On 5/15/2018 at 5:09 AM, TonyMac32 said:

Some feedback from @chwe and some improvements on the protection circuitry:

 

 

And for the record, electrical cleaner can/will dissolve your electrical connectors...   Also, lead-free solder is an angry and terrible thing.  ;-)

 

I'm charging my Pixel off of one of the power ports while I play music over bluetooth on the Tinker...  

I like the big cylindric capacitor that looks like a water tank. Reminds me of the old times... 

2 hours ago, TonyMac32 said:

oshpark is a bit spendy if you are doing iterations, but it's also not in China so for me it saves any customs delays

 

yep - oshpark is a bit spendy - 2 layer is $5/sq inch, 4 layer is $10... keep your boards small - last one I did was RPi Zero footprint

 

Still I like them for proto runs... 

 

I've got a couple of friends over in Shenzen, so production PCB's, or even short full builds with components, are easier, and they're outside of the regular mess of things... small boards like this fit in a suitcase/backpack for someone heading from HKG to SJC or LAX, and they can drop it off at the airport...

 

hint - keeping things on one side of the board reduces cost quite a bit in production - and makes things easier for home runs...

 

Yes - still have a few rolls of non-ROHS old-school solder in my little station,  non-ROHS is the spawn of satan sometimes... solder paste and a dedicated toaster oven for SMT is a win...

54 minutes ago, JMCC said:

cylindric capacitor that looks like a water tank

 the new one has 2, and a big inductor that looks like a small office building.  

 

4 minutes ago, sfx2000 said:

hint - keeping things on one side of the board reduces cost quite a bit in production - and makes things easier for home runs...

 

Yes, all of my components sit on the "top", other than the header to connect to the other board.  

 

6 minutes ago, sfx2000 said:

solder paste and a dedicated toaster oven for SMT is a win...

 

I've taken to using a hot plate with thick steel "heat diffuser" to keep the surface consistent, I don't get any charring that way, the only tricky part is removing boards to cool them acceptably, but I've more or less worked that out, and the hot plate was $25.  I will admit SMT is new for me (I hand soldered the boards in this thread, for instance), but so far I'm having decent success.

18 minutes ago, TonyMac32 said:

I've more or less worked that out

 

I had to work it out with the spouse on the toaster oven   

 

The one I have has decent thermal regulation, and heat on both sides - good for reflow even on BGA parts like my old Macbook Pro motherboard with the nvidia chip that had problems...

Well, Oshpark turned around my order in 5 days, so I was able to build one today:

 

 

So far no problems, 5.16 V open circuit voltage, nominal design intent was 5.20, so within tolerance.

5 hours ago, TonyMac32 said:

Well, Oshpark turned around my order in 5 days, so I was able to build one today:

 

Nice - I would open up the mask for the screw mounts and clear the plane there - but that's a minor nit...

 

Just sharing insight with an ESP32 board I was working on...

1 minute ago, sfx2000 said:

I would open up the mask for the screw mounts and clear the plane there

 

Not a bad idea, interestingly the KiCad template doesn't have that done, I'd kind of ignored it while trying to get the grounding of that buck converter sorted.  It's a bigger footprint than I wanted, but that's because those cute little Chinese boards you get don't follow any of the guidelines on the L and C values, or how the grounding should work...

 

I noticed my libraries are broken on my GitHub repo for this, giving anyone cloning it a ton of "?"'s, I'll have to mess with it to either move all the symbols I need into the repo itself, or whatever.  #1 gripe about KiCad is library management.

 

I'm working on an ESP32 board for my day job, I went ahead and got the wroom32 module (or something like that.  I'm not feeling my best today and am not going hunting for them.  )   Was thinking about one for fun on my own time, but that stuff really gets as much OSS/OSH attention as I think anything can stand.

12 minutes ago, TonyMac32 said:

I'm working on an ESP32 board for my day job, I went ahead and got the wroom32 module (or something like that.  I'm not feeling my best today and am not going hunting for them.  )   Was thinking about one for fun on my own time, but that stuff really gets as much OSS/OSH attention as I think anything can stand.

 

I went with a simple board first, and it was an epic fail on my part...

 

4 minutes ago, sfx2000 said:

I went with a simple board first, and it was an epic fail on my part...

 

I've designed some 315 MHz stuff, and honestly know I can't do higher frequency RF design well enough to show anyone. 

 

One of My Tinker S's sacrificed themselves during testing, I had the poor thing perched over some random crap, too much of which was conductive.    On the bright side, the converter can tolerate short circuits quite well.  The Tinker S, not so much.

12 minutes ago, TonyMac32 said:

On the bright side, the converter can tolerate short circuits quite well.  The Tinker S, not so much.

 

magic smoke holds many lessons...

4 minutes ago, sfx2000 said:

magic smoke holds many lessons...

  In my case, "Clean up your bench, you slob"  - Magic Smoke

18 hours ago, TonyMac32 said:

In my case, "Clean up your bench, you slob"  - Magic Smoke

 

Hehe - probably more than a few folks...

 

And SW guys - clean up your source and merge it forward...

 

Note - comment not intended for any single group, but SW in general - old cruft in code is like a messy workbench

18 hours ago, sfx2000 said:

I went with a simple board first, and it was an epic fail on my part...

 

 

@TonyMac32 - you've inspired me to pick that board back up and try again - but this time do it bigger... first one was an IOT focused HAT board using the WROVER module for development - it ended up perhaps too ambitious with feature creep...

 

(anyone interested in a QCA9531 base with 128MB RAM/16MB Flash that runs OpenWRT with N300 Wifi and 2 1G ethernet interfaces - aka the RouterHat)

Cool. I have an expanded version of the power hat in mind, and some other small utilitarian sort of hats intended to be low cost for those who want them, of course nothing I can't use will be created unless I get some kind of return, I have only so much time and money to donate...

Sent from my Pixel using Tapatalk

Hi @TonyMac32

 

You call this thread  "Pi-Factor power solution" your Website: Pi-factor Power HAT on Open HW: Raspberry Pi Power Supply Board while the Github: Tonymac32/Pi-Power  -  it does make sens to keep the 'brand'.

As you can see on page 1 of this thread my heat sink is large. So I would have remove quite a bit of the board (the part with the holes) up to the GPIO header. I have about 3mm to these pins.

Judging from the pictures here: https://www.openhardware.io/view/664/Raspberry-Pi-Power-Supply-Board  the area around the SoC can be removed easily?

 

By the way, for what reason did you add USB-A Connectors, there are already 4 ?

 

4 hours ago, Tido said:

By the way, for what reason did you add USB-A Connectors, there are already 4 ?

 

On 4/3/2018 at 6:00 AM, TonyMac32 said:

I will have moved one of the USB ports though since @chwe happily pointed out that Y-cables don't reach that far to get from the real ports to the power-only ones...  (oops)

 

 

 

On 3/9/2019 at 3:36 AM, chwe said:

I will have moved one of the USB ports though since @chwe happily pointed out that Y-cables don't reach that far to get from the real ports to the power-only ones...  (oops)

Yes, but no.  The target of my question was another. I will rephrase it for Tony.

 

@TonyMac32, your design is for 'only' 15Watt of which the SoC can consume most of it, if not all. For what reason did you add USB-A Connectors, there are already 4 on the ATB?

 

Because the board is intended to work on any Pi-shaped board that can be powered via gpio, Like a Pi, Le Potato, etc. A lot of those don't pull as much as the ATB.

Sent from my Pixel using Tapatalk

Hi @TonyMac32,

I just looked at the BoM and because I don't want to solder SMT myself I tried to identify THT versus SMT. While doing so I came across two minor bugs for identical parts I guess: 0603 vs 0604:

C2	CAP CER 0.1UF 50V Y5V 0603
C3	CAP CER 0.1UF 50V Y5V 0604

R1	RES 100K OHM 0.1% 1/8W 0603
R3	RES 100K OHM 0.1% 1/8W 0604

If I look at the bigger picture I guess we could aim for all of them to one size? :

R1	RES 100K OHM 0.1% 1/8W 0603
R2	RES 40.2K OHM 0.1% 1/8W 0603
R3	RES 100K OHM 0.1% 1/8W 0604
R4	RES 220K OHM 0.1% 1/10W 0603

C2	CAP CER 0.1UF 50V Y5V 0603
C3	CAP CER 0.1UF 50V Y5V 0604
C4	CAP CER 150PF 50V X7R 0603

Which size should I choose (I guess one-size-fits-all) ?

 

Yeah that must be a slip, 0603 is the standard package size. I need to go back and refresh that project to make sure there aren't any better options for parts now, thanks for the heads up.

Sent from my Pixel using Tapatalk

On 4/3/2018 at 12:57 PM, chwe said:

by the way, you constantly refuse my INA219 circuit on the board for a more advanced power-consumption monitoring on the board..  

no space for another 8pin smt, RN and a shunt wired to i2c and GND VCC of the barrel jack?

https://www.ti.com/product/INA219    this part is available at https://jlcpcb.com/parts/componentSearch?searchTxt=INA219

Nominal design intent was 5,2 Volt. Considering to over come the Shunt, the ugly pin-header and the aging this should be fine. However, running such a circuit all the time is not really sensible as it burns power over the Shunt.
https://www.rahner-edu.de/raspberry-pi/strom-messen-mit-ina219/ this person writes about: Shunt of 0,1 Ohm maximal 3,2 Ampere. Maybe one can change the value of the resistor, I don't know.

 

I guess for that reason @TonyMac32 wrote: An example, I could use a hall-based sensor to have a much better SNR and an effectively non-existent loss.

 

For example with:  ACS712  which is also available at JLCPCB https://jlcpcb.com/parts/componentSearch?searchTxt=ACS712 . I just don't know how to put that into the circuit.

 

I think I'll do a redesign, if anyone is interested. This one started out as a specific Tinker-based design, then kind of mutated. I found a 5A buck converter that might be nice.

For shunt resistors, something more like 50 milliOhms as opposed to 100 would probably be better. The leads between the shunt and the amplifier need to be equal, and as short as possible.

Sent from my Pixel using Tapatalk

10 hours ago, TonyMac32 said:

something more like 50 milliOhms as opposed to 100 would probably be better.

On the german website it says: this INA219 can read at maximum a voltage drop of 0,32 Volt. It will probably work with less, but maybe less accurate - I don't know, just a guess.

 

10 hours ago, TonyMac32 said:

if anyone is interested

HERE I AM     The 15 Watts are too low at maximum SoC consumption and something on the USB or Ethernet port or worse on both.

Well, one thing to be said: I don't wanna bake PCBs - is this mentioned buck converter available at  JLC ?

 

I'm not worried about the INA219. I don't want 0.3V drop across that resistor. Unless you out the feedback after that resistor, if the buck converter supports it...

Sent from my Pixel using Tapatalk