jens-eike

I'll be there. Yes, I had trouble posting, too. Since this post came through: go to about:config and look at the javascript options!

Hej Rasmus, the PCB is from German TI-Revue. Two EPROMs seem to be stacked in the GROM area, ther is provision for cartridge ROM at the empty 28-pin solder pad area, this can be modified to accept an 8K RAM chip. Back in the days I modified this PCB to a battery backed GRAM device with bank switching to run Extended Basic (my home made GRAM Kracker).

Another wild guess: is SWPB implemented as a (circular) shift? That could explain the sloooooow execution, even on the 9995.

Look for the TMS9228, it is the missing link with many features of the 9938 designed by TI. Most video modes like 80*24, 512 color palette, 256*210 bitmap etc were planned for this chip. I read a full preliminary data book once, this link shows only an excerpt: http://www.datasheetarchive.com/TMS9228-datasheet.html The 99/8 has a TMS 9118 VDP, the newer design with only two TMS4416 RAM chips (16k*4bit each).

Take a look at the snug SPVMC and documentation at: http://home.arcor.de/system-ninety-nine-user-group/spvmc/index_e.htm The card comes with the Speech Synthesizer's standard vocabulary plus a male voice (aviation vocabulary) and a female voice (talking clock).

This datasheet: http://ww1.microchip.com/downloads/en/DeviceDoc/11173G.pdf states an erase cycle time of 40minutes, that seems rather long, TI's TMS27C512 was specified with an erase time of 21 min (MOS Memory Data Book 1991), my experience is 15min should suffice (blank test OK). Oops, yes the 29C512 is 32pin, my bad. Same offer, if you send me the *.bin file, I can program the chip for you, Flensburg is probably closer thus cheaper and faster mail.

If you want to keep it with the TI-99: the ASCSI card allows to program Flash-EEPROMs with the DSR-Loader program: http://home.arcor.de/system-ninety-nine-user-group/util/dsrldr3_e.pdf (omit _e for German version) What size EEPROMs do you mean, 8k*8 is 8k bytes or 64 kbit EPROM: 27©64, EEPROM: 28C64 (see Thierry's page about EEPROMs), 64kbyte = 64k*8 = 512kbit is 27C512 (EPROM) or 29C512 (Flash-EEPROM, Atmel AT29C512 is usable in ASCSI!). I am not aware of an EEPROM of this size, but a 32kbyte version is available: 28C256. All of the above chips come in 28-pin packages and should be usable on the cartridge board, but check for connection of the -WE signal, if it is connected to the E(E)PROM, it could cause trouble (writing to the EEPROM while changing banks). Jens-Eike

Thierry's drawing is a little difficult to read, I prefer Fabrice's FAQ: http://www.ti99.com/texasinf.htm (answer 10) 5 = Sound out 6 = Ground The Mechatronic 80-col device needs a Y-cable, too. It merges sound from the 6-pin DIN socket and RGB-video into a SCART cable.

the modification of the modulator is described at: http://www.ti99.com/ti99r1.htm

The 6-pin output has Y, R-Y (Pr) and B-Y (Pb) signals. Y can be used as black/white output, but color needs serious encoding (see the color circuit in the Powetran Cortex schematics at powertrancortex.com. Modern TVs with component input (YPbPr) can display the color differential signal directly, and witout the rainbow color distortion seen on NTSC consoles :-) To answer your question: there is no such thing as 6-pin A/V cables (for the TI).

Robert van Veen from Rotterdam, haven't heard from him since the Treff in Venlo 2005 :-(

Another TMS9995 related project is at http://ti99.collosumus.net/ti99/project/hc/default.aspx with my schematics redrawn at http://ti99.collosumus.net/ti99/project/hc/step1.aspx

Yes, it reads the high byte (MSB) first, then the low byte (LSB). There are some good comments about the inner workings of the 9995, 9918 and ALPHA (99000) at http://spatula-city.org/~im14u2c/vdp-99xx/ especially: http://spatula-city.org/~im14u2c/vdp-99xx/e1/99-4_History_By_KG_in_answers_to_Matthews_Questions.doc

like the TI-99/4A: sequentially. The TMS9995 reads the bytes in logical order, address 0 before 1, while the TI-99 reads the low byte first. (that is the reason why Thierry's cards have a TI/Geneve switch for byte order) benefit: the 9995 can read or write single bytes, while the 9900 reads a word and writes back the word with the modified byte (the dreaded read-before-write that slows everything down in our TI)

The 12MHz are divided by 4 to 3 MHz, resulting in a 330ns memory cycle. Depending on the decoding logic, 100ns should be fast enough. The TMS9995 data book gives timing diagrams with necessary setup times. (I am going for 45ns EPROMS for a 99105 board w/ 167ns memory cycle) edit: just found the timing diagram, it has the access time t(acc)=3/4*t(c2)-135 = 3/4*333-135 = 115ns, so your memory allows 15ns for decoding.

The 'LS32 is an OR-gate, if either -Memen OR A0 is high, -ROM_CE is high (EPROM inactive). This is not the -RAM_CE select condition, as -MEMEN still needs to be low, else the ram could write false data at CRU accesses. An OR-gate is an AND for negative logic: if both -MEMEN AND A0 are low, the select is low (memory access at >0000->7FFF), for >8000->FFFF, invert A0 as in the schematics. Happy new year!

My favourite card is like a hybrid TI-99 and TI-59: the TM-990/189 http://www.vintagecomputer.net/ti/TI-TM990-189/