Forums

New Atmel ATtiny Microcontrollers have 6 Pins, ADC, and LOW Power Consumption

Started by Bill Giovino November 24, 2009
On Nov 26, 2:52=A0am, David Brown <da...@westcontrol.removethisbit.com>
wrote:
>=A0While that is not an unrealistic scenario for large > volumes, a single wire programming interface that worked across the > voltage range would have been much more useful.
Of course it would have been much more useful - it would also have been much more expensive, and larger. These tiny devices are a very careful trade-off, and usually they Do have larger, more featured siblings in the family. Those larger siblings have wider Vcc, more pins, and guess what ? - they also cost more too! Only using low volumes ? - then saving a few cents is not even on the radar, is it ? Using Millions ? then those few cents are very important. -jg
"Jim Stewart" <jstewart@jkmicro.com> wrote in message 
news:hejubo$i8u$1@news.eternal-september.org...
> Viktor wrote: >> On Nov 24, 8:17 pm, "Bill Giovino" <contac...@microcontroller- >> nospam.com> wrote: >>> http://www.microcontroller.com/news/Atmel_ATtiny_4_5_9.asp >>> >>> These Atmel ATtiny microcontrollers have 6 pins, and an A/D. Draws a >>> measly 100nA in >>> sleep mode. >>> >>> Article shows a useful pinout (what's the pinout on a 6-pin >>> microcontroller?), a block >>> diagram, and an interesting competitor analysis. >>> >>> Pretty funky little device! >>> >>> - Bill Giovino >>> Executive Editor >>> http://Microcontroller.com >> >> No debugWire :( > > That was the first thing I looked for. The > lack of debugWire will definitely rule it out > for low volume, fast turnaround projects.
If you honed your software skills and took a bit more care, you wouldn't need debug :)
On Nov 25, 10:58=A0pm, Viktor <vkes...@gmail.com> wrote:
> > No debugWire :(
That's a Die/cost trade off - use a larger variant for debug, if you really need to. Nice would have been a SOT23-8 pin debug version, and a 6 pin std model, but probably the die area impact would have meant it no longer fitted into SOT23-6.... I'm sure Atmel would have considered that feature combination. One omission that DID surprise me, was the lack of a temperature sensor. These are now very common, on any ADC uC, and the smaller ones are MORE likely to be "at the thermal coal face" The die-impact of that is very low indeed, so perhaps Atmel decided that the testing time was too costly ? -jg
-jg wrote:
> On Nov 26, 2:52 am, David Brown <da...@westcontrol.removethisbit.com> > wrote: >> While that is not an unrealistic scenario for large >> volumes, a single wire programming interface that worked across the >> voltage range would have been much more useful. > > Of course it would have been much more useful - it would also have > been much more expensive, and larger. > > These tiny devices are a very careful trade-off, and usually they Do > have larger, more featured siblings in the family. > > Those larger siblings have wider Vcc, more pins, and guess what ? - > they also cost more too! > > Only using low volumes ? - then saving a few cents is not even on the > radar, is it ? > > Using Millions ? then those few cents are very important. >
I understand all that. However, even in smaller quantities such a small and low power device could be very useful. As you say, in low volumes the last few cents are not important - many people would be happy to pay twice as much ($1 instead of $0.50) for a version of these devices with usable in-circuit programming and debugging support. It is true that such features add to the size and expense of the device - but not so much that it would not fit in the same package, or cost noticeably more to produce. After all, the difficult part of the in-circuit programming is already in place, and the debugWire block is already developed for other AVR's. I'll grant that asking for programming at all voltages could be too much - but at least 3.3V is not unreasonable.
On Nov 26, 9:38=A0am, David Brown <da...@westcontrol.removethisbit.com>
wrote:
> I understand all that. =A0However, even in smaller quantities such a smal=
l
> and low power device could be very useful. =A0As you say, in low volumes > the last few cents are not important - many people would be happy to pay > twice as much ($1 instead of $0.50) for a version of these devices with > usable in-circuit programming and debugging support.
Taking some examples, from Digikey, the rather more capable tiny13A is 57c/100+ [EEPROM. Debug. Wide VccPgm, wider Vcc operate ] - _but_ that comes in SO8 or QFN10(3mm), so that shows the die-cost of these features is quite a lot. Tiny10 shows at 0.49960@ 25+ [and also the bemusing 79c/5K ! oops... ] So, if 10c matters, then Tiny10 makes sense. I also see the Tiny13A lacks a temperature sense - ?? some newer Tiny's have it.
> > It is true that such features add to the size and expense of the device > - but not so much that it would not fit in the same package, or cost > noticeably more to produce. =A0
Err THAT claim seems a big call - how do you _know_ that it would still fit in SOT23-6 ? The Tiny13 package choices say otherwise: that charge pumps do have a large die impact, as do EE blocks.
> After all, the difficult part of the > in-circuit programming is already in place, and the debugWire block is > already developed for other AVR's. =A0I'll grant that asking for > programming at all voltages could be too much - but at least 3.3V is not > unreasonable.
Charge pumps =3D=3D die area. I'm sure Atmel looked at all their blocks, and the primary target on the Tiny10, looks like the SOT23-6 package. (and maybe some price point as well, but the package becomes vitally important to meeting the price...) There is some room between Tiny13A, and tiny10, so perhaps a ATTiny14 is in development ? -jg
-jg wrote:
> On Nov 26, 9:38 am, David Brown <da...@westcontrol.removethisbit.com> > wrote: >> I understand all that. However, even in smaller quantities such a small >> and low power device could be very useful. As you say, in low volumes >> the last few cents are not important - many people would be happy to pay >> twice as much ($1 instead of $0.50) for a version of these devices with >> usable in-circuit programming and debugging support. > > Taking some examples, from Digikey, the > rather more capable tiny13A is 57c/100+ > [EEPROM. Debug. Wide VccPgm, wider Vcc operate ] > - _but_ that comes in SO8 or QFN10(3mm), so that > shows the die-cost of these features is quite a lot. >
Not necessarily - without opening up the packages, we don't know if the die is significantly different in size even though the package is slightly larger. We also don't know anything about the process size (at least, /I/ don't) - it could be that the Tiny10 has a physically smaller die mainly because of a process shrink. EEPROM takes a fair bit of die area, so it could also account for any differences.
> > Tiny10 shows at 0.49960@ 25+ > [and also the bemusing 79c/5K ! oops... ] > > So, if 10c matters, then Tiny10 makes sense. > > I also see the Tiny13A lacks a temperature sense > - ?? some newer Tiny's have it. > > >> It is true that such features add to the size and expense of the device >> - but not so much that it would not fit in the same package, or cost >> noticeably more to produce. > > Err THAT claim seems a big call - how do you _know_ that it would > still fit in SOT23-6 ? >
I don't *know* - it's just a somewhat educated guess. Without concrete data from Atmel (unlikely here, I think), it's the best I can do.
> The Tiny13 package choices say otherwise: that > charge pumps do have a large die impact, as do EE > blocks. >
Yes, I thought of that, which is why I thought a 3.3V programming voltage might be more realistic - it's a smaller step from the full 1.8V range, and thus a smaller impact.
>> After all, the difficult part of the >> in-circuit programming is already in place, and the debugWire block is >> already developed for other AVR's. I'll grant that asking for >> programming at all voltages could be too much - but at least 3.3V is not >> unreasonable. > > Charge pumps == die area. > > I'm sure Atmel looked at all their blocks, and the primary target on > the Tiny10, looks like the SOT23-6 package. (and maybe some price > point as well, but the package becomes vitally important to meeting > the price...) >
I have no doubt that Atmel looked at their potential customers' needs here, and that the likes of us wanting better ISP and/or debugWire are not very relevant. There is no point in Atmel listening to customers who buy less than a million of these things at a time - the profits are too low to bother about anything less. But that won't stop wishful thinking!
> There is some room between Tiny13A, and tiny10, so > perhaps a ATTiny14 is in development ? > > -jg
-jg wrote:
> On Nov 25, 10:58 pm, Viktor <vkes...@gmail.com> wrote: >> No debugWire :( > > That's a Die/cost trade off - use a larger variant for debug, if you > really need to. > > Nice would have been a SOT23-8 pin debug version, > and a 6 pin std model, but probably the die area impact > would have meant it no longer fitted into SOT23-6....
sounds like you want another 6 pads on the top of the IC for programming and debugging. (via pogo pins or similar) :-) Cheers Don... -- Don McKenzie Site Map: http://www.dontronics.com/sitemap E-Mail Contact Page: http://www.dontronics.com/email Web Camera Page: http://www.dontronics.com/webcam No More Damn Spam: http://www.dontronics.com/spam 5% to 15% OFF all 4DSystems OLED Displays & modules. http://www.dontronics-shop.com/4dsystems-m1.html
On Nov 27, 4:41=A0pm, Don McKenzie <5...@2.5A> wrote:
> -jg wrote: > > On Nov 25, 10:58 pm, Viktor <vkes...@gmail.com> wrote: > >> No debugWire :( > > > =A0That's a Die/cost trade off - use a larger variant for debug, if you > > really need to. > > > =A0Nice would have been a SOT23-8 pin debug version, > > and a 6 pin std model, but probably the die area impact > > would have meant it no longer fitted into SOT23-6.... > > sounds like you want another 6 pads on the top of the IC for programming > and debugging. (via pogo pins or similar) > > :-)
A wry smile perhaps, but I actually do have here, emulation parts with connection 'out the top' !! So it has already been done :) However, perhaps more practical, is the 'more pins' choice, and that has also been done. Winbond and Fairchild are two suppliers I know OTTOMH that supply bigger-bonded emulation - usually adding 2-4 more pins. Atmel do it on some 80C51 variants. - but even bonding pads cost die area, and the die paddle in a SOT23 will be a VERY serious constraint. -jg
On Nov 27, 5:41=A0am, Don McKenzie <5...@2.5A> wrote:
> -jg wrote: > > On Nov 25, 10:58 pm, Viktor <vkes...@gmail.com> wrote: > >> No debugWire :( > > > =A0That's a Die/cost trade off - use a larger variant for debug, if you > > really need to. > > > =A0Nice would have been a SOT23-8 pin debug version, > > and a 6 pin std model, but probably the die area impact > > would have meant it no longer fitted into SOT23-6.... > > sounds like you want another 6 pads on the top of the IC for programming > and debugging. (via pogo pins or similar) > > :-) > > Cheers Don... > > -- > Don McKenzie > > Site Map: =A0 =A0 =A0 =A0 =A0 =A0http://www.dontronics.com/sitemap > E-Mail Contact Page:http://www.dontronics.com/email > Web Camera Page: =A0 =A0http://www.dontronics.com/webcam > No More Damn Spam: =A0http://www.dontronics.com/spam > > 5% to 15% OFF all 4DSystems OLED Displays & modules.http://www.dontronics=
-shop.com/4dsystems-m1.html debug pins on top of package? yes ceramic SO23-6 :) jokes beside, i happen to have both ATtiny13A and ATtiny10 samples on my desk the price difference is minimal, also the PCB area difference is minimal t13 3x3 QFN vs SOT23-6 what takes also about 3x3 mm I think t10 positions correctly where it was designed to fit it in: REPLACE all microchip sot23-6 devices it is designed to be drop in replacement, and to be programmable using the microchip isp connector (same pin arrangement) would be nice to see 3.3V and self programming, and, well yes it could be better but for dirt cheap places, it fits the bill, it little easier to assembly, little bit cheaper if the bom allows, then tiny13a is next choice Antti