"starter kits" disposition

I have a special shelf for them, where they sit around because they're too valuable to throw away, and too useless to, well, use.

I think I have them going all the way back to an ADSP2105, if not a MC6811F1 (with BUFFALO, whatever the hell that is).

--
www.wescottdesign.com

Same with me. I actually seldom use them for much other than trying the tools and getting my feet wet. The ones that are low cost are usually too simple to do much with and the ones that are useful I usually don't buy because they are too expensive. Some of the FPGA manufacturer's eval boards are the highest priced! Often over $500 or a grand. Even the DSP boards have come down in price enough to be practical... usually.

--

Rick

I probably have about a 20% hit rate of actually building a circuit around a starter kit. Even then, half the time the starter kit is sitting on my bench with a bunch of flying wires off to a protoboard, and it all gets torn apart when there's either a working circuit board or the customer abandons the project.

As soon as I feel I can trust a chip, it becomes more economical to just design revision 0 of the board. If the customer hasn't nailed down their requirements yet it may end up being a subset of the final product, but at least it'll be representative of whatever that subset is.

For that matter, the last Cortex M board I built was for a new-to-me manufacturer (NXP) and a new-to-me chip (obviously) that I just built from the data sheet because I knew the debugging interface would work. It did, the board works, and it was all with a minimum of muss, fuss, or bother.

--
www.wescottdesign.com

Ditto. I really don't see the point in them :< But, clients seem to like to purchase them so their staff can come up to speed on a particular processor before I've finished a design. I guess they enjoy "playing" so the fact that they don't have the same *real* hardware available doesn't bother them (I'm not going to waste my time designing a daughter-card for some COTS starter kit that undoubtedly has a very "vanilla" implementation)

"Look! I made the lights blink!!" :-(

Eventually, the starter kit gets shelved. Then, someone decides to clear *their* shelf and I get *another* package in the mail (receiving an NB-1 and a DK-LM3S9B96 recently triggered this post). Usually, I just pilfer the software and things like JTAG programmers -- which can be repurposed for other things besides *their* starter kit. Then, the boxes go on *my* shelves (or, I pull the antistatic envelopes and throw them in a big "COTS kits" box)

Displays are always tempting but usually a "compromise", like most other things in their design (the Stellaris kit has a QVGA which might make it worthwhile to repackage the board as a small media tank for my network speakers)

But, many of the others I've accumulated over the years almost seem silly to hold onto (if *ever*!). OTOH, if I can quickly repurpose them, then I'll feel better about the effort :-/

(snip on starter kits)

I have a Digilent S3E board that I got some years ago and finally got to try it out. One that slowed my down was that when I installed the Xilinx software it didn't install the program that talks to the USB port on the board. Once I got that in, it wasn't long before "I made the lights blink!"

I wrote verilog for the 74160 and 74161 counters, divide the 50MHz down to 1Hz, then an 8 bit binary counter to LEDs.

I also wrote verilog to use two DCMs to genrate 14.31818MHz from the available 50MHz clock.

But yes, the first one was blinking lights.

-- glen

My point was that they are pretty useless as the basis for prototypes. I.e., the days of wirewrapping a one-off daughtercard to leverage the components present on the starter kit are long past. Chances are, you're going to have to (or want to) layout a real board (so you can produce it more reliably and *reproduce* it when you decide you need a second one to tinker with).

So, the "development board" doesn't typically buy you much towards your

*real* project/product.

Their real use seems to be in places where good simulators aren't available (real hardware simulates real hardware pretty well! :> ) Years ago, I gave one to a friend who wanted to control the pumps and heater in his hot tub. I'm sure he could have purchased something for far less than the parts he had to glom onto the development board... but, he enjoyed the effort.

They're just so crippled that repurposing them for much of anything is more of an effort to figure out *what* to use them for than to actually *do* the repurposing!

[The QVGA panel makes the Stellaris kit far more tempting than most. And, the 1588 NIC makes use driving my "network speakers" obvious. If they had a touchpad, it would probably be perfect! :> ]

It's funny how simple things can mess you up. I was working at a place with some dozen or more FPGA coders while I was in the hardware department. They organized FPGA in the software department and even their more experienced people were not really that good at some things. They had a new board to load the bitstream into and couldn't get it to work in spite of tons of corporate experience in writing the software. I got pressed into service to help them.

I got there and they had *five* engineers all trying to work on this one problem, at least one of them a senior engineer. I asked a few questions which took some effort to get answers to (after all, who the heck was I to be interfering?) and make the point that while the configuration process provides little feedback as to what you are doing wrong, it is a *very* simple process to get right. I mentioned two things that are common mistakes and they assured me that they had tried both. One was to send a couple or three extra clocks after the end of the data to move the FPGA out of configuration mode. I was told they had sent 1, 2, 3 up to some huge number like 64K extra clocks and it still didn't work. I don't recall what the other thing was. But they didn't realize they had not tried them at the same time, lol. I finally got them to listen to me and do both together and it worked. The senior engineer was amazed. But I hadn't done anything except have confidence that configuration of an FPGA is a simple thing.

If they had first worked with an eval board they would have been sure of the hardware and been able to focus on their procedure rather than being in the dark as to what to do next.

--

Rick

(snip, I wrote)

I sometimes notice that I can get more error messages from a small program than a big one. For big ones, I try harder to get them right.

There is a story about networking telephone support asking people to be sure the power plug is in the right way. Not that it can go in the wrong way, but it works better than asking people if it is plugged in at all.

In the newspaper today, they have a "Car Talk" question. Someone with a car had to jump start it, and, after the battery had charge enough couldn't get it to start. Finally, they read the manual (rare) and find that you have to open and close the driver's side door after the battery has recharged. It seems that is when the car computer checks for the keyless key fob being nearby.

The starter board configures (or can configure) through JTAG, at which point the configuring program knows a lot about the board (like what kind of FPGA it is.) For other methods, there are more ways to get it wrong.

I might try to make my own board, and will have to figure out all those.

-- glen

Why, what is your purpose?

--

Rick

(snip, I wrote)

Hardware emulation of older computers.

I want a nice simple board, FPGA, serial port, keyboard, VGA, maybe a little more.

-- glen

I am certain you can find that in an existing board somewhere. What keyboard interface? These days they are all USB.

--

Rick

Can't understand the purpose of this. Software emulations these days even on weak hosts (Raspberry Pi etc) are much faster than the old hardware ever was.

Same. The top shelf in the lab, the one you can't get to without either the ladder or monkey climbing, has all the old dev kits next to the box marked "Power Supplies of Limited Utility".

I like them for an opportunity to try the tools out before I get too deeply committed, but generally once I can blink an LED and _maybe_ testdrive the one or two wacky on-chip peripherals I'm feeling dubious about. But I'd never dream of trying to put together any substantial portion of my circuit against one.

Relatedly, I've never bought any of those expansion boards that let you connect some specific ADC/transceiver/explosive device to the proprietary connector on the dev kit.

--
Rob Gaddi, Highland Technology -- www.highlandtechnology.com 
Email address domain is currently out of order.  See above to fix.

(snip, I wrote)

It is, but it isn't the same thing.

For one, what is the software emulation of a 1024 bit shift register?

Yes, you don't need one, but if you want one ....

-- glen

Easy to emulate in software. 1024 bit shift register is implemented in software by a 128 byte array. Rather than moving the data in the bits a pointer to the byte and bit in the byte are maintained and incremented. Much less power consumption too. :)

--

Rick