Options for Production


#1

Two very basic questions about moving from prototype to production:

  1. For the particle cloud side of things – if I don’t want to do fleet management, but simply want to be able to make cloud function calls/webhooks – does that require paying per device?

  2. If I am using a particle argon as the basis of my product – what are my options for moving that to production? I see there is a production board out there but not yet available for purchase. Also, what about the photon? I don’t see a similar production board (even hinted at) – what is one expected to do to move this into a product?

My current strategy is that I’ve worked the existing development boards into a PCB and just inserted it into a socket. This seems to work OK, but I’m wondering if there are better (most cost effective!) options.


#2

For 1: How many devices are you aiming for?

For 2: The Photon is available without pins and can be surface mounted.
For the Argon there will by a SoM called A-Series but they are not yet available for purchase.


#3

From my experience, you will want to have the flexibility of replacing the Photon or Argon until you are very certain that the board is working 100% and the connections. It is hard to get the board right and you may come across unplanned for conditions. If you end up throwing the board away or needing to replace it with a revision then the cost is a bit less if the Photon or Argon can be recovered! Also, having the micro USB connector and mode buttons makes loading software a lot easier in small volume production. I have just gone over 650 Photons and I am still using a PCB with headers - SMT!

There is a stage between Prototype and Production typically called Pilot and this volume wise might be quite of lot of units or only a handful. The Production version of the Photon is the P1. With these and with the Argon and A-Series you need to be aware that you will need to re-design your PCB because the connection is different and the power supply circuitry is not on the modules. Also, the M2 connector for the SoMs will no doubt cost a bit!


#4

@ScruffR

For 1: It’s a consumer product so potentially thousands. Of course that will have to be supported by actual sales. Is there a way to use the particle functions without having to pay for each device? My understanding is that they “pay” features are about fleet management. Purchasing the device entitles one to lifetime access to the cloud functions and ota updates. (This is the one thing that has me evaluating other options. Consumers will potentially have these in operation for years and over time that eats into the profit. Asking customers to pay a service fee isn’t going to be very popular. Running the cloud bits myself I can scale much more cost effectively – well less than .39 cents a device. More like .002 cents per device.)

For 2:
Any idea when the A-Series will ship? Any idea of the cost?


#5

Hi @armor – thanks that’s very helpful!

Is this the guy: https://store.particle.io/products/p1

Any idea if the wifi antenna comes with it? That looks like an excellent option for making a production board.


#6

Yes. There is an antenna built in and a uFL connector for a compatible external antenna. External antennas are not supplied with the P1. As you see the P1 is out of stock. It also requires surface mounting. I have not used one of these since the bulk cost difference between the P1 and Photon doesn’t justify using the P1 unless you are talking 000’s - IMHO.


#7

Hi @armor – thanks for that, super helpful!


#8

In reply to your For 1: For WiFi devices like the Photon and Argon the first 100 devices connected to the cloud are free whether setup as a product or not. The advantages of the product approach is that the console allows for easier fleet management and OTA updates. Over 100 devices you need to pay $0.39 per device per month [edit: as quantities go up you then need to talk to Particle Sales - a commitment to big volumes can bring big discounts on the published prices!]. If you have the time to manage the infrastructure and update it then DIY is an option. I have found managing the device application code and the web app enough and I am happy to have Particle ensure the secure transportation and cloud APIs work. There isn’t any customer differentiator in the the cloud API. I agree the customer value model is a tricky one. We have had to work hard to get customers signed up to a software agreement which provides them with the benefit of a connected product that is maintained and updated. Either you can ask for a monthly subscription or an up front 1 year or 3 year contract. Alternatively you could decide to roll the cloud cost into your purchase price of the device!


#9

If you’re already thinking about scaling that big, reach out to the sales team who can tell you more about potential contracts as opposed to the self-service tier. Might be worthwhile.


#10

Take note that the P1 is currently unavailable and has been for some time.


#11

The P1 makes more gpio pins available as well as jtag which might be the justification one needs.


#12

Hi @jiskandarshah

Even though I am not nearly close to 1000’s of units yet, I am faced with the same challenges. I support what @armor and @ScruffR said, they make good points.

From my perspective (as a newbie) I do not really see the need to upgrade to P1 just yet. I have multiple Photons in commercial and light industrial environments and they seem to do just fine, with the added bonus of being able to easily swop them out should something go wrong. I mount them on custom designed PCB’s (in headers or surface mounted depending on the profile) which eliminated the need of any wires and connections which already is a huge improvement for me.

I am not sure whether there are any downsides to using Photon and Argon modules instead of production boards, maybe someone more familiar with Particle can assist with that.

Due to the huge difference in our exchange rate vs US$, the cost per device is a huge challenge here as well. As @peekay123 mentioned, if you are looking at possible 1000’s of units I would most certainly recommend speaking to Particle sales team, well, at least, this will be my approach if our prototypes pass the test and we need to go into production :slight_smile:

Regards,
Friedl.


#13

@friedl_1977 – thanks so much for all the commentary. It’s great to hear all that!

I also had a question about Argon vs Photon for production – all I really need is the wifi component and the Photon seems to have that bit. If anyone has anything to say about picking the Argon over the Photon I’d love to hear it. The price difference is substantial. Other things I’ve looked/looking at include the NodeMCU (available for ~ 6 bucks and does wifi) and the TI MSP432 MCUs. Both of those options of course require one to roll their own cloud solution.


#14

Hi @jls

If this is all you are after, I would go with the Photon for two very obvious reasons:

  1. Smaller footprint which in turn will allow you to build a smaller IoT device :slight_smile:
  2. As you correctly observed, substantial price difference.

My opinion then is to seriously consider Photon unless of course;

  1. You require these devices to connect as a Mesh network (in which case you will require Argon and Xenon),
  2. Require cellular connectivity in which case you would have to go wit Electron on Boron OR
  3. You need the increased capacity of the Argon as apposed to the Photon (more RAM, more pins and speculated superior ESP32 WiFi performance, suppose it depends on what you need)

This is exactly the internal discussion we recently, also do to costs. I persisted that we go with Particle products for various reasons.

  1. Support… Support and did I mention Support :slight_smile:
  2. If you are going to scale that large, availability is a key factor. I would hate to have to buy via 3rd party vendor in these QTY’s.
  3. Fleet management included in Particle solution.
  4. OTA updates - This for me is crucial. I know it can be achieved with NodeMCU, but not nearly as seamless as Particle.
  5. In my humble opinion, superior performance to the other devices you mentioned.

I do not have a problem with all open source as it is, but scaling at the size you mentioned, I would hate have to troubleshoot and get support from open source forums. Coming from Arduino and NodeMCU environments, I can tell from my experience, I will not consider anything other than Particle when building products for clients :slight_smile: Lessons learned…

Might seem biased, but I always try to build products that require the least possible end user support, both for my sanity as well as customer satisfaction. My first project I mentioned here (the RFID reader) started with NodeMCU. I very quickly recalled the scanners and changed to Particle Photons :slight_smile:

I hope this helps,even though it might be less technical and more user opinionated advice. Have you thought of how you will display your data? Maybe have a look at Ubidots, very easy to get going.

Regards,
Friedl.

@mariahernandez


#15

Hi @friedl_1977 – this is a great discussion, I appreciate you taking the time to chat about it.

The photon would perfectly meet my requirements as I do not require mesh or extended pins.

This was my suspicion as well. Out of the two options (nodemcu or ti) I was was favoring TI because their support is also excellent and TI is a very experienced player to say the least. I’ve ordered some NodeMCU evaluation boards but who knows what the consistency in these boards is – I’ve never heard of any of the suppliers.

However, for the reasons you have mentioned, I am probably still strongly pro-particle. Before the bluetooth support was released I had to actually dump the firmware and write a small test application that relied directly on the internal APIs. My experience with this was pretty excellent and the tooling seemed fine; also the engineers I interacted with were super helpful (I found some committers on github who were helpful finding all the bits I needed). In general I prefer to just write the application in C++, so I wish that process were generally a little smoother (eg, OTA updates when working like this), but I’m hoping it gets better in time.

Again, this is excellent feedback. Thanks so much for taking the time to write it!


#16

hi @jls

No problem. There are some gentleman in this forum, that has bent over backwards to help me. I am sure I have put their patience to the test, nevertheless, they persisted and remained very helpful, so I am glad if can help someone in return.

I do not have experience with TI’s MCU’s, but whenever I am looking for some sort of IC, I find myself first seeing whether I can find one of theirs. I have had pleasant experiences using their IC’s. Having said that, in terms is MCU’s… I like the fact that this is a focus area of Particle.

I honestly don’t think you can wrong using either Photon or Argon. Depending on the setup, you might even consider setting up a Mesh allowing you to use Xenon’s which is priced ±US$5 less than Photons. Of course in this case you will need at least one Argon to act as the Gateway. I am busy with such a project designing some Automation for our Coffee Roastery (referred to in my original reply). Here we will have a fruit salad of devices, Argon, Xenon and some Photons all having to work together :see_no_evil:

In my experience, consistency is not great. It is fine for projects at home where you can restart the device if stops responding and fiddle with it to get it back up again, but where I had to scan 100+ RFID cards consecutively and posting to MQTT server, it was a disaster. I also built a Bluetooth reader to connect to an iPad, but if inactive for ±2 minutes, we have to restart and pair again.

So I am deathly allergic to the ESP 8266 chip… :grin:

Side note: One more reason (outside of the ones mentioned before) I am also considering Argon is it seems the footprints and pin-outs of Xenon, Argon and Boron is similar. If I am correct, this is a huge advantage if you are having custom PCB’s designed as it will give you some flexibility to use either at any given stage. Please confirm though, just had a very brief overview of the footprints.


#17

Question – can one still get 2nd gen devices (Photon/Electron)? I don’t see them in the Particle store anymore. I can find some on-line, but many are at below list price, which I take to mean that these vendors are dumping their remaining supplies (because they can’t get new stock from Particle?).

I’s love to see Photons continued to be manufactured and offered for sale by Particle. Some additional advantages of Photons over 3rd gen devices (if you don’t need mesh) are:

  • 5 volt tolerant pins.
  • diode connected USB power (so that you can apply +5 volt power to the VIN pin without worrying about someone connecting a USB cable (for signals) and causing a power short).

I really miss these Photon features on the Argon (which is otherwise an excellent product).


#18

Hi @BobG -

Photons are most certainly still available. I could not find any mention that this would be the case, but I guess I cannot speak on behalf of Particle :grin: I am sure they will give a heads-up. With regards to the on-sale Photons, it might be due to increased popularity of the Mesh devices?

I actually just bought some Photons from Particle store recently and just double checked now, they do still appear there. Having said this, we buy from the wholesale store section, not that I think make a difference :slight_smile:

When Mesh is not needed, Photon remains my MCU of choice.

Best of luck.
Friedl


#19

@jls Hmm, depends what you need. Just connectivity to some other platform (PC/Mobile)? MQTT is the way to go if you use AWS you do not need to host anything and it is cheap! If you need device-management similar to particle, yes you will need to do this yourself (but again, don’t underestimate what you can do quickly in AWS)


#20
  1. You must have gotten some quality support :slight_smile: I wish I could say the same.
  2. Photon’s and P1 are originally BroadCom, then Cypress now made by USI, not by Particle. I would rather buy directly from the source, no middle man.
  3. Fleet management from Particle is nice and has value if you need it.
  4. OTA comes with lots of other platforms and in some cases is better implemented/supported than Particle’s.
  5. Superior performance only relates to what is needed. The Photon/P1 is a 120Mc ARM cortex M3, the TI and Node chips are slower but that matters only if you need computational speed. Get a TI CC3220 which compares pricewise with Photon and you’ll have an ARM M4 at 80Mc and more RAM/FLASH.

Not denying your experience nor opinions, just showing that there are better options than Photon/P1 for a given problem - it is always best to spend time to research your platform because you’ll be sinking in lots of time to make it all work and it is nasty to find out much later that there are … issues…