideal to use something like the Molex 3.96mm connectors instead of the .1" for the power outputs, but those things are just gigantic.  I'd probably have to go to a 16-output MOSFET board if I used those.  That would also allow single-layer trace routing, so two birds, but it would double up the number of boards.  I'd prefer to keep it as compact as possible, and right now I'm fairly confident the denser setup will be safe.

 

(I'd actually use crimp wiring for the button inputs as well, even though they don't need it for power handling.  Ribbon wire is actually less convenient for those in terms of installation since each strand has to be routed to a different place in the cab.  Ribbon cable is ideal for the inter-board connectors, though.)

 

Keep in mind, too, that this is an open-source design, so you can feel free to use it as the starting point for your own version if you want to beef up the power handling.

[mjr]

I just posted the latest version of the board designs - they're in the usual place:

 

http://mjrnet.org/pi...sion-board.html

 

This includes the new chime board, with 8 timer-protected outputs.

 

I think this is a production-ready version.  Markus (mk47), who was the one who got me started on this project, is going to get a test batch made in the next couple of weeks using this iteration, and I expect I'll get a batch made myself soon as well, so we should be able to test the physical version soon.

Edited by mjr, 21 November 2015 - 07:47 PM.

[mk47]

I guess the usual place is http://mjrnet.org/pi...sion-board.html ;-)

 

mjr, thanks for all your hard work on the pinscape controller!

[BorgDog]

This is looking pretty awesome Mike, thanks!  I'm looking forward to seeing your test batches in action, and it looks like I may need to brush up on my soldering skills.. a lot

[mjr]

I guess the usual place is http://mjrnet.org/pi...sion-board.html ;-)

 

Oops, copy and paste error!  Thanks for catching it.

[Rappelbox]

Hey mjr,
Nice to see that you finished the design
I am in talks with mk47 about organizing a group buy like I did with the pin 2 dmd project (search on the other forums ;)it's not welcome here)
Hope it's ok for you, as it is an open source thing...
Pm me about everything else

Greets,
Dom
Hey mjr,
Nice to see that you finished the design
I am in talks with mk47 about organizing a group buy like I did with the pin 2 dmd project (search on the other forums ;)it's not welcome here)
Hope it's ok for you, as it is an open source thing...
Pm me about everything else

Greets,
Dom

[mjr]

I am in talks with mk47 about organizing a group buy 
Hope it's ok for you, as it is an open source thing...

 

Absolutely!  

[vampirolatino2]

Guys, did a test quote for the pcbs and we can get it as low as $6 for the 2 boards. Who is in for a group buy?

 

or $7 for just a 1 board solution

Edited by vampirolatino2, 21 November 2015 - 11:22 PM.

[mjr]

I updated the page with links to parts lists on Mouser.com, to help with ordering the components.  You don't have to buy the exact parts listed there - it's just meant to provide a comprehensive list of what's needed.  Equivalent parts from other vendors and manufacturers will work equally well.  You might want to check eBay for the MOSFETs and optocouplers in particular - Mouser's prices on those are reasonable per item, but there are so many of each that it gets pricey.

[BorgDog]

Thanks for the update, but now you are just scaring me!  ZOINKS! that's a lot of parts and it's not even all of them!  Not sure I'm up to this diy task.