Mike's Hobby Blog

Attached is a scan of the original SCELBI bill of materials for the boards and a consolidated version in excell format. I forget where I got the scan, and apoligize in advance for not giving credit to the person that did this work. I haven’t checked for accuracy, but it “feels” right.

The hardest part to find is the 8008, though there always seem to be a few on ebay (at crazy prices). I plan on offering a bundled 8008D as an extra cost option with PCB sets. The D part is a package type, not a speed option. Most original SCELBIs had C package type, but they are harder to find (and more expensive). American Microsemiconductor has thousands of 1101s in stock at good prices (under $4). I have some extra 1101s, and may offer a bundle of 8 as an extra cost option. 74XX parts are available from unicorn electronics in PA.

Each memory board supports up to 4 banks of 8 chips each bank, for a maximum total of 32 chips per board (1K of memory). You can add up to 4 memory boards into a 8H for a maximum of 128 chips and 4K of memory. The minimum is 1 bank of 8 chips on 1 SRAM board for a total of 8 chips and 256 bytes.

You need 6K of memory to run SCELBI BASIC, so it is not possible on a normal 8H. I’ve already started development of the additional/different 4 boards needed for a SCELBI 8B, which can run SCELBI BASIC. The 4 additional boards are a different backplane, 4K SRAM board, memory expansion board and a PROM board. The CPU, DBB, INPUT and front panel are exactly the same for both 8H and 8B systems.

I’m currently looking at edge connector options and will post information on that and other chassis parts in a few days. The edge connector is .156 pin spacing and .140 row spacing, same as Apple 1, though you want ears and longer pins, if possible. Some wires are connected to the edge connector from the back side of the backplane, which makes longer pins desirable.

SCELBI Parts List

BOM for SCELBI-boards

Here is an interesting bit of trivia – not including the chassis components, the 8 boards making up a fully populated 4K SCELBI 8H require 587 individual components compared to 221 making up an Apple 1 motherboard populated with 8K and 278 making up a 48K Apple II, rev 0.

I’ll bet that some of you thought that this would never happen. I just ordered a batch of six different PCBs that will enable building the SCELBI 8H. I started on the first board about 15 months ago. Comparing to the Mimeo, which I did in 4 months, this has been a huge task.

CPU
DATA BUS BUFFER
INPUT
FRONT PANEL
1K SRAM
BACKPLANE

I plan to offer a limited number of prototype sets at a significant discount for people that have built up one of my other boards in the past. These will be untested boards that will be shipped to you as soon as I get them in, before I have done complete bring and test, myself.

I’ll announce a price in a few days.

Now to get my act together and get the rest of the components that I need for bringup.

Stay tuned,
Mike W.

Since I’m almost to the point of building my prototype SCELBI, I have to figure out how I’m going to power the SCELBI during initial bring up. While, later on, I intend to put together a linear supply similar to what SCELBI sold, for now I’m going to make do with more modern supplies. First thing to do is to look at requirements.

The SCELBI 8H requires two supplies, +5 volts and -9 volts. Here is what the SCELBI “ASSEMBLY AND INSTALLATION INSTRUCTIONS AND SYSTEM DOCUMENTATION” says about power requirements.

“CONNECTION OF THE POWER SUPPLY

THE SCELBI-SH MINI-COMPUTER REQUIRES A POWER SUPPLY CAPABLE OF DELIVERING +5 VOLTS AND -9 VOLTS (PLUS OR MINUS 2%.) THE CURRENT REQUIREMENTS OF THE SUPPLY DEPEND ON THE AMOUNT OF MEMORY IN THE SYSTEM (AND ANY ADDITIONAL PERIPHERAL UNITS WHICH MAY BE CONNECTED TO THE SUPPLY.) THE AMOUNT OF CURRENT THAT THE POWER SUPPLY MUST BE CAPABLE OF DELIVERING WHILE MAINTAINING ADEQUATE VOLTAGE REGULATION BE DETERMINED FROM THE FOLLOWING DATA.

THE BASIC SCELBI CARD SET CONSISTING OF ONE EACH OF: SCELBI #1100 CPU CARD, SCELBI #1101 DBB & OUTPUT CARD, SCELBI #1102 INPUT CARD, AND SCELBI #1104 FRONT PANEL CARD; REQUIRES A MAXIMUM OF 1.5 AMPS AT +5 VOLTS AND 100 MILLIAMPERES OF -9 VOLTS.

EACH “PAGE” OF MEMORY IN THE SYSTEM (256 WORDS) REQUIRES 200 MILLIAMPERES OF +5 VOLTS AND -9 VOLTS.

THE CURRENT REQUIREMENTS FOR ANY ADDITIONAL PERIPHERAL UNITS UTILIZING THE POWER SUPPLY MUST BE ADDED TO OBTAIN THE TOTAL CURRENT THAT THE POWER SUPPLY MUST PROVIDE.

FOR EXAMPLE: A BASIC SCELBI-8H EQUIPPED WITH 1024 WORDS OF MEMORY
(4 PAGES) REQUIRES: 1.5 A + (4 X .2 AMPS) = 2.3 AMPS OF +5 VOLTS AND
.1 A + (4 X .2 AMPS) = .9 AMPS OF -9 VOLTS.”

Now, as far as the -9 volt supply goes, this rating appears to be very conservative, at least on paper. There are only two ICs that use -9 volts on the SCELBI 8H, the 8008 processor, and the 1101 SRAM. Reading the spec sheet shows that the 8008 consumes a max of 60 milliamps and the 1101 SRAM a max of 19 milliamps per chip. Typical values are 30 milliamp for the 8008 and 13 milliamps for the 1101.

If the spec sheets are correct, then the base board set would typically draw 30 and max of 60 milliamps. Each page of SRAM would typically draw 13×8 = 104 and max of 19×8 = 152 milliamps. My initial bring up will involve the 4 base board set and 1 SRAM card with 4 banks populated. Max value for this setup is calculated as 152×4 = 608 + 60 or 668 milliamps. The 900 milliamps SCELBI recommends for this config, would give plenty of margin, and just to make choosing a supply easier, I’ll round up to see if I can find a minimum of a 1 amp supply for the -9 volts. Since typical values of a system with 1K might be around 104×4 + 30 = 446 milliamps, I may actually be able to power 2K of SRAM with a 1 AMP -9 volt supply.

The same calculation on the 5 volt supply is much more involved, since it literally involves counting current consumption of every component in the system, so I’ll go with SCELBI recommendations of 2.3 AMPS, but for sanities sake I’ll round up to 3 AMPs as a minimum for the 5 volt supply.

This is for initial bring up. Eventually I’ll want to test a fully populated system, with accessories. Once I have bring up done, I can measure actual consumption and increase power supply requirements based on measured values. Just in case I can start with with power supplies that will handle a full system at initial bring up, I’ll use SCELBI’s recommendations to calculate requirements for a full 4K, 8H system.

That would be 16 pages x 200 milliamps of 5v and -9v for memory, plus the base of 100 milliamps of -9 and 1.5 amps for 5 volts. Total would be 16 pages x.2 amps = 3.2 amps + .1 = 3.3 amps for -9 volts. For 5 volts, 16 pages x .2 amps = 3.2 amps and +1.5 amps = 4.7 amps. I’ll round up the 3.3 AMPs of -9 volts to 4 AMPs and 4.7 AMPS of 5 volts to 5 AMPS.

Ok, so now I know requirements for 8H power supply for bring up and and have a target for full system testing.

Minimum (bring up) for -9 volts is 1 AMP and full system is 4 AMPs
Minimum (bring up) for 5 volts is 3 AMPs and full system is 5 AMPs

What is the most practical way to provide +5 and -9 volts, at these power levels.

I actually have all the components in my Mimeo parts stash, except the -9 regulator, to build a linear supply that would meet this requirement. I actually have the -9 volt regulator, but it would have to stripped off my 1101 SRAM tester.

There are some downsides to this approach as it uses up some hard to find Mimeo parts, it would not be sufficient for a 4K 8H system, and it would take some time to put together and test.

To me, the best answer, is to use off the shelf variable “bench” power supplies. These will be useful for a number of other projects down the road and decent ones provide readouts on current consumption, which a basic homemade supply will not provide. I did a little investigation and there is an easy way to supply both positive and negative voltages to a circuit, if you have a couple of bench supplies with “floating” outputs. Here is a link to BK Precision’s description about how to do it.

http://kb.bkprecision.com/questions.php?questionid=224

Micronta 22-123

For initial bring up, I need 2 bench power supplies (or a single dual supply) with floating outputs for bring up. I already have an old Micronta 22-123 1 AMP 0-25 volt variable DC supply. I’ve had this supply so long, that I don’t even know when or where I got it. It’s been through the wars, having been repaired a number of times. For a long time it operated intermittently, but a year or two ago, I reflowed the solder on the power transistor pins, and it’s been solid, since. Since it will supply 1 AMP, it should work as the -9 volt supply for bring up, assuming that it has floating outputs. Even though I’ve had it a long time, I’ve always run it independently of other power supplies, so I wasn’t sure about that.

After failing to find a manual online, I checked for connectivity between the AC plug and the terminals and there was none. This was a good sign. To prove that the outputs were floating I connected the plus side to the ground on an Apple II motherboard. I connected this through a 5 ohm, 5 watt power resistor to the negative terminal of the power supply. Since the resistor was rated at 5 watts, I could run 1 AMP through the 5 ohm resistance when the power supply was set to 5 volts. This would give me a good indication that I could supply the SCELBI at -9 volts at 1 AMP with this power supply. Next, I connected a voltmeter on each side of the resistor, so I could measure current and voltage.

Before turning on the Apple II or power supply, I cranked the voltage on the power supply all the way down. Now I turned on the Apple II and started running a 2102 SRAM test, that I already had it set up to run. I made sure that it was running correctly and turned on the Micronta 22-123 power supply with crossed fingers. Nothing happened, as I hoped. I made sure the Apple II continued to run the SRAM test, before starting to turn up the voltage. Everything performed as expected, the supply providing 1 AMP of current at -5 volts with no problems, at all. The Apple continued to run the SRAM test, with no glitches. It appears that this supply will work fine as the bring up -9 volt supply. Half the bring up power supply problem is solved.

The other half of the power supply problem is deciding upon a new bench supply to support 5 volts. I’ll blog about my decision process in an upcoming post.

Layouts for for the five boards needed to build a SCELBI 8 H are near final. I’ve done all the design checking using my software as well as cross checking with schematics and I’ve sent gerber files to an online fab house design checker. I’m also getting quotes for pricing.

Just to give you an idea of how much work this has been, I’ve gone through the following approximate number of revisions of each board.

CPU – 81 revisions – this was the first board I started on, so some basic SCELBI things were figured out while doing the work for this board
DBB – 39 revisions
INPUT – 34 revisions
1KSRAM – 35 revisions
Front Panel – 16 revisions – most basic board, by far

The grand total number of layout revisions I’ve made since I started in earnest in October of 2011 is over 200.

I’ve also been working on the backplane, but it is not quite ready. The SCELBI 8H can be built without a backplane by hand wiring, and I’m likely to build my prototype initially that way, but getting a backplane made sooner rather than later, is not out of the question.

Here is a picture of Curtis’s hardwired 8H (image provided by Curtis)

Just a quick update on SCELBI progress…

I have completed schematic layout comparisons of three of the five main 8H boards. What this means, is that I have compared the layout to the schematics and corrected any unexplainable differences between the two. This is a time consuming process, but I did find a few mistakes in my layout, so this has been worthwhile. I also found (for the second time) that there is one small mistake in the DBB board schematic. I also discovered that there were two versions of the DBB board made (I am laying out the later version) The boards I have completed this check on, are the front panel board, the CPU board and the DBB board. The CPU and DBB boards are fairly complex and took a while to do.

The remaining boards, input and 1K memory, are less complex and shouldn’t take quite so long. After these two boards are checked, I have to do some checking of pads and hole dimensions for discrete components. I learned some time ago that getting hole sizes right is as critical as any other single part of this process. After the dimension checks are complete, I have a final CAD check left to do, before getting quotes and the boards sent off to be made.

I also have done some work on the 8H backplane. The backplane is starting to shape up, but I’m not sure if I will have these built along with the five main cards or not, since with a SCELBI you can hardwire a backplane.

I spent a couple of evenings taking another look at the SCOPEWRITER. Followers of my blog will remember that last spring, I declared the SCOPEWRITER project a failure. At that time I put it away, knowing that eventually I would make another attempt at getting it going. Well, I have good news, at this point, I can’t call it a failure, it might even be called a qualified success.

Here is an image of the output, that I was able to create.

Scopewriter

I had to make two changes from where I left off, in order to get this output.

So now I have a working SCOPEWRITER. Come the think about it, this may be the only working SCOPEWRITER in the world. What else will I do with it – well at some point it will make a fun output device for the SCELBI 8H and I may just have to create a web page for it. And one other thing – my point of view is that this is really a SCOPEWRITER, not a reproduction – I used instructions and PCB layout from the original magazine article to create it, just like a hobbyist would have done back in the 1970s.