D1xx on both XL and XE computers

[Dropcheck]

I've been trying to discover if my suspicions are correct.

 

According to the Sam's Computerfacts for the 130XE the signal D1XX on the ECI port comes from the 74LS138P(U2) pin 14.

 

On Sam's Computerfacts for the 800XL the 74LS138N(U2) chip has a NC on pin 14.

 

Would that be the same signal, just not brought out to external access on the XL line?

[+bob1200xl]

It's the same. 1200XL, 600XL, 800XL, 130XE - all the same.

 

Bob

[ricortes]

Deranged ramblings on my part. I used to put two 74LS138 in some of my 8 bits. You will notice that 2<?> of the enable pins are not used in the schematic. All that was necessary was to bend up one enable pin on the existing '138 and connect it to an address line to make all the hardware enables go to the bottom half of the page i.e. $D100-$D17F. Piggybacking a second '138 on top and routing the same address line to an inverted enable pin gave enables for an extra 8 hardware chips. I think there may have been one or two conflicts with existing chips that did their own decoding, but the remaining usable !E(nables) were still more then a stock power supply could handle. I never came across software that didn't run with with hardware pages divided this way but then I didn't try every piece of software ever written.

Edited October 25, 2016 by ricortes

[Rybags]

The signal exists but isn't provided externally on XL which means the external hw has to decode it. Sort of crazy. IMO the logical path in the first place would have been to provide a copy of the MMU IO pin which external devices could have used to know when an access to $D000-$D7FF was taking place then they could choose to intercept or not by just looking at a few address pins.

[Dropcheck]

The signal exists but isn't provided externally on XL which means the external hw has to decode it. Sort of crazy. IMO the logical path in the first place would have been to provide a copy of the MMU IO pin which external devices could have used to know when an access to $D000-$D7FF was taking place then they could choose to intercept or not by just looking at a few address pins.

 

I'm not sure I'm understanding.

 

Are you saying that simply because there is no physical trace to the external connector on the 800XL that somehow external hardware has to decode it while on the 130XE it doesn't need it too? I would think it would need to be decoded coming from the 130XE as well.

[Rybags]

ECI has the /D1XX line so a device knows without further deduction that an access to that page is occurring.

 

PBI has the address lines only, plus Ras and Cas. To provide the same level of decoding as the /D1XX line, a PBI device has to look at the high 5 address lines to deduce an IO address, then A8-A10 to deduce which page is involved.

 

Though on the other hand, the ECI doesn't have much advantage other than also providing the CCTL line for $D5 page accesses. It's just as handicapped as the PBI e.g. if it wanted to shadow the PIA to provide a banked Ram expansion.

Edited October 26, 2016 by Rybags

[Dropcheck]

Yes, but...... There's a trace that goes directly from U2 pin 14 to the ECI on the 130XE. On the 800XL there isn't a physical trace but the signal is still present on U2 pin 14. Can't a jumper be run from the chip to an unused/reserved pin on the 800XL PBI. Effectively exposing D1xx on a modified PBI. If the signal is the same on both machines, the only difference is one is already exposed on ECI and on the other a jumper needs to be run.

 

At that point the question becomes does external decoding still need to done on both signals?

[Dropcheck]

By extension could not the CCTL and HALT signals also be put on the other two unused/reserved pins of the PBI?

[+tf_hh]

Yes, but...... There's a trace that goes directly from U2 pin 14 to the ECI on the 130XE. On the 800XL there isn't a physical trace but the signal is still present on U2 pin 14. Can't a jumper be run from the chip to an unused/reserved pin on the 800XL PBI. Effectively exposing D1xx on a modified PBI. If the signal is the same on both machines, the only difference is one is already exposed on ECI and on the other a jumper needs to be run.

 

At that point the question becomes does external decoding still need to done on both signals?

 

The main goal is... every PBI expansion which need $D1xx (what is mandatory if the expansion has an own ROM and want to be 100% PBI compliant) can´t count on the existence of a supplementary connected /D100 signal on the PBI. So every designer has to decode the A8...A15 lines byself to generate this signal. Newer expansions use a CPLD or a GAL, the older ones a TTL 74 chip (74xx682 as a 8 bit comparator) or a bunch of AND/OR chips.

 

Jurgen

[Dropcheck]

 

The main goal is... every PBI expansion which need $D1xx (what is mandatory if the expansion has an own ROM and want to be 100% PBI compliant) can´t count on the existence of a supplementary connected /D100 signal on the PBI. So every designer has to decode the A8...A15 lines byself to generate this signal. Newer expansions use a CPLD or a GAL, the older ones a TTL 74 chip (74xx682 as a 8 bit comparator) or a bunch of AND/OR chips.

 

Jurgen

 

I disagree. I believe that the developer has to assume and require that the user will modify their PBI. Every time a developer creates a new internal mod for a disk drive or the computer, there are anywhere from 1 to many extra connections that have to be made. Traces cut, parts removed etc. That's the nature of the beast. You always want to make the fewest, and the neatest you can, but there will be changes done.

 

Assuming that the signals D1xx, HALT and CCTL were brought out to live on the PBI reserved lines, what else would be needed? Would additional decoding need to be done? Or could these signals be fed to various cards internal to the connected black box to be acted on by the cards accordingly?

[+bob1200xl]

For one, you need to decode $D1FF and $D800-$DFFF for the PBI. $D1FF selects the PBI device and $D800-$DFFF is the ROM space for that device. Hardware registers also need decoding in the $D1xx page.

 

Of the old PBI devices, Supra, MIO, and Black Box, none required you to open your computer. With modern CPLDs, you can run the whole device with just a few chips.

 

You're going to lose a lot of folks if you make them modify their systems - is it worth it?

 

Bob

[Dropcheck]

Okay one last time:

 

Does the 130XE ECI D1xx signal need to be decoded or can it be used directly?

[Dropcheck]

You're going to lose a lot of folks if you make them modify their systems - is it worth it?

 

Bob

But that's just it....... I don't know of a internal mod made in the last ten/twenty years that hasn't required some kind of extra connection inside the computer. That hasn't stopped people from buying these mods or installing them. I don't think I will lose any customers.

Edited October 26, 2016 by Dropcheck

[+Stephen]

But that's just it....... I don't know of a internal mod made in the last ten/twenty years that hasn't required some kind of extra connection inside the computer. That hasn't stopped people from buying these mods or installing them. I don't think I will lose any customers.

The difference is INTERNAL mods vs an EXTERNAL plug in device. That's why some people will use a 1064 on their 600XL, rather than opening the machine and desoldering.

 

I have no issue modifying machines, which is why I keep about a dozen Ataris. But if I am using a PBI or ECI device, all I want to do is plug it into the machine, stock or otherwise.

[_The Doctor__]

Make an option like i eluded to in your other thread

[ricortes]

Okay one last time:

 

Does the 130XE ECI D1xx signal need to be decoded or can it be used directly?

Directly. Consider it is possible to make everyone happy. Just say "A some future date, I may make an 800xl to 130xe adapter." It would only take two chips, 74LS688 & 74LS138. Tell people they can either run a jumper from the existing MB or buy your adapter for $79. That should get you the desired result.

[Dropcheck]

The difference is INTERNAL mods vs an EXTERNAL plug in device. That's why some people will use a 1064 on their 600XL, rather than opening the machine and desoldering.

 

I have no issue modifying machines, which is why I keep about a dozen Ataris. But if I am using a PBI or ECI device, all I want to do is plug it into the machine, stock or otherwise.

Aren't there several devices that plug into the PBI external and still require an additional wires to be connected? Sys-check comes to mind. Some additional devices goe so far as to 'require' an OS mod etc.

 

I agree that some people don't want to do mods internal or external to their computers. They either have someone else do it for them or they don't buy mods. But what is new?

 

I don't see that as a reason not to do the project.

[+Stephen]

Aren't there several devices that plug into the PBI external and still require an additional wires to be connected? Sys-check comes to mind. Some additional devices goe so far as to 'require' an OS mod etc.

 

I agree that some people don't want to do mods internal or external to their computers. They either have someone else do it for them or they don't buy mods. But what is new?

 

I don't see that as a reason not to do the project.

Oh - please don't see this as an objection to the project (because I am interested and will be buying). Just pointing out that I think many users feel external plug in devices should just be plug&play, while internal mods are "anything goes". I'm also in the camp of loving to mod my machines, so no worries there.

[Dropcheck]

If Atari had continued for any number of years you would have continued to see different models come out with different external connections. They would mirror what happened with the differences between the 600/800XL and 65/130XE external connections. Sad to say the PBI is not standard, nor is the cart/ECI. They are not interchangeable and only occurred on 1 model apiece, so neither can be considered a standard. Even if you insist on saying one or the other is a standard, I don't know of any standard that doesn't evolve over time and inherently change in form and/or content or disappear.

The battle between Betamax and VHS comes to mind. Completely incompatible formats. Betamax came first and was considered superior to the competing VHS. But as VHS evolved and became better, longer recording time and cheaper, Betamax began to lose market share. Eventually VHS completely supplanted Betamax. It wasn't too much longer that VHS itself was supplanted by DVD and now the same thing is happening to DVD with Blu-Ray and streaming video. Each time a new standard was formed, the best features of both initial formats were combined.

I'm surprised at all the negativeness here at this idea. I'm not expecting rah rah go girl. But really...... Are we that afraid of change? So far the objections seem to fall into these categories.

(It'll be costly. You would have to decode this or that. CPLD do it better cheaper)

 

Duh... all of that is debatable and/or solvable and compared to what... for what? That's not a reason not to do something, those are problems to solve in the steps necessary to bring something to market. I'm not manipulating signals at this time, just physically routing them to a common output format.

(XZY product doesn't play well with TXT product)

 

So....? What does that have to do with standardizing the PBI/ECI bus connector? That's between XZY and TXT products. They didn't play nice together before.

(I might have to connect three wires internal to my computer)

 

The only way to avoid some modification is to redesign the computer from the ground up to include any and all current mods. But isn't that the biggest modification of all and the most costly. Simply because you now are rebuying not only the base computer but all mods too. And then you would have something come along that needed a internal signal that wasn't on the external connector. The key is the ratio between mod and gain. For instance 3 internal jumper connections, no external case mods to gain the most important signals coming from the 130XE ECI connector on the PBI connector.

(There are PBI devices out there that don't follow the standard)

 

Not sure what standard referred to since there's no physical and electrical backward compatibility between computer model bus connections anyway. Either bring these pbi devices into the new standard by modifying them or incorporating their custom connections/requirements or cut them loose. Worse case is that those who own these devices will need a clean/virgin 800XL or 130XE to use them, but I suspect that's the case for the most part anyway now. Best case is that these devices may run better and/or be more compatible with other devices.

(Existing multifunctional PBI devices: XYZ product was designed and built in 1983 for the 800XL PBI. It is the cat's meow, a wizzbang product. It doesn't matter that not many were produced and even fewer have survived the years and absolutely none have been upgraded to take advantage of the new technologies out there, we can't change the PBI connector because they would stop working.)

 

If they were properly designed then no changes I purpose to the 800XL PBI connection will affect them. Those pins shouldn’t be connected to anything on the XYZ product. If they weren't designed properly, then they are non-standard devices as well and the previous statement is valid.

 

---------

What I am trying to do is to develop a standardize PBI/ECI adapter for both machines. The adapter will physically route the signals to a common bus connection, not manipulate the signal. To that end I am collecting information on what is needed to allow the connection of the greatest number of existing devices or the modification of those devices. It is certainly doable. It may require some internal 800XL jumper connections. Perhaps on the 130XE too. There might even be connections that would need external case modifications, but at this time I'm not sure.

Shouldn't be that expensive either whether you have an 800XL or 130XE. If you are fortunate to have a 600XL that has a PBI connector than you shouldn't have that much cost either. If you have a 1200XL, 65XE or a European model w/o the PBI/ECI, then things get much more costly and hard to do, but then what exactly has changed. Most likely you weren’t going to put the PBI/ECI on these computers anyway.

Eventually I plan to develop a 1090XL like shell for multiple devices to coexist within, a physical framework to support them. That is where the manipulation of the signals will take place. I want to foster new development for years to come. Or at least what new development there might be. It certainly helps if you have a standardize card slot to pick the signals off of including power and ground without worrying how to run additional wires off the board or provide a container for the device. New developers might even finally be able to use the Atari provided method to allow multiple devices to co-exist on the same bus and auto configure.

But to get back to the first step. Would someone buy this standardized PBI/ECI connector? It depends. It may take the finished 1090XL shell to motivate people, it might even take the new wizzbang card device to motivate them to buy. I'm not even sure I'll release it until the 1090XL shell is done. But you have to start someplace. You can’t finish something if you don’t start.

[flashjazzcat]

I guess ECI and PBI are considered standards because it's twenty-five years since an Atari 8-bit computer was produced and the vast majority of them have an ECI or a PBI connector on the back. Certainly standards develop over time, and we're at version three of USB now, but they're still making computers with USB ports on them, just as they're still making DVDs and Blu-Rays.

 

With a couple of notable exceptions, I can't think of many PBI/ECI peripherals which can't already co-exist with one another perfectly well, providing sufficient edge connectors are made available. The whole New Device architecture is geared towards things playing nicely together, and it's only when - for example - one device decides to respond to all eight bits of $D1FF (assuming it's the only device on the bus) that things grind to a halt.

 

I was unsure, for the entire duration of the thread, exactly what new product was under discussion here, so I'm pleased you included a recap. In any case, some kind of homogenised parallel interface is probably a good idea, and I think the notion of populating the n/c pins of the PBI edge connector with missing ECI signals has been bandied about in the past. But are there enough spare pins? And I assume the assertion is that existing devices which happen to tie n/c pins to GND for whatever reason will be deemed "non-standard" (I'm not sure if any do, but it's just a thought that occurs to me)?

 

Anyway: great - go for it. For some reason I thought you were designing a PBI/ECI adapter with a bunch of lovely cart and ECI headers so we could plug in loads of existing stuff all at the same time. I'd forgotten about the 1090 re-imaging, or perhaps I'd confused the two.

[+mytek]

If they were properly designed then no changes I purpose to the 800XL PBI connection will affect them. Those pins shouldn’t be connected to anything on the XYZ product. If they weren't designed properly, then they are non-standard devices as well and the previous statement is valid.

This is the key to minimizing any objections, since in most cases this new PBI/ECI adapter shouldn't cause any problems with legacy hardware.

 

 

Eventually I plan to develop a 1090XL like shell for multiple devices to coexist within, a physical framework to support them. That is where the manipulation of the signals will take place. I want to foster new development for years to come. Or at least what new development there might be. It certainly helps if you have a standardize card slot to pick the signals off of including power and ground without worrying how to run additional wires off the board or provide a container for the device. New developers might even finally be able to use the Atari provided method to allow multiple devices to co-exist on the same bus and auto configure.

I was thinking about this, and what popped into my head was maybe doing something that would not just be a 1090 style expansion box. Because why limit yourself to Atari's vision of how things should have been?

 

So here are my thoughts... make an expansion box that also incorporates the A8 motherboard as an integral part. This could be done in a number of ways... a) have a slot for the motherboard to plug into via it's PBI right next to the other PBI related expansion slots, b) Incorporate the circuitry to re-create the A8 inclusive to the expansion slot back plane using standard A8 IC's, or c) same as option 'b' but this time do it with an FPGA instead for minimization of the footprint.

 

Just thinking outside of the 'Box'

 

- Michael

Edited October 27, 2016 by mytekcontrols

[Timothy Kline]

But to get back to the first step. Would someone buy this standardized PBI/ECI connector? It depends. It may take the finished 1090XL shell to motivate people, it might even take the new wizzbang card device to motivate them to buy. I'm not even sure I'll release it until the 1090XL shell is done. But you have to start someplace. You can’t finish something if you don’t start.

 

Good morning, Dropcheck!

 

Of course I can't speak for anyone else, but *I* would definitely purchase one. Even so, it's still a fair question, since you're not likely to invest hours of labor on something that only one person would buy. Nor should you have to.

 

As for the 1090 clone, I'm likewise very interested and intrigued, especially if the device would allow Atari developers to put together mod cards that took full advantage of the parallel computing access and allow for alternatives to the cartridge-based solutions we have right now for:

 

1. Bluetooth

2. USB

3. serial and parallel ports to replace the need for an 850.

4. SD and/or IDE drive access

5. memory expansion after the order of the 1064 module WITH battery backup during power-off (similar to BIOS batteries in PC?)

 

Ideally, the last would be available with the release of the 1090 clone to provide a non-invasive means of extending the Atari 8-bit's RAM, especially for the XE, since who wants to solder one of those mainboards?!

 

Factoring in what FlashCatJazz pointed out in his post:

 

The whole New Device architecture is geared towards things playing nicely together, and it's only when - for example - one device decides to respond to all eight bits of $D1FF (assuming it's the only device on the bus) that things grind to a halt.

 

 

...this would allow the primary feature of extended RAM, and presumably 4 expansion edge slots (if they used 2 bits each), which would be more than sufficient given the fact that the end user would be able to mix-and-match the features they want for their personal system. If anything, traffic could feasibly be handled similar to how the PCIe bus on a PC works, where the bandwidth on the bus is handled according to what is plugged into which slot, x16, x8, x1-- each slot affording a card a certain number of bits to access. At least in my way of thinking.

 

It seems like someone could even develop a card expansion to bring stereo sound to the Atari-8bit in addition to the built-in POKEY sound.

 

Anyhoo, if there's anything I can do to help encourage you forward and help support your efforts, please let me know.

 

--Tim

[+mytek]

Dropcheck are you only planning on only making the board, or will it also include an enclosure for the reimaged 1090 expansion board?

 

I'm thinking just the board is what you have in mind, since even 3D printing would likely be out of the question for something in the size required. So here's a thought (there I go, thinking again ), how about making it the size of a mini ATX motherboard so that it could be used in something like this...

 

 

- Michael

[Bikerbob]

Very cool thread, and Dropcheck I am always interested in Hardware for the A8, this is my primary interest, so I will be following this closely.

 

Moving my A8 (or one of them) to a ATX box has been a dream of mine for a while. use the transkey to use a STD keyboard and mouse, having something that can house expansions.. or even give me room to build platforms above the mb so that my U1mb could sit.. a card reader could be housed etc.. would be awesome. I would rather custom house my mod computer than hack the original cases.

 

James

[Mathy]

Hello guys

 

We'll need an Atari Logo to cover the name on the mini ATX box. And the RAL number for the colors Atari used on the XL and XE.

 

Sincerely

 

Mathy