universal programmer

On Fri, 23 Dec 2005 17:32:04 -0600 in sci.electronics.design, "Abstract Dissonance" wrote,

The way to limit yourself in the long run is to invest so much in some programmer that you feel stuck with it.

In the short run, pick a chip that is available and easy, and do something. My favorite are the Atmel AVR family.

Here's a starting point for a programmer that can cost you approximately $0.

Funnily enough, my first Eprom programmer was a Needham's unit.

I'm puzzled they haven't already done it. I can believe it's that difficult.

Graham

[snip...snip...]

Correct, the Atmel ISP does both the AVR family (AT90/ATtiny/ATmega) and the ISP versions of their 8051 family (AT89S).

One possible gotcha (although it's obvious when you think about it) is that the AT89 and AT90 use different polarities for the RESET pin. When the ISP dongle is in the quiescent state, it sets its RESET output to (IS NOT RESET) -- which, of course, is (IS RESET) for the other family. Had me wondering what the heck I'd done wrong, the first time that I used it across families and the chip wouldn't wake up ...

--
Rich Webb   Norfolk, VA

It looks like some kind of reprogrammable array. Maybe an FPGA could do the trick. I think I'm gonna dig more deeply... could be interesting!

Well, I am trying to do something.. make a programmer ;) Remember, I'm just doing it to learn. Maybe its a waste of time but I'm sure to learn something from it(hence its not really a waste of time). I got nothing better to do with my life ;)

yep... look how easy it is too ;) if all MCU's are about that easy then it shouldn't be any difficult to make a universal programmer like that(for most MCU's). There are two issues here though, one is connecting the computers parallel port to the MCU properly and adding the external components to the MCU... but there seems to be few components in this case and just a few pins(5) so if all MCU's are in general like this I can't see why there isn't a universal programmer that is cheap(i.e., < 100$)

What all probably do is just go ahead and do basicaly what you said though. Just get me a chip and start programming and do the programmer by hand like on the link you sent... Maybe I can do the "universal programmer" later(as it might require a MCU to use anyways).

Thanks, Jon

No, maybe every month or even week... atleast for a few months... since I am just learning about this stuff and there are tons of chips on the market. It may be the case that you are right but I don't see it is a waste of time but a challenge.

I already have a real world product waiting to be put together. I have even worked out most of the details. The only problem is programming the MCU(I did the project in a simulator). I have several other projects I want to do too and they all require MCU's. While I could easily just order the parts and put the project together and "hard program" the MCU I figure that I might as well see how hard it is to make a universal programmer.

Thanks, Jon

yeah, basicaly a "programmable matrix" one could view it as a matrix of transistors

• + + |+|+|

• + + |+|+|

or something like that. This way one could turn "make" paths by turning on and off certain combinations of transistors. For N wires there are N^2 transistors and hence N^2 "control" pins. But if one just wants permutations then this is reduced drastically since once one control pin is selected it should determin the state of many others.

now it might require 2N^2 transistors for analog(to get both polarities) and some other transistors for the switching logic but ultimately it will accomplish what I want. I'm sure there is a better way but this is what I thought up of trying to do it the "hard" way. I think its similar to an FPGA but the configuration is different(doesn't use and and or gates directly).

1 2 3 4 5 6 | | | | | |

-----------= a | | | | | |

-----------= b | | | | | |

-----------= c | | | | | |

-----------= d | | | | | |

-----------= e | | | | | |

-----------= f

basicaly one can "route" pin x into pin y by tracing a path from the start to finish. There are many ways to do this but some are easier than others. The problem is that one doesn't want two inputs to be "combined" so we would need to add transistors to "control" the flow. i.e., if I wanted to map pin 1 to pin f I could do it like this

1 2 3 4 5 6 |

- |

- |

- |

- |

- |

-----------= f

or pin 3 to pin b

1 2 3 4 5 6 | - | -------= b

Then its just a matter of combining the above to get different permutations of the input.

I think the FPGA is very similar in its idea except it is designed for digital systems instead of analog. In the above it should work with analog or digital(but the controlling is digital ofcourse). I'm still reading up on FGPA's and other devices like it though.

Maybe the idea could be extended from RAM where the flip flop holds a state. But the output of the flip flop is linked to the base of a transistor and controls that transistors state... so we can "write to memory" and it will allow us to program the array.

There might be other ways to do what I'm talking about but its just some ways I've thought about doing it.

Thanks, AD

OK, we are back to square one. Unless of course, your product is a programmer.

After many products and many programmers (twenty four years), I feal that a cheapo programmer is better than a generic do-it-all programmer.

But, it seems that you really want a do-it-all programmer.

Well, I for one can't help you.

I have two all-in-one programmers, and both are no longer used. One is dead (ChipMax), the other is ISA buss based (no longer supported by the manufacture), and I don't have an ISA buss on my laptop.

It took almost 10 years before they became obsolate.

Now I am buying in-circuit-programmers for the latest chips out there. I have an ISP programmer for TI, Microchip, Atmel. All together were cheaper than a single all-in-one device programmer.

Maybe someone else here still uses an all-in-one programmer.

Good luck, let us know what you end up with.

donald

I agree with you, Donald. The original poster could certainly learn something by putting together his own programmer (the very first MCU I ever used was an

8748, and not only did I build the programming hardware for it myself, I also wrote the assembler and programming software for it myself as well), but I think he's in danger of starting a project that'll never get finished when he sets his sights on tring to build some sort of universal programmer before he's even built one for a specific set of MCUs.

I've had two as well... a Xeltek Unipro and a Xeltek Superpro. I sold the Unipro sometime in the late '90s when it was still worth something; the Superpro I gave away in the early '00s becasue it has become pretty much worthless by then. I do see that Xeltek is very much alive and kicking, however, so clearly there's some market left for these beasts.

---Joel

Hi Joel,

I thought Xeltek went under, sometime in the 90s.

Weren't they bought be a larger company ??

donald

Could be, I don't recall; the "they're alive and well" comment comes from seeing that

appears alive. :-)

I remember calling them up once and they flat out told me that their stategy to keep the money flowing in was that, after a certain period of time, they'd quit supporting older programmers (releasing new programming definition files for them), even if the hardware was perfectly capable of programming various newer parts. Pretty annoying, although I suppose it's not all _that_ different from what happens with a lot of software these days. I would have preferred a model where you just had to buy software updates or could subscribe to updates for a period of time...