Flash Point - Ram Page
Rayman
Posts: 14,877
Here's my top-secret big idea (I'm such a blabber mouth):
I'll combine a SQI flash chip with 4 SPI SRAM chips. The SRAMs will have
SI and SO connected together (I may use a series resistor on SO for failsafe) and
all four sharing the same clock and chip select.
Will it work? Don't know... I hope to find out by Friday...
But, maybe you can convince me this won't work and save me time and money?
In this nibble configuration, the board will use 8 pins.
That will let it still work on the Prop Demo board.
In byte mode, it will need 12 pins. This is a good fit for Prop Platform Usb, so I'd like to do a board that just plugs into the Prop Platform in either upper or lower row.
The SRAM, although much smaller, allows unlimited writes.
The SRAM allows sequential write and the flash sequential read.
So, I think I can directly fill the SRAM from Flash with the Prop just providing clock signals.
If it all works out, this could make for some very exciting graphics for not much money...
I'll combine a SQI flash chip with 4 SPI SRAM chips. The SRAMs will have
SI and SO connected together (I may use a series resistor on SO for failsafe) and
all four sharing the same clock and chip select.
Will it work? Don't know... I hope to find out by Friday...
But, maybe you can convince me this won't work and save me time and money?
In this nibble configuration, the board will use 8 pins.
That will let it still work on the Prop Demo board.
In byte mode, it will need 12 pins. This is a good fit for Prop Platform Usb, so I'd like to do a board that just plugs into the Prop Platform in either upper or lower row.
The SRAM, although much smaller, allows unlimited writes.
The SRAM allows sequential write and the flash sequential read.
So, I think I can directly fill the SRAM from Flash with the Prop just providing clock signals.
If it all works out, this could make for some very exciting graphics for not much money...
Comments
I am certain it will work...
Tim Moore got 4 bit wide SPI ram (using 4 chips) working first, see: http://obex.parallax.com/objects/346/
Then I designed a set of stacking boards to host four single bit SPI ram/fram/eeprom/flash
Tim's drivers did not work for me with the 23K256 SPI ram's
I considered making a version of FlexMem for the W25Qxxx 4 bit wide SPI flash's, but decided not to as I could use four one bit high density devices on the standard FlexMem board to get 256Mbits onto one tiny board stacked on top of a RAM populated board.
I found the *3* unrelated problems that prevented my four bit driver from working, and now I am working on an optimized >3MB/sec driver for my board (potentially >7MB/sec with a multi-cog very strange driver)
See http://forums.parallax.com/showthread.php?121326-FlexMem-lt-25-Memory-expansion-for-ANY-propeller-board-with-a-10-pin-header!-%28F
You can see the initial prototype PCB's from April 2010 in the thread, I'll be receiving a batch of nice blue production boards next week
Bill, I'm glad what I'm doing is at least a little bit different from yours (so as not to step on your toes too much).
Anway, maybe there will be cross-fertilization to lead to better software for everybody...
For some applications the memory layout of the SQI chip is quite interesting, so I will keep an eye on how you can make it sing
It's apparently fine to step on my toes though. I consider the 8 bit wide 2 SQI chip solution a direct *RIP* of the SpinSocket-Flash usage idea which you were obviously aware.
Competition does serve end users well so I won't complain too much, but we all should now accept that all ideas subject to "open season" without reservation.
And to be honest, I think your SpinSocket concept is a lot different than what I'm doing here...
BTW: My main motivation is better graphics for my 3.5" LCDs. This idea of generic modules was just an afterthought...
I really like your stacking SpinSocket idea - I hope you sell a zillion!
I don't think that Ray's Flash Point will plug directly into your SpinSocket.
The idea of using eight one bit SPI ram's has been discussed on the forums about 1e6 times - the only reason no one has marketed a board like that is that financially it does not make sense due to the cost of the SPI ram's.
I thought of using quad bit W25Qxxx parts on a FlexMem stacking board months ago, but decided to stick with one bit parts and one PCB for cost issues.
I still think you should be able to sell a ton of 28 pin SpinSocket modules - its a great idea.
Doubling up of four bit memory to make an eight bit word has been done forever on every eight bit micro out there - but only you are doing it for your nice 28 pin SpinSocket platform!
I've found eight bit wide spi-ish flash devices quite some time ago, but again, prop pin conservation and cost considerations ruled out my using them at this time.
p.s.
I consider your 'we all should now accept that all ideas subject to "open season" without reservation' to be going overboard; I certainly won't accept that.
How do you expect me to believe you didn't read about SpinSocket-Flash at least once?
And, I thought yesterday you mentioned you were interested in a Windbond chip that isn't code compatible with the SST chips I'm using...
Don't worry. Nothing you've done so far interests me as having any value. I always love it though when you announce your cats getting out of the bag when someone else introduces something interesting. You have really bad timing and other attributes.
Needless to say, I disagree - however I won't descend to personal mud slinging
but I will agree I sometimes have terrible timing; life causes unexpected delays in my projects and drivers
I don't understand your animosity - neither mine nor Ray's modules are meant for your Spin Flash platform.
As I recall, The "cat slipped out of the bag" was a response to a question asking if I would use 4 bit wide flashes on FlexMem; where I explained why I won't, and I let the cat slip as it provided logical backup to my reasoning.
Seems I'm late to the game in a crowded field...
I already have some SST SQI interface code, if you want to see that...
I like your products, and I was glad to see you bring out yet another module for the prop.
Heck, I like Jazzed's projects too - just check the posts I always leave on his projects.
I know he is a good guy, I think he was just having a bad day - and without the personal attack, I would not have posted my last message. I was trying to smooth ruffled feathers in this thread, not ruffle them further.
Obviously now, neither of these ideas was as orginal as I thought. To be honest, I haven't paid much attention to external memory options in the past...
I definitely did not want to stir up trouble with frequent posters like you and Bill...
I haven't even opened up the Winbond datasheet because I'm afraid I might like it better...
I just took a closer look at FlexMem (which, I see now is a great design.) I hope this board is sufficiently different for Bill... First, there's no "Flex" at all. Also, I see he made a point of putting the data pins on P0..P3. In hindsight, that is probably better, but I'm going to leave it alone...
I think I've also decided that if I do a 8-bit flash board, it will be specifically for Prop Platform USB. I'll hope that will make it sufficiently different from jazzed's SpinSocket ideas.
Here's the current layout:
(also the last cat escaped in the TODO list due to David's question, which Steve may not have noticed)
Thank you for your kind words! I was told today I would get the production boards next Wednesday or Thursday... I can't wait!!!!
nice small clean board you have there... I might add a bit more space above the top trace for safety when they de-panelize.
I have NO issues whatsoever with your proposed layout / pin usage!
If you look at my previous postings I tried very hard to have everyone go to the Protoboard 10 pin standard (which I adopted) for the good of the Prop... and I also encourage everyone to use my 4 pin i2c pinout and the 5 pin "HCOMM" pinout (that I later found out the Ybox2 also used)!
Actually, as I told jazzed before... I like his SpinSocket idea, and I hope he sells a zillion.
The idea is to grow the prop market so that ALL of us can make decent money.
Second, I think I have a better idea... I'll just make a thin, long adapter that connects one or two Flash Point modules to the Prop Platform.
This I did this,so you can't do anything similar bit is a bit childish.
So the first person to use this chip or that chip now claims dibs and nobody else can touch it?
Folks on here spend a lot of time thinking up ways to get more people to use the prop, why would anyone want to when the first step is to make sure you don't tread on an area someone else has already stuck a flag in?
</rant>
I also think the SQI idea is revolutionary, which I why I moved so fast when I found it...
There's a saying: "Great minds think alike", I think that applies here (even if it is self-flattering...)
PS: I like it better that he vented here rather than holding a grudge against me for some reason that I couldn't contemplate...
I thought this was a new idea (I did a google search a few days ago, and the only hit was a Parallax Basic Stamp forum page regarding a different SPI chip). But, now I see it's a very old idea...
So, why doesn't Microchip promote this concept? Only thing I can think of is that their microcontrollers probably have dedicated SPI hardware that wouldn't support this mode.
Yes Rayman, I consider you my friend too. I expect to have a good little surprise for you by UPEW.
Actually, considering your Propeller Platform LCD board is on half a card (I have one), making another half board with big, fast storage is a good option. By fast I mean 8 bit wide.
We can leverage software gains on standardizing the data on at least P0..7 pin-out since that's really the only place where we can get the highest single-COG throughput (4.7Mbps to 6Mbps). Don't put anything on P8 ... I'm using P9 for SCK and P10 CS*.
I won't have time to make a Propeller Platform 2xSQI board, but I do have plans for a 40 pin DIP module. If I post further on 8 bit it will be in the other thread.