About board design decisions

[q2dg2b]

Hello.

I don't know if this place is the place to share this question, but I haven't found anything (forum, github...) more suitable.

The question is: I've seen new Arduino UNO boards manufactured by Adafruit don't have assembled ICSP pins to be able to (re)program ATmega16U2 chip. Former Arduino UNO boards from Italy had it (and still have it). Who (and why) has decided this change?? I'm very upset with this uncommented shortage.

On the other hand, now Arduino has achieved new manufacturers, it would have been a good oportunity to make some changes to some boards, like putting a microUSB connector instead of the typeB conector in Arduino UNO (calling it "R4" then, for instance). It is only one idea among many...

This google group is for developers, but I think some discussion about board design would have been very constructive during these so convulsed days

Thanks

[q2dg2b]

Well...I swear I wrote previous post before last news about new products (https://blog.arduino.cc/2015/07/09/discover-the-new-arduino-product-page). These recent news give a really different perspective of everything (for those who want more, there some interesting comments below that text explaining a bit all the fuss...)

But the question remains the same...Who (and how and why) decides the design of the Arduino products? How people can suggest improvements? There's a very interesting thread in the forums title "Suggestions for the Arduino project, but, unfortunately, it seems nobody from Arduino Team has time to read it.

Thanks

[bob cousins]

Arduino hardware follows the cathedral model of open source : it's not a community effort.

The reason for retiring boards is largely down to the split with SmartProjects. For new manufacturers, there is a significant setup cost for each product, so they don't want to support products with low sales. There is also a natural evolution of product ranges, it makes sense to retire less popular boards and concentrate on the popular ones.

The issue of a lack of responsiveness and feedback from Arduino to any form of communication is well known and remarked on regularly. But of course, Arduino are not obliged to respond or spend any effort on user requests, that might be an expectation setup by the "open source" tag but is not guaranteed.

In the end, we have to be grateful for what Arduino do provide (which is a lot), and if we want anything different contribute it ourselves.

[Massimo Banzi]

Dear q2dg2b

I decide. 

If you have any suggestions please let’s talk about it here. I’m happy to discuss the thinking behind some of our choices.

In general Arduino is more free than before, we are not constrained by the decisions of a single manufacturer, so expect more products to appear on the horizon based on partnerships with a variety of OSHW players.

The rationale behind some some of these decisions falls within these general categories:

* Products which do not make a lot of sense anymore: Either they are available from some of our friends and it makes more sense to recommend those than manufacture something which sells 500 units a year or something.

* Products that are ripe for being re-imagined. So , for example, the Leonardo is not “dead”, it will come back in an improved form.

* Products that are overlapping and/or too expensive. The new Zero and the Due have usb host capabilities, it makes more sense to enhance and document those instead of having a selection of products which are all supposed to be documented, maintained etc. 

Take the ADK, Google put zero effort in promoting that standard and got very very few projects made with that. I rather make sure you can use a Zero as an ADK than maintain an old expensive board.

* Clearing the product line to prepare for the arrival of new families of products

* Cleaning up the product line to make it much more understandable. Less products with good documentation and content rather than a lot of half-assed products.

From a software point of view they will still be supported by the IDE for a long time. We might just move support for VERY old boards (like atmega168 or something) into a dowloadable core so that new users don’t have to struggle to understand the boards menu. A semi-skilled user can add support for those boards in something like 3 mouse clicks 

As I said if you have any questions ask me anything :)

m

--
Massimo Banzi
m.b...@arduino.cc

--
You received this message because you are subscribed to the Google Groups "Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to developers+...@arduino.cc.

[q2dg2b]

Thanks A LOT for your response, Massimo. Glad to hear you!! (A little complaint: if these explanations had been written  -in this group, for instance, or in the forum- simultaneously with last blog entry, it would have been perfect. But never mind).

I just ask for one thing, though: would it be possible to create a place (here or in Github...) reserved to discuss about future improvements/designs of boards? As I said before, there is a "Suggestions for the Arduino project" which could do the trick, but the point is the interaction with someone accredited by the Arduino Team to respond. I mean, if someone could ask (like me), "why not replace USB-B connector by microUSB connector in UNO?", it would be very constructive getting a authorized response saying "it's too expensive" or "we will see if it is possible and we'll inform" or "we don't see the need", etc. Maybe asking "why don't make UNO compatible with 3,3V using a switch?" could have a response saying "it doesn't worthy because that of that...", etc. Maybe asking "why have assembled ISCP pins dissapeared in new UNO?" or "why Tinkekit connectors in motor shield still remains?" could have a technical meaningful response, etc, etc. If these discussions were even carried before prototyping new boards, it would be a very reliable system to know the needs of the community and to cover them more exactly. I think this kind of feedback would be very constructive to Arduino as a whole. But maybe I'm wrong, this is also possible.

Thanks

P.D: Sorry, last question! You say "[Some products] they are available from some of our friends and it makes more sense to recommend those than manufacture something which sells 500 units a year or something."  Could it be possible, then, to add a special "recommended products" section inside "products" page showing some of those products (or directly selling them in Arduino store)? I'm thinking of "certified" USB-Serial adapters, in "certified" ISP Programmers, in TFT screens (compatible with TFT library if it's possible), etc. Thanks a lot again

[Wayne Holder]

If you have any suggestions please let’s talk about it here. I’m happy to discuss the thinking behind some of our choices.

I'm curious if anything can be done to provide a way to get an official VID/PID from Arduino.cc, perhaps in exchange for a royalty, for those of us that have an interest in making compatible hardware products that are based on chips like the ATMega32U4 and ATSAMD21G18A, but can't afford to deal with the fees charged by USB.org?

Wayne

[David Zanetti]

USB.org take a very dim view of sublicensing a VID, if you wish to retain your VID as a formal allocation.

Doesn't stop people either using defunct VIDs that they will never reallocate, or simply squatting on a VID, bit I doubt either of these options are what arduino.cc would be keen on doing.

--
Sent from my Android device with K-9 Mail. Please excuse my brevity.

[Georgitzikis Vasilis (tzikis)]

FYI, i think OpenMoko has an extensive list of VID/PIDs they don’t intend to use, and they give them away for free to open source projects

or so i’ve heard

[Massimo Banzi]

RE: PIDs

We’ve been discussing this topic a lot internally. We’re trying to figure out a way to do that respecting the contract we have with USB-IF.

(other entities who have been selling or giving away PIDs all had to have a chat with USB-IF at some point)

It’s most likely something we’ll offer to Makers who register their derivative product with us in some way (more details in the future)

m

--
Massimo Banzi
m.b...@arduino.cc

[Massimo Banzi]

I’m happy to debate our HW design here and we tend to design our products based on our research with different groups of users already.

The key in product design is know how to stop adding stuff. 

Many “crowdsourced” approaches end up with “Frankensteins” where people keep asking to add features that are rarely used but are cool to have just in case. This makes products bloated and complex to understand. 

Each audience might need a slightly different product. The nice thing about OSHW is that if you feel we’re not taking care of a specific audience you can make your own version.

some quick answers:

* USB micro vs USB B: USB micro are super cool. I have lots of Micro cables around but USB B tends to be more robust to everyday abuse. the UNO is still used in a lot of education environments where robustness is a factor.

* 3.3v with a switch. If you look at the progression of arduino designs we’ve been removing as many connectors and switches as possible to reduce complexity and make it easier for people to understand what is going on.. if you could turn your arduino in a 3.3v device with a switch that would be a big issue. I can see a frustrated beginner starting to fiddle with all the jumpers and switches to try to understand why their code doesn’t work and burning something in the process.

If you want 3.3v I recommend you look at our new Zero board.. it’s much more powerful than uno etc.

* TinkerKit is not an Arduino product (although I came up with the idea in 2006, it’s now part of the products that our former partner “own”) TK Connectors were added to simplify creating simple motor based applications. They might be gone in the future. Why do they bother you?

Final comment: There no such thing as “the community” but there are several different communities with different needs. They people who hang out here tend to be pros and have needs that are more specific that the wider audience of Arduino users (who don’t participate in mailing lists like these)

When designing a product one of the questions is “who do you design for?” change the target and you get very radically different designs.

m

--
Massimo Banzi
m.b...@arduino.cc

[q2dg2b]

Thanks a lot (again!), Massimo. This kind of feedback is really really pleasurable. Your point about "key in product design is know how to stop adding stuff" is really interesting to assess.

Thanks also for your quick answers...they all are very reasonable. Only four (last, I promise) points more:

1.-A question I've already asked before: Why aren't new UNOs ICSP pins to program Atmega16u2 preassembled (like they were in older ones)? Personally it's an annoying detail.

2.-Answering your interpelation: Tinkerkit doesn't bother me, but now they are pretty useless. Maybe replacing with Grove connectors (now Seeedstudio is a partner) could be a step forward. But maybe I'm wrong.

3.-I insist on the convenience of facilitate the access from Arduino store to some accessories which aren't mass-buyed but are useful to medium-advanced Arduino users, like (as I said in former post) ISP programmers, USB-Serial adapters, TFT screens (compatible with TFT library), etc...in a similar way you already sell shields or components from other people.

4.-Another user (tuzzynio) has recently commented in last blog entry about the convenience of having LTS products...maybe it's an idea to explore. What do you think?

Thanks again for all your patience and your passion.

[Urja Rannikko]

Hi,

I thought I'd put out a few comments...

On Mon, Jul 13, 2015 at 4:31 PM, Massimo Banzi <m.b...@arduino.cc> wrote:
> I’m happy to debate our HW design here and we tend to design our products
> based on our research with different groups of users already.
>
> The key in product design is know how to stop adding stuff.
>
> Many “crowdsourced” approaches end up with “Frankensteins” where people keep
> asking to add features that are rarely used but are cool to have just in
> case. This makes products bloated and complex to understand.
> Each audience might need a slightly different product. The nice thing about
> OSHW is that if you feel we’re not taking care of a specific audience you
> can make your own version.
>
> some quick answers:
>
> * USB micro vs USB B: USB micro are super cool. I have lots of Micro cables
> around but USB B tends to be more robust to everyday abuse. the UNO is still
> used in a lot of education environments where robustness is a factor.

Understandable. Does make me have 3 types of cables all for 'arduinos'
(I personally like mini, so have some of those on FTDI Basic's and
such, and then there's clones with micro and the Uno R3 with USB B.)..

>
> * 3.3v with a switch. If you look at the progression of arduino designs
> we’ve been removing as many connectors and switches as possible to reduce
> complexity and make it easier for people to understand what is going on.. if
> you could turn your arduino in a 3.3v device with a switch that would be a
> big issue. I can see a frustrated beginner starting to fiddle with all the
> jumpers and switches to try to understand why their code doesn’t work and
> burning something in the process.

how about a trace cut place (similar to reset-en) for splitting
Vcc/Ioref from 5V?.
3V3 does need a change of fuses (CKDIV8 and so on) and bootloader anyways so
making it a fast switch is not necessary, but now doing it would be
quite invasive.
There is also the issue of the 16U2 and IO levels (which would be moot
with the FT232RL with VCCIO ...)
The 1k resistors that are there would make it practically safe though,
but still not right.

q2dg2b also asked about the 16U2 ISP header removal... my guess is you
answered it with this (repeat):

> If you look at the progression of arduino designs
> we’ve been removing as many connectors and switches as possible to reduce
> complexity

I think this is discouraging hacking of the 16U2 and not helpful at
all. Even if you dont use the ISP header to ISP, it made a simple
target to hit the reset line (to use dfu) with any metal tool.
If it was just a manufacturing cost measure, then that is sad but
sometimes you need to do that...

Now that we're talking about the *U2, how was adding that (compared to
FT232RL) a reduction of complexity? It added a new firmware to the
system and i dont have much good to say of the default one.
Btw, if you happen to know the author/maintainer of it there's a
thread on this ML (started by me) you could point them to (hint hint).

> If you want 3.3v I recommend you look at our new Zero board.. it’s much more
> powerful than uno etc.

Thats cool but not very compatible and people have the lowest common
denominator, namely the Uno R3.
My experience with it is maintaining an arduino-compatible (hardware
level so only the 328s) firmware for flashing SPI chips with flashrom
(flashrom.org) serprog protocol.

In this use case it'd be nice to be able to just switch to 3V3...
people dont understand level shifters or dont have them, and fail at
building the simple resistor dividers. Resistor dividers also are
really weak so not very suitable for ISP, and not really tristateable.
(I'd actually just point them to get something else than the R3, but...)

--
Urja Rannikko

PS. In general I think your reduction of complexity in hardware is
reduction of functionality and
in software it is hiding what people are really doing from themselves
so they wont understand what they're doing... will get them to blink a
led faster but without real understanding of the system, so not very
helpful in the long term, but now I'm rambling off about non-hardware
stuff so i'll end it here.

[Roger Clark]

Urja

Re: Micro USB

I tend to agree with Massimo about the strength of connectors that are necessary.

I have some STM32F103C8 boards which I use for Arduino, and which have micro USB connrectors.

However they are easily damaged because they are only held onto the board by 2 small solder pads and some USB cables are very stiff and apply a lot of force to the connector.

I have ended up covering / encapsulating my micro USB connectors with glue to prevent them breaking, and I generally do not abuse my boards, unlike what can happen in an educational environment.

Re: 3.3 on Zero etc

I think that the issue with all the 3.3V boards, like the Due and the Zero (and a load of other third party boards like the Teensy, PIC32 and STM32), is generally not that they are ARM based microcomputers, and a lot of the AVR libraries don't run on ARM unless they have been ported.

Also the ARM boards tend to run faster (generally at least 5 times faster - though the Zero may not be quite as fast as the Due and Teensy etc), and this can cause issues with peripherals unless the library has been written with the speed of the processor in mind.

Level shifting down to 3.3V from external devices is not a big problem, as its easily achieved with resistor dividers, and most newer peripherals now work on 3.3V already, so using them is easier on 3.3V devices than it is on the Uno.

[ro...@mail.com]

hi,

    in general i think the decision to trim back the arduino product line is sound, and the choices made as to what to retire are reasonable. i do have some experience in electronic design and manufacturing, along with a good appreciation of some of the less-obvious costs that one strikes in the process.

my own personal view is that arduinos can be divided up into three general categories:

1. "gumstick" format: pro mini, nano, micro - cheap, ideal for embedding in projects.

2. robust format: uno, pro - hard to break, easy to replace processor.

3. higher end ("enhanced features") - more computing power, but expensive.

first question: of the above categories, what (approximate) market share does each achieve today?

second question: are we likely to see a shift to using the 32u4 processor in preference to the 328p? the 32u4 seems to have an advantage with the onboard usb allowing for a reduce in component count. yet a disadvantage in there being no DIP version. is DIP still relevant with "gumstick" format arduinos available cheaply?

third question: which way does the design team lean when it comes to cost versus performance? is the future seen as being dominated with ARM-based arduinos, or with the small 328p/32u4 processors remaining useful for learning and creating small projects. when looking in the ARM direction, how much concern does the us$20 Raspberry Pi model A+ create?

fourth question: rather than trying to compete with chinese clones (mostly pro mini and nano) has anyone considered a logo licencing arrangement, policed (and perhaps collected) by ebay (as the main marketplace), that generates a revenue stream back to the arduino design team?

cheers,

rob   :-)

[Andrew Kroll]

One other point on the missing header...
There are shields, some of which are in beta, some of which are available now, that  actually require that reset is held low on the 16u2 in order to disable it and allow the shield to program the board. This includes various WiFi and Bluetooth shields that provide OTA firmware uploads to the MCU board. You simply pop on a jumper cap to accomplish this. Could not be any easier.

If the actual header holes are there for you to supply your own pins, that is fine and dandy, and I can live with that.

 If the capability has been removed, then we have a major problem, and not including it for at least the OTA uploading case is a huge mistake IMHO. Lets not even get into the ability to be able to make the 16u2 appear as an entirely different device, which goes without saying, happens to be very very handy.

--
You received this message because you are subscribed to the Google Groups "Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to developers+...@arduino.cc.

--

Visit my github for awesome Arduino code @ https://github.com/xxxajk

[Massimo Banzi]

The holes are there. We simply didn’t mount the 6pin connector on the 16U2 on the first batch of US UNOs

What shields are you referring to? Link? :)

m

[Massimo Banzi]

> On 14 Jul 2015, at 18:30, RO...@MAIL.COM wrote:
>
> hi,
> in general i think the decision to trim back the arduino product line is sound, and the choices made as to what to retire are reasonable. i do have some experience in electronic design and manufacturing, along with a good appreciation of some of the less-obvious costs that one strikes in the process.
>
> my own personal view is that arduinos can be divided up into three general categories:
> 1. "gumstick" format: pro mini, nano, micro - cheap, ideal for embedding in projects.
> 2. robust format: uno, pro - hard to break, easy to replace processor.
> 3. higher end ("enhanced features") - more computing power, but expensive.

99% correct, In general this is reflected in the new diagram we introduced to explain the different product families.

>
> first question: of the above categories, what (approximate) market share does each achieve today?

Market shares are shifting at the moment. we’ll see after we are done re-organisng the product line.

>
> second question: are we likely to see a shift to using the 32u4 processor in preference to the 328p? the 32u4 seems to have an advantage with the onboard usb allowing for a reduce in component count. yet a disadvantage in there being no DIP version. is DIP still relevant with "gumstick" format arduinos available cheaply?

The DIP 328 is still very popular and cheaper than many “stamp” modules. Some products are on the 32u4 because of the single chip solution but you’ll see more Cortex M0+ products as well.

>
> third question: which way does the design team lean when it comes to cost versus performance? is the future seen as being dominated with ARM-based arduinos, or with the small 328p/32u4 processors remaining useful for learning and creating small projects. when looking in the ARM direction, how much concern does the us$20 Raspberry Pi model A+ create?

There are different segments : one where cost matters and one where higher costs are justified by more performance.
The raspberry pi is a different device. It’s a computer that happens to be cheap. Some people use it instead (or alongside) Arduino on certain projects.
On the other hand with the Zero you can develop a solution you can them transfer to your own PCB at cost which is lower than the RPi and you can power it with a coin cell for 1 year… Different applications/scales = different platforms.

>
> fourth question: rather than trying to compete with chinese clones (mostly pro mini and nano) has anyone considered a logo licencing arrangement, policed (and perhaps collected) by ebay (as the main marketplace), that generates a revenue stream back to the arduino design team?
>

Well we have a number of partnership plans in place but chinese clones are not interested in any of that. They are usually fly by night operations who are making a few bucks on extremely tight margins. We are working with some reputable Chinese companies who, on the other hand, are interested in working with us on their derivative products.

On eBay we have a TM team who reports all the counterfeits and we got a lot of them to remove our logo from their boards. It’s a whack-a-mole game.. :)

m


>
> cheers,
> rob :-)
>
>
>

[Massimo Banzi]

> Understandable. Does make me have 3 types of cables all for 'arduinos'
> (I personally like mini, so have some of those on FTDI Basic's and
> such, and then there's clones with micro and the Uno R3 with USB B.)..

We actually use only 2 now Micro and B. We’re migrating legacy products to micro

>> how about a trace cut place (similar to reset-en) for splitting
> Vcc/Ioref from 5V?.
> 3V3 does need a change of fuses (CKDIV8 and so on) and bootloader anyways so
> making it a fast switch is not necessary, but now doing it would be
> quite invasive.
> There is also the issue of the 16U2 and IO levels (which would be moot
> with the FT232RL with VCCIO ...)
> The 1k resistors that are there would make it practically safe though,
> but still not right.

Well you’ve clearly explained how this could be safely done only at manufacturing… (or by and expert user willing to re program two processors etc)

>
> q2dg2b also asked about the 16U2 ISP header removal... my guess is you
> answered it with this (repeat):

The holes are still there. We just didn’t mount the connector. the percentage of people who want them is quite small compared to the bulk of UNO users.

>> If you look at the progression of arduino designs
>> we’ve been removing as many connectors and switches as possible to reduce
>> complexity
> I think this is discouraging hacking of the 16U2 and not helpful at
> all. Even if you dont use the ISP header to ISP, it made a simple
> target to hit the reset line (to use dfu) with any metal tool.
> If it was just a manufacturing cost measure, then that is sad but

> sometimes you need to do that…
>

The holes are available so entering DFU mode is still possible like before

> Now that we're talking about the *U2, how was adding that (compared to
> FT232RL) a reduction of complexity? It added a new firmware to the
> system and i dont have much good to say of the default one.
> Btw, if you happen to know the author/maintainer of it there's a
> thread on this ML (started by me) you could point them to (hint hint).

the 16U2 allowed us to remove the step of installing the FTDI drivers on mac and Win. Currently on windows we just need an .inf file which is just a short text file so nice and easy.

Remove one step from the installation process = remove a chance a beginner can make a mistake or be discouraged by a failed driver installation (they happen and on windows we have seen some crazy behaviours like computers re-installing the FTDI driver every time you connected the board)

>
>> If you want 3.3v I recommend you look at our new Zero board.. it’s much more
>> powerful than uno etc.
> Thats cool but not very compatible and people have the lowest common
> denominator, namely the Uno R3.
> My experience with it is maintaining an arduino-compatible (hardware
> level so only the 328s) firmware for flashing SPI chips with flashrom
> (flashrom.org) serprog protocol.

Cortex M0 processors can be equally easy to use given proper documentation and tools to simplify your life. We are at the beginning of a process of migrating libraries to avoid direct access to registers so that they can be easily ported between processors.
Migrating part of the community to ARM 32bit and Intel 32bit is a journey and we’ve just started it.

As for shields expect us to release a new spec doc on how to design shields which properly auto-configure. so we can have “XYZ” compliant shields like we have R3 compliant layouts on Unos etc.

>
>
> --
> Urja Rannikko
>
> PS. In general I think your reduction of complexity in hardware is
> reduction of functionality and
> in software it is hiding what people are really doing from themselves
> so they wont understand what they're doing... will get them to blink a
> led faster but without real understanding of the system, so not very
> helpful in the long term, but now I'm rambling off about non-hardware
> stuff so i'll end it here.

Let me ramble too :)

My experience is that you need to provide an environment where there are several layers that can be “peeled off” as you become an expert.
nothing is hidden, it’s all there but beginners start at an high level of abstraction and they are led to levels where abstraction is less and less as they progress in their learning

When we look at pictures of late 19 century cars we laugh at the driver wearing a fur coat and goggles who is required to step off the car, change several settings in the engine , maybe even warm up some parts using a flame, then proceeding to start the car by painfully operating a hand crank… We would find that unacceptable. Even professional car designers expect to unlock door, sit, fasten seat belt, turn key and drive off. Why would you want people to wear a fur coat, goggles and gloves while operating a micro controller? :) :)

This notion that “learning = pain” is so ingrained in the engineering culture to the point that it’s hard to have real innovation (unless you’re working with kids then you’re allowed to think simple ;) )


m

[Roger Clark]

As a follow up to Massimo's comments about enforcing the Arduino brand on eBay etc

I'm a bit curious what is the "brand"?

Is it anything with the word Arduino in the listing on eBay, in which case there are thousands of listings of devices with this in the title, even if the item in question has very little relevance to the Arduino boards .

A lot of the listings say "For Arduino" so I presume thats OK.

Looking at my own pet subject, I see the STM32 Maple Mini boards developed by LeafLabs as Arduinio compatible are also being listed as "For Arduino"

Hot Sold New STM32 ARM Cortex-M3 Leaflabs Leaf Maple Mini Module For Arduino

Even though they are not a module that is attached to an Arduinio, they are a board which can operate using the Arduino IDE and API

I'm not sure if Arduino.cc had any license deal with STM for their Nucleo range of boards (which have Arduino compatible headers on them), as on eBay they are being listed as

NUCLEOF411RE STM32F411RE Nucleo Arm Arduino 100mHz Cortex M4, St-Link mbed 401RE

So I presume this may violate your "brand" ??  So this vendor would probably end up renaming to "for Arduino" as well ??

I suspect this will get to be a more and more grey area, as the ESP8266 can now also be programmed as an Arduino and more and more boards are being developed to use the Wiring API and the Arduino API.

m


>
> cheers,
> rob   :-)
>
>
>

[Marco Brianza]

I also know a shield that requires that jumper...
Can you guess?

Marco