Particle generation woes

Hi all,
I’ve got a new product coming up and the customer suggests using Particle. I’m a hardware and software dev that have benefited greatly from the Particle platform in my consulting life. Given my experience with the Particle platform and ecosystem (overwhelmingly positive), I should jump with joy, but I’m not. I’ve pondered this for a long time and my main problem is the lack of focus on radios that work well. If you make a product based on WiFi, you can rest asured that it will “just work” in any controlled setting. If a client is using your product in a factory, you do not know if they have cellular coverage, but WiFi can usually be arranged even if you are 50 feet below ground.

I’m not counting the Spark since that was just a dev board. To me, the first gen of Particle hardware was the Electron, Photon and the P0/P1 modules. These all worked well and when you had prototyped using the dev boards, there was a clear path going forward to develop your products. Particle had many pages on their website dedicated to explaining how you went from prototype to fully certified board. Great stuff! I have developed 3 commercial products based on P0/P1 for 3 different customers. It all runs well on a daily basis and many people can depend on these products.

The next gen hardware was odd. There was the huge E-series module and then it was the Argon, Boron and Xenon. I purchased a few of these boards when they arrived to do some testing. I couldn’t get the Mesh part to work at all, despite following instructions very carefully. Based on this I concluded that I was too early, so I left it on the shelf for a year. I then tried again, had similar issues and decided to not use this in any product.

Not only that - the formerly great pages that explained how to go from prototype to product using Particle suddenly disappeared? I have not yet seen a replacement, so this certainly makes Particle loose a lot of potential customers. This was one of my main reasons for going with the platform initially. Newcomers to a platform just don’t understand how they can go from using Particle Dev Boards to making a product. Now, the best they can do is ask the friendly people on this forum for advice. Most users have a hard time grasping why they can’t just slap on an ESP32 module. They don’t understand the value proposition that Particle offers. We as seasoned Particle devs know how much time we save on each project, but newcomers don’t get it.

While Particle spent time on getting Mesh working, my customers started asking for 5Ghz wifi support. I asked, but nothing was ever announced from Particle. Competitors like Electric Imp have had this for many years, so Particle started to look kind of “old” despite trying to get Mesh working. As time went, more and more bugs were introduced in the firmware and it really seemed like there was so much focus on Mesh that testing with Gen 1 was skipped. We worked around the bugs, but it didn’t instill confidence with my customers.

The only thing that looked promising in terms of Particle hardware were the System-on-a-Module (SoM) with M2 format. Imagine if you could just click in the relevant module for any customer and know that it would just work, not matter the connectivity? This is something I really could use, but there is still no Wifi module. Cat-M1 is fine if you’re in the Americas, but the rest of the world has deployed NB-IoT https://www.advantech-cl.com/resources/news/44842bfb-ccc6-4642-aa53-c9d82386cbbb so with Particle only offering CAT M and no NB-IoT hardware, I really don’t see a clear path forward, given that the two main platforms for me is missing in the future plans?

It should be clear to anyone that you cannot make products based on Dev boards. You can’t just slap an Argon onto your custom PCB and be happy with that. A brief datasheet won’t get you through CE and CB testing. You’ll need modules that come with full documentation, down to the RoHS compliance for every part used on the module. For P0 and P1, this is available through the vendor of the module. For Argon, there is no BOM available that I can find.

Dare I bet on a company that seems to dismiss the only global radio there is (wifi) and only expands in terms of what is deployed in the US? Also - I’d appreciate to get some assurance that P0 and P1 still are good options for new designs and I would absolutely love to hear about the P2 module that supports 5ghz wifi, the P3 module that offers easy to use LoRa radio, the P4 NB-IoT module and the P5 module (a low piced STM32F205 without a radio) that lets you buy these chips with preloaded Particle firmware for ease of use. And then - all of these in M2 form factor so I can pick and choose what the customer wants. Please?

3 Likes

maybe @will can help here?

Hey there @jenschr – would love to have a conversation about some of your concerns. Particle’s path and direction have changed at least a couple of times since we first launched the Photon in 2015, which I think is common for growing startups trying to find the best path to achieve their goal, which for us is enabling product creators like yourselves to deploy real IoT products that solve real problems and create real value.

Let me address a few concerns that I pulled out of your post:

I hear you – with the majority of our enterprise customer base on cellular and 2G/3G sunsets rolling out across the world, our focus has been on building out a global LTE strategy that supports our customers in navigating that transition. That does not rule out the possibility of a Wi-Fi M.2 SoM, but wanted to provide an update on where our new product development resources have been going.

This is true, but the carrier landscape tends to be far rosier than reality, and there are some very real challenges that prevent NB1 from being a viable technology for IoT product creators:

  • It is designed for stationary objects and does not have support for mobile handoff, making it a poor solution for asset tracking use cases (a huge part of our customer base)
  • There is very little if any production-ready roaming agreements in place for NB1. The first agreements (between big carriers like AT&T and Vodafone are just being established and have not been broadly tested). This means that, if you’re deploying internationally, you’d need a different SIM (and thus a different carrier integration and SKU) for every carrier / country that you’re deploying into
  • The best solution for LTE-ready global deployments presently is LTE CAT 1, and we’ve recently launched a new SoM (the B523) that is targeted for production-ready deployments in Europe and leverages our M.2 form factor

I respectfully disagree. RoHS compliance is self-declared – we do the audits with our downstream suppliers to ensure our products are RoHS compliant and provide declarations of conformity to customers upon request. It is not necessary to have a full BoM to achieve certification and this is a process we support dozens of customers with each year.

Unfortunately I cannot give you that assurance, and we do not recommend the P0 and P1 for new designs. This is because those parts are relatively old in semiconductor terms, and ongoing availability is determined by USI, our upstream manufacturing partner.

We recommend that new designs leverage the Argon, which includes a more modern ESP32 modem.

We’re committed to the M.2 form factor for just this reason – forward compatibility. Glad to hear that this design format creates value for you as a product creator.

2 Likes

Hello @will ,

I have been a big fan for particle and have learnt alot from this community. However, with all the great help and support there have been a few set backs, and its coming to the point that forcefully and painfully i have to separate my paths.

  1. I had a product with Mesh, spent 3 4 months developing the prototype
    Result - Mesh got discontinued
  2. Spent some 6 months developing a P1 based production board
    Result - P1 is not recommended for new designs
  3. Wanted to create a BLE only based project
    Result - Xenon is discontinued, should use Argon
  4. Argon is not BLE only … but ok
    Result - Prototype is apparently the only thing you can do with a GEN 3 device. No way to build a production board.

The last point is important because a big part of IOT is size and power, which is the reason of having tech like BLE so you can build low energy devices, and hence use a smaller power source and together build something like a wearable. But right now all you can do is use the Argon as it is, which kind of kills the purpose of the previous two points. I remember there was some talk about an X series but that got discontinued as well?

Correct me if i am wrong, but the the way things stand the only thing that can be done if i want to build a product around WIFI or BLE is just use particle devices just for prototyping purposes? And for actual production development look for support outside of particle?

Right now i am looking into the ESP32 to replace my P1 designs . And trying to work with Nordic for my BLE product idea.

3 Likes

P2 module that supports 5ghz wifi, the P3 module that offers easy to use LoRa radio, the P4 NB-IoT module and the P5 module (a low piced STM32F205 without a radio) that lets you buy these chips with preloaded Particle firmware for ease of use.

Would add BLE somewhere there as well. But i doubt any of these would be coming any time soon? would this all remain a dream @will? :cold_sweat:

It’s fine to wiggle a bit, but I’m surprised that Particle would leave the niche they had? It was the #1 goto solution for any IoT device that you wanted to develop with just a small team or you wanted a fast development process despite higher price per unit. If you had more money, you’d just roll your own solution using low price Espressif modules. The same goes for anyone making price sensitive products. Particle have always been too expensive for many types of products. For those that only need a sensor or can use something kind of standard, it’s better to talk to Electric imp. I have a really hard time seeing where Particle is positioning themselves now? This used to be very clear.

I completely disagree regarding NB-IoT and Cat-M1. These two radios serve very different purposes. The profile for both data speeds, radio range and power usage are massively different. I’ve used Ublox modules for NB-IoT and it’s quite mature. Handover is not really an issue for a battery-saving device that only turns on every 15 minutes or less. If there is anything to connect to, it will happily connect to it and with better range, it enables completely different use cases than Cat-M1. It’s obvious that Particle would support Cat-M1 first, given that one couldn’t test it in the US until now. Not having devkits for NB-IoT or at least a plan seems short sighted to me.

It actually sounds from what you are saying that Particle are “stuck” business-wise and can only find good usecases for mobile? That’s a very odd situation given that Particle SIM’s and data pricing seem to be much more expensive than what I can get here. No sane customer would opt for a Particle sim over here in Europe. I guess we’ve got more competition?

As for the dependance on USI. If you look at the inside of the modules - it’s not rocket science? It’s two easily available IC’s and the required supporting electronics.


Given that all Particle customers were unable to get these for 4-5 months (not counting the earthquake that destroyed their factory, but after that too), I’d say solving that would be important. I’ve tried talking to USI directly (and their local resellers) and they’re not interested unless you by 6k at a time. Particle gets much better prices than I could ever get from them, so I’m suspecting that Particle’s volume is pretty good?

I’ve been sitting in so many customer meetings discussing why one should not buy the cheaper Espressif modules rather than using the Particle eco system, so it really hurts me to hear that you now recommend building based on ESP32 rather than the P0 or P1. Business wise, I just don’t see how that can succeed?

I find it incredibly frustrating that I have to explain my customers the advantage of using Particle. Formerly, the website was full of good pages I could reference when pitching, but these days the site is so corporate that customers won’t have any idea of where to look. “I can’t find any hardware on their website” was what one customer recently said. He didn’t want to click the links that aggressively pushed Buy this or that. He just wanted to read about the hardware and find what the “Particle advantage” was, compared to buying IC’s from “normal” vendors and design your own hardware.

Given that you said that P0 and P1 no longer are recommended for new designs, I guess I’ll better have my customer stock up on these while they can? We’re having a design meeting on Monday, so I really don’t see how I could recommend using Particle for the next generation? Please tell me I’m wrong here…

3 Likes

I agree that with the P0 and P1 sunsetting that there is a gap in a similarly priced WiFi module product, but I do not think they’re recommending people use a raw ESP32 module – instead the Argon that contains an ESP32 as a component.

1 Like

@HEng Replacing a $13.50 P1 with a $27 Argon is a definitive downgrade to me when they do the same thing. Did I miss anything?

1 Like

Yes, the Argon has BLE which allows a slicker user experience for device registration. There are also more resources on the device, which makes it a clear upgrade. I would hope that the module form factor will be available at some point at a lower cost though.

Nahh. The wifi process works fine without BLE. I’m not sayin that BLE isn’t nice, but it just isn’t neccesary? Plus BLE does come with it’s own set of challenges. I’ve used Particle Photons to teach IoT with 50-70 students per year for 5 years. It has never failed completely, though I do know all the LED codes by now and how to restore keys.

Bluetooth has however caused issues that were unsolvable when students didn’t have BLE 4.1 compatible devices. You cannot judge stuff like this based on the hardware that you (as a developer) have. You need to field test it. WiFi just works. BLE will hopefully “Just work” soon, but it’s not quite at the level of penetration of WiFi. Official sources state 90% BLE only in 2023. Wifi is in 99% of all networked devices today, including 8 year old devices.

1 Like

While you may have technically savvy users, exiting the app and reentering is clunky on a good day. Argon solves those device registration issues in iOS by “exploiting” the BLE “backdoor” to make a much better user experience.
You also need to consider that while BLE market penetration isn’t perfect, your students probably have crappier hardware than the average consumer. Apparently BLE support goes back to iPhone 4S in the Apple ecosystem, which was released nine years ago. There are not many of these still around.

Wifi itself isn’t without difficulty, and I know you can’t say you haven’t had a Photon that you’ve had to reset half a dozen times to get it to connect in Listen mode properly. There have been lots of little interoperability issues I’ve seen over the years with various routers and things that I hope a more heavy-duty wifi stack can handle more gracefully as well.

The lack of a lower priced Argon module is concerning at the moment, but I can’t imagine that one isn’t in the works. I like the fact that they’re focusing on fewer products because that means a lower configuration and QA burden and hopefully that translates into a higher overall quality.

I am curious to see what future wifi support looks like.

1 Like

We use https://setup.particle.io. Works great. None of the issues you mention, but I agree. The app is far from optimal, but little have been done to it through the years.

Hey @jenschr, it feels like you read my mind.
After seeing the pie chart that Zack showed at Spectra (half the sales are from tracking apps) it makes perfect business sense to give WiFi a lower priority. I hope some of that revenue goes to improve or advance the WiFi products, but they have to pacify the VCs as well.
Now I understand more about the death of the Xenon, although I still don’t see why not sell it as a BLE device… I digress.
For reading that lowly sensor, this platform has a higher starting cost (even for the P1, which being phased out, should cost less). I looked at what I get for that price and in most cases it’s justified.
I started from a place where I needed new skills to do my job, so the cost was not an issue. I acquired many skills in the last few years that I am now able to mix Particle devices with other hardware. So, I can think more about cost now.
I hope Particle will throw a few bones our way because it should also be about enabling people to learn, build, invent and create the need for new products. And also because, even after spending many late nights with a Photon, I think it’s fun.
I agree Particle feels “stuck”, but it’s only because they want to be. They showed incredible resolve with the release of three new devices (plus the SoMs, debugger, adapters, cloud support…now the tracker) in two years. So, we should not feel “stuck” either.

Hey folks,

Lots of good discussion since my last post, so want to try and provide some higher level comments rather than try to respond to each specific point. A few things to clarify:

  • We have no plans to deprecate or “phase out” the P0 and P1. We’re simply identifying with the NRND moniker that it is an aging part, that Particle is not in direct control of its production (as we are for all our other SKUs), and that customers who can leverage the Argon are encouraged to do so. We have many customers on the P0/P1, still release Device OS releases, and the product is fully supported on the platform.
  • We’re more committed than ever to support our customers at scale. What your statement omits, @jrjack, is that we’ve released the B402, B523, Tracker SoM, and Tracker One products which are all based on Gen 3 and are all designed for mass production.

    A more appropriate question might be, "why the focus on cellular?" versus Wi-Fi so far in 2020?

    The reasons span from strategic to very tactical. As an example, we know we need to deliver a clear path to global LTE coverage because 2G/3G networks around the World are being deprecated, especially in our primary US / European markets. None of this means that we are done investing in Wi-Fi, simply that so far this year our focus has been on cellular.

  • I have no issue with NB1. In fact, we’ve successfully tested Particle devices on NB-IoT networks! At the end of the day, I think M1 and NB1 are far more similar than different in their value prop to IoT product creators (lower cost, lower power, better RF penetration). At the end of the day, though, it’s simply not possible today to deploy a NB1 product with a single SIM within the entire European content (much less the World) like you can with LTE CAT 1. This may change in the future, and we’re tracking the maturation of global NB-IoT network deployments closely.

At the end of the day

much of what I’m hearing in this thread is disappointment that we haven’t invested more resources in early 2020 towards advancing our Wi-Fi (5GHz, A SoM), NB-IoT, and previously deprecated Mesh product lines. I hear your voices and deeply value your feedback, as we build our roadmap around what we feel will help make our customers successful.

^^ Absolutely – one of the hard lessons we’ve learned as a growing organization is that the road to hell is paved with good intentions, and that no one is served by our overpromising and underdelivering.

While we haven’t been able to advance the platform in early 2020 in all the areas we’ve discussed on this thread, we have been able to invest major resources into global LTE, a new branch for more reliable, predictable LTS releases for Device OS, improve the reliability core OTA updates features, and launch new products that better support our growing base of asset tracking customers.

We hope that these investments will put Particle and our customers in the best position to be successful for the long-haul. As Zach stated at Spectra, we expect COVID to separate the “vitamins” from the “medicine” in IoT, and are determined to show the unique value that an integrated platform like Particle can provide in this time when flexibility, time to market, and trustworthy partners are more important than ever.

6 Likes

Great reply @will!
This is what I had hoped to hear. For some reason, there still is no published video from Spectra and I got into the event a little late, so I missed the keynote. Reading the slides from the Keynote shows me that you’re still thinking the right way. 3G is already turned off in most parts of Norway, so putting resources towards solving mobile connectivity makes a lot of sense in that regard. LTS releases can solve some of the headaches we’ve had recently.

I also didn’t know that NB-IoT primarily was a SIM issue. As mentioned - pricing wise it still doesn’t make sense for me to use a Particle SIM, so I’d appreciate if the products were not forcefully tied to Particle.

Wifi still is the most important IoT radio to me, but I can obviously transition to new services such as the Tracker SoM which looks like a great kit. I think my woes could be solved if Particle would create the Argon SoM in M2 form factor. Alternatively if they could publish something on how to use Particle services & software on self-developed hardware (as in an Argon reference-design). As mentioned, it is very limited what devices you can put a $27 module into and still make money. The P0 & P1 is already kind of expensive at $13.50. An Argon design requires an ESP32 for $3 and a nRF52840 for $6.50 (in low volumes), so I guess one could do a custom board at a comparable price as the P0 and P1 - given that one had a reference design to start with?

Here’s the videos from the event, in case you wanted to circle back on the Keynote:
https://spectra.particle.io/recordings/

Thank you for taking the time to communicate that, as you’re not alone with the request. It is being seriously considered – I will provide more updates if/when we have them.