PGram+ Card --- Loading up the DSR

[Gary from OPA]

In my on going battle to get the PGRAM+ card to work perfectly without any glitching, I been studying the schematic, and tracing out the logic.

 

There is 4 things I going to try

 

#1 - First thing I going to try is to ground the UNUSED logic gates, after studying the schematic and looking at the PCBoard and tracing out the wires, I have found out the following:

 

U3 - Second '74 Flip-Flop is NOT used, it just left FLOATING the inputs.

U4 - 2 buffers on the '367, 3A and 4A are NOT used, it is just left FLOATING the inputs.

U7 - Second '139 2-line to 4 line, is NOT USED, it is just left FLOATING the inputs.

U17 - 2 OR Gates '32 (3A/3B & 4A/4B) are NOT ued, it is just left FLOATING the inputs.

 

Now, as many of you know, it is always bad to left unused logic floating, it effects the timing of the other parts of the chip you are using, and causes other issues, and in this case the U4 (367) and the U7 (139) is critical for timing to enable the groms and enable the firehose data bus, which could be why I am having problems when reading or writing a series of bytes in a row when in two peb setup.

 

#2 - Second thing to try, if the above doesn't help, I am going to order all new 74F logic chips for the PGRAM from mouser, these are the fastest versions, even faster than HCT chips, and much quicker than LS

 

#3 - Third thing to try, replacing the flexcable card and console firehose end with all new HCT '244 and '245 chips which are tiny bit faster and also have more driving power.

 

#4 - Last thing to try, screw all the small 74LS (now F) logic on PGRAM, and redesign it with a single chip to do all the complex decode/banking logic to get the fastest speed possible.

 

Tomorrow, I going to ground the unused inputs I found in step #1, and report back the changes or improvements. -- Before I do that, I want to see if I can make usage of some of the unused existing logic to expand on the features of the PGRAM, one issue that needs to be solved ASAP is the fact it only currently supports 16K of ROM/RAM at the >6000 cartridge space, it needs to be at least 32K to be able to run most modern cartridges like the new RXB2024 -- With the unused existing logic, I should be able to increase the number of banks, we will see.

 

Hopefully, #1 solves my issue, if not time to submit my mouser order and wait for those 74F parts and see if that does the trick.

[humeur]

The fastest are the AS  on the other hand the consumption is much higher to do with the voltage regulators if they heat up more.

 

Les plus rapide sont les AS  par contre la consommation est bien plus élevé a voir avec les régulateurs de tension s'ils chauffe plus.

 

Edited August 7 by humeur

[Gary from OPA]

39 minutes ago, humeur said:

The fastest are the AS  on the other hand the consumption is much higher to do with the voltage regulators if they heat up more.

 

Les plus rapide sont les AS  par contre la consommation est bien plus élevé a voir avec les régulateurs de tension s'ils chauffe plus.

 

It is tough to get new stock of AS, even. ALS as well not to mention the original S and SC versions.

 

Places like Mouser only stock LS. HCT and thankfully most of the 74 chips in the F version which is not bad in terms of speed originally invented by Fairchild but you can get TI licensed versions as well now.

 

We will see how it works out, today i am going to ground the unused 74ls inputs to see if that makes the timing better and reduces data errors, if not then I am going to order the F chips and hopefully they solves the problem.

[humeur]

12 minutes ago, Gary from OPA said:

It is tough to get new stock of AS, even. ALS as well not to mention the original S and SC versions.

 

Places like Mouser only stock LS. HCT and thankfully most of the 74 chips in the F version which is not bad in terms of speed originally invented by Fairchild but you can get TI licensed versions as well now.

 

We will see how it works out, today i am going to ground the unused 74ls inputs to see if that makes the timing better and reduces data errors, if not then I am going to order the F chips and hopefully they solves the problem.

Are you the only one having problems with this card??

 

Tu est le seul a avoir des soucis avec cette carte??

[Gary from OPA]

48 minutes ago, humeur said:

Are you the only one having problems with this card??

 

Tu est le seul a avoir des soucis avec cette carte??

There is not many out there with one any more being used on a daily basis like me.

 

But I trying to use it in dual PEB setup, so I ran into some timing issues with the original design, it works fine in single PEB setup.

 

The pgram has always been tight on timing, even on a one peb setup, it was found it be best to be used right next to the flexcable card in slot 2, and without any sidecars either, and those with upgraded console like 4mhz crystal ran into issues as well.

 

The amount of complex decode 74 logic on the P-GRAM is crazy which makes the timing more critical, so anything to speed up the logic helps.

 

Switching the original 100ns/120ns static rams to 55ns versions helped a bit, but not enough, the problem now is basically how slow it is in enabling the RDBENA line which tells the firehose to enable it's 245 chip to accept the data coming in or going to the 4a console.

 

The chain of 74 logic to make the RDBENA signal is huge on most cards it's only check if DSR space and CRU base matched, in the pgram case it has to match that, along with cartridge space and grom space (but only if not console GROMs) not to mention checking if 8600 onboard clock access as well.

Edited August 7 by Gary from OPA

[humeur]

I don't know if I can help, I'll look at the diagram whenever a brilliant idea comes to me.   😂

 

Je ne sais pas si je peut aider je vais regarder le schéma dé fois qu’une idée de génie m 'arrive .

[Gary from OPA]

4 minutes ago, humeur said:

I don't know if I can help, I'll look at the diagram whenever a brilliant idea comes to me.   😂

 

Je ne sais pas si je peut aider je vais regarder le schéma dé fois qu’une idée de génie m 'arrive .

Thanks. I was hoping to do some more research work today. But my left leg and ankle is killing me and the AC is broken as well. So I will get back to it on Thursday here and report back my findings, with first grounding out the unused logic gates, that should make a difference. If not, it will take about a week for me to get the 74F chips in from Mouser.

[humeur]

I do have a card but never use it given the number of components, does the voltage regulator heat up a lot, there could be a rapid drop in voltage which causes certain signals to have a threshold that is too low.

 

J'ai bien une carte mais  jamais utiliser au vue du nombre de composant le régulateur de tension chauffe t'il beaucoup, il pourrais se produire un baisse de tension rapide qui fait que certains signaux ont un seuil trop bas.

[Gary from OPA]

5 minutes ago, humeur said:

I do have a card but never use it given the number of components, does the voltage regulator heat up a lot, there could be a rapid drop in voltage which causes certain signals to have a threshold that is too low.

 

J'ai bien une carte mais  jamais utiliser au vue du nombre de composant le régulateur de tension chauffe t'il beaucoup, il pourrais se produire un baisse de tension rapide qui fait que certains signaux ont un seuil trop bas.

It doesn't seem to get too hot. I will test the voltages.

 

But it works perfectly fine in a single PEB setup, so if there was an issue with voltages it should have issues when being used in a single PEB setup, which it doesn't. It's only when two firehoses are running with it, and it's not other cards as I removed all other cards as well, and I tried multiple firehoses and consoles.

 

So it's basically just a timing issue, which some P-GRAM had anyway from day 1 even in a single PEB, others have found in the past it works best in slot 2 right next to the flexcable card in slot 1 and without any sidecars attached either on the console or upgrades like faster 4mhz crystal.

[+Ksarul]

Unicorn Electronics has quite a few AS and ALS parts in stock. . .they also have a lot of F parts.

[Gary from OPA]

31 minutes ago, Ksarul said:

Unicorn Electronics has quite a few AS and ALS parts in stock. . .they also have a lot of F parts.

Thanks I might use them instead of Mouser as their minimum order is only $30 instead of $100.

[humeur]

15 hours ago, Gary from OPA said:

So it's basically just a timing issue, which some P-GRAM had anyway from day 1 even in a single PEB, others have found in the past it works best in slot 2 right next to the flexcable card in slot 1 and without any sidecars attached either on the console or upgrades like faster 4mhz crystal.

I think it's more chance or a feeling of the user for the slot to use

 

je pense que c'est plus le hasard ou un sentiment de l'utilisateur pour le slot utiliser

[+dhe]

@mizapf any chance of the pgram making its way in to MAME?

[Gary from OPA]

13 hours ago, humeur said:

I think it's more chance or a feeling of the user for the slot to use

 

je pense que c'est plus le hasard ou un sentiment de l'utilisateur pour le slot utiliser

Not really, old MicroPendium articles and newsletters from users with the P-GRAM when it was new back in the day, mentioned that they had errors if it was located away from the flexcable card in the PEB that it worked best right next to it.

[humeur]

the printed circuit would be enough of the 15cm pebox to disrupt the operation and well this card is sensitive

I always come back to the consumption of the card for this poor regulator. it has a load which already consumes quite a bit just two 40ma max LEDs

 

Did you do a charging test without the backup batteries just to see if with or without there is a difference.

 

 

 le circuit imprimer serait suffisant de la pebox de 15cm pour perturber le fonctionnement et bien elle est sensible cette carte.

j'en reviens toujours à la consommation de la carte pour ce pauvre régulateur. il a de la charge  qui consomme pas mal déjà juste deux led 40ma max

A tu fait un essai de chargement sans les batteries de sauvegarde juste pour voir si avec ou sans il y a une différence.

 

 

 

[Gary from OPA]

1 hour ago, humeur said:

the printed circuit would be enough of the 15cm pebox to disrupt the operation and well this card is sensitive

I always come back to the consumption of the card for this poor regulator. it has a load which already consumes quite a bit just two 40ma max LEDs

 

Did you do a charging test without the backup batteries just to see if with or without there is a difference.

 

 

 le circuit imprimer serait suffisant de la pebox de 15cm pour perturber le fonctionnement et bien elle est sensible cette carte.

j'en reviens toujours à la consommation de la carte pour ce pauvre régulateur. il a de la charge  qui consomme pas mal déjà juste deux led 40ma max

A tu fait un essai de chargement sans les batteries de sauvegarde juste pour voir si avec ou sans il y a une différence.

 

 

 

It doesn't charge the batteries, uses just a large coin sized lithium for the battery. Earlier designs had issues with it draining faster than it should, but no problems like that on this card, do it's not putting any extra load on PEB.

[+mizapf]

13 hours ago, dhe said:

@mizapf any chance of the pgram making its way in to MAME?

 

Maybe have a look here... https://www.ninermame.org/info/scope

And https://www.ninermame.org/info/releases tells you that it is available since May 2020.

[humeur]

Sorry, I didn't pay attention to the battery.

The power supply part is a hack with the addition of a diode in the GND part of the regulator which when empty increases the output voltage, however the internal protection system no longer works.

 

 

Pardon je n'avais pas fait attention pour la pile.

La partie alimentation est une bidouille avec l'ajout d'une diode dans la partie GND du régulateur ce qui a vide augmente la tension de sortie, par contre le système de protection interne ne fonctionne plus.

[Gary from OPA]

1 hour ago, mizapf said:

 

Maybe have a look here... https://www.ninermame.org/info/scope

And https://www.ninermame.org/info/releases tells you that it is available since May 2020.

That is good to know. How would one enable it in the config. As the default batch files setup by the script only config the HSGPL card.

[Gary from OPA]

3 minutes ago, humeur said:

Sorry, I didn't pay attention to the battery.

The power supply part is a hack with the addition of a diode in the GND part of the regulator which when empty increases the output voltage, however the internal protection system no longer works.

 

 

Pardon je n'avais pas fait attention pour la pile.

La partie alimentation est une bidouille avec l'ajout d'une diode dans la partie GND du régulateur ce qui a vide augmente la tension de sortie, par contre le système de protection interne ne fonctionne plus.

I will look at that part at some point. But I installed a brand new strong battery, and I don't test it without the battery either. It stores the memory fine even after upgrading the two ram chips to 128k ones from 32k.

[humeur]

I measured the empty intensity around 300ma without use, the TTL voltages 4.86 Volts

 

J'ai mesurer à vide l’intensité environ 300ma sans utilisation  les tensions TTL  4,86 Volts

[+mizapf]

2 hours ago, Gary from OPA said:

That is good to know. How would one enable it in the config. As the default batch files setup by the script only config the HSGPL card.

Usually by "plugging" it into the PEB:

 

... -ioport:peb:slot3 pgram ...

 

(or any other slot). You can choose between PGRAM 72K or PGRAM+ 192K in the configuration menu. There are two DIP switches: CRU base selection, and card turn on/off. There is a disk image containing the "PGRAM Utilities"; I'll attach it here.

 

Edit: The PGRAM(+) is currently only available for the TI-99/4A and the TI-99/4A with EVPC.

Pgram_Utilities.dsk

[Gary from OPA]

1 hour ago, mizapf said:

Usually by "plugging" it into the PEB:

 

... -ioport:peb:slot3 pgram ...

 

(or any other slot). You can choose between PGRAM 72K or PGRAM+ 192K in the configuration menu. There are two DIP switches: CRU base selection, and card turn on/off. There is a disk image containing the "PGRAM Utilities"; I'll attach it here.

 

Edit: The PGRAM(+) is currently only available for the TI-99/4A and the TI-99/4A with EVPC.

Pgram_Utilities.dsk 180 kB · 2 downloads

Two questions:

 

Does mame remember what is loaded into the 192k, once I load the 4 modules in via the P-GRAM loader and shutdown the emulator when I restart is it all there still?

 

Finally, when I finish updating the DSR for it and expanding the memory to allow full 16 banks of gram instead of only 4 on my real life physical pgram, how hard is it to update the emulation to support the minor changes?

 

 

 

[+mizapf]

2 hours ago, Gary from OPA said:

Does mame remember what is loaded into the 192k, once I load the 4 modules in via the P-GRAM loader and shutdown the emulator when I restart is it all there still?

 

Yes, the RAM is buffered by a battery, so I implemented it as buffered in MAME as well. The contents are stored in <yourmamedir>/nvram/ti99_4a/ioport_peb_slotX_pgram_<subcomponent>

 

where slotX is the slot in which you plugged the PGRAM.

 

2 hours ago, Gary from OPA said:

Finally, when I finish updating the DSR for it and expanding the memory to allow full 16 banks of gram instead of only 4 on my real life physical pgram, how hard is it to update the emulation to support the minor changes?

Does this involve a change of the hardware?

[Gary from OPA]

Just now, mizapf said:

Does this involve a change of the hardware?

It mainly replacing the left 128k with a 512k, and middle 32k with a 128k. That part just expanding the number of banks to 16 from 4, no hardware changes just a new dsr.

 

But second modification I will be doing will be changing it from a fixed 16k ROM 6000 space to 32k of banking that is easy to do as will using two extra cru bits currently unused and on the physical card the unused OR gate and unused 74 flip-flop. It would also due to the way the hardware is designed expand the DSR space to matching 32k as well from it's current 16k, so the right most ram socket would be a stacked set of two 32k chips.

 

Lastly, I am looking at doing a third modification that would allow for multiple ROM banks to go along with the grom base. On the physical card that would require adding a new 512k chip on top of the original 32k chip instead of a second 32k and adding a latch to store the current grom base.

 

Once I complete the new DSR and hardware changes then I going to layout a new board as well, work with Jim on it if need be as he been looking at re releasing the pgram card as new hardware item.