Multi Propeller board.

BTX

Hi all !!

I'm involved in a design, where I need to put 20 Pchip, that·share·8 pins· of each. ( 8 bit bus ).
I think to use at least two 74LVTH244 to divide in '10 and 10' Pchips sharing the same pins, and then, another propeller must drive the two 244's.
Do you think if it could be·possible·using this configuration, without ·having any problems ??
Is it the correct way to do it ?
Any suggestions or comments will be apreciated.

Thanks so much.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards.

Alberto.

Kaio

Hi Alberto,

I think you need 20 of '244, each controls 8 pins of a Propeller.

Do you not need the bus bidirectional?

BTX

No Kaio, the bus is not bidirectional, only one way.
Why not all propellers input pins togheter ?

Thanks again.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards.

Alberto.

Kaio

Ok, then you need the '244 only as driver and you could connect the same pins of each Propeller together.

But what should be on the other side of the '244? You have said another Propeller? What is your intention of this circuit?

Harley

BTX,

Are these '20' slave Propellers on the same pcb?

If so, why any need for the -244s?

Just buss the 8 'data' lines from the 'master' to the slaves.

And let the master provide a 5 MHz timing, via Synth object, for their clocks; no need of 21 crystals.

But how does a slave know when to read? Looks like the master needs to provide 20 individual 'read' signals or enables.

I am using just a 2-Prop design. So have had to get more 'gray hairs' thinking of such things. Possibly a 3rd Prop can be used. More gray.

You'll probably use a programming header for each of the 20 Prop and EEPROM slaves.

My $0.02 USA.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Harley Shanko
h.a.s. designn

BTX

Kaio:
My intention is to have one pchip master, and 20 pchip as slave. Thanks for your help.

Harley:
I think to buffer the firts pchip with 244's, cause I'm totally not sure if I could drive all directly with one pchip.
Obviously, I need more lines to let the 'slaves' to recognize the data correctly, my intention was not to complicate the question, so I omited that. And the real question, is.... if one pchip could drive such quantities of slaves. (really this is not commonly job.....)

Talking about, using one pchip to drive the clock of the rest, is a very good idea !, but just not my case now, I'm involved in a big project, the time is going down, and I need to solve still a lot of things more, I'll use 21 Propstick USB for all, so programmiing, clock, and a more simple PCB is the better choice now.
I only hope, that parallax people, consider the ammount of money and let me a discount for quantities, if not ...... I must leave the 'sleep nights' for others days :-(
Thanks so much for your help too.

I would like too, to listen some answer from Parallax team people, until I do it. (Only to confirm or not the method...).

A complete internal schematics of the Propeller I/Os, would like so much too. These are the circumstances, where I miss the full Propeller datasheet.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards.

Alberto.

Harley

BTX,

If I recall, the Propeller is CMOS. So for the HI and LO levels there is negligible load; mainly on the transitions only to drive the capacitive loads. Since the data sheet isn't out yet, some areas are questionable.

I'd guess your master would have no problem driving 20 other Props for the 8-bit bussing. And the fact they claim 40 ma drive, that should take care of those cap. loads on transitions. (Even for a clock to the 20 Props, if you were to not to use PropSticks).

Don't know if the separate 3.3v regulator on each PropStick presents a problem with directly driving other Props without any current limiting resistor on the Prop to Prop lines. There could be some voltage differences, but hopefully not to where intrinsic diode clamping would occur at the high level. You might want to check into that situation with your 21 PropSticks.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Harley Shanko
h.a.s. designn

lairdt

Kind of sounds like the "WASP-Radial" project that I'm working on for a fun parallel processing project, but I'm creating a communication network with fewer pins since I don't need the bandwidth.

BTX

Yes Harley, I thought about that 30 or 40 mA drive capacity, must be possible.
Good point, about the regulators diffrences, I'll check that, a bit more close at the test level of the design. Thanks.

lairdt: my project is to get read files from a USB drive with a vinculum chip , and then distribute all of them to another electronic system. (20 separate units).

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards.

Alberto.

BTX

I insist.
Please, I would like to hear some oppinion about this, from Parallax People Team.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards.

Alberto.

Harley

Maybe more, better detail might elicit more comments on this design. From Parallax and non-Parallax 'xspurts'.

Like how you plan for the master Prop to 'address' the 20 other Props.....

Like do the 'slaves' ever communicate back to the master.....

Any block or schematic diagram to help us/them understand better....

Myself, I'd say it was possible; that is until better details of the design is known; then maybe NOT.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Harley Shanko
h.a.s. designn

Graham Stabler

I think Alberto is looking for specific info on the drive capacity.

Graham

Newzed

With what I have learned from my Prop over Prop piggyback, I have no doubt I could add more to the system.· If I want to establish communication with the Slave, the Master sends a "C" to the Slave and it responds.· With 52 upper and lower case commands that can be sent, I'm sure I could add 20 more Props if I desired.· I think it would be appropriate to have a power supply for each two Props.· The only thing I'm not sure of is how many Props can you send data to using one I/O port.· You might have to use two different I/O lines for each 10 Props.· However, the point is academic as far as I personally am concerned.· I don't plan anything like that in the forseeable future - I have my hands full just handling two Props.

Sid

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Yesterday is history, tomorrow is a mystery, and today is a gift.

That is why they call it the present.

Don't have VGA?
Newzed@aol.com
·

BTX

Sorry, don't get angry with me....
I respect all oppinions, but I only want to see, the view point of the Pchip creators, for that application.
I still didn't do any tests of it.

Talking about master and slaves, perhaps, If I've time in the application, I'll put all lines together, still the control inputs, then each prop, could receive two bytes, one could be the direction, and the another the data for one of them. So it takes 40 bytes to get all 20 bytes of data charged in the slaves. In that way, I wil save I/O pins from the master.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards.

Alberto.

cgracey

Alberto,

If you want to have an 8-bit bus, I would pick a particular 8 pins, say 0-7, and wire them common to all involved Propeller chips.

A single·Propeller is plenty capable of driving into an array of 20 others, or even many more. Only speed would have to be compromised as the array grew, and this would be·due as much to circuit board capacitance as pin capacitance.

I would also connect at least one extra pin common to all Propellers to serve as a 'start' signal that the master could control. Then, I would put the same assembly routine into each slave, with some way to differentiate each one so that it would know·its unique 'address'.

It would also be important to drive the entire array with maybe the same 80MHz oscillator pack (8-pin device) to keep their clock edges aligned. This would allow you to push transfer rates to perhaps 1/2 of the instruction rate (10MB/sec).
·


▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


Chip Gracey
Parallax, Inc.

Smartguy

This will help with my Biped project.

(this name is being changed, sorry)

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Intellibrain 2 w/ atmega= 265 dollars
Propeller Proto board= 24.95 dollars
Building a robot using the Propeller= Priceless (Well, almost)

Smartguy

Are you using the DIP model for the slaves or the QFP?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Intellibrain 2 w/ atmega= 265 dollars
Propeller Proto board= 24.95 dollars
Building a robot using the Propeller= Priceless (Well, almost)

QuattroRS4

Atmega,
Would you consider maybe changing your name ? - It seems too long and I don't like the reference..

Quattro

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Necessity is the mother of invention'

rjo_

Alberto,

Did you ever notice that when things get real interesting... a very knowledgeable guy... in this case, Chip, steps in and warns you to be careful about either inductance or capacitance?

I don't want you to stop what you are doing... and I hope that you will shovel information to us as you go along... BUT

I would like to note that the demo board has 8 LEDS... and I don't think they were put there by accident. With the right software and a way to conveniently link boards together with some fiber optic and PRESTO you have an expandable network of propellers... which has no capcitance or inductance problems.

Rich

rjo_

Alberto,

I am also personally convinced that the electromagnetic pollution around us is not exactly good for us and the pollution is getting worse... it might be a significant co-factor along with our chemical environment causing epidemics as diverse as breat cancer and autism.

Inevitably we will be moving to fiber optic networks, anyway. The Propeller community should lead the way.

Rich

BTX

OK guys ...thanks so much to all for the replies. !!!!

Chip:
Your answer is very clear, it seems that I'll need an oscillator pack to do it, and I will need, only 1.6 Mbytes/sec.... So it sounds very well, if you think that I could get about 10MB/s. But putting some 244's in the way ..will not that reduce the final capacitance between them (two 244's -> 10 slaves Pchips instead 20, and only 'two' for the master) ?????

ATmega: I will use QPF package (I suppose that is much better than DIP in this case). (ATmega .....not cool name...for this forum..)

rjo_: I don't know any fiber optics that could solve this case.... could you name some parts about it ?

Suggestions about ..How to minimize the PCB capacitance ?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards.

Alberto.

cgracey

BTX said...

Chip:


But putting some 244's in the way ..will not that reduce the final capacitance between them (two 244's -> 10 slaves Pchips instead 20, and only 'two' for the master) ?????

It will reduce capacitance, but it will also introduce delay that will probably be greater than the delay resulting from the original capacitance of all 20 Propellers in parallel. You don't need to worry about adding '244s. It just makes things unnecessarily complicated.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


Chip Gracey
Parallax, Inc.

BTX

Thanks Chip !!

No more doubts about the schematic, I'll do what you say so.

Suggestions about ..How to minimize the PCB capacitance ? Short lines ?, which distance between them ? minimal or prefered not so close ?, ground plane is good for this case ? or absolutely not ?

Thanks again.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards.

Alberto.

BTX

Chip:
Could you be so glad to answer me this ?

Just need to design the board this week.

Suggestions about ..How to minimize the PCB capacitance ? Short lines ?, which distance between them ? minimal or prefered not so close ?, ground plane is good for this case ? or absolutely not ?

Thanks again.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards.

Alberto.

cgracey

Alberto,

Just keep the chips close together. I would let the wires go where they naturally want to. Planes are good for ground and power, but I wouldn't worry about their effect on the wiring. The wiring should present its own solution. Be sure to have wide VDD and VSS paths with local 1uF caps near the Propeller pins. That's all.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


Chip Gracey
Parallax, Inc.

BTX

Thanks so much Chip.
I want to understand .1 uF caps... Or you mean 1uF Tantalium ?.. We're talking SMD of course.

I hope all things work fine, and I'll post some good picts of it when I've them.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards.

Alberto.

cgracey

BTX said...
Thanks so much Chip.
I want to understand .1 uF caps... Or you mean 1uF Tantalium ?.. We're talking SMD of course.

I hope all things work fine, and I'll post some good picts of it when I've them.

Sorry, I meant 1.0uF ceramic, if possible. In surface mount, these can be found in tiny 0603 packages. Tantalum caps would work fine, too, but ceramics have the lowest ESR.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


Chip Gracey
Parallax, Inc.

BTX

OK. Chip.

.......Oh I forgot.
Does Parallax have that 80Mhz. oscillador pack that you recommend me to use in this ?
I only see in the web 3.3V oscillators with a maximun load about 15pF !! (Is this suitable for 16 Pchips in parallel ?).

It will be difficult for me to use P0..P7 for the bus, is it strongly necessary ? Why this ? Perhaps that informationn would be usefull for future· projects.



Does somebody knows about it ?


Sorry ...I never finished to ask·

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards.

Alberto.

Envio editado por (BTX) : 4/8/2007 4:33:17 AM GMT

BTX

Some suggest about it ?
Thanks.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards.

Alberto.

BTX

Hi all.
I've some doubts still, about to do the multipropeller board, like these:

Chip suggest to me to use some "oscillators" to drive all propellers with the same clock, but i was looking them in the web, and all I can found is with a maximun load of 30pF, Like each propeller (If I don't understand bad) has a 6pF capacitance on each XIN pin....how is it possible to drive all 16 props with a single oscillator ?

I'm just designing that boards, I'll share all the info that I get, no problem that for me (I hope to be usefull for somebody).

But I'll appreciate so much, some help about this point.
Thanks.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Regards.

Alberto.

QuattroRS4

You could go the route of the following -

www.ak-modul-bus.de/cat/documentation/CY27EE16.pdf

Cypress do a number of variations..

http://www.onfulfillment.com/cypressstore/(S(eiujala3c22wlp55xaoce245))/Product.aspx?d=13&p=499&sid=205&AspxAutoDetectCookieSupport=1


Quattro

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Necessity is the mother of invention'

Post Edited (QuattroRS4) : 4/11/2007 3:28:20 AM GMT