An actual project for a newbie biologist

[randallmeyer2000]

I thought maybe my zillion posts (approx 40 in a month) should include at least one mention of my progress since joining your group.

When I started I (1) owned 6502 and other comp. sci. books that I had partially read/digested, (2) owned a 65816 chip WDC, (3) owned odds and end, hardware stuff, but not nearly enough to get started.

I am still in about the same place, but have learned a lot about how to move forward. I have a memory map, though details will work out in the very near future. I have found my part #'s and prices for RAM and ROM and I/O chip (probably VIA) and have an idea of how complex I want my first system to be. I have also taken up the foolish task of continuing my previous hardware project; my "fool's errand", a Focal Plane Array of 5 QTY (or 9 QTY?) CMOS image sensors (forgot to design the lens for it, but I think I can do that , no problem!).

Was searching the internet yesterday to see if anybody has scooped my focal plane array idea yet? Apparently not! ALL the FPAs pics/papers I see are radio telescope and/or infrared, government type projects. A few NASA and other space agency imagers have CCD arrays, but none like mine, and none with my purpose in mind. I should post pics, ASAP, as scientists who place second, barely exist!

I have pictures of my partially completed FPA, and will post them here (forgot to put them on my flash drive this morning! How very cro magnon of me, to carry my 1s and 0s, by hand/foot, instead of "by wire".). Maybe with this new impetus, I can see the project through to completion? They (i.e. the CMOS imagers, in question) use an I2C bus, so, I might need to think very critically about how the 6502 project might dovetail with this previous insanity.

Well, anyhoo, this thread is for capturing my own self progress, and hopefully, might even generate a comment or two from the wizened elders/fellow newbies. Pictures of various parts, in various states of array and disarray, will follow. Hopefully more gets put together, than falls apart! Cheers! (and thanks!).

[randallmeyer2000]

A sessions of teaching binary to a certain ten year old has got me behind in my memory decoding scheme; but, soon, he will have his own chip and it will be time us BOTH to work on it!

Here is the rough draft, as I have no time to look at it anymore today; must go home and work on my "air-gap" computer! Maybe draft 2 will post tomorrow, and maybe an inkling of a PCB pattern, too!

(I must check the actual diagrams for the 74138; Maybe Wakerly, or "Weste and Eshraghian" will tell me the propagation delays for the 74XX logic in my schematic? I can calculate those and see if it is too cumbersome.)

Scanned files are from Weste and Eshraghian, and probably somewhat out-of-date. It seems to be a standard undergrad textbook, from its wide availablity on ebay/amazon/etc. @ like 1$ price! I haave not read/fully-cogitated yet, but plan to.

[GARTHWILSON]

For each logic IC, the manufacturer's data sheet will tell the propagation delays. Use these, not some generalization from a theory book. Any given logic IC (for example, 2-input NAND '00, inverter '04, etc.) will have different times and other characteristics from one family to another, like 74HC00, 74AC00, 74AHC00, etc.. I definitely recommend using CMOS, not TTL. TTL (like 7400, 74LS00, etc.) offers virtually no benefits to outweigh its deficits. See the 6502 primer page on 74xx logic families and timing margins and the links to related ap. notes at the end.

[randallmeyer2000]

Thanks Garth,

I just made my order, which means (1) I won't be buying/shopping for parts in the near future (always broke, I am!) and (2) I will post less, and solder more.

1QTY AT28C256-15PU-ND IC EEPROM 256KBIT 150NS 28DIP
2QTY 428-2158-5-ND IC SRAM 256KBIT 15NS 28DIP
2QTY 296-8230-5-ND IC DUAL 2TO4 DECOD/DEMUX 16-DIP
2QTY 296-1575-5-ND IC 3-8 LINE DECODER/DEMUX 16-DIP
1QTY 296-4225-5-ND IC 3-TO-8 LINE DEC/DEMUX 16-DIP
1QTY 296-32945-5-ND IC DUAL 2-4 DECODER/DEMUX 16-DIP

I hope to have some parts left over for PandaPro; This project is probably over his head, but at least we have more in common (more to talk about) when we are talking binary! It's a stretch, for a lil' fella just learning his times tables, to start in on "exponents of 2". He's doing well. Decoding memory will take a bit of time to explain, but I'm sure he'll get that too. I think he got the reset vector thing, already.

I have to try to read ahead (your primer) while he's at school, so that I have good info to give him on I/O!

When I make a Mouser order (perhaps after XMAS, but maybe not until my B-day in July!), I will have to see about getting him the 65C02. Or, perhaps, I'll get the 65C02 for myself (or a vintage NMOS'02 or maybe even a"4004-almost-museum-piece"), and he can get the '816. Kids expect advanced functionality, and will be disappointed with less, I think.

I was looking up Federico Faggin--no Nobel for him yet???? Smith and Boyle (CCD) got theirs! When I got out of college (summer 2002) Faggin was still with Foveon, an interesting company, to say the least. I was not "smitten" with their technology, but it WAS/IS interesting. And I liked that this guy was still active. Like, maybe he had a checklist (1) change the known universe, and everything in it. check. (2) Conquer markets and minds with a simple device. check. (3) re-invent the image sensor. check.

I was just reading today that he also was the brains behind the Zilog (Z-80?). I mean, the whole company, too! Busy life. Must be tough to pack it all in! They don't give those Nobel's posthumous, and he is 73. They better hurry up! Unless somebody else deserves it? Some rumors of microprocessors stretching back to F-14 Tomcat (arguably, not-so-Nobel worthy?) and an earlier (about one year) company/ I forget who?

Well, I have some posting to do, tomorrow and the next day, but I must organize the workbench too. Too many books, and not enough tools!

I did find my PCBs! But must unearth the Datak developer.

Thanks again, for all advice. Here's photos of my molex connectors; still finding the ribbons; misplaced them (I hope they are big enough conductors for my image sensor focal plane array projects "I-partial" and "II-still-a-dream"). And photos of my '816, in foam. And tomorrow have picture of latest on 6502-o-poly; will post then, but must go find lunch now.

Oh yeah, and my calculations of the petzval/coddington curvature of the human optical system. Seems to be off by a factor of 2X? Oh well, I triple checked it, back in 2011. I guess I'll leave it a mystery; maybe "hack it" in the "hardware stage" of optical "Lens Design"?

[BigEd]

Federico Faggin gave quite a good lecture on his story, which includes the history of microprocessors:
https://www.youtube.com/watch?v=hugZii_eX30

[Dr Jefyll]

randallmeyer2000 wrote:

my "fool's errand", a Focal Plane Array of 5 QTY (or 9 QTY?) CMOS image sensors [...] Pictures of various parts, in various states of array and disarray, will follow.

I'll be watching with keen interest. In all my life I've never seen an array in disarry! (Isn't that like the sound of one hand clapping?) Cheers,

Jeff

[randallmeyer2000]

Well, Dr. Jefyll, lots more to say on the science of eyes/optics/cameras, and not exactly the kind of thing that gets batted around the 6502 forum (unless, it does, and I am too new to notice?).

I have studied, alot, from books and papers and academic perspective. I have poured my biological head into the project, and tried to stretch back to my father's point of view--i.e. New England Tech assoc. degree, telephone wire tester/one-dot-five-troubleshooter--to get the electronics knowledge. It is a bit of a schizophrenic/multiple-personality research project, but I can see no other way forward than to study it ALL!

Here is a link or two, I meant to post yesterday and in the previous week; but my head is brimmming with thing that must be seen by people who have interest in this subject.

1989 van der Spiegel, Kreider et. al. "Foveated CCD Sensor". Good paper, low resolution. Sandini et. al; Wodnicki et. al. Many others, but most still have a somewhat low resolution, and all seem to be planar (i.e. focal plane).

To curve the image surface is interesting. Nature article 2008-ish, describes a guy named Rogers doing it with flexible electronic matertials. Not sure this is the right way, but it is one way to do it.

When photography was invented, the images were curved. (Kingslake, "A History of photographic Lens"). They would line the people up on a street corner, curved, in order to get a focused image in the periphery!

Then, science, glass, schott, abbe and zeiss entered the picture; circa 1860-1880-1900. Sure, Petzval helped, with theory and practice, but new glass was developed, to the extent that, when WWI broke out, the U.S. was separated from their sup[ply of good optical glass. They had to ramp up the research in this area.

OK, less of my "re-telling" and more posting links (I am still waiting for lunch.... I guess I'll just start calling it "dinner").

http://www.nasa.gov/mission_pages/keple ... embly.html

https://www.iee.et.tu-dresden.de/iee/an ... fovea.html

https://www.iee.et.tu-dresden.de/iee/an ... node1.html

And, one I have just found tonight, and have not acquired oir read, but which looks somewhat promising;

"Concept of a Dynamic Fovea for Early Data Dimensionality Reduction in Vision Systems and its First Silicon–Implementation in a CMOS Image Sensor" By Jan Skribanowitz and Andreas König

I'll have to track it down tomorrow or the next day.

Cheers

[randallmeyer2000]

Hey, BigEd, thanks for the Faggin lecture; I only watched half, but I watched that half twice! Immediately, I had to show "the kid" (i.e. PandaPro); two GREAT anecdotes about (1) His first two days at Intel and (2) His first set of 4004 wafers. (I am such an idiot! I had a chance to buy a 4004, with one broken pin, but otherwise operable, for about 20$ +S and H, and I didn't do it! Well, maybe the Garrett Air Research/Navy/Army/U.S.A.F./F-14Tomcat chip will happen onto ebay sometime? probably not, but a guy can hope!).

Someday, when things are a little more settled, maybe PandoPro will watch the rest with me and I can explain some of the details. I mean, I was not really "up to speed" on "self-aligned gates", but from my study of CCDs I know that switching from aluminum to doped-polysilicon gates was a big deal. (Its like a college lecture, so I don't worry if we only watch ten minutes at a time! "Small chunks" helps, to absorb the lesson!).

I dl-ed a few more lectures, i.e. Ted Hoff, Carver Mead, etc., and hope something interesting arises from them.

Remember George Gilder? I hesitate to mention him, for fear the politics and money and other unmentionables would sour this string, but I am not a fan of Gilder. No fan of Intelligent Design. Don't care for inherited wealth, no matter how "noble" or how "intelligent" or "honorable" said wealth is. I like meritocracy, a quantity in short supply in the 21st century. I only mention him because he wrote a book "The Silicon Eye". I found the book an "interesting" introduction, but having already read a bit about foveated image sensors, I thought Gilder focused (pun intended) too much on the "Neuromorphic" angle.

I will be interested to hear Carver Mead, in his own words. It will be nice to separate his actual ideas, from the filtered version I probably got from Gilder (I read "The Sil. Eye" a while ago, so, luckily, I remember very little of it).

I soldier on, with the arrayed-disarray; the 9-image sensor project. Dug it out of storage; probably fried the CMOS imagers, as they are 3.3V power supply, and I stored them--partially soldered to a partially etched PCB--in a supermarket plastic bag. DUMB DUMB DUMB. Oh well; I have seven more image sensors, still in original packaging, so the two that I probably broke might just have to stay that way. Maybe I'll dump the PCB I have and build a new one; I did dig out my silver/graphite, plated through hole prep., and my H2SO4 and HCl (I forget what those are for? electroplating, I presume? Time to bust out my chem books!). I found a few presensitized boards, with protective coating still on, but have not unpacked my developer solution, yet? My workspace is cramped.

I found my documentation on the original evaluation boards.. I guess Kodak bought these imagers from National Semiconductor (I think I knew that, but forgot?). Or maybe Nat. Sem. just made the Eval. Boards? I think the former is "more-true". In any event, they are discontinued; Kodak sold (circa 2011) to a venture cap. firm in CA, who, in a year or two, flipped it to ON semiconductor. Kodak has subsequently been in and out of bankruptcy; so its been wild a ride, since George Eastman first took a holiday from his bank teller job, and brought his messy chemicals with him, and decided "there has to be a better way to make a photograph". He happened upon film, and "mechanical advance" of film, and then much work in making cameras and lenses, and much chemistry into color filters and plastics and organic chemistry. And now Kodak is what? Lucky to be alive and kicking! (still making film, I understand, for the motion picture industry).

But I digress....too much business, not enough science and history!

Thanks again, for the link,

Cheers,

Randall

[randallmeyer2000]

Right; Aluminum shorts were the problem with early CCDs. With polysilicon, this problem was reduced (I think I read this in Benyon and Lamb copyright? circa 1980?). One needs at least two phases, preferably three, for a CCD, and poly-Si helped there.

I have been trying to fit the EEPROM, RAM , '816, and the 65C22 onto the same board, with one decoder 74138 (and maybe some XNORS? I haven't worked ALL the logic out, just yet!) with a double sided, plated thru hole, model, but it has been annoying. The pins don't seem to line up very easily, and don't "want" to let me minimize the distance between them.

Anybody know, is it OK for me to put some DIPs on one side of the board, and some on the other side of the board? Does it matter which side of the pin is soldered to which side of the board? Can a pin be attached to one trace on one side of board, and another trace on the other side of the board; i.e. "common-node"? I assume this is OK, but I don't know.

These are very "newbie" questions, I know. I should just hack apart some old electronics, but a professional board will look too neat, and I won't learn anything about what is possible.

Also, I have started to build a little (Industrial fiber Optics) digital transmitter-fiber-optic circuit, but stalled because I needed a pinout diagram for my Quad 2N2907 transistor (PNP? I think). Maybe a trip to radio shack and $0.23 will solve the problem; just get a single transistor instead of the quad IC? Or, I could plug it in and "try it or fry it", till I find the pins I want? I'll build the digital fiber optic receiver tonight, I guess. The transmitter is TTL compatible; for serial comm; I forgot the frequency specs, but it should be interesting to send pulses with light.

Maybe when the F-O circuit works (in a day or two) I'll post a pic. a few more odds and ends on my PCB design, and I'll post that too; along with my decoder circuitry (If I remember right $0000-$5FFF, RAM; $6000-$9FFF I/O ; $A000-$FFFF EEPROM). That'll be 24KB RAM, 16 KB I/O, and 24 KB ROM.

Also, I should check all the other parts of 6502.org, to see if other PCB designs are already available? (It would save me time on the artwork! Though, not nearly as satisfying!).

[GARTHWILSON]

randallmeyer2000 wrote:

Anybody know, is it OK for me to put some DIPs on one side of the board, and some on the other side of the board?

That's what I did on the little computer board shown in the last two pictures at the bottom of http://wilsonminesco.com/6502primer/addr_decoding.html, with the EPROM going on the back. You can see it in the last picture, with that board upside down, and the EPROM on the top, with the white label on it. I wanted the EPROM socketed, so the rails connecting the two rows of the socket were cut out to later make it possible to solder the ICs that go on the other side, and the socket rows were soldered on first. The socket can only be soldered on the opposite side, and the pins would come out where they would later be covered by other ICs.

Quote:

Does it matter which side of the pin is soldered to which side of the board? Can a pin be attached to one trace on one side of board, and another trace on the other side of the board; i.e. "common-node"? I assume this is OK, but I don't know.

Commercially made boards of two or more layers always have the holes plated through, so the hole connects all layers at that point. The hole also wicks the solder through when you solder, regardless of which side you solder from. If there's room to get the soldering iron in, you can solder from either side, except that some things like sockets and relays won't expose the pins to reach on the top with a soldering iron like ICs do. ICs, resistors, and various other parts can be soldered from either side. For the rest, you have to plan how you'll get this thing together, as discussed above, so you don't wind up with a situation where you have something left to solder on and there's no way to access the place to solder because other things are already covering.

You can also put ICs inside a larger socket, so smaller ICs go under bigger ones, as shown in the last picture at http://wilsonminesco.com/6502primer/MoreOnPCB.html .

Quote:

but stalled because I needed a pinout diagram for my Quad 2N2907 transistor (PNP? I think). Maybe a trip to radio shack and $0.23 will solve the problem; just get a single transistor instead of the quad IC? Or, I could plug it in and "try it or fry it", till I find the pins I want?

If you're getting serious about this hobby, it's worth getting a good DMM. It will include a transistor and diode tester which you can use to figure out the pinout and measure gain, too. I have a paragraph on DMMs in the middle of the page on basic workbench equipment, at http://wilsonminesco.com/6502primer/BasicWkbench.html . The DMM will last decades, as long as you don't abuse it. Mine is about 30 years old.

The 2N2907 is indeed a PNP. You will probably save board space using discrete ones in a TO-92 package rather than a quad of them in a DIP.

[randallmeyer2000]

Thanks a million. I have tried to make the PCB layout two or three times, but each time I sort of stall and start from scratch again. The good news is, your advice is in line with what I was thinking and where I was going. I am getting closer to actually getting a working design (i.e. one that is possible for me to implement successfully at my skill level).

Parts are ordered; in the mail as I write. I still need (1) an oscillator can (2) a reset button (though, my local radio shack will have one) (3) a 3-input And gate, (4) a pin/holder-thingy for your expansion board and (5) your expansion/memory board; ( I haven't looked at the schematic yet for the expansion, but the '816 chip suggest I should!). Oh yeah, and 6522. And power supply (which I haven't thought much about! I don't suppose a 6 VDC, wall-wart and 7805 with a few electrolytic capcitors would work fine, would it?)

I am not ready to buy, yet (broke, as always) but in a month or two I could be ready.

So, a few small question; buses, both address and data, generally "fan-out", right? I mean, the whole bus trace is "driven" but because of the chip select, only one chip is "listening" at any one moment? So, my PCB traces for each address pin, or data pin, will generally branch into 3 or 4 or 5 different locations, right?

Regarding clock, your primer didn't seem to comment much on the '816. I will read it over again, but from the '816 schematic it seems there is only one clock pin, and it is an input (called Phi-2, I think?).

My question about 2n2907 was for a slightly different project; a fiber optic, TTL signal, LED driver. I bought a discrete one at radio shack on Monday; for 1$ I couldn't say no.

I think my decoder circuitry will require a 74138 (ordered and paid for, in the mail) and a 3-input And gate. Texas instruments offers $0.50 or so chips, with something like 8 to 10 ns propagation delay, so I'll probably order that soon.

I will try to post a schematic of my memory decoding, for perusal/correction. I just thought up the answer, while waiting for the radio-shack bus on monday! (if A,B,C, of 74138 IC is A15, A14, A13 of the 65816, then, the OutPut0=>OP2 (3 pins) address the 24KB of RAM; OP3=>OP5 (2 pins) address the 16KB I/O space; and the OP6=>OP8 (3 pins) address the 24KB of ROM).

Thus, for each, when one of the first three OutPut pins goes low, a 3-input-AND will go low, to enable the RAM CE pin; for I/O a 2-input-AND can chip enable (but does 6522 work like that? is there a CE pin.... don't answer that, i'm just thinking out loud); and, finally, the last three OutPut pins of the 74138 would use a 3-input-AND gate to pull the CE low, and enable the ROM.

It will make more sense when I put it on a PCB.

[randallmeyer2000]

One last question; possibly the most crucial (and I will read your segment on "Avoiding AC problems", rather than peruse it, as I did last time!). I should keep the clock signal far away from the address and data buses, right? I mean, its not even good to pass it over one of the A or D lines, even if they are on the other side of the board, unless its a brief pass (I mean, you wouldn't run the clock trace on one side of the board and a Data or Address trace directly oppposite it, on the underside, due to capacitive coupling, right?).

So, if a clock trace MUST pass over these other traces, it is best to (1) go right over quickly and (2) leave much room between the clock trace and other traces, regardless of what side of the board these other traces are on. Is that right? Or am I over-obsessive, at these frequencies?

Ooops, I have to go... local library is closing and computer time is over! (I "jacked" your website, and took it home to my "air gap computer". i will read the primer, again/more/thouroughly/etc.

Thanks again

[GARTHWILSON]

randallmeyer2000 wrote:

I still need (1) an oscillator can (2) a reset button (though, my local radio shack will have one)

Radio Shack will have a few things when you need it right now and can't wait for an order to come; but generally you'll pay way too much there, if they have the part at all. If you can wait long enough to receive an order and can buy enough parts to make the few dollars of shipping worth it, I would suggest getting familiar with Jameco, Mouser, and Digi-Key. You can do online-only if you like, but they do all have paper catalogs, unlike many industrial distributors. Jameco's catalog is the most oriented toward hobbyists and is only about 170 pages, with lots of color pictures. The other two have paper catalogs of about 2500 pages, with much smaller printing, and nearly everything is in stock, and in fact they stock a lot of things that aren't in the paper catalogs! We (the company I work at) use Mouser and Digi-Key a lot for production parts. For prototyping and hobby parts I sometimes use Jameco as well. I suppose that after I form parts lists, the company buyer probably collates them and sends a spreadsheet to the distributor, but I like to call them up with credit card in hand and list with exact stock numbers and quantities and prices in front of me. The process is very quick, especially after they have my address and other info on file. (Having the price there in front of me serves as a cross-check to make sure I got the right stock number. If the price they tell me doesn't match, I can see where the error is.)

Quote:

(3) a 3-input And gate, (4) a pin/holder-thingy for your expansion board and (5) your expansion/memory board

If you mean the memory board, I can supply the sockets as well when you order. I have both soldertail and wire-wrap. I just pass them on for my cost (which will probably be less than your cost since I buy in larger quantities).

Quote:

And power supply (which I haven't thought much about! I don't suppose a 6 VDC, wall-wart and 7805 with a few electrolytic capcitors would work fine, would it?)

The 7805 requires a little more voltage to be able to regulate. You shouldn't let the input drop below about 7V. Put monolithic ceramic capacitors at the input and output too, super close to the 7805.

Quote:

So, a few small question; buses, both address and data, generally "fan-out", right? I mean, the whole bus trace is "driven" but because of the chip select, only one chip is "listening" at any one moment? So, my PCB traces for each address pin, or data pin, will generally branch into 3 or 4 or 5 different locations, right?

Yes. "Fanout" generally refers to DC loads though, and CMOS presents essentially no DC load, only a little input capacitance (a few pF). If you had a huge number of these loads, it can slow things down a bit, but in your case it probably won't affect your design significantly.

Quote:

Regarding clock, your primer didn't seem to comment much on the '816. I will read it over again, but from the '816 schematic it seems there is only one clock pin, and it is an input (called Phi-2, I think?).

The 6502 primer is just that, a primer, so it's pretty basic. Most people going on to the '816 have already gotten their feet wet on the 6502. I also don't have much experience on the '816 as far as hardware goes. I have the 65802 in my workbench computer which is an '816 made to drop into an 6502 socket; so although it still offers tons of benefits, it is confined to the 64KB address space of the '02. I wrote a rather full-featured Forth kernel for the '802, and found it to be actually easier to program than the 6502 when constantly dealing with 16-bit values.

Quote:

My question about 2n2907 was for a slightly different project; a fiber optic, TTL signal, LED driver. I bought a discrete one at radio shack on Monday; for 1$ I couldn't say no.

If you're there and see it and you need it now, perfect. Get it. But again, realize that RS charges way too much otherwise. The PN2907 is basically the same thing in the cheaper molded plastic case instead of the can, and we buy the PN2907A, 2N3906, 2N4401, MMBT3906, MMBT4401, etc. for anywhere from one and a half to four cents each in quantity. Even in singles, you'll pay about one-fifth of the RS prices if you buy from a place like Mouser.

Quote:

I think my decoder circuitry will require a 74138 (ordered and paid for, in the mail) and a 3-input And gate. Texas instruments offers $0.50 or so chips, with something like 8 to 10 ns propagation delay, so I'll probably order that soon.

If there aren't any letters after the "74," it will be the old-fashioned TTL which, worst of all, presents an extremely heavy load at its inputs. Use CMOS, like 74HC, 74HCT, 74AC, 74ACT, etc.. If you need it right now and RS has it but not in CMOS, at least get 74LS, not just 74. I'd still advise going with CMOS as much as possible though.

Quote:

for I/O a 2-input-AND can chip enable (but does 6522 work like that? is there a CE pin

The '22 has two CE inputs, one being positive logic and one being negative. Use that to your advantage. It simplifies the glue logic.

About the clock line in your last post: Don't worry about running it close to other stuff. When it matters most, you'll use a ground plane anyway, and also observe the other things written there about groundbounce and things that are at least as much if not more of a concern.

[BigDumbDinosaur]

randallmeyer2000 wrote:

Regarding clock, your primer didn't seem to comment much on the '816. I will read it over again, but from the '816 schematic it seems there is only one clock pin, and it is an input (called Phi-2, I think?).

The 65C816 has only a Ø2 (Phi-2) input, which is the system clock on which all timing is based. It doesn't have Ø1 and Ø2 outputs likes those of the 65C02. Those outputs generally should not be used in new designs, as they slightly lag Ø2 in.

The "ideal" way to generate the Ø2 clock is with a can oscillator whose output is directed through a D-type flip-flop, such as a 74AC74 or 74ABT74.

Ø2 Clock Generator w/Flip-Flop

The above circuit can be used with a 65C02 (ignore PHI1) or the 65C816. If using bank latching with the '816 the PHI1 output would be used to gate the latch. The Ø1 and Ø2 outputs lag the oscillator output by 6-10ns. For this reason, no reference should be made to the oscillator output.

The resulting output is a very sharply-defined square wave with an exact 50-50 symmetry. The oscillator frequency must be twice that of the desired Ø2 clock frequency with this arrangement.

Quote:

I think my decoder circuitry will require a 74138...

Be sure to use CMOS devices with your CMOS microprocessor. Hence, you would want a 74AC138 or 74HC138 decoder (74AC138 recommended). There is no good reason to use 74ACT or 74HCT devices unless they are being driven from the outputs of 74LS devices—a 74xCT device will usually be slower than the 74xC equivalent.

[randallmeyer2000]

Sorry, Garth, I can be a bit "informal" when talking electronics and part numbers. I have to stop doing that. I forget what I mean (I was correct when I ordered the part? 74HC138, I think? I am pretty sure. Actually, I think I even bought a 74AC138 because, upon reading your primer--as I recall--I saw I might squeeze some extra performance out of them.)

I think the TI 3-input AND gates I wanted (for decoding) were also HC. My next digikey/jameco order will be that, and a switch, and maybe an oscillator can. I know RS can gouge, a bit; but the bus is cheaper than shipping and handling. but it takes about a day drive (weird schedules; transfer), so I catch up on reading.

Thanks BDD; "clock stuff" , I need!

Garth; I read about ground planes once; misplaced all my "good info about PCB design". My library is a mess.

I'll give the primer another view (I didn't since my last post; was watching videos on microproc. history and read very little. On the upswing, I did much CAD design on my PCB-drawing, and it is close to finished. Things still not "close" or "tight" enough, but it is a pretty thing, so far. I really should have had all the power and ground stuff figured out, already, but I have sort of saved it for last. Mostly, I've just been routing the address and data and decoder and CE pins.).

I suppose I am a bit distracted by my Focal Plane Array. I very much want to take a crack at finishing it. (not a "great" idea, because its 9 X VGA resolution, which is approx. 3 MP; Old sensors; old technology (KAC-9628; I think it came from Nat. Semiconductor product line, originally. I did not design the lens for these things, but it will probably be Gi-Normous!).

download/file.php?id=2946&mode=view

download/file.php?id=2947&mode=view

If you look at the second link, you'll see the line width is small, and they are, in some cases not quite separate. You will also see that I, in the haste, naivite and desperation of my 5-years-ago-self, have hacked out some of the copper with a razor blade, to separate the traces. Not a pretty solution, but effective, I think.

I do need to invest in a decent DMM, and some decent oscilloscope probes, and I might just check "continuity" of the traces, as they presently are, with my voltmeter/ohmmeter (though sharper probes are needed? The traces are small.).

If I am lucky, I might find the right ribbon connector to slide through those holes in the PCB and back to some other PCB. That's shouldn't be too tough.

Think I'll do that tonight; go home and look at molex, SMT, land-footprint. Also, see if the ribbons I have carry sufficient current. If not, my original "hack" plan, was to slide PCBs into the slots between the sensors.

If I start over and design one from scratch, I would start with the lens (like a human eye; as close a design as possible). Then move to the design of a curved PCB surface (i.e. approximate a curve with some sort of geodesic/polyhedron). Then, buy seven/nine Qty of the cheapest imagers available.

This can be a cheap "foveated array" if the central sensor is given a higher resolution than the outer sensors. No ASIC design required!

One final possibility involves the fiber optic image conduits I have. But that is another matter, for another day and another thread. I should stick; solely; to descriptions of THIS (i.e. 65816) project, here.

Next time, I'll post my PCB pattern (in its state of near completion) just as soon as I figure out which type of file is lossless (hint; NOT JPEG!). My CAD program is funny about that, for some really strange reason? Why would a graphic software give poor graphics? Oh well... Cheers and Thanks! (P.S. Never did figure out the pin out on the PN2907 (now I wonder if I paid enough attention to the part number?); or even the single one I bought from radio shack. Maybe I burnt the latter? It didn't seem to work like I thought it should? I guess DMM would fix this, and then I could use my brain on other things; but I should be able to figure out how a transistor works?? I am forgetting easy things; like PNP versus NPN. Back to the books , I guess!).