F18A Lockup

[sparkdrummer]

Help!

I fired up my main system this morning and everything was fine until I hit F7 while in BOOT. The system locked and I reset. Now I only get the F18A green screen and a loud system honk.

 

Any reset to get my system back up?

[+Schmitzi]

uih, maybe reseat the F18A-board and everything ?

[Iwantgames:)]

Did you reset by turning off the system or by reset button? If I get this screen and hit the reset it doesn't go away. If I turn the system off for about 10 seconds it goes away.

[sparkdrummer]

I have powered down completely and disconnected the console from the PEB with the same results. I cannot understand why selecting F7 from BOOT would create this issue. I was hoping maybe there is a reset on powerup that would clear the F18.

[sparkdrummer]

Fixed.

Disassembled console. unplugged F18A. Reassembled and tested - still green screen.

Disassembled console. unplugged F18A. Plugged in 9918A. Reassembled and tested. Black screen.

Shit, my old faithful motherboard has taken a dump.

And then my light went on - "Wait a minute dummy, start with the basics"

Disassembled console. Tested switching power supply I had put in a few months back. 12v - good, +5 - good, -5 - 3 volts and change. Tried this power supply in another console - same result - black screen.

Apparently the F18A does not need -5 to power up and that's what threw me off.

Anyway - went back to an old egg cooker power supply and we're back in business.

[Iwantgames:)]

Sweet!

[matthew180]

The F18A, like the original 9918A VDP, only requires +5V. I suspect the 12V and -5V in the console were initially added for the DRAM which requires all three voltages, but once added to the system other subsystems started using the 12V and -5V (the reset circuit comes to mind as something using the -5V.)

[globeron]

a shock --Today I had the same thing:

 

 

 

Steps which solved it (2048K Games module was inserted, Speech Synthesizer + NanoPEB)

 

1. Turn off TI-99/4A (Beige / non QI in this case, not sure it matters, but the Beige QI behaves sometimes different with the nanoPEB)

 

2. take out the Module

 

3. take out Synthesizer and nano peb (thus standard console)

 

4. turned it on, luckily the normal blue screen came back

 

5. then one by one connect everything, was wondering if the Nano-PEB and F18A are maybe interfering somehow.

 

 

 

Furthermore I hope it does not happen too much

(as I am building a standalone "gaming" console, so my kid can operate everything on its own)

 

1. Why a beige console (I like the silver/black), but my kid cannot turn on the power switch

(only downside of a beige console is there is no light indicator to show it is on)

2. Built 32K inside (so no external stuff, except a Speech Synthesizer

(downside, cannot connect other devices that have 32K in it, like nanoPEB or PEB box with a 32K card)

so have to be careful.

 

4. Speech Synthesizer actually it should be in the console as well)

(I believe in France someone did that)

3. F18A and VGA built-in, to get great graphics on LCD screens

[kl99]

Quick Reminder on Users of F18A, new cartridge boards and Navarone Expander. The Navarone Expander is not compatible with the new cartridge boards and can cause damage to the port.

Maybe the green screen is happening here as symptom.

[Tursi]

The "green screen" actually means that the TI did not boot - it's the equivalent of getting a black screen with a 9918A (as in, the VDP was never initialized). Knowing this can help troubleshooting efforts.

[matthew180]

Adding to what Tursi mentioned, the original 9918A powers-up with all registers initialized to zero. This has the side affect that the screen is blanked and the background color is transparent. Also, the VRAM is in an unknown state. For the F18A I decided that it would be nice to have some indication that the F18A was alive when things were not going as expected, so I default several of the VDP registers to have the screen not blanked and the name and pattern tables configured. Since the VRAM for the F18A is inside the FPGA I can also control the initial values of the VRAM, so I set up patterns for a character set and set the name tables with the test you see. This has come to be known as the "green screen" (I chose the colors because it reminds me of the old green monochrome monitors.)

 

To see the green screen the F18A needs power (5V), ground, and reset pulled high (i.e. reset is NOT being asserted.) If you see the green screen on a system with an F18A it means you at least have 5V and the power-on-reset circuit is finished and the system is NOT being held in reset. If the F18A is getting power but the reset input is low, the display will be blank since the video generation circuits are affected by the reset input.

 

The sound IC in the 99/4A powers-up generating a tone, so the "beep" you hear at power-on is actually the amount of time it takes the computer to finally getting around to shutting the sound *off* during the power-up routines. If the console is screaming when you power it on, then the CPU never even got far enough into the power-up routines to shut the sound off, i.e. something is wrong with the CPU, ROM, or related circuits (power, data/address bus, power-on-reset, etc.)

[+mizapf]

This is quite interesting, Matthew. I can well imagine that we add this start-up behaviour to the sound chip emulation in MAME/MESS so that we can reproduce this BBSOD (Black Beeping Screen Of Death). I still wonder how it gets to its initial values for the sound and noise generators. Don't seem to be random values.

 

This can help to diagnose more problem by means of emulation (as we did with VRAM failure already).

[Tursi]

I had that emulated in Classic99 for a long while, and it does add to the nostalgia... but it got annoying after a while and I went back to defaulting to muted. Classic99 is a closed system so you're less likely to build a configuration that doesn't at least boot GPL anyway.

 

Edit: as far as I know the values /are/ random -- certainly setting random parameters sounded much the same.

Edited December 17, 2015 by Tursi

[+mizapf]

I thought the sound is somewhat similar each time, but my memory may be wrong here. In that case it could be that the internal registers of the sound chip always start with the same value.

[+OLD CS1]

Sometimes there is a white noise component, sometimes there is a periodic noise component, most of the time there is just a single tone, but sometimes multiple tones.

[Tursi]

Yes, it's definitely different every time.

[matthew180]

Nice details about the chip here:

 

http://www.smspower.org/Development/SN76489?sid=ae16503f2fb18070f3f40f2af56807f1

 

I can't find the exact reference about the chip powering-on with tone generation active, but after reviewing the datasheet again I see why this is the most probable case. The chip needs power and a clock input to generate a tone, and these conditions are met when a system powers-on. The four 4-bit attenuators that control the volume use '1111' as "off", so if the device does start up with random register values there is only a 1-in-16 chance that a channel will be silent at power-on (assuming there is no strange bias for a flip-flop to be in one state or the other at power-on.) Compound that over the four channels and chances are petty low that the chip will be silent ('1111' in all four attenuators) at power-on.

[Tursi]

It's also worth noting that the SN76489 is a later chip than most TI's had (I think I heard once that some of the later ones used it??) There's not much difference, although a notable one is that the tone setting >000 is the lowest pitch (corresponding to a count of >400), where on the SN76489 this value doesn't count and outputs a flat line (as described in the 'playing samples' section). For noise the TI chip uses the 15-bit feedback register, but I haven't gotten to trying to confirm the tap bits. At some point I decided that they matched the BBC Micro in that spec ($0003).

[+mizapf]

the tone setting >000 is the lowest pitch (corresponding to a count of >400), where on the SN76489 this value doesn't count and outputs a flat line (as described in the 'playing samples' section).

 

yes, this also caused some headaches in MAME. Seems as if this is an indication of a DEC, JEQ instead of the proper JEQ, DEC (if we would try to implement the sound chip in TMS9900 language).