JonnyMac said...
Now that SX/B is in fact locked, I'm back to work (mixed in with other projects). The SX is going out of production, not going away. I see Delorian (cars) on the streets of Los Angeles all the time! [noparse];)[/noparse]
yeah, but you don't use Delorians in every little gadget you want to build. I need a few more SX chips than one...· [noparse];)[/noparse]
-Dan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
...which is why Ken gave fair and ample warning for any of us (including EFX-TEK [noparse][[/noparse]I'm part of]) to order what we think we're going to need. Like most, I wish that Parallax could gain control of the SX IP, "fix" it (flatten the memory, add more RAM, correct the power consumption issue, remove the PIC-compatibility mode) and have a SuperSX; but that's not going to happen. Where we (EFX-TEK) can, we're migrating to the Propeller. Where it doesn't make sense we're looking at other options.
The SX has been my favorite general-purpose microcontroller for a long time, for a lot of reasons. Its small size, high clock speed, low cost, and simple instruction set made it one of the best possible chips for a great many applications. Unfortunately, it seems that if the SX goes, so will my respect and loyalty for Parallax. The Propeller is a nice chip, but it serves a completely different market that the SX did. Replacing the SX in my designs with a Propeller would be like putting a mainframe in a coffee maker. So I will pray tonight that by some miracle, Parallax obtains the necessary assets from Ubicom to keep the SX alive. But if the SX is destined to die, then I am afraid that unless Parallax has a completely new product up its sleeve to fill this gaping hole, one of its biggest supporters will be forced to tearfully browse the product lines of the likes of Microchip and Atmel. [noparse]:([/noparse]
You do what you need to do. The SX, after this last manufacturing run, is gone. When the on-hand stock is gone in a few years, there will be no more. Parallax has no control of this situation. There is no way for Parallax to acquire the right to make more nor even design their own version. The issues are legal and are between Ubicom and Microchip.
I don't think you're going to find another chip similar to the SX48 in all the factors you mentioned. There are a lot of microcontrollers on the market with the number of I/O pins you want, but there are none really with the high end clock speed of the SX48. On the other hand, some of the functions handled by the SX48 using interrupt handlers would be handled by dedicated I/O units so the clock speed might not be necessary for most applications. Some of the Atmel AVR devices have the right number of I/O pins and a similar price, but run about 1/2 the speed of the SX48.
The Propeller is about twice the cost of an SX48. You'll need an external EEPROM which adds about $2.50 in moderate quantities. Unless your application is very very cost sensitive, that's about $6 total over an SX48. I think you'll find that the additional development costs associated with an AVR or PIC microcontroller over a Propeller may balance that for a lot of applications.
Post Edited (Mike Green) : 10/12/2009 2:26:44 AM GMT
Mike Green said...
I don't think you're going to find another chip similar to the SX48 in all the factors you mentioned. There are a lot of microcontrollers on the market with the number of I/O pins you want, but there are none really with the high end clock speed of the SX48. On the other hand, some of the functions handled by the SX48 using interrupt handlers would be handled by dedicated I/O units so the clock speed might not be necessary for most applications. Some of the Atmel AVR devices have the right number of I/O pins and a similar price, but run about 1/2 the speed of the SX48.
The Propeller is about twice the cost of an SX48. You'll need an external EEPROM which adds about $2.50 in moderate quantities. Unless your application is very very cost sensitive, that's about $6 total over an SX48. I think you'll find that the additional development costs associated with an AVR or PIC microcontroller over a Propeller may balance that for a lot of applications.
Silicon Labs has a 8051 clocking at 100Mips (C8051F36x). I haven't used one, but targeted as a replacement for some of my apps. When I run out of SX's I'll have to start looking at these. It's the only 8 bit micro with similar clock speed that I'm aware of.
Again, I'd rather see parallax sell SX's, but since that won't happen forever, I need to start looking at other micros.
Very unfortunate situation...·· ...And, I think I·know who to blame...
-Dan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
I am actually using a dsPIC30F4013 now and you can't just compare MIPS.
Using a 20MHz resonator the sx can set/clear a pin in 1 instruction (= 1 cycle), the dsPIC
requires 3 instructions to make that happen. Running internally at 80MHz it thus takes 12 cycles
to set/clear a pin ( = less than 7 MIPS).
Running VP's on a dsPIC is also quite demanding and I must yet see to have
8 uarts running (that's why I used the sx).
So why don't you see next generation SX chips, or 8051s or other 8/16 bit processors?
The answer is based on Moore's Law, the number of transistors in a given area doubles every 2 years.
Most small embedded SOCs are pad limited these days, which means for a given number of pads the die will not get any smaller.
Add to that the size of the CPU is becoming a smaller fraction of these chips, with peripherals and memory taking up most of the space.
Add to that the cost of a mask set around $1M these days for even a 4 year old process.
What this all implies is it's expensive to develop a new chip let alone a lot of varieties. The not so secret, dirty little secret is many of the variations of these chips are actually the same die, its just feature disabled at the factory.
The difference in area between an 8 bit CPU and a 32 bit CPU gets to be a rounding error. With chips like ARM, in thumb mode even the advantage of code space is eliminated (thumb code is a 16 bit instruction version of ARM). And the big advantage of a 32 bit chip over the 8/16 is the need to page memory goes away, as paging is a huge hassle. Most code is now written in standard languages, mostly C, so portability becomes simpler.
Hence it makes no sense to update an 8/16 part.
FYI a 70MHz ARM with 32KB Flash, 8KB RAM, and some peripherals now goes for < $4, quantity 1
I know this question is very premature........but has any thoughts gone into a possible replacement for the current SX chip in the current Parallax products?
I know there is not a direct replacement, I just want to get some of the ideas which may have passed across the table.
We understand that none of the possible ideas would be the final decision.
@James - I know I'm not Ken, but I've read the many discussions on this issue on these forums involving Ken, Chip, Beau, Paul, etc. It's very clear that Parallax is using the Propeller as an SX replacement for many of their products and, given the fact that they own the IP, this will only continue. There are some products where the programming is fixed, where there's no intention of consumer access to the microcontroller, the microcontroller is simply used as another component, and component size and cost is paramount where Parallax will use some other manufacturer's microcontroller. External to Parallax it won't matter at all which one they use. They could even use several different manufacturers' products. It won't matter and it's no more our business than Parallax's choice of a resistor or voltage regulator to use.
Mike Green said...
@James - I know I'm not Ken, but I've read the many discussions on this issue on these forums involving Ken, Chip, Beau, Paul, etc. It's very clear that Parallax is using the Propeller as an SX replacement for many of their products and, given the fact that they own the IP, this will only continue. There are some products where the programming is fixed, where there's no intention of consumer access to the microcontroller, the microcontroller is simply used as another component, and component size and cost is paramount where Parallax will use some other manufacturer's microcontroller. External to Parallax it won't matter at all which one they use. They could even use several different manufacturers' products. It won't matter and it's no more our business than Parallax's choice of a resistor or voltage regulator to use.
Well, that is a good point. I guess I'm prying where it is not really acceptable to pry.
In some of our products, the SX is simply another component and it can be replaced with something else which is newer, cheaper, and probably just as fast. We've scheduled out redesign of any SX-based product except for the BASIC Stamps (we're holding many years of SX chip inventory for them ~15-20 years for BASIC Stamp manufacturing). Something like the Ping))) could use another processor just as easily, so it will. Anything else you want to know, just ask.
Too early to do that right now since there exists ample demand and an eventual short supply. The prices have been the same for a year, however, and we didn't raise them even though we are in this EOL situation. Many companies actually view EOL as an "opportunity" rather than the PITA we find it to be for our customers and our company. Done our best to keep this one reasonable for customers.
Near Term: Keep using SX, I've got a bunch around and like SX/B, maybe buy some more protoboards
- hoping the forum for SX doesn't die out anytime soon.
Most Projects: Switch to Propeller, could use the PROPBasic by Bean but will probably keep trying to learn Spin
Really Simple Projects: Considering PICAXE - like 8 & 18 pin DIP formats., oh, I still have a small supply of Basic Stamps too.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tom Talbot
New Market, MD, USA
Now, this is a serious disappointment.
I just got started with these chips and I think they're incredible.
[noparse][[/noparse] They speak BASIC. [noparse]:)[/noparse] ]
I guess I'll just have to order a good 'hand-full' of these ICs
so I can finish a couple of copies of my projects.
Yes, I'm already learning all I can about the propeller. Jon's article's in Nuts and Volts are helping. I agree with you Electronegativity-- Parallax has been good to me as someone learning. I'll stick with them, they've earned it!
Wow. This really just hit me today. I think I subconsciously ignored the single line EOL thread summary when I was here a month or so ago.
This is really a shame. The SX is a great little device. It was my introduction into microcontrollers, and with the books, good documentation, tech support, and of course everyone here made the learning process fun, easy, and rewarding.
Like others here, I've got some decisions to make about how/if to migrate my current SX hobby project to another uC. I've been playing around with Xilinx FPGA's, and add on 8-bit Picoblaze uC and 32-bitMicroblaze uP's soft cores. I'm seriously thinking about going down that path.
The last time I mentioned this, probably because I'm too naive to understand the whole market, I was shush'd up quickly. I would love a softcore SX! While I'm no expert, given the size of the instruction set, limited built-in hardware peripherals, etc, creating a softcore for it wouldn't be an impossible task. I know, I know, Ubicom won't release the IP and all that. Just dreaming.
I will say that given my retro-computing interests, FPGAs are really neat for future-proofing stuff, and bringing stuff back to life. Custom silicon made years ago by Atari, Commodore, etc can be reverse engineered (or at least functional equivalents produced), using Verilog or VHDL, and then can loaded into just about any FPGA. So you can do the clone thing where you emulate/duplicate the original hardware (full on, including, say a 68030 processor) --- but you can also bring old hardware back to life, by interfacing the new FPGA with an old computer when replacement chips are no longer available.
Keith, give the Propeller a Spin. I haven't picked up a BASIC Stamp or SX since I gave it a whirl. Lots of fun, nice to code in smaller pieces that fit in objects or cogs, and amazing power with the high-level drivers for video, mice, etc.
The support in the Propeller forum is 100x what you've seen for the SX. If you've been happy with the SX, you'll love the Prop. Never mind the suggestions to use PIC chips; stick with us for the highest success and satisfaction!
Keith,
Do look at what's been done already with Z80 and 6502 emulation using the Propeller. An FPGA will usually get you more speed, but is harder to work with unless you're starting with ready-to-go Verilog or VHDL. Even then, a Prop can make a great peripheral controller.
The web site quantities only show what we have in stock, not what we have in process (wafers being fabricated and chips being tested).
As for the SX28AC/DP-G, we have 99,890 units (+/- 1000) in process. They've already shipped from Taiwan and should be reflected as in-stock on our web site within a week or a bit longer. After these run out, we're out of stock of this particular chip. According to the past usage values of the SX28AC/DP-G, we should have inventory for 60 months, but that doesn't account for any panic attacks that cause certain customers to buy all of them.
Ken Gracey
Parallax Inc.
Post Edited (Ken Gracey (Parallax)) : 1/3/2010 1:34:27 AM GMT
Thanks for information Ken, especially given the day of the week (Saturday). I will try not to panic, but it's hard not to. I was so happy when I found the SX and it worked well for my application.
yllawwally, yes sir! Many more are trickling in every day. See the picture below and look in the "Future Free" column. We're currently receiving 10-25K units per week, and shipping them out as they arrive. At some point we catch up with the backorders and "Future Free" becomes our lifetime supply. So, to get some just place an order (even if it is a backorder). We'll ship them to you but you may not even see a positive quantity on the web site. This is a very dynamic situation, but the short story is there are plenty of chips (about 250,000 units) for which there is no buyer! And this supply should be fully received within 30 days or so.
Ken Gracey
Parallax Inc.
Post Edited (Ken Gracey (Parallax)) : 1/5/2010 8:57:52 PM GMT
Comments
-Dan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
The Propeller is about twice the cost of an SX48. You'll need an external EEPROM which adds about $2.50 in moderate quantities. Unless your application is very very cost sensitive, that's about $6 total over an SX48. I think you'll find that the additional development costs associated with an AVR or PIC microcontroller over a Propeller may balance that for a lot of applications.
Post Edited (Mike Green) : 10/12/2009 2:26:44 AM GMT
Again, I'd rather see parallax sell SX's, but since that won't happen forever, I need to start looking at other micros.
Very unfortunate situation...·· ...And, I think I·know who to blame...
-Dan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Using a 20MHz resonator the sx can set/clear a pin in 1 instruction (= 1 cycle), the dsPIC
requires 3 instructions to make that happen. Running internally at 80MHz it thus takes 12 cycles
to set/clear a pin ( = less than 7 MIPS).
Running VP's on a dsPIC is also quite demanding and I must yet see to have
8 uarts running (that's why I used the sx).
regards peter
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
The answer is based on Moore's Law, the number of transistors in a given area doubles every 2 years.
Most small embedded SOCs are pad limited these days, which means for a given number of pads the die will not get any smaller.
Add to that the size of the CPU is becoming a smaller fraction of these chips, with peripherals and memory taking up most of the space.
Add to that the cost of a mask set around $1M these days for even a 4 year old process.
What this all implies is it's expensive to develop a new chip let alone a lot of varieties. The not so secret, dirty little secret is many of the variations of these chips are actually the same die, its just feature disabled at the factory.
The difference in area between an 8 bit CPU and a 32 bit CPU gets to be a rounding error. With chips like ARM, in thumb mode even the advantage of code space is eliminated (thumb code is a 16 bit instruction version of ARM). And the big advantage of a 32 bit chip over the 8/16 is the need to page memory goes away, as paging is a huge hassle. Most code is now written in standard languages, mostly C, so portability becomes simpler.
Hence it makes no sense to update an 8/16 part.
FYI a 70MHz ARM with 32KB Flash, 8KB RAM, and some peripherals now goes for < $4, quantity 1
I know this question is very premature........but has any thoughts gone into a possible replacement for the current SX chip in the current Parallax products?
I know there is not a direct replacement, I just want to get some of the ideas which may have passed across the table.
We understand that none of the possible ideas would be the final decision.
James L
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
James L
Partner/Designer
Lil Brother SMT Assembly Services
Are you addicted to technology or Micro-controllers..... then checkout the forums at Savage Circuits. Learn to build your own Gizmos!
Well, that is a good point. I guess I'm prying where it is not really acceptable to pry.
James L
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
James L
Partner/Designer
Lil Brother SMT Assembly Services
Are you addicted to technology or Micro-controllers..... then checkout the forums at Savage Circuits. Learn to build your own Gizmos!
What Mike said.
In some of our products, the SX is simply another component and it can be replaced with something else which is newer, cheaper, and probably just as fast. We've scheduled out redesign of any SX-based product except for the BASIC Stamps (we're holding many years of SX chip inventory for them ~15-20 years for BASIC Stamp manufacturing). Something like the Ping))) could use another processor just as easily, so it will. Anything else you want to know, just ask.
Ken Gracey
Parallax Inc.
Too early to do that right now since there exists ample demand and an eventual short supply. The prices have been the same for a year, however, and we didn't raise them even though we are in this EOL situation. Many companies actually view EOL as an "opportunity" rather than the PITA we find it to be for our customers and our company. Done our best to keep this one reasonable for customers.
Ken Gracey
Parallax Inc.
Near Term: Keep using SX, I've got a bunch around and like SX/B, maybe buy some more protoboards
- hoping the forum for SX doesn't die out anytime soon.
Most Projects: Switch to Propeller, could use the PROPBasic by Bean but will probably keep trying to learn Spin
Really Simple Projects: Considering PICAXE - like 8 & 18 pin DIP formats., oh, I still have a small supply of Basic Stamps too.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tom Talbot
New Market, MD, USA
I just got started with these chips and I think they're incredible.
[noparse][[/noparse] They speak BASIC. [noparse]:)[/noparse] ]
I guess I'll just have to order a good 'hand-full' of these ICs
so I can finish a couple of copies of my projects.
What a drag... [noparse]:)[/noparse]
The small size, fast speed, high I/O count, wide range of operating voltages, and onboard non-volatile RAM really made the SX rock.
Parallax has been good to me over the years so I ordered up a bunch of Propellers rather than bail immediately to PIC.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
I wonder if this wire is hot...
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Shawn Lowe
When all else fails.....procrastinate!
This is really a shame. The SX is a great little device. It was my introduction into microcontrollers, and with the books, good documentation, tech support, and of course everyone here made the learning process fun, easy, and rewarding.
Like others here, I've got some decisions to make about how/if to migrate my current SX hobby project to another uC. I've been playing around with Xilinx FPGA's, and add on 8-bit Picoblaze uC and 32-bitMicroblaze uP's soft cores. I'm seriously thinking about going down that path.
The last time I mentioned this, probably because I'm too naive to understand the whole market, I was shush'd up quickly. I would love a softcore SX! While I'm no expert, given the size of the instruction set, limited built-in hardware peripherals, etc, creating a softcore for it wouldn't be an impossible task. I know, I know, Ubicom won't release the IP and all that. Just dreaming.
I will say that given my retro-computing interests, FPGAs are really neat for future-proofing stuff, and bringing stuff back to life. Custom silicon made years ago by Atari, Commodore, etc can be reverse engineered (or at least functional equivalents produced), using Verilog or VHDL, and then can loaded into just about any FPGA. So you can do the clone thing where you emulate/duplicate the original hardware (full on, including, say a 68030 processor) --- but you can also bring old hardware back to life, by interfacing the new FPGA with an old computer when replacement chips are no longer available.
Sorry to hear about this.
Keith
The support in the Propeller forum is 100x what you've seen for the SX. If you've been happy with the SX, you'll love the Prop. Never mind the suggestions to use PIC chips; stick with us for the highest success and satisfaction!
Ken Gracey
Parallax Inc.
Do look at what's been done already with Z80 and 6502 emulation using the Propeller. An FPGA will usually get you more speed, but is harder to work with unless you're starting with ready-to-go Verilog or VHDL. Even then, a Prop can make a great peripheral controller.
The web site quantities only show what we have in stock, not what we have in process (wafers being fabricated and chips being tested).
As for the SX28AC/DP-G, we have 99,890 units (+/- 1000) in process. They've already shipped from Taiwan and should be reflected as in-stock on our web site within a week or a bit longer. After these run out, we're out of stock of this particular chip. According to the past usage values of the SX28AC/DP-G, we should have inventory for 60 months, but that doesn't account for any panic attacks that cause certain customers to buy all of them.
Ken Gracey
Parallax Inc.
Post Edited (Ken Gracey (Parallax)) : 1/3/2010 1:34:27 AM GMT
Ken Gracey
Parallax Inc.
Post Edited (Ken Gracey (Parallax)) : 1/5/2010 8:57:52 PM GMT