Propeller data sheet here!!!

cgracey

Thanks to mainly Paul and Stephanie here, we have a preliminary data sheet ready. This has some important graphs of power consumption that have not been released before, plus some electrical specs. We've just·specified a·commercial temperature rating for now, until we do some more tests.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


Chip Gracey
Parallax, Inc.

Skogsgurra

Thanks! I was about to ask what typical pin capacitance is - found it immediatly in the data sheet. Great!

BTX

Great !! Thanks.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards.

Alberto.

Tom Bampton

Excellent!

One suggestion, it would be really useful if you could start adding bookmarks to your PDFs as a table of contents in a similar way to how the PIC datasheets do it. The datasheet and propeller manual would be significantly more accessible with them.

I have lost count of how many times I've sworn looking for info in the prop manual due to having to repeatedly return to the contents (which doesnt have hyperlinks) and then repeatedly use goto page til I find what I'm looking for. Conversely, with the PIC datasheets, I can always jump to within a couple of pages of the information I need using just the bookmarks pane in Acrobat Reader.

The propeller manual is huge and I really don't expect you to waste tons of time adding bookmarks to it, but the datasheet is much smaller and it would be great if you could add them there.

Anyway, nice work Paul and Stephanie, the datasheet is going to be extremely useful.

T.

Jeff Martin

Hi Tom,

I'll look into doing this... seems to me the last time I tried, Adobe crashed and I had to move on to something else, but since then I have a new version.· I will post update notifications here.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
--Jeff Martin

· Sr. Software Engineer
· Parallax, Inc.

MikeK

In section 7.2 (DC Characteristics), shouldn't Vil (max) be 0.3 Vdd, rather than 0.3 Vss?
Mike

CJ

section 3.3 has a typo in the amount of cog ram there is, it says 16KB but it should be 16Kb

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Parallax Forums - If you're ready to learn, we're ready to help.

Jeff Martin

MikeK said...
In section 7.2 (DC Characteristics), shouldn't Vil (max) be 0.3 Vdd, rather than 0.3 Vss?
Mike

<section 6.2>

Yes, you're right, 0.3 Vdd.

CJ said...
section 3.3 has a typo in the amount of cog ram there is, it says 16KB but it should be 16Kb

Yes, or more clearly stated, 2 KB.


Thanks for your input!

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
--Jeff Martin

· Sr. Software Engineer
· Parallax, Inc.

Post Edited (Jeff Martin (Parallax)) : 3/31/2007 11:03:48 PM GMT

Tracy Allen

Thanks. It is a wonderful summary of Propeller capabilities.

Under maximum ratings, it states 0 to 70 degrees C for ambient temperature under bias, but then in the AC and DC characteristics it states a temperature range of -40 to +125, and it is unclear how that applies to the Min and Max values. There is a clear contradiction there that should be worded in a different manner. It states that the typical values apply at 25 degrees C. IMHO, the maximum rating should simply state TBD, to be determined, and the AC and DC characteristics should reflect testing that has actually been done for variation over temperature and state "preliminary".

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

Paul Baker

Hi Tracy, the reason for the discrepancy is that the chip hasn't been temperature stressed while biased. But the maximum and minimum values provided in the data reflect more than just temperature stress, they also represent the most extreme corner cases of the process variation. As such, the only means for obtaining the data is to simulate the corners of the process and the junction temperature. This is how the industry achieves these numbers considering how much money it would cost to produce wafers at the corner cases only to be thrown away after the test because they are marginal on thier performance due to the skewing of process variables.

Our intent is to test the chip at automotive, and rather than resimulating the datapoints when we move to the full datasheet (which would shift all the numbers and likely cause a lot more confusion than this discrepancy), we decided to provide the final number for the sake of consistancy wherever possible.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Steph Lindsay

Hello all, and thank you for your feedback!·

Version 0.2 corrects the items noted above, and is attached.· This and future versions will be availabe from the Propelller downloads page.

http://www.parallax.com/propeller/downloads.asp

You can email me directly with any errata you find: editor@parallax.com

-Stephanie Lindsay

Editor, Parallax Inc.

Tracy Allen

Hi Paul and Stephanie. I just want to say, good job, again.

Paul, I don't want to argue about the temperature specs, but I am of those people who jump to the electrical specs. I found your exlanation uhhhh, slippery, and you need to be prepared for the question. Usually there are typical values, min and max, given for operation at room temp., and separate values given for variation over temperature extremes based on statistcal techniques that take into account both real measurements and the design parameters. The statement in that part of the data sheet strongly imples that at least cursory test has actually been done under bias at those temperatures, and that contradicts the statement of maximum values.

There is language that can hedge the bet, which applies admirably to the Propeller: "The xxxxx are designed, characterized and expected to meet these extended temperature limits, but are not (yet) tested at -40°C and 85°C. Guaranteed I grade parts are available. Consult the factory." (that language, except the "yet" are lifted from a Linear Tech data sheet.) To the engineer, that is reassuring without overstating the case, especially in a "preliminary" data sheet.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

Tracy Allen

Sorry to harp on the subject, I see you already addressed that concern in the revised data sheet. Still, the meaning of "simulated" might not be obvious, and IMHO the same "simulated" range really should apply to the maximum ratings for sake of consistency.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

mahjongg

Good work! It's very important to have a datasheet like this.

One small comment, the logic level thresholds as stated at page 3 (paragraph 1.9 pin Descriptions, P0-P31) seem to be contradictory to those stated at page 20 (Paragraph 7.2, DC characteristics, Vih and Vil). Also page 3 mentions the term "VDC" which is not used in the electrical characteristics, or maximum ratings.

Mahjongg.

Phil Pilgrim (PhiPi)

One item I can't find in the current (PropellerD50.3) datasheet is the brownout voltage. Will this be spec'd in a future rev?

Thanks,
Phil

Paul Baker

The brownout voltage is between 2.5 and 3.0 V. While there is wide variance from chip to chip, a specific chip's brownout voltage is very stable over time and temperature.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

viskr

I was looking up something in the most recent spec, and thought I'd point out something that is a bit misleading.

When you mention memory sizes they are referred to as KB, which most people assume means KBytes. It is customary for CPU specs to reference memory sizes in bytes even if there natural word width is 64 bits. For instance in section 3.3 while the memory is also described as 512 longs, it really should be 2KB or 16Kb, with 2KB being the most common.

Memory sizes in bits are normally reserved for describing memory devices or embedded memory of FPGAs.

Ale

viskr: The pointers inside the cog are to longs, not to bytes, so sort of makes sense to speak about longs. Maybe an aclaration like: 512 longs (2 KBytes), is not a bad idea.

evanh

"Word" is the correct term. Long is nominally undefined. So, it should be listed as 512x32bit words if it isn't.

Mike Green

"Long" has been used in several programming languages to refer to a 32-bit or 64-bit value where the base value is 16 or 32 bits. Given that the 3 operand sizes used in Spin are 8 bit, 16 bit, and 32 bit, it's reasonable to use "byte", "word", and "long" for them, particularly with the history of having 8 bit bytes and 16 bit words in Parallax's other two lines of processors and the extensive documentation that uses those terms.

Duffer

I guess nobody uses the term "Double Word" any more to refer to a 32 bit intity. Maybe it's just an old mainframe term. Bit, Nibble, Byte, Word, Double Word was what I was used to, but getting used to thinking in Longs hasn't been a problem.

Steve

deSilva

As this a purely academic discussion I shall be glad to join it

Evanh is of course correct in his interpretation of the term "word". A "word" is the basic addressable unit of a memory absolutely independend of its size, as well as the basic unit to be handled by the ALU.

Early mainframes - generally used as "number chrunchers" - used 30 to 60 bits, as it makes no sense to have less precision for numerical computations.

In certain applications even higher accuracy was needed, and the concept of "DOUBLE" entered early programming languages like FORTRAN IV. These operations were performed with software support on (single) word ALUs.

The next generation of "mini computers" used 16 bits as their "natural" units, mainly driven by cost considerations. Note also that the size of the addressable units was generally coupled with memory width.

In the meantime it had become popular to address memory as "bytes" as non-numerical applications had outplayed the number crunching. As always marketing considerations played their role, as you can announce much larger numbers in bytes than in words

With the advent of double precision ALUs (64 and 80 bits) and ultra-wide memory there is no natural system any longer.

I personally still understand the term "word" in its original context. So I generally speak of words in a COG sometimes creating confusion....

Post Edited (deSilva) : 8/11/2007 9:11:35 AM GMT