Noisy video problem, plus...

[Geoff Oltmans]

Well, maybe spoke too soon... maybe a loading issue on the data bus instead. I came back after a couple hours and tried again and while things are better than they were, it's still a bit flakey.

[Geoff Oltmans]

Update: The Speech Synthesizer was a bit of a red herring. There's definitely issues with this Alpiner game while it's plugged in. But, turns out that there's also a problem with the game when the PEB interface is plugged in as well!

[+FarmerPotato]

Is your console a later one, QI or displays V2.2?

 

 

[Geoff Oltmans]

5 hours ago, FarmerPotato said:

Is your console a later one, QI or displays V2.2?

 

 

Nope, it’s a black and silver. 

[+FarmerPotato]

1 hour ago, Geoff Oltmans said:

Nope, it’s a black and silver. 

Ok, then it has the weird -0.72 Vss rail. Check that?

 

TI (that is, Mike Bunyard in his Hardware Manual) believed it was necessary for correct functioning of GROM, but the QI console dropped it so it’s a mystery.
 

(We had a Zoom discussion on:  how the diode failure fries the p-code card. Surely not that drastic here!) 

https://www.unige.ch/medecine/nouspikel/ti99/groms.htm#Pinout

 

There’s another statement in the TI PES and Console Technical Data. “The GROM are poor line drivers at best” which is how 3rd party peripherals are able to override them (GRAM Kracker for example.)

 

Very interested in how your effort turns out!

 

 

 

[+FarmerPotato]

Duh. Didn’t see mention of this so far: have you cleaned the contacts of the cartridge port?
 

Do you have a replacement L to swap in? 

[Geoff Oltmans]

12 hours ago, FarmerPotato said:

Duh. Didn’t see mention of this so far: have you cleaned the contacts of the cartridge port?
 

Do you have a replacement L to swap in? 

Thanks for the link to the Vss info... I was unaware of that. I'll have to check that out.

 

Yeah, I've thoroughly cleaned both the console GROM port connector and the L connector and edge. I have also reflowed the solder joints to the GROM card edge on the console PCB.

 

I haven't done my check on the PIO port of the RS232 interface card yet.

 

[Geoff Oltmans]

22 hours ago, FarmerPotato said:

Ok, then it has the weird -0.72 Vss rail. Check that?

 

TI (that is, Mike Bunyard in his Hardware Manual) believed it was necessary for correct functioning of GROM, but the QI console dropped it so it’s a mystery.
 

(We had a Zoom discussion on:  how the diode failure fries the p-code card. Surely not that drastic here!) 

https://www.unige.ch/medecine/nouspikel/ti99/groms.htm#Pinout

 

There’s another statement in the TI PES and Console Technical Data. “The GROM are poor line drivers at best” which is how 3rd party peripherals are able to override them (GRAM Kracker for example.)

 

Very interested in how your effort turns out!

 

 

 

Just measured Vss at the GROMs in the console and by my multimeter it reads -0.705 so I think I'm good there.

 

I replaced a few components in the console now too.... I replaced the two 74LS138s responsible for address decoding (U504 and U505) and the bus drivers at U614 and U615. I don't have 74LS244s at the moment so I can't replace that component at U616. I have reflowed the GROM sockets, the GROM port connector, I've tried a different VDP since I have a few of those on hand and it's socketed already. It seemed at first that removing the TI BASIC GROMs helped, but later test that didn't make any difference.

 

So far replacing U504 fixed my PEB DSR access in the >4000-4fff range on the FDC. Before that I was getting garbage in that address space. Everything else hasn't really helped matters.

To summarize, where I'm at currently: Standalone the console seems to work A-OK. Any loading on the expansion bus connector (PEB or Speech) will make a couple of games not work perfectly (Parsec and Alpiner). DSR access >5000-5fff no bueno on the FDC. Memory expansion in PEB tests out fine, so I think that eliminates address bus issues since >2000-4000 and >A000-FFFF should wiggle all the Address lines appropriately. Data bus likewise seems to work okay to the PEB Memory card as reads/writes are all okay. CRU write access to FDC appears to be okay, all signals from the latch outputs are good (card select, motor strobe, drive selects, etc). I haven't completely checked CRU read from this access space, but I did see that the DS1*Strobe input does read a '1' when the motor strobe is on, so I presume that should all be okay. 

 

 

[Geoff Oltmans]

Oh yeah, I've checked all the power supplies early on and even bench tested with an old PC supply instead of the TI one, no difference there either.

[+FarmerPotato]

I wonder what happens if you replace the LS245 with HCT245 (or even ALS245)

It doesn’t sound like the address bus is a problem, but the data bus that is glitching. 
 

Access to the external bus should incur 2+2 wait states for each word. GROM is especially slow (ISTR 30 cycles?) Perhaps something is wrong in the wait state generator. 

 

 

[Geoff Oltmans]

1 hour ago, FarmerPotato said:

I wonder what happens if you replace the LS245 with HCT245 (or even ALS245)

It doesn’t sound like the address bus is a problem, but the data bus that is glitching. 
 

Access to the external bus should incur 2+2 wait states for each word. GROM is especially slow (ISTR 30 cycles?) Perhaps something is wrong in the wait state generator. 

 

 

Yeah I don’t see the address bus as a problem either since most of the accesses to from shouldn’t involve it at all. It feels like an electrical issue rather than a pure logic issue… too slow a state transitions between Voh/ol, but you might be onto something with the wait state generator. I suppose I could check for that with the scope.

[Geoff Oltmans]

Looking at the schematic, I'm kinda wondering if, like the Vss thing mentioned already, if the problem could be related to the circuitry connected to the resistor network connected to the data bus outputs of the GROMs and GROM port. I don't know what the AC characteristics are of these GROMs. (is there a datasheet for the GROMs?) GROMs must have some sort of open drain drivers in them I guess. I haven't poked around there yet.

 

 

Edited July 14, 2022 by Geoff Oltmans

[+arcadeshopper]

On 7/10/2022 at 5:36 AM, Geoff Oltmans said:

Nope, it’s a black and silver. 

this is a misnomer, some qi and 2.2 consoles are black and silver, always check the motherboard and side port to ensure you aren't dealing with a QI  I don't know if the TI service center was doing this OR it's from motherboard swaps but I've had 3-4 black/silver QI's come through here 

[Geoff Oltmans]

It’s definitely a non-QI board

[AwkwardPotato]

9 hours ago, Geoff Oltmans said:

Looking at the schematic, I'm kinda wondering if, like the Vss thing mentioned already, if the problem could be related to the circuitry connected to the resistor network connected to the data bus outputs of the GROMs and GROM port. I don't know what the AC characteristics are of these GROMs. (is there a datasheet for the GROMs?) GROMs must have some sort of open drain drivers in them I guess. I haven't poked around there yet.

AFAIK the datasheet for the TMC0430 GROMs has never been located. If you want to verify the pull-up/pull-down circuit, probe the collector of Q500. It should swing between 0-5V and follow the DBIN# signal. If the low level is greater than a couple hundred mV, the transistor is bad, or its base resistor is open. If the high level is significantly greater than 5V, one of the diodes/R520/R521 clamping the collector voltage is open.

[Geoff Oltmans]

13 hours ago, AwkwardPotato said:

AFAIK the datasheet for the TMC0430 GROMs has never been located. If you want to verify the pull-up/pull-down circuit, probe the collector of Q500. It should swing between 0-5V and follow the DBIN# signal. If the low level is greater than a couple hundred mV, the transistor is bad, or its base resistor is open. If the high level is significantly greater than 5V, one of the diodes/R520/R521 clamping the collector voltage is open.

Thanks... I measured these yesterday and it appears that Vbe swings from 0-0.7ishV which should be okay, Collector is 0 and clamped to approx 5V. It is a bit messy looking though, not very square, about 200ns.

 

[atrax27407]

You can find a section with the specs from TI GROMs here:  http://www.unige.ch/medecine/nouspikel/ti99/titechpages.htm

[Geoff Oltmans]

Update:

 

I was able to borrow another console from a coworker and using the FinalGROM99 cartridge I have I noticed the same strange behaviors with certain games with that. I used the tests from the github repo for running on-target tests for GROM page accesses from there and found that several of those tests fail with anything plugged into the expansion port on the console, but doesn't if nothing is plugged in. In the other thread in here for the FinalGROM99 Firmware for QI consoles I installed that firmware and my problems went away! So, I have a perfectly functioning console now.

 

Turning my attention back to the PEB floppy disk controller, I eliminated the flex cable interface as being a problem, tested my 32KB RAM expander and that works, the TI RS232 interface card works, and the FDC did not. I was able to probe using Mini Memory from the FinalGROM cartridge the various CRU bits on the FDC, and check the DSR ROMs on the card. The page >4000-4FFF worked, but >5000-5FFF returned all zeros. I probed with an o-scope the chip select lines from the PAL chip on the card, and the lower bank chip select worked, but not the upper one. You could see if wiggle just a little, but it was more of a sag. I suspected that the ROM itself was bad, so I pulled that from the board and probed the chip select line again. This time the line was able to be pulled all the way to 0. I happened to have a blank 2532 EPROM on hand, so I burned that EPROM with the DSR code and installed it in the FDC, and now I'm in business! I was able to initialize a disk successfully, and then save and load a BASIC program.

 

 

Edited October 4, 2022 by Geoff Oltmans

[Geoff Oltmans]

Oh, and the noisy video thing. I think there may be something up with my monitor now. The other console does the same thing with predominately white displays. I guess maybe the white signal is "hotter" than my other machines maybe? I haven't checked the output levels relative to my other machines yet.