nilrods

generic development board thoughts

21 posts in this topic

I, like probably most on this forum, have a small collection, maybe 20 or so, development boards of different types, processors, FPGA, plain chips, etc.

 

Most of them have a similar collection of peripherals/connectors//etc attached to them.

 

I am curious why someone hasn't come up with a sort of generic development board where you could plug in different fpga, arm, arduino, etc via something like a PGA sub-board/socket with the chip.

 

I figure it is probably a pipe dream. I know nothing at all about circuit board design and I figure there must be some technical reason for this but thought I would see if anyone on this list knows of good reasons why something like this doesn't exist.

 

I realize the main board would be more expensive than say a single development board, but I for one would definitely pay more for a board I could reuse. Might even be more practical to add some of the more advanced features I see people mentioning they would like to see on boards since your only buying one. I also think in the long run it would probably be cheaper if the person buys multiple chips.

 

I would think whoever made them would still have a ongoing revenue stream since people would still need to buy additional chips on the PGA from them for other devices they want to use.

 

Just kind of looking for thoughts from the folks more in the know around that area.

 

Thanks,

Chris

 

Share this post


Link to post
Share on other sites

IMO it's incredibly hard to decide what to put on an FPGA board and what to leave off.  If you put on too much, it's too expensive.  It you put on the wrong things, it won't sell.  If you put on too little it's hard to compete with a board that has "a little bit more".

 

Having generic connectors is often a good idea.  I really like the Papilio One and recommend it as a great board for learning about FPGAs and it's a great buy for US$38.  Papilio One brings out virtually all the FPGA's I/Os to 0.1" connectors, and only has a handful of dedicated pins.  OTOH, you have to plug in a wing to do anything useful.  OTOOH, wings are cheap and it's easy to make your own from 0.1" perfboard.

 

While generic 0.1" connectors are great, what if you need something fast like 3 Gb/s differential pairs?  Those kind of connectors are expensive, and how many do you put on?  What about having PCI or PCI Express connectors?  Suddenly the choices multiply, and trying to decide what to do can be paralyzing.

 

A boards is often created for a specific purpose, such as to provide hardware for an FPGA development course.  Now, you could go with a vendor's evaluation board from Xilinx or Avnet, or go with an independent board such as Papilio and adapt your course to that board.  The risk is that the board could become unavailable (like the very nice US$40-50 Spartan-3A board Avnet used to sell) and suddenly you need to rework your course.  Or you can create your own board that matches exactly the needs of your course and avoid that risk.  That's what they did at San Diego State University: http://www.eetimes.com/document.asp?doc_id=1280958

 

If you do go with someone else's board, be sure it's open-source hardware like Papilio.  That way if the board becomes unavailable you can clone it.

 

And then you have boards to plug into a specific base board, such as the LOGI-Pi (for Raspberry Pi) and LOGI-Bone (for BeagleBone).  IIRC, ValentF(x) at first tried to design a board that would work for both, but that obviously wasn't feasible.

 

I'll close with a bit of humor.  In the Midwest you often see a bumper sticker that reads "She who dies with the most fabric wins."  I'd like to see a version for "development boards" :)

Share this post


Link to post
Share on other sites

Well, the very first version of the Papilio, before the Papilio One, was designed with that concept. Here are some pictures:

 

post-29509-0-63139700-1432067009.txt

post-29509-0-25748600-1432067011_thumb.p

 

The idea was to have stacking modules and you could swap out the module with a Xilinx Module or a Altera Module, the USB and power module were self contained. Or you could develop other modules such as ARM boards.

 

I actually put a lot of time and thought into it... At the time I didn't know anything about running a business or promoting a product. So I made a store, made some boards, put them in the store andddddd. Crickets... Nobody was remotely interested.

 

So I went back to the drawing board and came up with the Papilio One after trying to figure out what it was about the Arduino that people liked and what could be done better. When I released the Papilio One is when I started getting some traction. I'm not sure why no one cared about the modular approach. At the time I felt that a modular approach is something that the more technical non-casual audience desires and unfortunately there are less people in that category then the casual audience. The casual audience that wants to check out some new technology wants something not complicated and easy to understand. When you start talking about stacking modules I think it cuts down the people who will actually buy it. At the time it seemed like a reasonable explanation. Looking back I wonder if it was just my complete lack of experience with promoting and explaining a device.

 

Jack.

Share this post


Link to post
Share on other sites

John,

I agree with what your saying. I think the papilio is an excellent choice myself and being open source as you mention.

 

 

Jack,

Interesting, so it sounds like less of a technical issue then. Being new to the board I had no idea that was what the original Papilio was. Do you still have any floating around?

 

I have to admit I think it was/is a great idea, but of course the marketplace speaks when in business. Do you think it was more an issue with the concept or more with how the business side was implemented initially, not meaning to salt any wounds? I just know now you know have the DesignIDE and ZPUino and all in addition to just the boards themselves which I think helps also.

 

I think you are right that people who are just dabbling or looking into something want something simple and easy. Of course, the arduino/raspPi compatibility can have a big impact on sales since they have such a large user base. I am sure the non-casual user base is smaller also.

 

I had just read on some of the background on designing the DUO and how the feature list was large and things kept getting cut from the list. I have seen similar threads on other sites when people discuss boards also. And when you look at a single board that may be out of date quickly those thoughts are true.

 

I was just thinking that if you had a bare processor/fpga/etc board, like it looks as the original papilio was, you could spend a little more money on peripherals knowing they would not be worthless inside of a couple years when the latest XYZ came out and you just had to buy it all over again. I know I would spend $2-400 easy for just a well featured base board, without the fpga/cpu/etc if I knew I could add a upgraded one when either I needed more.

 

I understand you would not be able to keep pace with the latest, XXXXX super high speed interface, but wouldn't mind spending a few more buck for some high speed connectors for special purpose. Want sata, HDMI or super high speed USB, special memory, etc? I know I would feel better spending a few more bucks if I knew I could use on any number or FPGA's/ZYNQ/Arm/etc board.

 

Also I wouldn't mind the board being a bigger if I didn't have 20 or so lying around :)

 

 

Just my thought.

 

Thanks,
Chris

Share this post


Link to post
Share on other sites

Some of us who have the modular Butterfly Platform boards beg to differ. There were not enough of us though! Countinued thanks for your thoughtful designs and efforts Jack. We are behind you!

TrapperBob

Share this post


Link to post
Share on other sites

Trapperbob,

Just curious, do you say that due to technical issues or just because the design did not continue on?

 

Thanks,

Chris

Share this post


Link to post
Share on other sites

i like that original butterfly board.

 

its very similar to one of my projects (based around the Microchip PIC 18F4550)

Share this post


Link to post
Share on other sites

@nilrods,

I had no technical problems with the butterfly board. I thought the modular approach was the way to go. I still believe that modularity is a good way to go. As has been said it would certainly help keep down the number of boards one collects because of of being able to mix and match various peripheral circuits with the core FPGA and being able to upgrade the core when appropriate. I think Jack has done a fine job supporting the peripherals and would like to see an approach to make the core modular as well. Once again , I thank Jack for all he has done to support the community and look forward to being able to continue supporting him in his efforts.

TrapperBob

Share this post


Link to post
Share on other sites

rpflaum,

 

Ah, the good old days! You have been supportive since the very beginning and I wanted to say thank you! Some of your early purchases made a big difference, the first year of GadgetFactory I only made less then $10,000 in sales and went through my savings pretty quickly! So every sale had a big impact, I remember that you were one of the top supporters. :)

 

Thanks!

Jack.

Share this post


Link to post
Share on other sites

Ok, guys, so this got me thinking about my last trip to China. Jingfeng, the CEO of LinkSprite, showed me a cool modular design that they had for the pcDuino. He also suggested that we work together to make an FPGA module. I just talked to him and he pointed me to the system that he showed me. What do you guys think? Is this along the lines that you were thinking?

 

Here is the base board with a slew of cool peripherals:

http://store.linksprite.com/evaluation-board-a-for-pcduino3c-core-board-pcduino3c-not-included/

 

Here is an allwinner ARM Cortex module that plugs inthe the base board:

http://store.linksprite.com/linksprite-pcduino3c-core-board/

 

I can work together with LinkSprite and we can make an FPGA module for this base board as well. Then, from there we could make new module's such as the ATOM processor, an Altera FPGA, Lattice FPGA module etc. I'd actually love to hear ideas for other modules...

 

Now, this could turn into quite a bit of work on my part so I'd have to carefully think about whether I'm biting off more than I could chew with this. But I can get some of these boards from LinkSprite to sell in the GadgetFactory store if there is enough interest. And Jingfeng and I have been looking for an excuse for me to go back to China, working on an FPGA module might be just a good enough reason. :)

 

Jack.

Share this post


Link to post
Share on other sites

Jack,

You know I think I had seen this one before. There are a few others also like it on ebay and other websites. The only reason I hadn't jumped already on one was that the ones I had seen only had one module available(ex. an Arm or an altera fpga). To me it just kind of defeats the modular purpose. I think expanding with new modules would give it a competitive edge.

 

I think that is a great start and may even be the solution.

 

Looking at the summary. wifi, hdmi, lcd, nic, sd, usb 2.0, ir, jtag, camera. Nice feature list.

 

 

Couple questions if you know the answers.

 

I see the LCD and sata connectors. Would those be considered high speed connectors? So if used with an FPGA that could also be used for other high speed things also, not just LCD or sata right, of course with the appropriate connector/wing/breakout board?  If not, some sort of expansion connector might need to be added, just thinking.

 

Does the Network connector support gigabit or would that depend on what it was connected to? Couldn't tell if there was a network chip or just a physical connector.

 

Looks like about 30+ gpios if I am counting correctly, of course plus the LCD and Sata if those can also be reused for other purposes when not needed for the specific function. Does that sound about right? Not sure how many pins are in the Sata and LCD connectors.

 

What would be on the module board, only the module chip or the module chip and memory? Not sure if they have to be together on the module board.

 

I like the sounds of all of the modules you mentioned and would probably get most of them. Since the module board would be relatively stark, does that make the idea of using BGA devices on it easier?

 

If so, that would open up a lot of options. Just some ideas. I would like to see some more powerful things on them. I understand they would cost more since the chips themselves cost more, but as I said your leveraging on your existing purchase. Zynq chip would be a great one. Larger Spartan 6 for more complicated projects and still be able to use designlab. Maybe one of the Vivado based like a Artix, which seem to be inexpensive or something like that. Of source, some Altera also. Maybe even CPLD like a coolrunner for prototyping some of the small things. Maybe a shield or some connector to attach an arduino or raspPi to pull in some of that market? Could it be like a super shield for an arduino, not sure how many pins would be able to mapped.. Definitely an ARM cortex R for realtime applications.

 

 

I can say if there start to be more modules for a board like this I will definitely buy one.

 

Just my thoughts.

 

Thanks,
Chris

Share this post


Link to post
Share on other sites

IMO the biggest challenge facing FPGA development boards is cost.  A Raspberry Pi 2 or ODROID-C1 is US$35 for a desktop-capable quad-core ARMv7 with 1 GB RAM, quad (or more) USB 2.0, and Ethernet.  The cheapest Spartan-6 board I have seen from a reputable vendor is about US$70, twice as much.  I had hopes for the Hackaday Arduino-Compatible FPGA board when it was first announced at US$50, but now it's up to US$70.  US$35 is a great price point, because it's at a level that a hobbyist can buy without spousal approval and something students can afford.  US$25 would be even better for an FPGA board.

 

If you look at Spartan-6 LX9 pricing, it's not the chip cost.  The problem is that FPGA boards aren't popular enough to manufacture in 10K or larger lots, like Raspberry Pi.  The fact that there are so many different FPGA boards diluting an already small market makes the problem worse.

 

So I would recommend designing an FPGA board that really drives the price down.  I really like the Butterfly Platform that Jack posted upstream.  Very simple board, with just an FPGA, bypass caps, and connectors.  Spartan-6 is nice because you only need 3.3V and 1.2V -- you don't need 2.5V like the Spartan-3E.  While I like having an FT2232H on board for programming, the fact is that those chips are pricey.  For US$15 you can get an Adafruit FT232H breakout board which makes a dandy JTAG programmer, and you can use it for all your FPGA boards.

 

The other big challenge is the steep learning curve of vendor FPGA software, but a number of us are working on that with various approaches.  If it the software was a lot easier and there was a really cheap board we could help a lot of new FPGA users get started.

 

JMO/YMMV

Share this post


Link to post
Share on other sites

John,

I would definitely agree with you for many users the low cost is the bottom line. You are 100% right on why the Pi and such are so cheap. It is business so much about quantity is affecting prices people pay. If you want to sell hundreds of thousands or more that is probably the bottom line. Also the cheaper it is the more people will try it.

 

But of course there is much competition in that area also coming from businesses who probably have easier access to materials to build being foreign, but I am only speculating here.

 

Being new to FPGA's I can definitely agree with you assessment of the tools. I have used software development tools for 20+ years and thought there was really some issues with some of them. But wow I have to admit that some of the FPGA tools are so much worse. I can't believe some of those buggy things/versions were ever let out of the door. Who writes those things...

 

There is actually a cheap Spartan-6 for sale on ebay along with some bare cheap alteras. Search ebay for item 111640406811. I actually picked 1 up when it was at like $26 with free shipping, looks like it is more now but still like $31.

 

I bought to use for a dedicated purpose so didn't need lots of the other stuff from a development board like my papilio duo(to use for an encoder driven power feed for one of my mills). Now I know nothing about board design so can't speak to the quality of the board, but looks fine to me and I have not had any issues with it yet. Does require a jtag programmer to use.

 

There are a bunch of altera versions in the same price range also.

 

 

EDIT: Just an edit after re-reading my post. On the part about FPGA software I was not referencing Designlab or anything open source like that. I was referencing ISE and Vivado specifically in case anyone may have misinterpreted my comments.

Share this post


Link to post
Share on other sites

The Pi is so cheap because (at least initially) they were able to buy the Broadcom parts at a discounted price normally only offered to high volume customers. The actual volume of Raspberry Pi boards produced has far outpaced predictions though so they may actually be in that volume range now but initially they weren't. Also it's a SOC designed for smartphones which are cranked out in absolutely HUGE numbers. I would speculate that the chip on the Pi is probably made in greater quantities than all current production FPGAs combined.

 

As far as connectors, I really like the standard 0.1" headers used on the Papilio and many other boards. Combined with a bunch of those jumper wires and a breadboard you can plug in almost anything. For higher speed IO I recall seeing someone using SATA connectors. They have relatively few pins but they're designed for high speed data, and they're used in huge numbers so the connectors and cables are cheap. I think the one I saw used a row of SATA connectors for a LVDS bus of some sort.

 

Those inexpensive FPGA mini boards are great. I have several of the Altera EP2C5T144 boards that I think were $16 each. I did a vintage computing daughterboard for those that has SRAM, micro SD, composite and RGB video, two LVTTL serial ports and an audio output. I made it primarily for the Multicomp microcomputer but it makes a good general purpose dev board.

Share this post


Link to post
Share on other sites

James,

That makes sense on the raspPi. I always wondered how the got the pricing down so cheap. Good info.

Share this post


Link to post
Share on other sites

John,

I thought a little more about the points you have brought up in your emails. I think you bring up some really valid points in general around pricing and users and students.

 

Unfortunately as you mention the prices of boards don't coincide with the price consumers are willing to pay.

 

Just an idea, as my business side is always thinking although this would probably be a not-for-profit venture, maybe we should set up a electronic development board exchange.

 

The idea would be that you could join by either paying an annual fee of say $35-45(or whatever makes it viable) or by donating a recent development board to the exchange(that meets certain criteria) that has a new value of say $70 or more(since once they are used everything devalues). Plus you would have to pay shipping for each board you wanted to try. You would only be able to "check out" 1 board at a time at that price level though. Of course, everyone would have to have credit card on file and if you break it you buy it.

 

The money would be used to buy additional boards and all.

 

Students could "rent" for a semester. People could try new boards and each would have an agreed price if they wanted to buy it up front. Kind of a try before you buy thought. As I am sure other folks have bought boards that looked cool only to later find out things that made using them more complicated or not for the use they had in mind.

 

Here is an article that implies this sort of thing is what the younger folks are looking for.

 

http://www.businessinsider.com/millennials-are-renting-instead-of-buying-2015-5

 

Just a thought. Let me know anyone else ideas on the subject, even if it is just that I am crazy :)

Thanks,
Chris

Share this post


Link to post
Share on other sites

I think the idea of a board rental service is interesting... Someone would have to do a lot of free work to keep it running though.

 

On another note, I just got one of these Linksprite boards in the mail today. It looks pretty cool, I'm meeting with Jingfeng next week and will talk to him about making an FPGA module.

 

Jack.

Share this post


Link to post
Share on other sites

I think you might have problems with damaged boards. ESD or people carelessly hooking up other things to them and frying IO pins. Some damage is not readily obvious and it would be hard to determine whether somebody fried a board or it was damaged before they got it. Maybe this would turn out to be rare, I don't know.

Share this post


Link to post
Share on other sites

Jack,

I am sure your probably right about the effort. I was thinking it would need an application to automate as much as possible, but I am sure your right. Of course it would need to have some papilio boards if it was done, so you would get some extra sales probably.

 

Sounds great on the linksprite board. I have to admit I am excited about the prospect. Let us know what your initial thoughts are once you have a chance to look it over some.

 

Any idea of what specific FPGA your looking to put on it initially? Are you looking to stick with one of the modules you used before on the papilio to reduce the design effort initially? I think the idea of having an ARM and an FPGA to begin is good. As those are 2 useful modules to start with. I will be watching updates. Count me in for one once you get some running.

 

 

 

Thanks,
Chris

Share this post


Link to post
Share on other sites

James,

Excellent points. I can see it being difficult to identify every little items that could break.

 

That was why I thought I would just throw the idea out there to see what others thoughts were on it.

 

Thanks,
Chris

Share this post


Link to post
Share on other sites

Jack,

I think there is a lot to like about the LinkSprite board. It would be a good base to start from. I certainly would like to help support this effort by buying some if you put them up in the store. FPGA modules ( Xilinx and Altera) would be a good place to start to expand the idea. Let us know how we can help!

Bob

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now