Parallax Forums
  HomeLog InRegisterCommunity CalendarSearch the ForumHelp
   
Parallax Forums > Public Forums > Propeller Chip > SixBladeProp - A new 6 propeller pcb  Forum Quick Jump
 
New Topic Post Reply Printable Version
99 posts in this thread.
Viewing Page :
 1  2  3  4 
[ << Previous Thread | Next Thread >> ] | Show Newest Post First ]

Cluso99
We live onboard



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Apr 2008
Total Posts : 4114
 
   Posted 1/27/2009 4:57 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
Here is the block diagram of my new SixBladeProp (TM) board. I hope to have it available in the next few weeks.cool
It can have up to six (6) Propeller circuits which I am calling "Blades".
 
Blade #1:
A typical Prop Proto Board circuit - PropPlug connector, VGA or TV (shared pins) connectors, PS2 Keyboard connector, PS2 Mouse connector, I2C EEPROM (24C256 or 24C512 or 24C1024). This prop also boots all other Blades (propellers).
 
Blade #2 and Blade #5:
These are identical. They each can have 1 or 2 128Kx8 or 512Kx8 Static Rams connected. The second static ram may be replaced with an SD or microSD connector or SPI Flash of 8Mbits-64Mbits.
Blade #2 serial may be directly connected with Blade #3 (in parallel with Blade #1).
Blade #5 serial may be directly connected with Blade #4 (in parallel with Blade #1).
These Blades are designed to run large programs, such as microcomputer emulations (see the PropAltair thread), large LMM programs, large C programs, etc.
 
Blade #3 and Blade #6:
These are identical and are designed for I/O peripheral use. Each has a 40 pin (30 bit) I/O connector.
 
Blade #4:
This has VGA or TV (shared pins) connectors. Also, a 128Kx8 or 512Kx8 Static Ram may be connected. A latch is used to hold the upper 6 or 8 address lines. The lower 11 address lines are connected directly to the propeller. This ram is designed to aid program or video memory, or both.
 
General:
A comon clock (oscillator) is used so that all the propellers (cogs) may remain "in syc" for high speed serial transfers over 2 wire interfaces.
Almost all components will be thru' hole, making it easy for hobbyists to assemble. An SMT version may be made later if sufficient demand exists.
 
Postedit: Answers to questions below:
  • All 6 props are on the one pcb - load only as many as you need. I anticipate the size to be 5x6". At one stage I thought about stacking a small generalised pcb.
  • The 20MHz is easily obtained, so a PLLx4 would be used. I am looking for a cheap 5MHz because it makes existing software easier - and less emi.
  • There is no ability to add an SD/microSD to Blade #1. There is however 1 LED (not shown).
  • Blade #1 boots all other Blades (#2...#6) after it boots (from its' eeprom or download). Remember, it can have a 128KB eeprom (24C1024). I am writing pasm code to access the eeprom above 32KB (which has been done by others anyway). There is no way to program the Propeller chips ROM.
  • Since there seems to be a lot of interest I will sell the bare pcb, plus some parts if required.
  • I am hoping to be able to place the SD socket over the microSD socket - then you fit either one.
Any comments are welcome cool
 
One question:
Is the pair of 2x20 Pins for the I/O the best connector or do you want a single 2x32 Pins (60 I/Os) or something else like Spin Studio sockets or Diligents connectors or a mix or something again ???
 
Postedit 4 Feb 2009:
The actual specifications and size have changed since this original proposal. See the full descriptions, schematics and pictures of the pcb which are contained in the following posts. cool   

 


Links to other interesting threads:
 
My cruising website is:  www.bluemagic.biz

Post Edited (Cluso99) : 2/4/2009 10:19:40 AM GMT


Image Attachment :
Image Preview
ClusoPropSchematic001.jpg
  101KB (image/pjpeg)
This image has been viewed 743 time(s).
Back to Top
 

heater
Registered Member



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Feb 2008
Total Posts : 3345
 
   Posted 1/27/2009 5:47 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
Bejusus!


For me, the past is not over yet.

Back to Top
 

Ale
Registered Member



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined May 2007
Total Posts : 1534
 
   Posted 1/27/2009 5:58 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
And which is the PCB idea ? to have just one module you can use for all or two three-propeller modules ?
Back to Top
 

Dr_Acula
Registered Member



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Dec 2008
Total Posts : 1640
 
   Posted 1/27/2009 6:03 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
Looking very interesting. Is this going to be a board? If so, I would presume you could leave parts out if you wanted to? Lots of memory. Lots of flexibility. Ability to build a one prop solution then expand as you go. I think this has a lot of potential. Please put me down for a few boards.

Heater, you got 8080 in one cog. Could you fit Z80 if you had 48 cogs?!
Back to Top
 

TreeLab
Registered Member

Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Nov 2006
Total Posts : 136
 
   Posted 1/27/2009 6:28 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
This is impressive. Do you have an estimate of the PCB footprint size?

In the interests of trying to have Blade 1 'look' like a familiar prop system, is there any way to add an SD interface to this one as well?

What is the advantage of running the clock at 20 MHz (and pll4x I presume) ?

I have no idea how to harness that much of a controller, but it would be an amazing device to work with. With some good ADC/DAC interfaces hanging off Blades 3,6 this would make an extremely powerful laboratory data acquisition and control system.

If they are available, I would take two such boards.

Cheers!
Paul Rowntree
Back to Top
 

Rsadeika
Registered Member

Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Feb 2005
Total Posts : 801
 
   Posted 1/27/2009 7:19 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
This sure looks impressive, but I am not quite sure as to how this is going to work. Since you do not have any EEPROMs connected to the other blade's, how do they retain a program? Now, if you are working on something to directly program the prop chip ROM, then that would make the difference. Am I missing something here?

Ray
Back to Top
 

Cluso99
We live onboard



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Apr 2008
Total Posts : 4114
 
   Posted 1/27/2009 7:37 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
I have answered questions above in the original post (keeps the specs together) :-)


Links to other interesting threads:
Back to Top
 

william chan
Registered Member



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Jul 2004
Total Posts : 914
 
   Posted 1/27/2009 9:24 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
What application can this board be used for?

Number crunching?


www.fd.com.my
www.mercedes.com.my

Back to Top
 

Cole Logan
Registered Member



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Nov 2008
Total Posts : 139
 
   Posted 1/27/2009 9:50 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
This looks very interesting I've always wondered about using multiple props and this would make a great test bed
 
 
EDIT
In the future How expandable would this be. do you think it would be possible to have 10 or even 100 connected together.
Back to Top
 

Mike Huselton
QuantumMax



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Nov 2007
Total Posts : 716
 
   Posted 1/27/2009 9:53 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
Jeebus!

Please carry on with your design. It answers all of my questions. The supercomputer thread Chip was involved in is implemented nicely here.
Also, the board can be populated with one Prop chip or six Prop chips, which is extremely flexible.
I would vote for a 5mhz master clock for EMI considerations and stability.
This is a very conservative design - which makes it a successful design.

Please go for it, and don't get too immersed in forum wish lists. Changes can be made later, as this is a truly Open Source design.


JMH - Electronics: Engineer - Programming: Professional

Back to Top
 

TreeLab
Registered Member

Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Nov 2006
Total Posts : 136
 
   Posted 1/27/2009 10:00 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
Since these blades have no eeprom, it is very important that they be robust against brownout-induced resets. One blade down could cause the central communications cog to hang. Would putting extra large bypass caps on the blade's power supplies be useful?

On a related idea : can a prop burn its own program out to the eeprom, for execution after reset?

I would think that it would be easier to make an adapter from the wider 2x20 format to the SpinStudio interfaces than the other way around. 8 bits per interface is a bit small (although Brian is considering multi-connector daughter boards)

Post Edited (TreeLab) : 1/27/2009 5:19:54 PM GMT

Back to Top
 

Mike Huselton
QuantumMax



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Nov 2007
Total Posts : 716
 
   Posted 1/27/2009 10:08 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
Treelab,

The rule of thumb is one 0.1 mf bypass capacitor for every logic group. Logic group is defined conservatively as one IC. A 10 mf tantulum cap is sufficient for the board.


JMH - Electronics: Engineer - Programming: Professional

Back to Top
 

Cluso99
We live onboard



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Apr 2008
Total Posts : 4114
 
   Posted 1/27/2009 10:13 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
I am very aware of decoupling and brownouts. However, I cannot control the external power.

There will be an object to program the other cogs data into high eeprom (above 32KB), and also to read it to program the other props after reset.

There is no reason why multiple boards could not be used to make a super super prop design.


Links to other interesting threads:
Back to Top
 

heater
Registered Member



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Feb 2008
Total Posts : 3345
 
   Posted 1/27/2009 10:22 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
Cluso this is crazy. However I love it. Please put my name down for a couple of boards.

Can we please insure that the RAM is accessible with the minimum number of PASM instructions? No superfluous shifting of bits into the right port positions.

Dr_A, I should think a number of z80's could fit in there. I don't need to get MP/M running just run many copies of CP/M. Multiuser Prop system! 1976 would be so jealous.


For me, the past is not over yet.

Back to Top
 

hinv
Why are some things obvious, yet unthinkable?



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Jan 2007
Total Posts : 622
 
   Posted 1/27/2009 11:40 AM (GMT -7)    Quote This PostAlert An Admin About This Post.
Cluso,

You cannot have a 20MHz clock and still use the PLL. 160MHz is tops of the spec for it.
The PLL always runs at 16x and then divides down to get 8x 4x 2x.

I think it is a great idea. Do you plan on using Beau's high speed serial protocol to connect them together?
http://forums.parallax.com/forums/default.aspx?f=25&m=233212&p=1&ord=a
Back to Top
 

Mike Huselton
QuantumMax



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Nov 2007
Total Posts : 716
 
   Posted 1/27/2009 12:56 PM (GMT -7)    Quote This PostAlert An Admin About This Post.
Cluso,

I would appreciate it if you could supply a finished block diagram and board schematic. The actual printed circuit I can handle separately.
I realize that this is a lot of work and you have limited time.

Anything we can do to help? I think that this is an open source joint forum effort.
Maybe we could just leave you alone to concentrate on finishing the block diagram and and schematic.

Thanks again.


JMH - Electronics: Engineer - Programming: Professional

Post Edited (Quantum) : 1/28/2009 4:19:03 AM GMT

Back to Top
 

Kye
Kwabena Agyeman



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Feb 2008
Total Posts : 1029
 
   Posted 1/27/2009 1:18 PM (GMT -7)    Quote This PostAlert An Admin About This Post.
Wow,

That's a mega design...

Why not wait until the propeller two? Seems like your trying to get more power out of it that it can actually do.

Add complexity makes it all so much hard to maintain and easier to fail...

But, wow. Goodluck.


Nyamekye,

Back to Top
 

Bob Lawrence (VE1RLL)
Registered Member



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Aug 2007
Total Posts : 506
 
   Posted 1/27/2009 2:33 PM (GMT -7)    Quote This PostAlert An Admin About This Post.
cool Very nice. can't wait to see what you guys program it to do. yeah


Aka: CosmicBob

Back to Top
 

mctrivia
Registered Member



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Jan 2009
Total Posts : 3482
 
   Posted 1/27/2009 6:03 PM (GMT -7)    Quote This PostAlert An Admin About This Post.
i was thinking of making a stackable design like this when the prop 2 came out. Though I was planning on putting 8 surface mounted on each board and letting you stack up to 256 boards. can you imagine over 300GIPS that beats intels best.
Back to Top
 

Paul Rowntree
Registered Member

Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Feb 2007
Total Posts : 49
 
   Posted 1/27/2009 8:49 PM (GMT -7)    Quote This PostAlert An Admin About This Post.
Cluso, can you give an idea of how many cogs on Blade 1 are used for links to Blades 2-6, and how many remain free for a fully loaded board?
Back to Top
 

mctrivia
Registered Member



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Jan 2009
Total Posts : 3482
 
   Posted 1/27/2009 8:53 PM (GMT -7)    Quote This PostAlert An Admin About This Post.
that depends on your code and how fast they need to comunicate. the fastest methods are 1 cog per chip per direction.
Back to Top
 

Mike Huselton
QuantumMax



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Nov 2007
Total Posts : 716
 
   Posted 1/27/2009 9:13 PM (GMT -7)    Quote This PostAlert An Admin About This Post.
Treelab,

The rule of thumb is one 0.1 mf bypass capacitor for every logic group. Logic group is defined conservatively as one IC.

A 10 mf tantulum cap is sufficient for the entire board.


JMH - Electronics: Engineer - Programming: Professional

Back to Top
 

Peter Jakacki
thinker tinker



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Sep 2006
Total Posts : 1389
 
   Posted 1/27/2009 9:47 PM (GMT -7)    Quote This PostAlert An Admin About This Post.
Ray, you sure you've got both oars in the water mate? eyes (sorry, couldn't help but use a nautical term)

Six Prop chips might be fun but it makes the software a little bit awkward, don't you think? Don't get me wrong, I like the idea of toying with this many Props all strung together but I reckon it will be a lotta lotta hard work and in the end it's still kinda kludgy.

When it comes down to real-time control and soft peripherals the current Prop shines but with it's memory limitations it makes it a little difficult as you know to run larger applications. This is the reason why I have used an ARM chip teamed up with the Propeller in my earlier iCONSOLE design. The ARM chip is very cost effective (cheaper than a Prop) for running a large application and there are variants that allow you to address large amounts of memory. Even the LPC2148 that I use has 512KB of Flash but I also have the LPC2478 which features an External Memory Controller with full support for SDRAM. This is not to toot the horn for ARM chips but simply to mention that I reckon this would be far more cost effective and practical rather than trying to use Prop chips imPROPerly. We have seen how PIC and AVR chips are stretched beyond their capabilities as simple 8-bit microcontrollers and made to handle ethernet for instance when there are better solutions at a similar price.

I have a design in which I used 4 Props with one co-coordinating the other three but that was a specialized application.

Now, when son of Prop comes out it will a totally different story, I can't wait but in the meantime I am having a lot of fun with the current Prop which I suspect will still be the most popular for general jobs and projects.

*Peter*
(part-time stirrer)
Back to Top
 

hinv
Why are some things obvious, yet unthinkable?



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Jan 2007
Total Posts : 622
 
   Posted 1/27/2009 10:10 PM (GMT -7)    Quote This PostAlert An Admin About This Post.
mctrivia said: "that depends on your code and how fast they need to comunicate. the fastest methods are 1 cog per chip per direction."

Has anybody beat Beau's 4wire serial connection at 14.5MegaBaud? I believe it uses only one cog for transmit and 1 for receive also.
Back to Top
 

mctrivia
Registered Member



Email Address Not AvailablePersonal Homepage Not AvailablePrivate Messaging Not AvailableAIM Not AvailableICQ Not AvailableY! Not AvailableMSN Not Available
Date Joined Jan 2009
Total Posts : 3482
 
   Posted 1/27/2009 10:15 PM (GMT -7)    Quote This PostAlert An Admin About This Post.
personally i am of the opinion of replicating the cog hub on the large scale. the reason I chose 8 for my design was so each of the 8 chips could share a block of 2MB of ram each having access to it in turn. no serial direct between them just sharing memory space. Have not figured out exact specifics because realistically it would cost a fortune and be only useful for cracking encryption algorithms, computing digits of pi, or other similar mathematical indevers.
Back to Top
 
[ << Previous Thread | Next Thread >> ]
New Topic Post Reply Printable Version
99 posts in this thread.
Viewing Page :
 1  2  3  4 
 
Forum Information
Currently it is Thursday, July 29, 2010 5:23 PM (GMT -7)
There are a total of 462,441 posts in 62,066 threads.
In the last 3 days there were 90 new threads and 802 reply posts. View Active Threads
Who's Online
This forum has 20143 registered members. Please welcome our newest member, ME01.
59 Guest(s), 17 Registered Member(s) are currently online.  Details
John Abshier, Erik Friesen, RossH, Kevin Wood, simpsonmichael1, David Betz, Julian800, Martin Hodge, RDL2004, Harley, Sapieha, Timmoore, Gene Bonin, wiresalot, Ravenkallen, T Chap, Tubular