How about a Digispark for the ATTiny84 series of chips?

[digistump]

I guess it should be a compliment that I thought you were formally trained! at least since I didn't mean it in the "oh he must be an engineer" sense. I have a degree in Economics - I've yet to do anything related to it, except of course run a business, but I was doing that well before the degree.


I'll beat my head against the app note and USB spec some more - it is long over due - thanks for the clarifications, they make far more sense than any spec docs!


The zeners are about 2 cents together - you're right about the 5v reg, it would be a bigger effect, but still only 22 cents (with external parts) - I'd like to do a no regulator version, an onboard led version (with the attiny85), and a version with a mini/micro usb plug - if the no regulator version approaches the price of the Digispark it won't make it obsolete - I'll just make the original Digispark cheaper (which I can do if I bring the next batch of production in house, as I plan to for the pro).


I'll try the capacitor and removing the resistors on the o-scope as soon as things settle down - thanks for helping me to finally really understand all that.


@everyone else - If anyone else wants to patch micronucleus to try it with no pull-up I'd be grateful (and probably send some small gifts), if not I should have time when the DigiX stuff settles down to give it a go if no one else gets to it first


off to continue to buy every sam3x8e in the USA...

[gogol]

Hi,

On the zeners - am I correct that even with all of this - without them when running at 5v it would be exceeding the USB spec of 3.3v (3.6v) max? If there are solder jumpers to disconnect the zeners, do you still hate them? How about you gogol?

Solder jumpers would a big step forward, as it would allow to reuse a digispark again with zeners, after disabling them for some projects. Before I went to program plain attinys, I destroyed the zeners on one spark with a scalpel, which is irreversible.  The solder-bridge would be reversible.

In the meantime, i played with self-constructed USB-adapters, where I used through-the-hole zeners, which made more problems, than the tiny smd-zeners of the digispark. I have now two digisparks, where I removed the (broken) controller. I can use them as well as USB-adapter to core-attinys, and they work in some scenarios, where my own adapters don't work.  However, when I remove the zeners from my adapter, they are working as well.

However, and there comes the point, where I agree with Erik:
I have not tried connecting my adapter without zeners to my workstation and current working notebooks. I tried that only on some 8yr++ computers.

I haven't found up to now clear statements in the USB documentation stating either, that 5V is OK or that there is any danger over 3.3V.

From there I agree with Erik, that you need to be on the safe side, when he is releasing products.  So I think, that we need to find either very clear statements (in addition to the proof of concept) or we have to live with a solution like the solder bridge.

Replacing the pullup with a software solution is form my point of view much higher priority!

As USB will work with cutted zeners (in some cases maybe even better), it will not work with a cutted pullup!  Furthermore the pullup destroys the use of the pin even more, than the zener.

regards

  gogol

[CBcracker]

off to continue to buy every sam3x8e in the USA...

I know its too late to change the digiX now, but maybe for a "digiX lite" this chip would be worth considering: MK20DX128VLF5.  They're at least half as powerful as the sam3xe for 1/3rd the price (266c/q100 @digi).

[digistump]

@gogol - that pretty much follows my thoughts exactly regarding both the zeners and the pull-up and priorities and caution.


@CBCracker - lost of great ARM chips that are much much cheaper than the sam3x - but also a boatload of work to add full arduino support for one and maintain it - and then libraries etc would all have to be ported. The goal of the DigiX was 100% compatibility with the Due so we were stuck with at the very least the SAM3 series. I hope to work with ARM chips more and make more products with them - but those likely would be a departure from "Arduino" as I've been talking with other folks about new IDE options that are more full featured/more user firendly.


To get back on topic - I ordered a bunch of parts to start prototyping the Pro in the next month or so!

[digistump]

@CBcracker - one thought I've had about pulling the resistors - we're making the assumption it is usually plugged in directly with no cable - which might be wrong seeing as every Digispark I have attached to a computer right now is on an extension (some several meters long). SO perhaps the resistors need to be there but just a lower value to find an inbetween ground when plugged direct vs on a cable?

[CBcracker]

@CBcracker - one thought I've had about pulling the resistors - we're making the assumption it is usually plugged in directly with no cable - which might be wrong seeing as every Digispark I have attached to a computer right now is on an extension (some several meters long). SO perhaps the resistors need to be there but just a lower value to find an inbetween ground when plugged direct vs on a cable?

If the termination is not perfectly matched, you'll get reflections; the further it is off, the stronger the reflections.  However as I posed before about electrical propagation delays, it will be way too fast to affect 1.5Mbps communication.  You shouldn't even be able to see the difference (with vs. without terminating resistors) on your 100Mhz scope.  If you add caps for slew rate control, then the resistors *will* make a noticable difference of what goes out to the host USB port; they will be much higher than the ESR of a ceramic cap, and will help the caps dampen the voltage swings.  It should be pretty obvious if you put two scope lines on the t85 pins for D+/D-, and two more scope lines on the USB female connector's D+/D-.
In the end the resistors shouldn't make much difference to what the host sees, since it's transceiver will have series resistors on it, but they'll probably be internal to one of the ICs so you won't be able to trace it on your scope.

[defragster]

 Just saw this from kick starter fall 2011: http://moderndevice.com/product/extracore/#!prettyPhoto - Yes, I'm new to this area.  Here’s hoping to add something:
 
That XCore is $15 and but very minimal - excepting the high cost/pin processor? My ideal Pro: as many uniformly usable pins as the Spark could be cleaned up to have, not wasting any 'runtime' power on dev time hardware so a battery could last longer with a small footprint at a cost low enough to put them anywhere they could be of use.
 
The XCore loses that on cost and uneeded pins for tiny jobs, but has simplified/reduced so much else that it is only marginally larger than a Spark because of the USB tab.  That USB tab is ‘the cool part’ with power and completeness – but also the non-standard part bringing overhead.
 
If the Spark lost any of the USB tab/support and gained usable pins - and had lower run-time power that would be a Pro. The XCore uses external USB/Serial - is that what was meant on this thread [gogol:4228] by using a Spark for a programmer? If not then a standard USB/Serial card would complete my idea of a Pro. Any possible reduction in the micronucleus giving more code space is also a Pro. Especially if the standard IDE interface could be used for programming without the software serial requiring the user to jump through hoops on every code edit.
 
The user can add a pilot light when needed. Does dropping power count as a full reset? There seemed to be other wins if this didn't plug direct into USB port (thinner, gold free board), if the $4 programming tool became smart with a Spark, or just split off the development support pieces, the Pro would be a much smaller XCore. 
 
If a programmer and a Pro were $16 – but ~$30 for a Pro four pack – it could have needed support for a beginner as a Pro upgrade and a better cost per I/O utility factor for real world jobs.

[CBcracker]

Just saw this from kick starter fall 2011: http://moderndevice.com/product/extracore/#!prettyPhoto - Yes, I'm new to this area.  Here’s hoping to add something:

$15 for a small ATmega328 board is about 3x overpriced.
http://www.fasttech.com/products/0/10004915/1380906-pro-mini-microcontroller-circuit-board-module-for-
The pro mini is also better for breadboarding since most of the pins run down the sides.

[defragster]

Wow, 5 for $23.70.  I posted then saw ATmega328 in 10 count for $2 each, about the same as the Attiny85 - I expected more $ diff.

[digistump]

A Digispark will never be unable to program itself - that is the premise of the Digispark and also a personal thing for me - I hate "development" boards that you need another board/programmer to use - to me then it is just a breakout board (which certainly have their place). I've also decided somewhere in this all that the Digispark will always be an Attiny device.


Will we ever produce an atmega device? - most certainly, but it will be its own line. Atmega328 devices have been done so many times, by so many talented people it just isn't a market we want to compete with right now - we couldn't make the extra core for less, and comparing their prices or our prices to China will never match up - heck we've yet to pay ourselves a dime and we still can't match Chinese prices, we never will be able to because even if we were to have it made there we still pay to get it to the US and pay taxes on it when it comes in and pay taxes on the income, don't have subsidized shipping, have much higher costs for utilities, buildings, etc. Someday we'll also have to pay people for providing support, community, instructions, etc - as my 9am-3am days are getting long. For this reason I do and will continue to ignore comparisons between our prices and Chinese prices. If someone ever says "Hey sparkfun/adafruit has this for less" that will have my attention, we carefully set our prices to be significantly more affordable than other quality US companies who provide similar levels of support, community, schematics, board files, instructions, etc.


When we make an atmega device it will be a breakout board or simple dev boardand it will be to compliment our flagship products like the DigiX and Digispark or it will be our own take on the standard Arduino aimed at lowering the price while still providing a US made, quality, and paying royalties to Arduino if requested. Which is another reason we've stayed away from the atmega arduino compatibles - we feel that innovation is just as important as affordability, etc - sure I can make an atmega board smaller, cheaper, etc - but in the end, at best, it will only be marginally better - the Digispark makes the attiny series far easier to use and inspires real innovation in using less resources, smaller boards, etc - we feel this is likely a reason for the interest it generates (it's not just another clone/derivative). If you're in need of an atmega based board I strongly recommend the extra core or femtoduino - they are both great boards made by great folks!

[defragster]

Good points indeed.  I was looking at the extreme minimization for power savings - looking for some gain in the middle.

[digistump]

One thing the pro will have - if it isn't obvious is solder jumper to detach anything that uses power - these will be on the zeners, LEDs, etc. That combined with sleeping the chip and clocking it down should allow for extremely low consumption

[CBcracker]

Wow, 5 for $23.70.  I posted then saw ATmega328 in 10 count for $2 each, about the same as the Attiny85 - I expected more $ diff.

If you're gonna buy 5, last week I posted an AliExpress link to a seller with a price of $17.50 for 5 with free shipping.

[defragster]

Indeed you did - That link is going for US $16.62 right now.  I wasn't looking for 5 - just went to the 5 count price break.  I was on the end of that 8/8 post . . . should have gone back.

[defragster]

One thing the pro will have - if it isn't obvious is solder jumper to detach anything that uses power - these will be on the zeners, LEDs, etc. That combined with sleeping the chip and clocking it down should allow for extremely low consumption

I missed that and just caught the discussion as proposals to go that way.