Stuart

Some drives spin up even when they're not being accessed. So if you're copying files between 2 drives, you might have all 3 of them spinning and consuming power ...

The 28-pin speech synthesiser memory chips only have internal connections on pins 1 - 14. They appear as one-bit (serial) or four-bit (parallel) output.

I'm after a CTRL key cap for a black 'switch' (i.e not Mitsumi) keyboard. Should have a hollow square peg on the back. If you have a spare one kicking around then I'd be very grateful. I'm in the UK and will obviously pay postage plus a beer or two. Thanks, Stuart.

If you solder the 4 corner pins first then you can check everything is straight before doing the rest of them.

Correct. Ditto for the data lines as well.

But how would I fix that to my shorts or t-shirt?

Nnnnnnoooooo! We've already split this forum once for "Software and hardware development for the TI-99/4A Computer". Couldn't much of what the OP is proposing be handled by way of a sticky/pinned thread?

If you open the case and remove the PSU board, I'm pretty sure that the pins that the motherboard power cable plugs onto are labelled on the component side - should be +12V, +5V, -5V and GND. Don't know off-hand what the order of the pins is though as I'm away from home at the moment and don't have a spare power supply to look at. You should be able to solder a new cable onto the back of the relevant pins. I'll stress again - you need a (cheap)voltmeter to confirm that you've got the cable and power connector round the right way before you plug in into the nanoPEB - if you've got the polarity reversed then you'll be looking for a new nanoPEB. Bad idea trying to charge a smartphone. A USB port should be able to supply 500mA. That's half as much again as the motherboard draws (~950mA) and may well damage the power supply or transformer.

But if you're going to fit it inside the console, what's the point of using a 12V>5V regulator when you could just tap into the existing 5V supply directly?

A resistor would drop the voltage but it's a very bad idea for this application. The voltage (in Volts) across the resistor = the current going through it (in Amps) * the resistance (in Ohms). The problem here though is that the current drawn by the NanoPEB won't be constant - it will probably increase when you read/write to the card for example, so the voltage dropped across the resistor won't be constant, and so the voltage you're supplying the NanoPEB with will vary. Whether it will go high enough to damage it or low enough for it not to work, I'm not sure. You'd probably also need a fairly high power (Wattage) resistor to drop enough voltage and supply enough current. A 7805 voltage regulator is what you'd need here. (Using a pair of resistors as a voltage splitter to generate a reference voltage is fine as long as you're pulling a very low current.) What I've done for my nanoPEB is to solder a cable with connector direct onto the internal PSU output, and feed it out through the side port slot. To do this you'd need to be able to solder the cable together, and have a multimeter to verify you've got the voltage and polarity right before you connect it to the nanoPEB.

If the new PCB is going to be smaller, could you have it horizontal, with the 'extension' for the speech synth to plug onto as part of the PCB? Your current connector that plugs into the console - would the PCB fit between the rows of pins on that and you solder the pins to pads on either side of the PCB? (Not sure if Ksarul was suggesting something similar or not?)

Have you taken the GND from a couple of side port pins and connected them together, to make sure you've got a really good GND for the circuit if one of the contacts is a bit dirty? The LS245 buffer will 'tidy up' a noisy signal but with the small board and few components, things shouldn't be that bad. Have you got a multimeter to check the voltage at the ICs? When you say "it only sort of works", what's it doing?

The two pots also get a mention in the FDC data sheet - there's a copy at [http://lab.nethence.com/frederic/Tavernier/Driver/WD2793%20datasheet.pdf]. For the WPW, it seems a little vague as to what value you need (but reading more of the datasheet might help - I only had a very quick skim through).

If you look at this page ... [http://www.s100computers.com/My%20System%20Pages/ZFDC%20Board/ZFDC.htm#Data Seperator Adjustment] hyperlink not working - copy/paste the bit between the []'s) ... it has the procedure for adjusting the two pots. But you need to be able the program the FDC to a certain mode though (possibly a bit of assembly) and you'll need an oscilloscope or logic analyser. Or a trial and error approach seems possible ...

Here's an interesting pair of TMS9900's ... http://www.ebay.co.uk/itm/lot-of-2-ti-tms9900jdl-ic-chips-/331924930101 They appear to be manufactured by Rochester Electronics (www.rocelec.com) in 1999 rather than being TI originals.

I don't believe a bad VRAM will normally stop it booting to the menu screen (but that will be corrupted). Culprits for the long beep might be bad GROMs or GROMs need reseating, and IIRC someone recently found one of the scratchpad RAMs had gone bad, or a host of other potential failure points.

It's me at the weekends. (Just to clarify - that is a joke!)

Powered directly off the side port. Draws around 8mA when running a simple XB program - that will vary a bit depending on how bright your LED is (mine is fairly dim). Well within the 50mA max guideline in the console spec. Nice looking PCBs!

Matt, have you had a chance to progress this? I've knocked up the circuit on a breadboard with no buffers and it works OK**. Just the LS138, LS21 and the 62256 RAM. LS21 pin 6 is connected to the RAM /OE and /CS. The LS21 should be able to directly drive a small LED (with current limiting resistor) as well. (** Frustrating to start with. My console has 32K 16-bit internal RAM running with no wait states, and a switch to disable. But the switch disabled the RAM but didn't disable the 0 wait states when accessing the 32K RAM area. So with the breadboard RAM circuit attached, it was somehow reading the address back as data. For example, stepping through the bytes from address >A000 using EasyBug would give the data >A0, >01, >A0, >03, >A0, >05 and so on. Very confusing. Had to rewire the switch so it gave the normal wait states as well when internal RAM disabled.) Stuart.

Don't think this is a 4116 memory problem - with a bad memory chip you'll get a stable display but corrupted characters. Suggest removing the motherboard and checking if the video connector has come loose and broken the soldered connection to the video signal. (Alternatively try wiggling the video cable while it is powered on and see if the picture appears. Next I'd try replacing the TMS9918A and next the crystal. Or fit an F18A.

Comparing the AT27C040 and W29C040, the pinout is slightly different (look at A18, /WE, VPP).

The book that first got me into the 9900, more years ago than I care to remember. Bought a copy from Maplin.

Interesting interpretation of "New" when the TIM9904 clearly has the remains of heatsink compound in the two indents at the ends of the chip.

As you've been rewiring, make sure you've still got address bus A0-A2 reversed when connecting to the LS138 A0-A2, like on your original schematic. Easy to get that wrong, and it would give strange results.

'B' is for 'Bus'. 'Data Bus In'.