Professional Serie « LeafLabs Garden

LeafLabs Garden » Development

Professional Serie

(18 posts) (5 voices)

Started 4 years ago by Math
Latest reply from gbulmer

12 Next »

Math
Member

I'm still wondering why all the arduino, Leaflabs & cie are still stuck in some kind of hobbyist paradigm.

While I do appreciate the massive upgrade from Arduino, and recognizing that a great working Maple board is somewhat new/in progress...

I'm really going to get sick if I receive again a proposal from Siemens and friends for thousand $ boards that have the same functionally as a 30$ 8bit Arduino Uno.

What I want to see is: a Professional Serie line of Maple boards

I could see about 3-4 models such as High quality data acquisition, Home automation, Industrial PLC, etc.

What I personally what:

- Extended temperature ranges, immunity to electrical noise, and resistance to vibration, impact and condensation
- Protecting against overvoltage, reverse voltage, overcurrent, and ESD
- All solid state caps
- 16bit+ ADC
- board layout that is NOT Arduino compatible, but designed for making easy connectors (screw-on?)
- some input with pull-up resistor bridges designed work with higher voltage inputs (jumpers to by-pass?)
- Mini-PCI board format ? (easy storage box with power supply)
-build in Ethernet port with TCP/IP (small linux kernel?)
- PCI-E style expansion board instead of sheild?

Actually, kind of like a Gator but with a nice ARM MCU

Posted 4 years ago #
gbulmer
Moderator

- Extended temperature ranges, immunity to electrical noise, and resistance to vibration, impact and condensation
- Protecting against overvoltage, reverse voltage, overcurrent, and ESD

IMHO, those are $100's/board extra for development time, testing, liability insurance, plus handling warranty returns. Oh, and some electronic parts

16bit+ ADC

Are you saying you want an external ADC, or an ARM with 16-bit ADC on board?
What sample rate?

- board layout that is NOT Arduino compatible, but designed for making easy connectors (screw-on?)

How big are you prepared to have the board?
If this were a professional board, I'd expect to be able to get an oscilloscope to every signal pin.
Screw connectors are typically 5mm/0.2inch pitch, the smallest I have used are 3.81mm/0.15inch
With 51 I/O's (so ST say) on a 64-pin chip, (not including power connections) would be 7.65 inches to put around the edge.

-build in Ethernet port with TCP/IP (small linux kernel?)

This is three requirements, do you want all three, or would ethernet and a TCP/IP stack be sufficient?

I don't understand what these might mean:

- Mini-PCI board format ? (easy storage box with power supply)
- PCI-E style expansion board instead of sheild?

You're going to have to explain.
I don't think a PCI-E, or PCI bus interface is a good fit for a low-cost microcontroller, and IMHO PCB edge connectors are even less user friendly than header plugs and sockets (I think edge connectors are less reliable too)

The ECAD is in the repository.

Posted 4 years ago #
Math
Member

[quote]IMHO, those are $100's/board extra for development time, testing, liability insurance, plus handling warranty returns. Oh, and some electronic parts[/quote]

You can get a full gator for about 50$. I have not checked if they include some kind of warranty on that stuff. Reverse voltage is not an hardware problem but a user problem. Still, quite handy on the field.
And yes, I would pay the extra 100$ easily. I mean, that's spare change for a big project but a huge liability to have your equipment fail.

[quote]Are you saying you want an external ADC, or an ARM with 16-bit ADC on board?
What sample rate?[/quote]

I do not know what solution would be easier/cheaper. In fact, I'm quite curious to know the answer. Sample rate: that's a bit harder because for the project I have in mind that's not important. Solution would be to look for a good average based on most industrial sensors.

[quote]How big are you prepared to have the board?
If this were a professional board, I'd expect to be able to get an oscilloscope to every signal pin.
Screw connectors are typically 5mm/0.2inch pitch, the smallest I have used are 3.81mm/0.15inch
With 51 I/O's (so ST say) on a 64-pin chip, (not including power connections) would be 7.65 inches to put around the edge.[/quote]

Size is the last of my concerns.
It does matters for consumer electronics.
As far as connecters go, even trough holes would be better than the Arduino female headers...
That why I said having 2-3-4 models depending on market. Different size and connector types.

[quote]
This is three requirements, do you want all three, or would ethernet and a TCP/IP stack be sufficient?[/quote]

Just trowing some ideas around. Most industrial equipments are switching from Serial RS232 to TCP/IP, so always handy.

[quote] I don't understand what these might mean: [/quote]

Mini-PCI board format: Sorry, I meant to say MicroATX.... I was talking about board size, so that you could fit it into a standard PC box.
- PCI-E style : A way to connect expansion boards (into your PC box) using a PCI style connector. Instead of stacking sheild, you add expansion card (network sheild, storage sheild, wifi sheild, motors controller). An other idea would be ATI crossfire type of connectors

Even if the total was 200$+ for the package, that is still a huge bargain for people stuck paying thousand $ for board designed for a single duty. Sure, if you do volume work, you can afford the time and expenses to design your own custom boards. In my case, working in a small engineering shop with customers from about every field, having a good reliable/easy to program (open source, C or "Arduino" language) affordable (ever worked in the nuclear industry? :D) controller/data acquisition board would be a Godsend.

As an exemple, I was quoted this week 1200$ per board to connect a single position detector sensor. Thats about 2000$ per board that I must charge my customer. All that for a 12bit ADC with an RS232 connector? Paying 200$ for a board that could acquire quite a few PSD, plus do the output on a LCD, switches and everything? HELL YEAH

Posted 4 years ago #
gbulmer
Moderator

Interesting. I should add, I am not a LeafLabs employee.

I haven't worked in the nuclear industry, but a friend/colleague did, and some of the software I've worked on runs in manufacturing plants in production systems.

I did a board which had some input protection for a University, using a similar technique to the Gator+, but it cost quite a lot to test, and it did reduce the performance of the inputs and outputs. It was only designed to be 'safe' upto 18 volts on the basis that the board was powered by a 9V supply, so it only had be protected from accidents with that magnitude of voltage.

Is that sort of level sufficient or are you asking for something more?

IMHO, liability and testing is a big part of the cost.
I worked with Ed Yourdon in the early 00's, and he said several software product companies spent more on lawyers to handle product liability than software development. He wasn't talking about nuclear industry, or even manufacturing, he was talking about normal software applications.

Gator has no liability cover, it has the normal 90 day warranty - it should switch on. http://ruggedcircuits.com/html/policies.html

Rugged Circuits LLC makes no other warranty of any kind, either expressed or implied, including any warranty of merchantability, fitness of our products for any particular purpose, even if that purpose is known to Rugged Circuits LLC, or any warranty relating to intellectual property. Rugged Circuits LLC shall not be liable for any injury, loss, damage, or loss of profits resulting from the handling or use of our products.

So, would that level of liability (i.e. standard 90 day) be okay?

Posted 4 years ago #
Math
Member

Open source design, open source software. If it fails based on bad design then I'm the sucker. Liability is based on the concept of black boxes. Plus, if your boards suffer from poor reliability, people will know from mouth to mouth.

Plus, all the ''tricks'' for power protection are well known, so its not like asking to develop some new bleeding edge concept.

What is needed is to figure out exactly what would be the most universal concept. I could easily see a general purpose board, with ATI crossfire type of connector for expansion cards. Or even some kind of casing with already pre made connectors so that you only slide in new cards. 16+bit ADC/DAC card, network card, motor drive control.

You could even connect a few gecko drives to the board to a powerfull ARM board, load EMC2, create a web interface and run a kickass CNC controller.

If Red Hat managed to get a billion $ a year in sales, imagine what could be done with open hardware... people free Labview, ABB, Siemens, Fanuc...

For disclaimer, I should add that I'm a ME, so yes I do lack on the EE geek speak.

Posted 4 years ago #
gbulmer
Moderator

Math - if you think there is a need, why not do it?

How much effort do you think it'll take?

Posted 4 years ago #
Math
Member

Well, thats an easy one to answer.

I am not a wizard about hardware design. I simply assumed that a team that managed to develop a product such as the Maple had the knowledge to design a real product.

What I could do is design a business plan and hire people to develop the product. Trouble is that would be a huge finantial risk.

A more logical step would be having at least a somewhat working prototype. The next genius step would be finding a way to wedge the product into the market. I assume the standard open market pull could do wonders.

Posted 4 years ago #
gbulmer
Moderator

Just to avoid any misunderstanding, I am not a LeafLabs employee.

What I could do is design a business plan and hire people to develop the product. Trouble is that would be a huge finantial risk.

Yes, I could do that business plan too.
Yes it is a large financial risk.
What is your risk mitigation?

LeafLabs have already made a significant investment in Maple.
I think the board you seem to ask for is several times more complex, i.e. several times more investment.

A more logical step would be having at least a somewhat working prototype.

A somewhat working prototype won't earn any income to offset the risk.
If it were Open Source, it might provide a basis for someone else to bring something to production, but that just worsened the odds for the "huge finantial risk".

I worked at a large Cambridge MA based software company, and our CTO estimated it cost 10x to convert a working prototype into a viable product. I'm an optimist, and would hope for a factor of only PI.

The next genius step would be finding a way to wedge the product into the market.

Agreed.
What is the thing that converts this from an idea into a viable product?

I assume the standard open market pull could do wonders.

So where does the money come from to fund the "huge finantial risk"?
Once it is Open Sourced, the margin can't be too big, or the design will end up in a Chinese factory, undercutting the original.

Posted 4 years ago #
mbolivar
LeafLabs

Math,

Speaking as a LeafLabs employee, I will say that it's currently not a priority for us to develop hardened versions of our products. As it's been said, our designs are open; you are free to pursue this goal yourself, but it's not on our horizon.

Posted 4 years ago #
Math
Member

gbulmer:
1. Yes, I could do that business plan too:
Huge financial risk is if you pay people to do it. IF you have the skill to develop a working prototype yourself, its only a matter of time = money. One way to get free labor, is to get undergrad students to work on it. So much ressources is wasted at technical universities redoing the same exercises while this energy could be used working on opensource projects.

2. LeafLabs have already made a significant investment in Maple:
Perfect, just keep the same ARM 32 architecture. And knowledge gained...

3. If it were Open Source, it might provide a basis for someone else to bring something to production:
Well, nobody has managed to kill Arduino or RedHat even if it is opensource. The trick is to go low margin high volume. By going high volume, you reduce a lot your production cost thus sheilding you from simple copies. Also, do not underestimate the value of a brand. If you offer high quality, good service and low cost, there is no reason for a customer to look for an alternative.

4. What is the thing that converts this from an idea into a viable product?
Find a niche market. Since I'm the kind of retard that design CNC machines as an hobby I can give the the gecko drive exemple. These guys pretty much managed to get themself a pretty stellar reputation by hanging off the cnczone.com forum, answering questions and taking advices from customers. My idea would be to spot a few serious hobbyist meetings, forums and the like and become the defacto brand name (like arduino for hobbyists). When you manage to acheive and acceptable level of sales and quality, you move to universities, research labs, hospitals, etc. By that point, the whole software stack must be great enough to simply trash trough the industrial business.

5. So where does the money come from to fund the "huge finantial risk"?
You keep the cost down by working by yourself at first. 5000$ should be enough to cover prototype boards, test equipment, etc.

mbolivar:

Sure I know. If, I was actually planning on doing it myself, I would not be posting here. At least, not witouth a well defined business plan. My goal was simply to start a discussion to get people thinking.

Posted 4 years ago #
poslathian
LeafLabs

Sure I know. If, I was actually planning on doing it myself, I would not be posting here. At least, not witouth a well defined business plan. My goal was simply to start a discussion to get people thinking.

It is a very interesting suggestion, and you definitely succeeded in that goal!

My intuition was always that to achieve any level of success on the industrial side would require a runaway amount of support/testing/certification - starting with FCC and UL certs and ending with electronics ovens for performance/temperature curves, vibration testing, certified connectors, etc. Add in industrial-corporate sales team and support staff, and 2k$ for an "industrial arduino." I'd like to hear what you consider to be sufficient polish to breaking into your market.

Find a niche market. Since I'm the kind of retard that design CNC machines as an hobby I can give the the gecko drive exemple. These guys pretty much managed to get themself a pretty stellar reputation by hanging off the cnczone.com forum, answering questions and taking advices from customers. My idea would be to spot a few serious hobbyist meetings, forums and the like and become the defacto brand name (like arduino for hobbyists). When you manage to acheive and acceptable level of sales and quality, you move to universities, research labs, hospitals, etc. By that point, the whole software stack must be great enough to simply trash trough the industrial business.

This is good thinking! So what other forums does everyone else hang out at? ;)

Posted 4 years ago #
crenn
Moderator

thebestcasescenario.com :P

That's pretty much it! But the Leaflabs Maple name has been going around my uni because of my project! Hopefully soon it will be the Maple Mini!

Posted 4 years ago #
gbulmer
Moderator

Math

If it were Open Source, it might provide a basis for someone else to bring something to production:

Well, nobody has managed to kill Arduino or RedHat even if it is opensource.

Redhat invested in quality control, packaging and support for an existing body of software. IMHO their real creation was marketing their quality control and support. Redhat does do some software development and testing, but the core of the product, Linux, and the applications like Apache web server, GNU GCC, PostgreSQL, etc. were not developed by Redhat. They wrap quality control and support around Open Source, which is itself not Open Source.

In fact, I would suggest that Redhat did exactly take Open Source products developed by others, and make money out of them. I am not saying that is bad, just the basis of much of their product is other peoples (freely published) work.

... and become the defacto brand name (like arduino for hobbyists).

Just to ensure people understand the genesis of Arduino.

The Arduino started in University, and, AFAIK, initial development was funded by the university. It's development costs were subsidised by a PCB manufacturer, who is a friend of Massimo. It is a beautiful piece of design, cleverly aimed at a niche, but was funded at low risk.

So, I don't see that either Arduino or Redhat are examples of the approach you advocate.

I had a quick look, and AFAIK Geckodrive has no Open Source products.

Posted 4 years ago #
poslathian
LeafLabs

a little piece of history about the risk of the original arduino project - and im not knocking, this was great serendipity - Arduino took advantage of an Italian law where they could operate with net-90 payment to their local fab company. This means that they could build large volumes of arduino on 90 day credit. This is a pretty huge advantage then say - having to find $10k up front for the first fab run. Furthermore, with this scheme, you could continually grow the volume of each run without having finished selling the first run!

Without this, it takes a fairly long time in order to build up the cash to do runs of 1000 or even 4000 pcs (4000 is where the prices really start coming down because you can often buy direct from the manufacturer of the components rather than a distributor), even if your demand could sustain those high volumes.

No point here - just interesting....

Posted 4 years ago #
Math
Member

---poslathian---
-support: Sure, we have a huge community and a free forum :). Need design help? our engineers can help develop your application for a small fee:D
-testing: open source style testing...
-certification: Only required for highly regulated industries (nuke, aero, etc), dont aim for those markets at first.
-FCC and UL certs, ovens for performance/temperature curves, vibration testing, certified connectors, etc: That was my 5000$ investment for testing benches... can't you find that sort of stuff at the MIT? Must be a nice EE/ME grad project...
-industrial-corporate sales team and support staff: Start with free marketing, aka peer to peer

-I'd like to hear what you consider to be sufficient polish to breaking into your market: I spent a lot of time (read $$$) last year working for a crown corporation (country known for its maple syrup) in the nuclear market. I swear to God, those guys were stuck programming their controllers in DOS because of software lock in. That's how bad it is. Over looked into home automation? Major mess. I could go on and on... Just make in open source, reliable, professional friendly (nice connectors, easy to use software) and it will sell.

---gbulmer---
-In fact, I would suggest that Redhat did exactly take Open Source products developed by others, and make money out of them. I am not saying that is bad, just the basis of much of their product is other peoples (freely published) work.

That's called innovation. Innovation is the technological leapfrogging of existing technologies. What is the Maple if not an Arduino ripoff? In the FOSS world, you cannot called an invention ripoff, simply innovation. What I'm proposing is no technological marvel invention, simply innovation on a business level. Leaving the low level pure hobbyist market and working on bigger and better. By itself, the Maple was easy to develop: take the Arduino concept, software, reputation, market but port it to a 32bit microcontroller. But it stopped there since that was the easy path for engineers. No need for sale, marketing, market development, etc.

Since the Maple is technically Arduino V2, its build on all its past innovations (''It's development costs were subsidised by a PCB manufacturer, who is a friend of Massimo'', etc etc). The Maple R5 is now working prototype board. In my mind, that's a halfway done project. And that was the easy part.

-I don't see that either Arduino or Redhat are examples of the approach you advocate:
They are not. But they are examples of projects that first looked impossible but succeed.

-and AFAIK Geckodrive has no Open Source products:
I know, that why they will never achieve world domination. On they other hand, do you really need an open source drivers? But, they managed to created products based on customer inputs, flawless reputation, low cost high quality products, etc.

Posted 4 years ago #

RSS feed for this topic

12 Next »

Reply »

You must log in to post.