help on DC-AC inverter for Public art Project. Thks

This is the place for any magazine-related discussions that don't fit in any of the column discussion boards below.
Post Reply
User avatar
Viking
Posts: 87
Joined: Mon Apr 12, 2004 1:01 am
Location: Aylesbury, England
Contact:

Post by Viking » Sat Aug 09, 2008 10:21 pm

Hello Dyarker, I've just realised that in the original drawing for the display, the panels are 15x10 (150 pixels) and not 20x10 (200 pixels) as I had originally thought! Not sure now how I thought it was 200 pixels.
So I will revamp the drawings I posted above. I'm also going to invert the layout of the pixel position on each panel so the data is sent sequentially starting at the top left and moving to the bottom right. This is because, being shift registers, you send the last data first.
So ignore the diagrams above (I will see if I can remove them from the post).
Regards
Rob

Additional:- Actually that doesn't make sense either.
If the pixels are 5â€
LIFE….the crappy bit between birth and death

User avatar
Viking
Posts: 87
Joined: Mon Apr 12, 2004 1:01 am
Location: Aylesbury, England
Contact:

Post by Viking » Sun Aug 10, 2008 1:57 am

Hello Federico/Dyarker,
I'm having difficulty mapping the pixels on the current version of the display because it is not totally square. You get an oval unless you let the circle not come up quite to the top and bottom edge.
For a 28' by 28' square using 28 panels, each panel is actually 4' by 7' and each pixel is 4.8" by 4.2".
I will let Federico confirm the exact makeup of the display before I go any further.
Regards
Rob
LIFE….the crappy bit between birth and death

dyarker
Posts: 1711
Joined: Fri Aug 22, 2003 1:01 am
Location: Izmir, Turkiye; from Rochester, NY
Contact:

Post by dyarker » Sun Aug 10, 2008 3:35 am

Viking,

Slow down, take a breath. Panels 10 wide x 20 high were fine. In center two at top, and center two at bottom, only 15 pixels in vertical direction are used. That gives 70 pixels vertical and 70 pixels horizontal, a fine circle is possible.

Did someone say 4ft by 8ft? 5inches by 10 is 50inches (not 48inches), 5 inches by 20 is 100 inches (not 96inches). I thought the 5inch squares was driven by what is available from film manufacturer. (film can be cut, but the edges must be resealed).

Federico,

I thought you were clear enough in your image on page 2. I only got lost on change from A6818 to 25 8 bit channels per panel of pixels.

Now I've got to halt too!

===================
Going to 200 pixels per panel (used or not) was a good idea for having standard spare driver boards just in case. It was good for me too, because I wrote code to make the pixel to channel/bit conversion. With variable number of A6818s, I would have create a lookup table by hand.
Dale Y

User avatar
Viking
Posts: 87
Joined: Mon Apr 12, 2004 1:01 am
Location: Aylesbury, England
Contact:

Post by Viking » Sun Aug 10, 2008 5:04 am

Hello all,
maybe I lost the thread somewhere along the way, but the panels were always 20x10 pixels? and arranged as shown, with four rows of 7. But in Federico's original drawing, if you count the pixels in a panel, there are only 15 by 10. So the square is a true square (actually it is 60 by 70 pixels), but only by changing the number of pixels in the panel to 150. If you try fitting 4'x8' panels together, the result is not a square.

Image

So four lots of 20 pixel panels stacked end to end is 4x20=80, so display is 80x70.
Regards
Rob
LIFE….the crappy bit between birth and death

User avatar
Viking
Posts: 87
Joined: Mon Apr 12, 2004 1:01 am
Location: Aylesbury, England
Contact:

Post by Viking » Sun Aug 10, 2008 5:08 am

Yes, definitely losing it now! Just counted pixels again and realised that the top and bottom rows have 15, but the other are 20!!
Muppet!
Take an hour to lie down in dark room, then start again!
Rob
LIFE….the crappy bit between birth and death

User avatar
Viking
Posts: 87
Joined: Mon Apr 12, 2004 1:01 am
Location: Aylesbury, England
Contact:

Post by Viking » Sun Aug 10, 2008 6:34 am

OK, hopefully got this one right.
BUT display is still 70x80, as all panels circuits are identical and have 200 pixels of data sent to them. The actual display area is 70 by 70.
And the data to each panel must be sent in reverse, as the first bit of data to be clocked into the shift register ends up at the end of the shift register chain. So other drawings, waveforms etc are still correct.
Regards
Rob


Image
LIFE….the crappy bit between birth and death

dyarker
Posts: 1711
Joined: Fri Aug 22, 2003 1:01 am
Location: Izmir, Turkiye; from Rochester, NY
Contact:

Post by dyarker » Sun Aug 10, 2008 7:57 am

To save money Federico was only going to install the film pixels to form a circle. (his drawing on page 2)

Original post - film is 5" x 5". Last post on page 1 - he mentioned 4' by 8', and drawing on page 2 labeled that way. That's a oops in arithmatic by Federico that we also missed at the time.

His last post was 7 Aug, let's give him a day or two to digest this. It is his project after all.
Dale Y

User avatar
Viking
Posts: 87
Joined: Mon Apr 12, 2004 1:01 am
Location: Aylesbury, England
Contact:

Post by Viking » Sun Aug 10, 2008 8:25 am

Hello Dyarker, yes, you are right, I tend to get carried away sometimes. Must be all that caffeine?
Will wait for Federico to read through and digest.
Regards
Rob
LIFE….the crappy bit between birth and death

federico muelas
Posts: 67
Joined: Thu Jul 17, 2008 7:54 am
Contact:

Post by federico muelas » Mon Aug 11, 2008 7:21 am

Hi guys,
I was out for the weekend. Whoa guys, thanks a lot! give me few minutes to digest this (truly) and I´ll go back to you.
Thanks a lot!!!!!
Fede

User avatar
Viking
Posts: 87
Joined: Mon Apr 12, 2004 1:01 am
Location: Aylesbury, England
Contact:

Post by Viking » Mon Aug 11, 2008 9:54 am

Hello Federico,
Hope you had a good weekend away.
I won't throw any more mud in the water, but I've come to the conclusion that Dyarkers idea of making all 28 panels identical is the best one, despite the extra cost in terms of hardware which would not be used in all panels.
This would mean that all receiver boards would be identical and interchangeable, and the only setting would be the DIP switch to select the particular panel number. And that makes the code to turn a pixel pattern to a bit stream easy as well as the receiver PIC code.
I will let you catch up.
Regards
Rob
LIFE….the crappy bit between birth and death

federico muelas
Posts: 67
Joined: Thu Jul 17, 2008 7:54 am
Contact:

Post by federico muelas » Mon Aug 11, 2008 4:41 pm

Hi Guys,
Sorry about the mess, I should have been more specific.
I totally agree, it makes sense to make all the controllers the same , despite the real number of pixels and size of panel...

The pixel is not really 5â€

User avatar
Viking
Posts: 87
Joined: Mon Apr 12, 2004 1:01 am
Location: Aylesbury, England
Contact:

Post by Viking » Tue Aug 12, 2008 4:11 am

Hello Federico,

There are two ways of configuring the PIC within each receiver.

The simple way:- The only setting that the code needs, is which panel it is in. This is because all panels are 200 pixels or 25 x 8 bit words. So panel 1 is always 1-25 bytes… panel 28 is always bytes 676-700 and so on. So the DIP switch just needs a 5 bit word (binary 0-31). The receiver will always clock 200 bits, regardless as to whether they are connected to pixels or not. If not all the shift registers are fitted (only one 6818 for example) the ‘blank’ pixels must be sent first followed by the pixel data in reverse order. This is because the last bit to be sent ends up at the number one shift register output. All data between 1 and 168 will drop out of the shift register and only the last 32 bits will be stored.
This approach complicates the bit map to data stream transformation.

The complicated way:- Only the data which codes for actual pixels is transmitted. The PIC must then need to know a) what is the first byte of data and b) how many bytes to clock in. There are 3744 pixels in the display, which is 3744/8= 468 bytes. So the start setting DIP switch must have 9 bits (binary 1024) in order to set the start data position and a 5 bit data length DIP setting to set the number of bytes to be clocked (maximum number will be 25, minimum number 1).
This approach means that the pixel data is transmitted sequentially and should make the bit map to data stream transformation easy. In the receiver, pixel number 1 is connected to the last output in the shift register chain and so the data is reversed within the receiver.

Regards
Rob
LIFE….the crappy bit between birth and death

federico muelas
Posts: 67
Joined: Thu Jul 17, 2008 7:54 am
Contact:

Post by federico muelas » Tue Aug 12, 2008 11:48 am

Hi Guys,
I don’t know how helpful this truly is but this is the pixel map for each of the panels.
Thanks a lot
Fede

Image

federico muelas
Posts: 67
Joined: Thu Jul 17, 2008 7:54 am
Contact:

Post by federico muelas » Tue Aug 12, 2008 11:51 am


federico muelas
Posts: 67
Joined: Thu Jul 17, 2008 7:54 am
Contact:

Post by federico muelas » Tue Aug 12, 2008 3:52 pm

Hi Rob,

So it seems both methods are suitable.

A couple of questions, on the first method, does each PIC has to count the incoming bytes or channels to find out when its first data byte arrives in order to start feeding the 200 bits into the A6818s? or the PC would send a “specialâ€

Post Reply

Who is online

Users browsing this forum: Google [Bot] and 18 guests