AFAIK, the built-in Prop2 ADCs are single-order bit-streaming only. A counter is required to capture samples. They can't go all that fast unless you just want 4-bit resolution or similar. Should be okay for audio though. Anything else is too bulky.
That's a good point - there is a whole family of "single-order bit-streaming" ADCs available, designed for metering and easy isolation. ( AnalogDev, TI, Avago etc)
These typically have a CLK and Data-out pins, and some models source the clock, and some expect the clock to be provided.
A P2 could share the clock as common on many channels, and then it needs to accept the multiple-Serial-Data.
A benefit of supporting this ADC structure in development & P2 silicon, is it also has commercial applications.
There will be plenty of P2 Control apps, that need external ADCs, for more accurate I / V measurement.
I think the following SAR ADC should actually be fairly similar to what Chip has talked about for the P2: ...
A bit optimistic me thinks, not that I've seen Chip say anything on this matter. That AD7641 requires 19 capacitors per input pin (38 caps differential) with a large scale of cap sizes. I'm pretty confident there will only be one or two capacitors per pin in the Prop2 design. They have to fit in the space between the bonding pads so as not to consume too much of synthesis space.
At low resolution (bit-depth) the Prop2 will be able to pull video speeds with the bit-streaming sigma-delta ADC. I don't know the maths but, as an example guess for 4-bit resolution, first-order capture is simply 16 clocks. Eg: 200 MHz / 16 = 12.5 MHz sampling.
Ken, what are the 0.1" headers on the FPGA board compatible with, if anything?
Another approach would be to make a cable to connect to Linear Tech's "quickeval" series. Its straightforward 0.1" DIL 14 pin with SPI and I2C buses and power rails included. A quick digikey search shows over 130 eval boards, including SAR and Sigma Delta ADCs, DACs, and lots more.
Linear don't make MCUs and I think would benefit from having something like the Prop to talk to. It would be kind of analogous to the way the BeMicro FPGA boards connect to Analog Devices eval boards through an "interposer" pcb.
This module was announced by Analog Devices as a generic board for microcontrollers (Renesas), and it was also announced by Digilent Inc. as a generic module for their FPGAs.
When I asked Digilent (maybe more than 1 year ago) about the release date they said that this was a low priority project. I also asked on Analog Device's forums and they didn't replied back. Digilent still has the device in their website: http://www.digilentinc.com/pmodpack but it was never released.
One of the best parallel ADCs was AD7671 (capable of 12.6 Vp-p Vin and 1 Msps !) but with high cost $50 USD.
Manufacturers are currently moving from Parallel-SAR ADCs into low power Serial-SAR ADCs (like AD7980, LTC 2383, TI ADS8329 ...) All of them around $15-$20, and around 10-20 mW operational power.
My preference was for Analog Device instead of TI or LTC due to its overall characteristics (independent Vref up to 5V, Pseudo-differential input, low INL/DNL, datasheet examples, etc ...) but I think that they are all very similar in performance and price. All are top low-power Serial-SAR ADCs.
There are higher speed ADCs like the 6 MSPS AD7625, or 10 MSPS Serial SAR AD7626. But I am not sure if the P2 will handle LVDS and also they cost around $50 USD.
10 MSPS is currently the top speed for (non-pipelined) SAR ADCs. If higher speed is needed then you need to use Pipelined ADCs.
Pipelined ADCs like AD9266 can handle 20, 40, 65 and 80 MSPS (from $28 to $40 USD).
Higher speed than that I have currently no interest due to high price and low analog input range (they are all around 400 miliVolts p-p, and I am currently looking for ADCs capable of 5V or 10V input range) but I understand that may be other people working with Software Defined Radios or USB oscilloscopes that need 250 MSPS - 2 GSPS (and maybe lower resolution 10-12-14 bits).
While there are other choices for lower resolution ADCs. My preference if for (at least) 16 bits ADCs, 1 MSPS or better, and low cost. I think that AD7980 is currently the best choice.
(I do not work for AD. If anybody knows similar devices, please reply with substitutes.)
So what about the maximum code size limit? Is it still a puny 32k for spin & 64k for c++/asm? With the need for a driver for everything, the code space limit REALLY sucks. By the time you get all the drivers in, you are out of code space. I am constantly having to re-write multiple cog drivers into a single cog or pack 2 drivers into one. It is nearly impossible to use with multiple sensors, even if they are all on the same bus. Even the arduino can handle larger programs! It would be a sad day when to have to downgrade to a slower, less powerful chip, just for a bigger code space. Either that or onto a real cpu that costs hundreds of dollars.
Hi zlantz,
The proposed ram size for P2 is 512K. FPGA testing of the last P2 image had 256K with hub execution mode available as well.
This well and truly solves the 32K code barrier of the P1. The new P2 also gains an extra 8 cogs too! :)
This was done back in late 2006 after only spending a month or so with the Propeller it did use an external uController to grab the A/D channels and also some slow moving I/O. But the Propeller did all the rest including the graphics. It also talked to the radar detector and mirrored all the LEDs from it. It also was able to monitor Speed, RPM, Calculate gear, as well as catching every Fuel Injector pulse and measure the fuel used. All in an $8 uController. And I think I had 2-3 cogs left over. :-)
Another thing to note...This was done in a weekend from start to finish. Maybe 30 hours of work. I think that sums up what the Propeller really is.
To say I'm excited to what will be able to be done on the P2 is an understatement.
So what about the maximum code size limit? Is it still a puny 32k for spin & 64k for c++/asm? With the need for a driver for everything, the code space limit REALLY sucks. By the time you get all the drivers in, you are out of code space. I am constantly having to re-write multiple cog drivers into a single cog or pack 2 drivers into one. It is nearly impossible to use with multiple sensors, even if they are all on the same bus. Even the schitty arduino can handle larger programs! It would be a sad day when to have to downgrade to a slower, less powerful chip, just for a bigger code space. Either that or onto a real cpu that costs hundreds of dollars.
And, to make more flexible use of that 512KB, Chip hopes to have an extended memory model known as HubExec. And there is hints of another possible extension for external RAM also.
PS: The Prop2 won't be able to run the exact same binaries of the Prop1, ie: Prop1 sources will need recompiled for the Prop2. And will also need at least small modifications to work on the Prop2.
How about a DIP board (Basic Stamp-style) that is pin-compatible with the P1 and only exposes port A? But will the P2 die be small enough for this?
It will likely come, but will also need some creativity around the physical pins.
I have one DIP module here, that looks to use blind vias on 4 layer board to manage the pins.
That allows full width of top-side placement/routing, and gives some location for soldering pins.
Also of appeal are the modules that can also be flush mounted.(ie do not have to be DIP )
To cover that .would need a PCB with blind vias, and edge-plating, with free pins included.
It does bump to 4 layers, but on a small PCB, the price impact should be quite low.
We have used 0.1" std square pins, and pushed into a DIP socket - it is a tight fit in the socket, but gives a very cheap 'sacrificial pin' DIP footprint, that then plugs into std DIP ICs.
Others have machined pins, but they give a failure point that is hard to fix when damaged, and machine pins are quite expensive.
This was done back in late 2006 after only spending a month or so with the Propeller it did use an external uController to grab the A/D channels and also some slow moving I/O. But the Propeller did all the rest including the graphics. It also talked to the radar detector and mirrored all the LEDs from it. It also was able to monitor Speed, RPM, Calculate gear, as well as catching every Fuel Injector pulse and measure the fuel used. All in an $8 uController. And I think I had 2-3 cogs left over. :-)
Another thing to note...This was done in a weekend from start to finish. Maybe 30 hours of work. I think that sums up what the Propeller really is.
To say I'm excited to what will be able to be done on the P2 is an understatement.
I always thought that was really cool what you did Jim, actually inspiring because it's not just a static demo, and all in a weekends work! +1
Does this mean there won't be a DIP version of P2? What packaging is planned?
I think they're planning on packaging the new FPGA image in a red box with a green bow since it will probably be delivered just in time for Christmas! :-)
Well that's nice, but the FPGA image I don't really care too much about... I do care about being able to use the P2 when it releases though, and if the only form I can get it in is surface mount or something I can't actually plug into a socket then I can't use it...
Well that's nice, but the FPGA image I don't really care too much about... I do care about being able to use the P2 when it releases though, and if the only form I can get it in is surface mount or something I can't actually plug into a socket then I can't use it...
If you look at the first post in this thread you'll see a picture that Chip posted of the pinout. It certainly doesn't look like a DIP package to me. However, I believe Parallax has said that they will have modules that can be used in through-hole designs.
I suspect that's not going to work for me - unless Parallax (or someone else) produces a DIP board I suppose, but that will likely add too much to the cost. Guess I'll start looking for an alternative...
I suspect that's not going to work for me - unless Parallax (or someone else) produces a DIP board I suppose, but that will likely add too much to the cost. Guess I'll start looking for an alternative...
Modules are getting cheaper all the time, so the cost you imagine may be wrong..
A good example is the UMFT234XF, which FTDi can assemble and sell retail for 1+ @ $5.95 (50+ @ $5.46)
I suspect that's not going to work for me - unless Parallax (or someone else) produces a DIP board I suppose, but that will likely add too much to the cost. Guess I'll start looking for an alternative...
good luck - please tell us if you find one
Today you are lucky if new chips still have pins and not only balls, especially in the class of the Prop 2.
The worst are these tiny chip scale packages with balls in 0.4mm pitch. No clue how this can be handled.
The problem is that I use Props to teach year 7-10 kids. I have a class set of Boe-Bots and a class set of SumoBots which are great, but for the advanced kids I use the Prop. The focus is not on electronics, so we don't want to be soldering surface mount stuff (I can't anyway, and I doubt any of the kids could - and besides we don't have the tools). With the current Prop I can have them prototype stuff on breadboards, then build some simple boards of their own. They're kids and they tend to kill chips, so we put the Props in sockets on any boards they build so that when they accidentally put 9v or 12v through the Prop they can just swap it out (I do try to discourage that, but they are kids...). If the new Prop can't be used the way I'm using the current Prop then eventually I'll have to find a different chip to use I guess. I don't know what - the Prop is a good fit. If someone sells the Prop2 on a board I can put in a socket then great, but I don't know whether the cost of that will be prohibitive - every time one of the kids kills a Prop we'd have to toss the board out and put a new one in.
The old prop will still be around. Does the robotics class need the prop 2 to achieve its goals? If the kids outgrow the prop, then they should be advanced enough to either solder smt or use the prop 2 in whatever breakout/demo footprint it comes in, I would think.
The problem is that I use Props to teach year 7-10 kids. I have a class set of Boe-Bots and a class set of SumoBots which are great, but for the advanced kids I use the Prop. The focus is not on electronics, so we don't want to be soldering surface mount stuff (I can't anyway, and I doubt any of the kids could - and besides we don't have the tools). With the current Prop I can have them prototype stuff on breadboards, then build some simple boards of their own. They're kids and they tend to kill chips, so we put the Props in sockets on any boards they build so that when they accidentally put 9v or 12v through the Prop they can just swap it out (I do try to discourage that, but they are kids...). If the new Prop can't be used the way I'm using the current Prop then eventually I'll have to find a different chip to use I guess. I don't know what - the Prop is a good fit. If someone sells the Prop2 on a board I can put in a socket then great, but I don't know whether the cost of that will be prohibitive - every time one of the kids kills a Prop we'd have to toss the board out and put a new one in.
The new Propeller would be able to be used the same way - don't worry about this detail right now. Though it has more pins we could put it into a proto-friendly package. As a matter of fact it'd be a huge mistake if we somehow, in some way, didn't do this. You are a target customer so we'll provide it the way teachers want to buy it.
so we put the Props in sockets on any boards they build so that when they accidentally put 9v or 12v through the Prop they can just swap it out (I do try to discourage that, but they are kids...).
So why not create a breakout board especially for kids? There are simple profound means to protect chips: a diode in parallel to Vxx will help from exchanging + and -. And overvoltage at pin inputs can be prevented with diodes to ground and vcc ( will also replace the first mentioned diode). Those components are available to protect inputs from overvoltage, e.g. in telecommunication and they are cheap.
So why not create a breakout board especially for kids? There are simple profound means to protect chips: a diode in parallel to Vxx will help from exchanging + and -. And overvoltage at pin inputs can be prevented with diodes to ground and vcc ( will also replace the first mentioned diode). Those components are available to protect inputs from overvoltage, e.g. in telecommunication and they are cheap.
Indeed. But that's not what the focus of our class is, nor do we have the means. Happily Parallax is coming to the rescue.
Unfortunately availability of DIP is one of the things that keeps me from going to XMOS instead of Propeller (well that and I don't like the weird custom C dialect they have, I'd prefer just to use C++ on it). For me as a hobbyist, being able to easily breadboard is a big thing. And then being able to solder it myself when I decide to make a PCB. Not sure having a breakout board is really the same, I'd feel weird about soldering a breakout board into a PCB.
Well, we'll see. I'm still stoked to see what comes out with the P2. I'll definitely buy a board or two to play with. I hope Parallax does well with it.
The old prop will still be around. Does the robotics class need the prop 2 to achieve its goals? If the kids outgrow the prop, then they should be advanced enough to either solder smt or use the prop 2 in whatever breakout/demo footprint it comes in, I would think.
No, we don't really need Prop2 to achieve our goals, but it does open up more possibilities. And on the point of soldering SMT parts - there is virtually no chance the kids could do that even if we did have the right tools.
Year 7 kids are 11 years old. If I get one student per year who has actually heard of a resistor, much less seen one, I'm lucky. These kids aren't electronics enthusiasts wanting play with robots - they're just kids who think robots might be fun to play with. I have them for 1 hour per week for the year - maybe 40 weeks. The focus of the class is to introduce them to the possibilities of microcontrollers, servos, sensors etc, through robotics. Learning about electronics and how to solder SMT will come later if that's the way they choose to go (though I confess I have trouble seeing those things, let alone actually soldering them). My focus is on showing them what they can do and how, not teaching them electronics (besides, my background is software and I dabble with the electronics - I know enough to make me dangerous). I do try to teach them some fundamentals, but in the end if they come out being able to put a few simple sensors and servos together with the Prop to make something work (of their own design - so they have to understand the concepts), but not really understand the fundamentals of the components, that's ok. They learn by doing, and they pick up a basic understanding along the way.
I'm a volunteer at the school (my kids go there). It took me a long time to get the Boe-Bots and SumoBots into the curriculum - for the first few years we were using 6 of each that I cobbled together from various places and donated to the school, but the school did buy class sets of each at the start of this school year. They did have some old Lego things that the teachers were used to and didn't want to let go of. I ran some classes for the teachers to show them the Parallax stuff, so now they can teach the "mainstream" robotics classes, and I just take the advanced kids in a separate stream (eventually, hopefully, I'll get the Props into the mainstream curriculum). The Props and associated bits (breadboards, wires, LEDs, servos, Pings, Sharp IR sensors, RTCs etc., even GPS chips for some of the kids) are all mine at the moment, but eventually the school will buy their own. I'd hate for the school to do that then find out that the roadmap didn't include chips the kids could actually use... Being able to prototype with breadboards is essential in the school environment, and soldering SMT devices just isn't feasible (in my experience anyway). Thanks again to Ken and Parallax for recognizing that.
Comments
That's a good point - there is a whole family of "single-order bit-streaming" ADCs available, designed for metering and easy isolation. ( AnalogDev, TI, Avago etc)
These typically have a CLK and Data-out pins, and some models source the clock, and some expect the clock to be provided.
A P2 could share the clock as common on many channels, and then it needs to accept the multiple-Serial-Data.
A benefit of supporting this ADC structure in development & P2 silicon, is it also has commercial applications.
There will be plenty of P2 Control apps, that need external ADCs, for more accurate I / V measurement.
A bit optimistic me thinks, not that I've seen Chip say anything on this matter. That AD7641 requires 19 capacitors per input pin (38 caps differential) with a large scale of cap sizes. I'm pretty confident there will only be one or two capacitors per pin in the Prop2 design. They have to fit in the space between the bonding pads so as not to consume too much of synthesis space.
At low resolution (bit-depth) the Prop2 will be able to pull video speeds with the bit-streaming sigma-delta ADC. I don't know the maths but, as an example guess for 4-bit resolution, first-order capture is simply 16 clocks. Eg: 200 MHz / 16 = 12.5 MHz sampling.
Another approach would be to make a cable to connect to Linear Tech's "quickeval" series. Its straightforward 0.1" DIL 14 pin with SPI and I2C buses and power rails included. A quick digikey search shows over 130 eval boards, including SAR and Sigma Delta ADCs, DACs, and lots more.
Linear don't make MCUs and I think would benefit from having something like the Prop to talk to. It would be kind of analogous to the way the BeMicro FPGA boards connect to Analog Devices eval boards through an "interposer" pcb.
Yes, AD7980.
Thank you Ken for asking feedback.
For general purpose, I have preference for SAR ADC instead of Delta-Sigma.
My vote is for the never released AD7980 module (based on the AD7980 16-bit, 1 MSPS SAR):
http://wiki.analog.com/resources/tools-software/uc-drivers/microchip/ad7980
This module was announced by Analog Devices as a generic board for microcontrollers (Renesas), and it was also announced by Digilent Inc. as a generic module for their FPGAs.
When I asked Digilent (maybe more than 1 year ago) about the release date they said that this was a low priority project. I also asked on Analog Device's forums and they didn't replied back. Digilent still has the device in their website: http://www.digilentinc.com/pmodpack but it was never released.
One of the best parallel ADCs was AD7671 (capable of 12.6 Vp-p Vin and 1 Msps !) but with high cost $50 USD.
Manufacturers are currently moving from Parallel-SAR ADCs into low power Serial-SAR ADCs (like AD7980, LTC 2383, TI ADS8329 ...) All of them around $15-$20, and around 10-20 mW operational power.
My preference was for Analog Device instead of TI or LTC due to its overall characteristics (independent Vref up to 5V, Pseudo-differential input, low INL/DNL, datasheet examples, etc ...) but I think that they are all very similar in performance and price. All are top low-power Serial-SAR ADCs.
There are higher speed ADCs like the 6 MSPS AD7625, or 10 MSPS Serial SAR AD7626. But I am not sure if the P2 will handle LVDS and also they cost around $50 USD.
10 MSPS is currently the top speed for (non-pipelined) SAR ADCs. If higher speed is needed then you need to use Pipelined ADCs.
Pipelined ADCs like AD9266 can handle 20, 40, 65 and 80 MSPS (from $28 to $40 USD).
Higher speed than that I have currently no interest due to high price and low analog input range (they are all around 400 miliVolts p-p, and I am currently looking for ADCs capable of 5V or 10V input range) but I understand that may be other people working with Software Defined Radios or USB oscilloscopes that need 250 MSPS - 2 GSPS (and maybe lower resolution 10-12-14 bits).
While there are other choices for lower resolution ADCs. My preference if for (at least) 16 bits ADCs, 1 MSPS or better, and low cost. I think that AD7980 is currently the best choice.
(I do not work for AD. If anybody knows similar devices, please reply with substitutes.)
The proposed ram size for P2 is 512K. FPGA testing of the last P2 image had 256K with hub execution mode available as well.
This well and truly solves the 32K code barrier of the P1. The new P2 also gains an extra 8 cogs too!
This was done back in late 2006 after only spending a month or so with the Propeller it did use an external uController to grab the A/D channels and also some slow moving I/O. But the Propeller did all the rest including the graphics. It also talked to the radar detector and mirrored all the LEDs from it. It also was able to monitor Speed, RPM, Calculate gear, as well as catching every Fuel Injector pulse and measure the fuel used. All in an $8 uController. And I think I had 2-3 cogs left over. :-)
Another thing to note...This was done in a weekend from start to finish. Maybe 30 hours of work. I think that sums up what the Propeller really is.
To say I'm excited to what will be able to be done on the P2 is an understatement.
You could have just read the topic title.
And, to make more flexible use of that 512KB, Chip hopes to have an extended memory model known as HubExec. And there is hints of another possible extension for external RAM also.
PS: The Prop2 won't be able to run the exact same binaries of the Prop1, ie: Prop1 sources will need recompiled for the Prop2. And will also need at least small modifications to work on the Prop2.
How about a DIP board (Basic Stamp-style) that is pin-compatible with the P1 and only exposes port A? But will the P2 die be small enough for this?
It will likely come, but will also need some creativity around the physical pins.
I have one DIP module here, that looks to use blind vias on 4 layer board to manage the pins.
That allows full width of top-side placement/routing, and gives some location for soldering pins.
Also of appeal are the modules that can also be flush mounted.(ie do not have to be DIP )
To cover that .would need a PCB with blind vias, and edge-plating, with free pins included.
It does bump to 4 layers, but on a small PCB, the price impact should be quite low.
We have used 0.1" std square pins, and pushed into a DIP socket - it is a tight fit in the socket, but gives a very cheap 'sacrificial pin' DIP footprint, that then plugs into std DIP ICs.
Others have machined pins, but they give a failure point that is hard to fix when damaged, and machine pins are quite expensive.
I always thought that was really cool what you did Jim, actually inspiring because it's not just a static demo, and all in a weekends work! +1
Thanks for the kind words Peter. :-)
Does this mean there won't be a DIP version of P2? What packaging is planned?
Modules are getting cheaper all the time, so the cost you imagine may be wrong..
A good example is the UMFT234XF, which FTDi can assemble and sell retail for 1+ @ $5.95 (50+ @ $5.46)
good luck - please tell us if you find one
Today you are lucky if new chips still have pins and not only balls, especially in the class of the Prop 2.
The worst are these tiny chip scale packages with balls in 0.4mm pitch. No clue how this can be handled.
Andy
The new Propeller would be able to be used the same way - don't worry about this detail right now. Though it has more pins we could put it into a proto-friendly package. As a matter of fact it'd be a huge mistake if we somehow, in some way, didn't do this. You are a target customer so we'll provide it the way teachers want to buy it.
Ken Gracey
So why not create a breakout board especially for kids? There are simple profound means to protect chips: a diode in parallel to Vxx will help from exchanging + and -. And overvoltage at pin inputs can be prevented with diodes to ground and vcc ( will also replace the first mentioned diode). Those components are available to protect inputs from overvoltage, e.g. in telecommunication and they are cheap.
Indeed. But that's not what the focus of our class is, nor do we have the means. Happily Parallax is coming to the rescue.
Well, we'll see. I'm still stoked to see what comes out with the P2. I'll definitely buy a board or two to play with. I hope Parallax does well with it.
No, we don't really need Prop2 to achieve our goals, but it does open up more possibilities. And on the point of soldering SMT parts - there is virtually no chance the kids could do that even if we did have the right tools.
Year 7 kids are 11 years old. If I get one student per year who has actually heard of a resistor, much less seen one, I'm lucky. These kids aren't electronics enthusiasts wanting play with robots - they're just kids who think robots might be fun to play with. I have them for 1 hour per week for the year - maybe 40 weeks. The focus of the class is to introduce them to the possibilities of microcontrollers, servos, sensors etc, through robotics. Learning about electronics and how to solder SMT will come later if that's the way they choose to go (though I confess I have trouble seeing those things, let alone actually soldering them). My focus is on showing them what they can do and how, not teaching them electronics (besides, my background is software and I dabble with the electronics - I know enough to make me dangerous). I do try to teach them some fundamentals, but in the end if they come out being able to put a few simple sensors and servos together with the Prop to make something work (of their own design - so they have to understand the concepts), but not really understand the fundamentals of the components, that's ok. They learn by doing, and they pick up a basic understanding along the way.
I'm a volunteer at the school (my kids go there). It took me a long time to get the Boe-Bots and SumoBots into the curriculum - for the first few years we were using 6 of each that I cobbled together from various places and donated to the school, but the school did buy class sets of each at the start of this school year. They did have some old Lego things that the teachers were used to and didn't want to let go of. I ran some classes for the teachers to show them the Parallax stuff, so now they can teach the "mainstream" robotics classes, and I just take the advanced kids in a separate stream (eventually, hopefully, I'll get the Props into the mainstream curriculum). The Props and associated bits (breadboards, wires, LEDs, servos, Pings, Sharp IR sensors, RTCs etc., even GPS chips for some of the kids) are all mine at the moment, but eventually the school will buy their own. I'd hate for the school to do that then find out that the roadmap didn't include chips the kids could actually use... Being able to prototype with breadboards is essential in the school environment, and soldering SMT devices just isn't feasible (in my experience anyway). Thanks again to Ken and Parallax for recognizing that.