Stuart

If you're laying out a new board, might want to look at using a Dallas DS1210 (8-pin DIL) to control the RAM power and /CE. Less likely to get RAM corruption if it is accessed while power is dying when you switch off. Stuart.

I suspect the "requires 20% loading on all outputs" won't be achieved on the -5V line (unless you add a load resistor). Current draw on the -5V line is only a couple of mA.

The processor needs -5V as well.

Nice. Are you still using ExpressPCB to make the boards, or someone else?

It needs a proper 99/4A power supply which produces two output voltages. Which country are you in exactly - I don't think you've said? If you're in the UK, I've got at least one spare 'power brick' which can be had for the cost of postage, should you need it.

There's also this ... http://www.angelfire.com/electronic2/ti99/memupgrade.htm ... which replaces the internals of the speech synth. The schematic link doesn't work but it may give a few ideas.

Availability: 0 Supplier lead time 15 weeks Minimum order quantity 432! I've had trouble finding these connectors as well, and have in the past done exactly as Jaime has done - cut down a longer connector, and use epoxy to mould an 'end cap'. Works fine, but visually may look a bit of a bodge. I've just ordered a 'proper' pack of 5 though from that link that someone posted a couple of days ago. If you need a male part for this connector, I think someone in the past suggested cutting the edge connector off an old PC ISA card ... BODGE = Bit Of Damn Good Engineering

Needs a 9-30V dc supply with a standard 2.5mm(?) dc power plug. Should be able to get a suitable adapter for a few $'s.

I'm intrigued! On the bq4852, pin 21 is D5 (according to the datasheet that I found). The battery is disconnected until the first application of Vcc on pin 36. What are people breaking pins off for after initial startup? Stuart.

Most CF<>IDE adaptors allow you to route +5V power via pin 20, which is the polarisation key position on the IDE cable. This does away with the need for a separate power cable. Stuart.

Good point, well put! Video performance looks pretty good to me.

A little project for any willing MESS user ... Some people have expressed doubt about whether a decent wireframe display could be done on the TMS9918 due to the port-mapped interface bottleneck. Another computer that used the TMS9918 (actually, the later but equivalent TMS9129) was the Tatung Einstein, which had a port of Elite made for it, and for which a MESS emulation exists. Anyone fancy the challenge of trying to get it running in MESS, and seeing what sort of video frame rate was achieved? There are details from someone who got it running back in 2010 on this thread: http://forums.frontier.co.uk/showthread.php?t=1788. Down near the bottom that page there are some details about it needing a specific MESS version. The link near the top of the page to the Einstein Elite disk images is still working. Stuart.

Well, that was a bit of a tease! But the recent talk of Elite in one of the other forum threads, plus finding a 22-frame animated GIF of the opening screen, plus Tursi's Convert9918 program, made this too good an opportunity to miss. No vector wireframes involved I'm afraid. The process was basically: (1) Started by using Convert9918 to produce the bitmap pattern and colour files for a full screen image from the GIF. Needed a bit of tweaking of the GIF and iteration to get the colours right and aligned as far as possible with the 8-pixel colour restrictions on the VDP. Passed the resulting raw pattern and colour data files through a little Visual Basic program to convert them to assembler-compatible DATA statements. (2) Manually tweaked the pattern and colour files in the rectangular area of the spinning ship so that the colours are consistently white-on-black. (In this area, for each row of 8-pixels, Convert9918 sets the colour to either white-on-black or black-on-white depending on whether that 8-pixel row is mostly white or mostly black. For the following animated frames, I only wanted to have to update the bitmap pattern data (not the colour data as well), so the colour has to be consistent.) (3) Then took each of the next 21 frames of the GIF, tidied a few pixels here and there on the spinning ship, and passed them through Convert9918. Passed the resulting pattern and colour files through a Visual Basic program to extract just the data for the rectangular area of the spinning ship and convert these to DATA statements. Then used another little Visual Basic program to go through the colour file and where an 8-pixel row was black-on-white, invert the corresponding byte in the pattern file so that it corresponded to white-on-black. (4) Assembly program to display the first frame in full, then update just the rectangular area of the spinning ship for the following frames. Check for a keypress after each frame, and if a key is detected, load the pattern and colour data for the text. Glad I was able to catch a few of you out. Stuart.

Correct. TI reversed MSb<>LSb compared to the rest of the industry.

I’ve been burning far too much midnight oil over the past couple of months doing a port of Elite to the TI-99, and today, after a bit of final polishing, I think it’s ready for release. Contains all the functionality you know and loved, except that the Universe is not quite as big until I find a better way of compressing the planet data (someone did recommend to me the Flair Pool algorithm, but I’m having problems finding information about it). Versions attached for Win994a and Classic 99. Load using E/A option 3, DSKx.ELITE. The program should auto-start. Haven’t tried it on real hardware, but I don’t anticipate any problems. Elite.zip Enjoy, Stuart.

For those interested in tinkering with hardware, I bought an 8-channel logic analyser the other day - one of these http://www.ebay.co.uk/itm/130993520591?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649 - for the rather stunning price of ... £10. It's a cheap clone of those made by Saleae (http://www.saleae.com/logic) so you cheekily download their free software, plug in your clone analyser, and off you go. Screenshot below of a TMS9995 going about its business ... Stuart.

What version of the CF7 have you got? It is one of the versions where you need to plug in an external power supply?

I've got my UDS-10 plugged into an Edimax EW-7228APn configured as a wireless access point (to my Netgear ADSL router). Works fine - although some reviews say the AP is a pain to configure. I didn't have a problem.

Just a little preview of the simple TI web browser that I've been working on. This is running on my TM990 connected to a Lantronix UDS-10, but will be portable to the TI-99. Got as far as being able to specify a URL and retrieving the HTML for that page. Next need to start looking at a few simple tags (character underline, character small caps), plus a sprite pointer to click links. Video: http://www.avjd51.dsl.pipex.com/z_ti_browser.mp4

As the TI would be so limited in what it can actually display (text - no problem, but text in tables? text in frames? different fonts? different font sizes? font formatting? etc.), how many HTML tags do you think would need to be supported? I think I would be surprised if 32K wasn't enough.

Assuming you know that the RS-232 card is good so it's not a hardware problem ... Check what the value for the ConnectMode configuration parameter on the Lantronix is. This determines if and how the Lantronix echoes characters back. Look at the Lantronix manual pages 28 ("Modem Mode" at the bottom of the table) and 31**. Configure the Lantronix over Telnet rather than using the Web Manager if possible - I found the latter would sometimes not set/reset some values properly, and ISTR that the ConnectMode was one of them. Once you've found a set of configuration settings that work, make sure you write them down somewhere safe. ** Page numbers in the full UDS-10 manual available here: http://www.lantronix.com/pdf/UDS10-UDS100_UG.pdf.

I believe so, but it may require different/more configuration settings as it's a more advanced device.

Be aware that I found configuring the Lantronix using the built-in Web Manager to not be very reliable - it would sometimes reset settings that I hadn't asked to be changed. Better off configuring it using Telnet. Looking through Ben's instructions in the link above, near the end it shows an example of retrieving a page from the Yahoo web site. Many websites are now hosted on web servers which serve many websites (that is, many websites are on a web server which has a single IP address). The GET command therefore needs to be followed with the Host: parameter to specify the host name from which to retrieve the specified page. So a sequence to connect to a server to retrieve a page might be: C198.136.54.70/80[>0D] (wait for response) GET /<name of page> HTTP/1.0[>0D][>0A] Host: www.<host name>[>0D][>0A][>0D][>0A] (wait for web page, which is returned with a standard HTTP header)

As you say, "the bits past 16K are dropped", so it looks like the VDP tables are actually ending up in the same addresses as using the values from the manual. So the manual is keeping things simple, without the reader having to work out that bits are dropped. So why the 'complicated' values that are*actually* written? Is that data used elsewhere in the GROM for other purposes maybe?

Where are you getting those values from? The values printed have always seemed correct to me (apart from the typo that VDP R7 in BASIC and EX BASIC is >07, not >17 as printed).