Where can I find smaller versions of components?

Hi guys, I’ve got a pretty simple device but I’d like to make it smaller.

The largest components are:

  • Pi zero - which I’ll replace with a nanopi or some other cheap/small SBC.
  • W5500 Ethernet board - which maybe could be replaced with something like this but I don’t know if you can just wire that to the gpio(?).
  • A relay - this one looks smaller.
  • USB/AC/DC power adapter (just disassembled one I had lying around) - I’m really lost on this one. I got a little help here but I’m still not sure what to buy.

Finally, if possible I’d like to just get all of these things put onto one board (with the possible exception of the nanopi). Obviously cost matters so for example, I wouldn’t want to spend $10 more on a slightly smaller relay.

This hasn’t been a straight forward process so if you’ve helped me out in the past I did listen but as I’ve been moving through the prototyping process I’m seeing that I need different things which has changed my plans. Anyways, thanks for any help.

Key decision time:

  • If you base your work on the ESP32 (not capable of running Linux, but has a RTOS) it’s easy (as these things go) to grab an existing design and modify it. This is what https://project-hermes.com/ has done to create a single integrated board for ocean temperature sensing.
  • If you want Linux, look at complying with the standards for breakout boards for your platform. For Raspberry Pi, look at the PiHAT; for BeagleBone, look at capes

Note that using Linux on ARM means also learning device tree, not just “big banging”, typically. I recommend getting a copy of Derek Molloy’s Exploring Beaglebone for BeagleBone, or Expoloring Raspberry Pi for Raspberry Pi.

Getting started is a bit easier on ESP32; you can start with Arduino if you are already familiar with it, but can also choose lots of other environments, and all you need to do is explicitly and directly set up port mappings. And you’ll get wifi and bluetooth out of the box, and you can add ethernet by connecting a PHY to the correct ports; see https://kazkojima.github.io/esp32-phy.html for example schematic. See https://leanpub.com/kolban-ESP32 for more information on ESP32.

ESP32 is also easier to integrate into a single board, in part because it’s designed for that purpose.

1 Like

Dang, well that definitely is something to think about (ESP32). My device does use some apps that will be easier on Linux (since they’re already made) but it could get by in a simpler form with just some basic TCP/IP knowledge on my part.

In any case how would you go about procuring the components? You seem pretty savy with electronics so would you design it all yourself (including modifying an existing design) or use whole off the shelf components?

I hesitate to give more specific advice because I really don’t know your constraints. This goes back to my CrowdSupply suggestion in another thread (which would require open hardware) — they have supply chain experience that I don’t.

I make for my own pleasure and not to sell, so sometimes I’m running wires on perfboard in an ugly tangle just to try out an idea and sometimes I’m designing my own boards in KiCad and having them made at OSH Park or one of the many inexpensive fab facilities oversees, but in either case I’m often optimizing for parts I’ve already purchased as part of other projects. I often purchase extras because between having extras in case of letting out the magic smoke and cost breakpoints, I often find it sensible to buy a few more than I need. Some parts I expect to use a lot of and will buy larger quantities, like 100 PMV20XNER because they are a good match for a lot of my projects, or a whole book of multiples of common through-hole resistors.

If I were designing for sale, I’d be looking instead at current part availability and cost. I think I suggested circuithub for that. It’s certainly something I’d consider if I were kicking off a commercial project in small quantities. A lot of choices would depend on the target market. Small quantity sales, high price, lots of value in the software? Optimize for spending less time saving pennies on the cost of each physical unit. Only makes sense in massive quantities at lower price? Worth spending more of your time to figure out how to save every cent you can on reducing cost. It’s just economics…

When it comes to electronics knowledge, I spent a lot of my free time for several years getting to the point I’m at, which is still the “tyro” stage. So either good news (someone who started out wondering what units applied to an RC time constant got to the point of initial successful circuit design in a few months of fairly dedicated learning and then progressed from there) or bad news (after years of experience, still a tyro).

1 Like

I’d say it’s encouraging because you can make it happen and in the end that’s what counts.

I’m keeping mum for two reasons. One, I don’t want anyone to tell me it’s a bad idea or that I can’t do it because either will be demotivating. Two, it could get stolen by someone.

I guess for now I’ll stick with what I’ve got/found and hack where necessary to get a production looking prototype done.

Hopefully I can get it small enough with just the components I’ve found so far. No sense in going crazy about production details until I can get some external feedback. The only two I’m super worried about are the Ethernet board and power supply but I’ll check tonight.

I’m decidedly just in this one for money. …with the mindset that if it doesn’t work I’ll try something else.

You could reduce size by combining processors and device libraries. Most of these single board configurations are built for generic use. Customizing the hardware, removing unnecessary components, can reduce space. To provide advice the amount and type of processing necessary would need to be defined.

Smaller relays are available and you can replace the relay with a small Mosfet. The characteristics of the load your are driving would be needed to design the driver or substitute the relay.


Hi Sean, you could outsource the smd design to a contractor and have a NDA agreement in place. I went the whole route from idea to custom production and Alibaba is a good start to ask for prices via an RFQ. For example a laser cut case from acrylic cost me in Australia $2400 for 150 cases. In China $5/case but I found a supplier who delivered for $1.80. Sometimes it’s a hunt and a lot of going fowards and backwards. I definitely encourage you to give the process a try.

1 Like

Hi Paul, Thank you for the encouragement. I actually just logged in to do exactly that. It’s weird when things synchronize like that sometimes. I mean I could design the board if I spent a ton of time learning how to do it but I don’t think that makes sense. …I’m just here to PM another user and say thanks to you and @donkjr for the tips. I’ll definitely keep them in mind.

I’m just a hobbyist here giving advice.

The hard work here isn’t laying a component on a board and routing some traces. It’s designing the schematic for the circuit, looking for flaws in the schematic design, working out a board shape that will hold the parts and meet your physical constraints, selecting appropriate design rules for the application domain (which, given that you are talking about a relay, includes being specific about voltage and current capacity, and current capacity depends on physical constraints for dissipating heat), choosing components that meet your needs (including connectors that will be safe given your constraints), laying out the board, checking the design for defects including thinking through potential failure cases, etc. You bring up a relay, so you are probably doing this in order to switch a high-voltage and/or high-current circuit, possibly switching AC.

As a hobbyist, I’m inconsistent in knowing what could go wrong — there’s a reason for engineering certifications here! I know some of the things that can go wrong here, but I’m not trained in all the things that are reasonably likely to go wrong. I can imagine some things? What happens if your relay welds closed and doesn’t turn off? Does it create an unsafe condition? Have you properly separated low-voltage and high-voltage sides of your circuit board? If you are switching AC and decide to replace a physical relay with a solid-state relay (SSR), do you know that its common failure mode is to fail on? I’m not giving legal advice of any sort here, but if I personally were selling electrical devices I would be getting advice on liability…

Reality-check, since you say you are in it for the money: The high-value things you can do in the market with an embedded network-connected computer and a relay are vanishingly small. Make sure you are aware of what people are doing with sonoff switches, which are based on the 16-bit ESP8266 predecessor to the ESP32, before you invest.