New PCB

[Dakkaron]

This is a discussion around the new PCB which should replace the Arduino and the adapter PCB.

The point of this new PCB is to reduce the necessary manual soldering and also increase durability of the circuit.

In issue #6 @dumblob and I suggested the following features of the new PCB:

• Capability to use an USB-C, Micro USB or reversed Micro USB plug
• Plug soldered directly to the PCB or connected via wires to through-holes on the PCB
• Should be as small as possible to not waste space

A few more things from my side:

• SPI interface should be exposed
• USB pass-through charging?

What other ideas do you have?

[dumblob]

I've given it a quick thought and for my use case I am kind of indifferent on whether to have a USB pass-through or not.

The pro of having a pass-through USB is I don't need to use wireless charging (which I'm not the biggest fan of, but it's not a deal breaker for me) and can charge my phones as I'm used to.

The con is I have to put off the case - but this shall be easy if we end up with "flexible parts" made from a sturdy non-flexible material. The pro is it won't affect power consumption and it'll be a little cheaper to buy parts for and a little less trouble to assemble (but we should anyway aim for automated assembly & soldering of ICs and small parts).

So I'd probably consider the power demand and if that was really negligible as a second parameter I'd consider the size of the PCB - if it would allow things mentioned below.

Regarding SPI I agree - though only as an additional interface and it shouldn't replace the usual over-USB flashing (which can potentially be done also from the phone itself - e.g. changing the layout - imagine using Termux on Android or iSH on iPhones wouldn't even require one to mess with Android SDK).

Oh and if the PCB space will allow for it, then I'd just put there the most capable unused GPIOs as their break out in the form of hand-solderable pads (or even holes if there already will be any other holes on the PCB - I don't want to increase the complexity nor price). One never knows how and when this might come handy. This includes GPIOs from Arduino but maybe also unused ports from the BB keyboard itself.

Last but not least the PCB could be kind of "universal" and thus maybe allow for certain non-negligible leeway (depending on the PCB size, but my guess is 10 mm should be enough) with regards to the BB keyboard piece to accommodate for different USB port positions (in all 3 axes) as they're not always in the middle. The rule of thumb would be to "shrink" the PCB from full width:

to shrinked width:

IDK, this is all just quick ideas.

[Dakkaron]

Sorry for the late answer, I thought I had already answered. Too many places to check ;)

Since there were some developments on the case front, also the requirements for the new PCB could change.

In the newest design, I screwed the keyboard to the case. This makes removal/attachment really fast and easy, since I just have to slide off the bottom part. Because of this, I would drop pass-through charging from the list of requirements, since it's really complicated to implement and it's actually pretty easy to remove the keyboard for charging.

Adding pass-through charging would take quite a bit of effort, and there is always a bit of risk, since if it's implemented wrong, it might fry the keyboard's electronics. So I am fine now with removing that requirement.

Regarding SPI I agree - though only as an additional interface and it shouldn't replace the usual over-USB flashing (which can potentially be done also from the phone itself - e.g. changing the layout - imagine using Termux on Android or iSH on iPhones wouldn't even require one to mess with Android SDK).

Sure, the SPI interface is only for emergencies and/or initial flashing. I guess, if I get the board made from a place like JLCPCB, the ATMega32u4 will probably not have the Arduino Bootlader preinstalled, so there needs to be a way to get it on the chip in the first place.

I had a look at flashing from the phone. There are a few Arduino IDEs for Android. The first one plainly didn't work with the board (it requires the Reset button to be pressed during flashing, but the Arduino Pro Micro doesn't have one). There is another one that looks better, but I haven't really tested it. But that one is pretty complicated to setup, so I am not sure if that's something that I would generally recommend to beginners.

Oh and if the PCB space will allow for it, then I'd just put there the most capable unused GPIOs as their break out in the form of hand-solderable pads (or even holes if there already will be any other holes on the PCB - I don't want to increase the complexity nor price). One never knows how and when this might come handy. This includes GPIOs from Arduino but maybe also unused ports from the BB keyboard itself.

Good idea! We already need holes for the USB wires, It shouldn't increase the price by too much. Assembly alone will probably already cost ~€15 per piece, so another euro or so for the PCB wouldn't be too much.

[Dakkaron]

This is supposed to be the first version of the PCB schematic. Do you all have any feedback?

Go easy on me, it's the first time I've done something like this.

PS: One thing I just noticed: I probably could have separated out a decently usable SPI- and I2C header (meaning signal wires + VCC + GND). Right now it's all bunched up rather randomly...

[dumblob]

I don't have a dedicated ground plane, since I wanted to stick with two layers for price reasons.

Dedicated ground plane is not needed for such low-speed designs. Two layers is more than enough. Just to avoid misunderstanding I really meant copper fills/zones and not a ground plane 😉.

And yes, LEDs with different colors solve the issue with being close to each other elegantly 😉. RGB is IMHO unnecessary (JLCPCB though surprised me they don't have it 😮).

And yes, the latest render with copper fills/zones in the blue layer looks good. Now also copper fills/zones in the red layer and you are probably good to go 😉.

[Dakkaron]

Dedicated ground plane is not needed for such low-speed designs. Two layers is more than enough. Just to avoid misunderstanding I really meant copper fills/zones and not a ground plane 😉.

Ok, that makes sense, thanks for the input!

RGB is IMHO unnecessary (JLCPCB though surprised me they don't have it 😮).

Yeah, my thinking was that I'd get more different colours with less space requirement. JLCPCB does have RGB LEDs, but only as Extended Parts. And each different extended part adds an additional $3 setup fee. And that's just not worth the very limited advantage.

And yes, the latest render with copper fills/zones in the blue layer looks good. Now also copper fills/zones in the red layer and you are probably good to go 😉.

Since there is only very little space on the red layer, and almost all red-fillable areas already overlapped the blue filled areas, I thought it wouldn't do much. I think, I now filled pretty much every single possible space on the board with fill:

Do you have any other input?

If it's all good, I'll let the design rest for a few days, in case I come up with something else that needs to be changed, and then I'll order 2pcs for a first prototype run.

When that's done and working, I'll see how many people I can get aboard and make a first real batch.

[dumblob]

Do you have any other input?

You asked for it! 😆 It is actually just nit pics from my side.

1. I thought the clearance from all the edges to the closest trace or SMD piece or a hole or via shall be bigger (but I did not open the design in KiCAD and thus didn't measure how much it is - so it is all guess work from the pictures you posted).

2. All corners (especially the "inner" one near the western corner of the Arduino chip) shall have ~2mm radius (1.5mm radius shall still be acceptable).

3. The traces look a tiny little too thin for cheap manufacturing - but again, I did not measure them, so it is just a huge guess work.

4. The vias look also a tiny little too thin for cheap manufacturing - but again, I did not measure them, so it is just a total guess work on my side.

With (2), (3), and (4) we shall check at JLCPCB (and ideally also one or two others) about their preferences.

If it's all good, I'll let the design rest for a few days, in case I come up with something else that needs to be changed, and then I'll order 2pcs for a first prototype run.

👍

Do I remember correctly that the minimum quantity at JLCPCB was 2 or even more pieces? If so, then feel free to send me a private message on mygithubnick at gmail dot com to find out the cheapest delivery path between our places (I have friends who travel a lot across Europe more often than not).

Do they still also pack some random other PCBs in the package so that one gets a surprise and at the same time can broaden her perspectives among HW?

When that's done and working, I'll see how many people I can get aboard and make a first real batch.

I'll be one of them (admittedly in my country in Europe 75 EUR is quite a lot, so I'm not sure my friends who wanted to try it would consider this trying any more and not an investment which they would be quite understandably somewhat reluctant to accept).

[Dakkaron]

Thanks for the feedback!

(1) I tried to get the clearances as small as KiCAD would let me, to save some space. What kind of clearance should I have?
(2) What do you mean by the radius? Should I draw rounded corners?
(3) I set the traces to the minimum width recommended by JLCPCB
(4) The vias are actually a tiny bit larger than the minimum, that JLCPCB requires

Since the production cost of the PCBs is very small (€1.90 for 5pcs) and the main part of the cost is assembly, it doesn't seem like the traces/vias drive up the price.

Do I remember correctly that the minimum quantity at JLCPCB was 2 or even more pieces? If so, then feel free to send me a private message on mygithubnick at gmail dot com to find out the cheapest delivery path between our places (I have friends who travel a lot across Europe more often than not).

I'll send you a message.

Do they still also pack some random other PCBs in the package so that one gets a surprise and at the same time can broaden her perspectives among HW?

Last time I ordered 5 boards and got 25 instead.

I'll be one of them (admittedly in my country in Europe 75 EUR is quite a lot, so I'm not sure my friends who wanted to try it would consider this trying any more and not an investment which they would be quite understandably somewhat reluctant to accept).

What country do you live in?
If you want to, we can talk per email in a bit more detail in regards of my price considerations.

[dumblob]

(1) I tried to get the clearances as small as KiCAD would let me, to save some space. What kind of clearance should I have?

IDK how much it is in your design but to be on the "safe side" I would probably choose 1.5 mm (it highly depends on the manufacturing process where they cut it along straight edges etc. and on the mounting in the smartphone case where you might want to have some "mounting click supports" around the PCB as shown below).

(2) What do you mean by the radius? Should I draw rounded corners?

Yes. JLCPCB will anyway make a radius on the one inner corner near the arduino chip. So it is better to define the radius by yourself than to let them choose one for you 😉.

(3) I set the traces to the minimum width recommended by JLCPCB

Hm, I for myself wouldn't go for the minimum width (I know etching nowadays is very good but something from the past still tells me it might be unnecessarily risky) but YMMW 😉.

(4) The vias are actually a tiny bit larger than the minimum, that JLCPCB requires

Ok. Usually PCB shops tell which vias are easy to drill (they drill several PCBs at once with thicker drills unlike with thin drills and thus the price goes quickly up). I did not check JLCPCB.

Since the production cost of the PCBs is very small (€1.90 for 5pcs) and the main part of the cost is assembly, it doesn't seem like the traces/vias drive up the price.

The width of traces should not influence price - it's really about reliability of the final PCB. Vias diameter though usually does influence the price - but maybe not at JLCPCB. So if it doesn't in this case, then leave them as they are 😉.

I'll send you a message.

I got it and will reply soon. Thanks!

Last time I ordered 5 boards and got 25 instead.

🤣

[spARTacusFromCB]

So the latest thought is the board will be no wider than the actual Q10 Keyboard? Do the latest thoughts include extra points on the board for if someone in the future wanted to explore adding some extra keys/trackpad/belt? I realize there probably isn't suitable space above the keyboard for a belt, but my Samsung A21 is a bit wider than a Q10 Keyboard. So, some folks might be able to squeeze in extras on the left / right sides of the keyboard, and still not exceed the width of the phone.

[dumblob]

Yes, that is exactly the reason why we try to make the new PCB as small and narrow as possible. You can then easily add other PCBs (with buttons, joystics, etc.) and solder them (either using a wire or using castellated holes) to the new PCB from here (I am privately in contact with @Dakkaron where we discuss using pads instead of through-hole connections - if pads will be used in the end - and I hope they will - then castellated soldering is possible unlike with the new PCB revision above).

Or you can take the new PCB layout and simply extend it in KiCAD in any direction you want and just place the buttons etc. there (because it is cheap to order assembled e.g. from JLCPCB).

So again, making the PCB smaller covers all the possible future use cases.