3 Port 45W USB-C Charger (Formarly: pcie USB card.)

[thDel]

Hello

 

I have decided that instead of buying this USB C header expansion card,

https://www.scorptec.com.au/product/Controllers/USB-&-Firewire/76403-SST-ECU02?gclid=Cj0KCQjw0Mb3BRCaARIsAPSNGpUG46HiQzUE1l8eOH4QryudtLSEiH9_gcw1gSvZ25R-7dP8INCYaIwaAqQTEALw_wcB

I will make it myself.

 

I know I will need to get a PCB printed and add components. I plan to make an exact clone of the linked card. Here are my questions.

 

1. Is this possible?
2. What is the best way to get a PCB printed?
3. Where can I source the components?
4. Will I need to write any software (drivers ect.) for it to work?
5. How much will it cost?

 

Thanks, Del

[WereCatf]

34 minutes ago, thDel said:

Is this possible?

Yes

34 minutes ago, thDel said:

What is the best way to get a PCB printed?

One of the billion different PCB-manufacturing services, like e.g. JLCPCB.

 

34 minutes ago, thDel said:

Where can I source the components?

LCSC, DigiKey, Mouser, Farnell Element14 etc.

 

35 minutes ago, thDel said:

Will I need to write any software (drivers ect.) for it to work?

If all the controllers are standards-compliant, no. Though, the controller may need you to program any default settings on it.

 

36 minutes ago, thDel said:

How much will it cost?

Do you really expect us to check the cost of the PCB, all the components, shipping to your location (which we don't even know) and so on? Fuck that. It'll cost a lot, lot more than just buying a readymade device, though.

[thDel]

1 minute ago, WereCatf said:

Fuck that. It'll cost a lot, lot more than just buying a readymade device, though.

I know it will cost more but BC of civic I won't be able to buy one for months and I thought this would be fun, my question isn't about specific price just if it will cost a stupid amount like $500

[WereCatf]

3 minutes ago, thDel said:

I know it will cost more but BC of civic I won't be able to buy one for months and I thought this would be fun, my question isn't about specific price just if it will cost a stupid amount like $500

Yeah, but there isn't really any good way of saying how much it will cost without knowing all the shipping-costs and the prices of the components and so on. If I had to just simply guess, pulling some number out of the hat, it'd be something like $100 - $200 per device.

[thDel]

Just now, WereCatf said:

guess, it'd be something like $100 - $200 per device.

Thanks, 

[Kilrah]

1 hour ago, thDel said:

I know it will cost more but BC of civic I won't be able to buy one for months

How could you not buy one but have no problem sourcing all the components for making your own?

 

You'll also need to count a couple of months actually learning how to design high frequency PCBs...

 

If you're not already equipped it'll cost much more than $500 by the time you buy 5 sets of components to cover for mistakes, get all the tools needed to actually populate the board and debug why it doesn't work, reorder PCBs once or twice after making corrections...

[Unimportant]

A 'scope with the bandwidth to make eye diagrams at pcie frequencies (to check signal integrity for your layout) cost as much as a nice car... 

[Hackentosher]

If you're not already intimately familiar with PCB design, specifically for high frequency like PCIE, it'll cost waaaaaaaaaaaaaay more than the $60 for that silverstone card you linked in your time alone to learn how to design such a device. PCIE is a pretty complicated standard, and as other people mentioned, you'd need thousands of dollars of test equipment to see what it's doing and why it's not working. The actual components themselves probably wouldn't be that bad, I'd guess maybe a $50 BOM cost in components, then your boards can be cheap assuming they can be on just two layers.

 

If you want to learn PCB design (which I highly encourage), I don't think this is not a good first project. Choose something simpler, maybe a USB 2.0 breakout or a small Arduino. 

[thDel]

7 hours ago, Hackentosher said:

you'd need thousands of dollars of test equipment to see what it's doing and why it's not working

 

12 hours ago, James Evens said:

Could also easily be $100+test equipment

 

13 hours ago, Unimportant said:

cost as much as a nice car

 

17 hours ago, Kilrah said:

You'll also need to count a couple of months actually learning how to design high frequency PCBs...

 

If you're not already equipped it'll cost much more than $500

OK i think i will give up and just do a USB charging station :). 

[Hackentosher]

1 hour ago, thDel said:

OK i think i will give up and just do a USB charging station :). 

I feel like we all just took massive dumps on your project idea, sorry about that. USB charging would actually be a great starting point, all you need is a stable 5V source and some voltage dividers on the USB data lines to enable charging on iOS as described in this blog post https://learn.adafruit.com/minty-boost/icharging.

[WereCatf]

Just now, Hackentosher said:

I feel like we all just took massive dumps on your project idea, sorry about that.

Nah, it's better to be realistic about something like this, instead of throwing money down the drain and realizing you ain't got what it takes. Besides, OP can always pick the project back up once they're better equipped to handle the requirements here. In fact, I totally approve of the idea, myself: it's both a very nice learning experience and can also be incredibly fun and rewarding to take an existing device and make your own, completely custom clone of it -- and possibly tack some extras of your own on it as well.

[Unimportant]

8 hours ago, Hackentosher said:

I feel like we all just took massive dumps on your project idea, sorry about that.

Nah, this is actually a nice example of the Dunning Kruger effect, where someone has too little knowledge about a subject to be able to realize how little he/she knows.

 

I'm not trying to discourage someone from learning and trying to build whatever he/she wishes but I think it's healthy to make them realize they're embarking on climbing a mountain, not the molehill their flawed perspective makes it out to be.

[thDel]

18 hours ago, Hackentosher said:

I feel like we all just took massive dumps on your project idea, sorry about that.

No, as others have said, if I had persued this i would have just wasted tonnes of money. 

My new project is a 50w 3port type c charging hub. I have finished the schematic but with my current expertise I assume there are problems with it.

[WereCatf]

41 minutes ago, thDel said:

I have finished the schematic but with my current expertise I assume there are problems with it

The text is illegible, so I can't tell what that module is you have for power in, but I assume it doesn't contain a fuse. You should always have a fuse, when dealing with live AC from the wall. Also, this design doesn't limit current per port, so if the AC-module in your shematic can actually supply 5V 10A, one could draw 10A from a single port -- that's way overspec and such a high amperage as to be a fire-hazard, as even the USB PD-spec only goes up to 5A and as such the cables aren't designed to handle double that.

 

Add a fuse and implement current-limiting per-port.

[thDel]

4 hours ago, WereCatf said:

The text is illegible, so I can't tell what that module is you have for power in, but I assume it doesn't contain a fuse. You should always have a fuse, when dealing with live AC from the wall. Also, this design doesn't limit current per port, so if the AC-module in your shematic can actually supply 5V 10A, one could draw 10A from a single port -- that's way overspec and such a high amperage as to be a fire-hazard, as even the USB PD-spec only goes up to 5A and as such the cables aren't designed to handle double that.

 

Add a fuse and implement current-limiting per-port.

Cool thanks,  is the resistor thing for the CC ports correct, they were 15k resistors if you can't read it.

[Hackentosher]

13 hours ago, thDel said:

Cool thanks,  is the resistor thing for the CC ports correct, they were 15k resistors if you can't read it.

You need 5.1ks on cc1 and 2 not 15

[Pastorcasper 2.0]

On 6/24/2020 at 7:16 AM, Unimportant said:

Nah, this is actually a nice example of the Dunning Kruger effect, where someone has too little knowledge about a subject to be able to realize how little he/she knows.

 

I'm not trying to discourage someone from learning and trying to build whatever he/she wishes but I think it's healthy to make them realize they're embarking on climbing a mountain, not the molehill their flawed perspective makes it out to be.

this is very true lol, too many times have i wanted to do something and luckily realizing that i was in WAAAAAAY over my head (thankfully i am yet to waste a ton of money trying to do something i couldnt)

[thDel]

4 hours ago, Hackentosher said:

You need 5.1ks on cc1 and 2 not 15

Cool thanks

 

[thDel]

Here is the updated screenshots with the resistors, is this correct?

[Unimportant]

4 hours ago, thDel said:

Here is the updated screenshots with the resistors, is this correct?

Simply adding a resistor in series with VCC is not going to work as current limiting as there will be a voltage drop across the resistor proportional to the current going trough it. For example, with the 1.6 ohm resistors you added, if one attempts to draw 1A, 1.6V will drop across the resistor, leaving only 3.4V at the output.

 

You need an active current limiter circuit.

 

Also, I'd try to use way thicker traces, or polygons for the power traces on the PCB layout, and use a star configuration or power planes rather then daisy chaining.

[Kilrah]

Didn't look at everything but a few comments:

 

- You should put the AC in and fuse on the other side of the power supply block for good separation and to avoid long HV tracks.

- Your DC traces are too thin for 10A.

- If the blue pads on top are the input, make them through holes since otherwise the heavy gauge wires you'll solder there are likely to rip the pads off easily.

- Your resistors inline with VCC are going to drop your output to nothing. At 3A a 1.6 Ohm resistor will drop 4.8V, so leave 0.2 to the device (and the resistors will burn...)

[thDel]

2 hours ago, Unimportant said:

polygons for the power traces on the PCB layout, and use a star configuration or power planes rather then daisy chaining.

what do you mean by this? i am a complete noob.

[Unimportant]

55 minutes ago, thDel said:

what do you mean by this? i am a complete noob.

You want the best possible, lowest resistance and inductance, connections for power and ground. Typically this is achieved by using power planes, which means covering the entire surface of the board on one layer for ground, and another layer for VCC, and then using via's to connect to the required plane wherever you need it, rather then running individual traces.

 

However, this requires 2 layers for ground and VCC alone. If you're limited to a dual layer then I'd suggest making the bottom layer a ground plane and using a star configuration for VCC on the top layer. A star configuration looks like this:

 

The idea is that each load has it's own separate trace to the power source and not trace from the power source to one load and then on to the next and so on. This way, any noise generated by one load due to the trace's parasitics does not affect the other loads.

 

Polygons are irregular shapes that allow you to better use available board area to make better, low resistance, connections where it matters. For example, here's a board I designed that uses polygons extensively on the top layer:

[thDel]

12 hours ago, Unimportant said:

The idea is that each load has it's own separate trace to the power source and not trace from the power source to one load and then on to the next and so on.

 

12 hours ago, Unimportant said:

Polygons are irregular shapes that allow you to better use available board area to make better, low resistance, connections where it matters

Thanks very much.

[thDel]

16 hours ago, Kilrah said:

- Your resistors inline with VCC are going to drop your output to nothing. At 3A a 1.6 Ohm resistor will drop 4.8V, so leave 0.2 to the device (and the resistors will burn...)

 

16 hours ago, Unimportant said:

Simply adding a resistor in series with VCC is not going to work as current limiting as there will be a voltage drop across the resistor proportional to the current going trough it. For example, with the 1.6 ohm resistors you added, if one attempts to draw 1A, 1.6V will drop across the resistor, leaving only 3.4V at the output.

How would i add the resistor so that each port recieves 3a?