Stuart

It depends on the size of the card. Remember that the CF7+ only uses half of the memory (the interface to the CF is 16 bits wide, but the CF7+ only uses 8 bits). So 39 volumes, each of 1600 sectors * 256 bytes, = ~15MB, which is ~half of 32MB. Stuart.

The third generation TMS99000 series processors could be operated like this. The processor has three 'bus status' pins which indicate what's going on on the bus for each cycle - getting instruction, getting operand, and so on. So by decoding these with the memory access signals, you can direct memory accesses to separate instruction and data memory banks. This technique, plus a 17th address bit, lets the processor access 256 kbyte of memory. Stuart.

There's a complete list of the tags in this document <http://bitsavers.tra..._usersGuide.pdf>, starting on page 149. But I see it doesn't include some of the tags you seem to be getting ... Stuart.

Not my 99/4A, but I've hooked my TM990 system up to the Internet so it can be used remotely. No practical purpose whatsoever, just wanted to have a play doing it. I've used a UDS-10 so far, but just got a SBL2e-100 from Netburner which also provides some LVTTL connections which you can control using a Telnet connection. So I'll be able to reset the TM990 remotely as well. Stuart.

I done the current measurements quite a while ago (and noted them). I didn't note the current without the XB cartridge, but from memory the difference was only around 50mA or so. The +5V circuit in the 'normal' power supply already is a switching supply, but the +12V (and -5V) supplies are plain regulators. Most of the heat I think comes from the +12V regulator which is mounted on that heatsink strip.

A few figures ... The console with the XB cartridge fitted draws around 950mA on the +5V line. The console draws around 330mA on the +12V line. The main output transistor on the power supply 5V circuit is rated at 3A. The 12V regulator is rated at 1.5A. The diodes in the bridge rectifier supplying the 5V and 12V circuits are rated at 1A each (so if two are in use at any time, that's a rating of 2A?) The transformer is rated at 1.6A, shared between the 5V and 12V supplies. I would think that anything 'digital' (within reason) you want to hang off the port is likely to be OK, but motors, relays and so on are not. You could of course do away with the transformer and power supply and run the console off an ATX supply, which would give all the power you're likely to need ...

If the tracks on the PCB start to glow, then you need to back it off a bit. No idea of a value as such, but consider that you can plug in a Gramkracker plus a cartridge, and a Gramkracker is pretty much stuffed full of chips (many of them are low power RAMs though?).

Adam, Attached is the assembly routine from Cortex Basic for plotting a line between two points. So this works out pixel positions along a line, which is pretty similar to what you need to do. May or may not be of interest. The three sequential XOR commands are a nifty way of swapping two registers. Stuart. PLOT.txt

It's more likely I think that one or more of the video RAM chips is faulty. It's displaying a picture OK, but the data is corrupted. If you remove the board you'll see a row of 8 off 4116 (or equivalent) RAM ICs. With everything connected back up again, you *might* find that one or more are running noticeably hotter or colder than the others - that might be a clue as to which needs replacing. Otherwise replace one, test, replace another if still faulty ... Replacements are fairly cheap and common on Ebay. To replace, I find it easiest to cut all the pins on a chip and throw it away, unsolder the remains of the pins and clean up the holes, then fit a socket.

A Maxim DS1210 (http://datasheets.maximintegrated.com/en/ds/DS1210.pdf) would make a good replacement for all those diodes, transistors and resistors in providing backup power for the RAM. It's a pretty foolproof little 8-pin chip that I've used with no problems in the past.

<Pedant> Note that these are the parallel port version ... so they're technically CF7+s rather than NanoPEBs. </Pedant>

Received the message below from a Randy Kaufman who was happy for me to post the details here. Contact Randy directly if interested - his e-mail is at the bottom. =============================================================== I found some TMS9900 chips, and wondered if there was any interest. I have 4 in ceramic packages, with no printing on them. In pencil, hand written on the ceramic, is "TMS9900JL". I also have one ceramic TMS9900JL, with a date code of 7643. There is one, also ceramic package, marked "TMX9900JL" with a date code of 7640 (note, 'TMX' instead of 'TMS', which indicates prototype). There are also 4 plastic packaged TMS9901 chips. All appear unused, although they are not in normal packaging. Don't know what they are worth, but I'll be reasonable! Randy Kaufman randyk357 (at) gmail.com

Just tried the following scenarios with my NanoPEB: (1) NanoPEB power switch off: console boots irrespective of whether a CF card is inserted or not. (2) NanoPEB power switch on: console displays a blue screen if no CF card is inserted (or presumably the CF card is duff or not detected for any other reason). (3) NanoPEB power switch off, switch on console, then NanoPEB power switch on: using MiniMem to do C1100=1 then M4000 will read out the DSR, irrespective of whether a CF card is inserted or not.

As you're using a CF card which seems good, I would suggest that your CF7 is damaged, and you'll need to get in contact with Jaime. Whatever data/formatting is already on a CF-card is pretty immaterial to the CF7.

Does the TI now boot with the card inserted? If no, you still have a problem. If yes, boot into BASIC, then CALL FORMAT(1), then CALL MOUNT(1,1), then try to save and load a line of code to DSK1. There's no 'valid config' on the card as such** - the nanoPEB will ignore all formatting and partition information that a PC might have put on the card - it just reads and writes sectors. (** Possibly apart from the disk<>volume mapping that is stored near the start of sector 0. A card filled with random data could possibly be mapping a disk to a non-existant volume, so make sure you do CALL MOUNT to set the mapping up.)

This is a simple little robotics project inspired by the project I saw here: [www.electronicsweekly.com/blogs/gadget-master/2010/10/eyeballing-for-donations.html]. The device is controlled by a TMS9900 processor in a modular TM990 system. Three analogue proximity sensors are connected to a multi-channel A-D converter. The sensors are scanned back and forth and enable a 'target' to be detected and tracked by the eye which swivels horizontally and vertically. When the target gets close enough, the money tin shakes. The project uses four stepper motors, with the stepper motor controller board control inputs connected directly to a TMS9901.

About the same length as a piece of string. (Sorry, couldn't resist ...) In practise, you're not going to actually need it longer than a few inches, and that will be fine.

Just to separate the CF adaptor from the main CF7 board. Prevents straining the connector if you're unplugging/plugging the CF card a lot.

Have you got a spare CF adaptor and CF card to try? From memory (excuse the pun), I seem to recall that it won't boot if it can't find a CF card. Then might be worth giving Jaime a ping via eBay and see if you can get a replacement Xilinx chip (which is socketed) and EPROM (which is soldered). If you know those are good, the other (one? two?) TTL chips on the board you can probably poke around yourself.

I've nearly finished a project I'm working on with my TM990 system. Three analogue proximity sensors, four stepper motors ... and a baked bean tin. Should have a video up over Christmas.

It also easy to add 32K inside the console using two memory chips and 3 TTL chips - excellent instructions at http://www.mainbyte.com/ti99/16bit32k/32kconsole.html

An easy mod is to connect a 74LS154 4-to-16 line decoder to the top 4 bits of the address bus (and /MEMEN applied to one of the /Enable inputs), and use that to drive a 16 segment bargraph LED. Each LED segment will correspond to a 4K block of memory, and gives you an indication of where significant numbers of memory accesses are happening. No point putting LEDs on the data bus IMO - it will just be a meaningless blur.

Not so much "successor" - one has a parallel port, the other a serial port. Other than that they're pretty much the same. The serial port version is the most popular and they're still being made in batches of 5 or so and sold on ebay. The chap who sells them has the ebay account "schmutzig1952" - you can contact him via ebay and enquire when the next batch is available. Stuart.

Has anyone got the second Commodore book by the same author - "The Amiga Years" - that was published this month? Any good?

I'd also check that the two rows of pins are not being reversed. That is, row 1 on one connector goes to row 2 on the other. ISTR that depending on the connectors and which way round they're fitted, it is possible for this to happen. Personally I'd put the NanoPEB on the table next to the cable, then you can see how the NanoPEB connectors mate, and check the corresponding pins on the cable, and make sure that nothing is reversed. You didn't flip the NanoPEB over when fitting it to the cable I assume? I not sure if the connectors are keyed so it would be possible to fit it the wrong way round.