ECS Program Expander was planned... in brown!

[Bamse]

Now I can't wait to open mine up and it hasn't even arrived yet ?

[Intymike]

And my last one, highest no. on the outside, but lowest on the inside...

 

Grey, with connector, glued trap door.

 

Outside no. P3084066

 

Inside no. 2 L00001900

 

Chips from 83 08 (AY) until 83 35 (U52)

 

 

 

 

[+Lathe26]

Thanks! That is all fantastic information.

 

Here's a quick table I threw together. Let me know what I missed. Not seeing a clear pattern here. More data will help greatly.

 

[+intvsteve]

ooohhhh... gotta go rummage through my ECSs now...

[intvnut]

I'll get my external serial number to you later when I get home.

[+Lathe26]

Is the 00509 the internal number? I didn't find another one inside.

 

 

I'm ok with assuming it is the serial number. In looking at my 4 ECSs which all have stickers, none of them had printing close to what yours looks like. Also, the 1 with a hand-written sticker uses a 5 digit number that starts with several zeros.

[Intymike]

I'm ok with assuming it is the serial number. In looking at my 4 ECSs which all have stickers, none of them had printing close to what yours looks like. Also, the 1 with a hand-written sticker uses a 5 digit number that starts with several zeros.

 

 

One other of mine has at this place some kind of Quality Proof stamp.

 

And I pulled the board of the first one out and there was no sticker at the under side like the others.

Edited March 8, 2019 by Intymike

[intvnut]

OK, the ECS I posted about earlier is P3 070913.

 

 

I will need to collect my other ECSs from storage for more data points.

[intvnut]

FWIW, I looked up the FCC record for BSU 9RD4182. It gives an issue date of 6-17-1983.

 

This seems consistent with most of the date codes we've seen so far indicating a late summer release for the ECS.

 

Week 34 ends just before my birthday, and week 35 starts with my birthday (Aug 29) in 1983.

 

8-29-83 is special for me for a different reason: That's the year I got my first computer (TI-99/4A). I think we got the ECS that Christmas. My brother and I both got Sharp Linytron TVs that year as well.

 

Too bad the big crash happened.

Edited March 9, 2019 by intvnut

[intvnut]

Here's a different question: How likely is it that the ECS and Intellivoice lost their expansion connectors at about the same time? When Mack Morris came in, a bunch of expansion peripheral projects apparently got cancelled, and that was mid-year 1983.

 

It seems pretty likely to me, IMHO. The counter proof would be to find Intellivoices without expansion connectors that were issued much before the ECSes appeared.

[intvnut]

OK, so I have to ask, does anyone else have a joker ECS with slotted screws instead of Phillips?

 

 

In any case, I happened to have P3 081145 in my closet. It has the connector, and it has internal serial number L00018628.

 

 

[intvnut]

Oh, and that last one I posted also has its expansion slot glued.

[intvnut]

So here's an even different observation, theory, and/or question: What are the odds that the units weren't built all in one go, but rather built in stages?

 

Related to this: What range of date codes do folks have on U52? And does anyone have a non-socketed U52?

 

Most of the date codes for the ECS components are fairly early in 1983. The AY-3-8917 is consistently in weeks 6 through 8. The MMI PAL is around week 15. The RAM and TTL bits don't seem to have date codes most of the time.

 

The ROM (U52), though, seems to be the long pole. It's also the only socketed IC on the board.

 

So, what if a bunch of ECS boards were manufactured, waiting for FCC approval (mid June) and final production ROMs (late August)? And, what if somewhere in that build-up, Mack Morris came in (late July) and brought the ban-hammer down on the Program Expander?

 

That would explain why we have a bunch of units with glued-down expansion doors, despite having the connector, and then later units without the connector. It would also explain why the date codes have such a wide skew, even for parts that are specific to the ECS (such as the MMI PAL chip). It might also explain anomalies between the numbers on the boards vs. the serial numbers on the outsides of the unit.

Edited March 9, 2019 by intvnut

[Intymike]

That's what I have in mind too.

The built the stuff in early summer and put it on shelf. Maybe they had to wait for the finished or altered software. In between came the order that the expansion will be no longer a thing (maybe for that the rom had to be redone).

And they finished the ECS's after the rom arrived, mixing all the boards they had on shelf. Even mixing the US and international ones as they are not different.

[intvnut]

That's what I have in mind too.

The built the stuff in early summer and put it on shelf. Maybe they had to wait for the finished or altered software. In between came the order that the expansion will be no longer a thing (maybe for that the rom had to be redone).

And they finished the ECS's after the rom arrived, mixing all the boards they had on shelf.

 

Apparently the departure of Stav Prodromou and Josh Denham followed by the arrival of Mack Morris is the punctuating event. Mack Morris was hired July 12th, and nearly everyone in HW development was laid off August 4th.

 

(EDIT: Stav's name is misspelled on the Intellivision Lives site.)

 

 

 

By the time ECS was released, however, the focus at Mattel Electronics had shifted. After the June 1983 CES in Chicago, Josh Denham and Stav Prodomou, Mattel Electronics' President and Senior Operations VP, resigned. Josh and Stav had been blamed for pushing the company too far into hardware production; hundreds of millions of dollars had been tied up in the development, beyond the original Intellivision, of the Keyboard Component, Intellivoice, Intellivision II, the System Changer, ECS, Aquarius (and peripherals), Intellivision III and the top-secret Intellivision IV. On July 12, 1983, Josh was replaced by Mack Morris, a marketing man famous for putting the blue dot on Breath Savers mints. (The gimmick or hook that set a game apart quickly became known as its "blue dot" around Mattel.) Under Mack Morris, the emphasis was put almost entirely on software (nearly everyone related to hardware development was laid off on August 4, 1983). The ECS received very little marketing push and further game development for it dropped to almost nothing.

 

This fits with the ranges of date codes we see to some extent. The ROM date codes are mid-to-late August, after the hammer came down.

 

Amusingly, one of the ECSes I opened up indeed had a blue dot on the RF shield. I didn't add it. I wonder if someone with a perverse sense of humor added it on Mack's behalf in manufacturing?

 

 

And on this point:

 

Even mixing the US and international ones as they are not different.

 

It turns out the brown and grey ECSes are slightly different. I had already suspected they were different, since I analyzed some tapes from an Italian ECS club and wondered why they were at the same baud rate as NTSC, despite both ECS units deriving their clock from the CPU clock on the cartridge port. (3.579545MHz on NTSC units, and 4MHz on European units.)

 

Lathe26 has been digging into this rather heavily, so I don't want to take any wind out of his sails. He can give more details.

Edited March 9, 2019 by intvnut

[+Lathe26]

That's an interesting theory, @intvnut, and it makes a lot of sense that the new management killed the ECS expansion port.

 

Here's some data that might help or hurt theory (not sure which). I've created BOM lists for 3 of my ECSs, including recording what the labels say. Looking at the earliest, 2nd latest, and latest data codes, here's what I found:

• P3 042331 (white) - earliest is 8220 (uber capacitor C5), latest is 8330 on 2 parts (volume potentiometer R6 and the ROM U52)
• P3 040814 (white) - earliest is 8303 (uber capacitor C5), 2nd latest is 8324 (MMI chip U42), latest is 8330 (ROM U52). Note that the potentiometer is unlabeled in this unit.
• P3 06820 (brown) - earliest is 8224 (uber capacitor C5), 2nd latest is 8332 (volume potentiometer R6), latest is 8335 (ROM U52)

As for the Intellivoice Stacking Connector (had a typo where I called it the Stanking Connector ), I'm not so sure that new management killed it off. I did a quick scan of opened up Intellivoices in the internet and found a tiny handful. Ones with chip dates around 8228 had the connector while those with chip dates around 8246 did not. This was quick and dirty research so I expect someone will point out a research flaw. We should get more Intellivoices opened up. If it turns out that this is marginally correct, then my theory is the connector was killed off for one of a few possible reasons:

1. It could be that that the ECS was internally revealed to the rest of Mattel Electronics sometime around August-October 1982 (date based off the chip codes). The ECS started as a secret project away from most of the Intellivision part of the company. An early internal reveal was necessary so that the regular Intellivision folks could start writing games for it prior to the ECS's release. This had the side effect of killing off the Keyboard Component.
2. The Keyboard Component was killed off before the ECS was internally revealed.
3. The remote controller add-on for the Intellivoice was independently killed before the ECS was internally revealed or the Keyboard Component was killed off.

The original intention for the Stacking Connector was to override the built-in controllers and replace them with wireless remote control controllers. My understanding is that of the overriding circuitry was built into the Keyboard Component (mainly with the KC's reverse buffers, if I'm remembering the name right). While other kinds of hardware could theoretically be plugged into the Intellivoice, I'm not aware of any other plans Mattel had for the Stacking Connector.

[intvnut]

Supposedly, the internal pitch was that the Keyboard Component and LUCKY (the ECS) were supposed to coexist in the marketplace. That internal story was supposedly to keep Dave "Papa Intellivision" Chandler from using his political clout from getting Mattel to cancel the ECS. You can see in some of the later memory maps (CCF_10232011_00026.pdf) that both the Keyboard and ECS were shown in the same memory map plan, alongside the Intellivision III planned EXEC II.

 

 

 

On a separate note: Apparently VLSI Technology was a big source for Mattel ROMs by the end, and according to the oral history at the Computer History Museum they had to come up with their own design for the ROMs. I wonder if the paged ROMs came from VLSI Technology and not GI? (The Mattel material starts around page 14. Apparently they had ROMs by mid-to-late '82.)

Edited March 9, 2019 by intvnut

[intvnut]

As for when the stanking... err stacking connector was killed off, maybe the decision was actually just to roll the wireless controllers forward to the Intellivision III and not offer them on the Intellivision / Intellivision II, to add more of a differentiating characteristic to the Intellivision III, especially since it wouldn't have the game catalog yet.

 

(And LOL on "stanking.")

[+Lathe26]

Supposedly, the internal pitch was that the Keyboard Component and LUCKY (the ECS) were supposed to coexist in the marketplace. That internal story was supposedly to keep Dave "Papa Intellivision" Chandler from using his political clout from getting Mattel to cancel the ECS. You can see in some of the later memory maps (CCF_10232011_00026.pdf) that both the Keyboard and ECS were shown in the same memory map plan, alongside the Intellivision III planned EXEC II.

 

...

 

On a separate note: Apparently VLSI Technology was a big source for Mattel ROMs by the end, and according to the oral history at the Computer History Museum they had to come up with their own design for the ROMs. I wonder if the paged ROMs came from VLSI Technology and not GI? (The Mattel material starts around page 14. Apparently they had ROMs by mid-to-late '82.)

 

That's some cool stuff. I noticed the document was dated as Sept 1982. Do we have an idea when the Keyboard Component was finally killed off?

 

I had no idea about VLSI's involvement with Mattel ROMs. It would be a smart move by Mattel to try to reduce the number of single-source parts. VLSI would be the perfect company for GI's unusual bus interface with GI's automatic memory address decoding (typically no chip selects).

 

As for when the stanking... err stacking connector was killed off, maybe the decision was actually just to roll the wireless controllers forward to the Intellivision III and not offer them on the Intellivision / Intellivision II, to add more of a differentiating characteristic to the Intellivision III, especially since it wouldn't have the game catalog yet.

 

(And LOL on "stanking.")

 

That makes a lot of sense. Companies like to have differentiators for product sequels. Why buy the Intellivision III if your Intellivision I or II is good enough with upgrades? Look at Microsoft Office; for many years Microsoft Office's single biggest competitor was the previous version of Microsoft Office.

 

Do we know when work started on the Intellivision III, even with the limited known history of it? The ColecoVision was released in August 1982 so there might have been an earlier reveal of the CV at some convention like CES or similar. My understanding is that the ColecoVision caught Mattel a little off-guard since Mattel was used to having the most advanced console in the market. This resulted in the Intellivision III and/or IV design work getting kicked off (or suddenly accelerated or shuffled) to retake the "most advanced console" claim.

[+Lathe26]

And on this point:

 

It turns out the brown and grey ECSes are slightly different. I had already suspected they were different, since I analyzed some tapes from an Italian ECS club and wondered why they were at the same baud rate as NTSC, despite both ECS units deriving their clock from the CPU clock on the cartridge port. (3.579545MHz on NTSC units, and 4MHz on European units.)

 

Lathe26 has been digging into this rather heavily, so I don't want to take any wind out of his sails. He can give more details.

 

Ok, time for some thunder.

 

For the past several weeks, I've been deep diving into the internals of the ECS. This started out as just trying to figure out how the UART worked and things exploded from there. I will present more finding soon (life has slowed me from presenting already), but here's what I'll say now:

 

White and Brown ECS's differ in how their baud rates are generated. The chip responsible for the UART (AUX and TAPE IN/OUT) is driven by MCLK which runs faster in PAL Intellivisions than NTSC Intellivisions. Since the brown ECSs were intended to be hooked up to PAL Intellivisions, they use a different value to divide MCLK to produce the baud rate. For example, if you set a brown ECS to output at 1200 baud when it is hooked up to an NTSC Intellivision, it will actually produce a slower baud rate of about 1075 baud because it was expecting a faster baud rate. Brown ECSs have a different part number of 4629 instead 4182 like the white ECSs. If you look at the edge connector daughterboard (board 4182-4239), pin 30 of the glob-top chip U41 is solder bridged to ground in the brown ECSs. Pin 30 is unconnected in white ECSs and appears to have an internal pull-up to +5V.

 

So... if you hook up a brown ECS to and NTSC Intellivision or hook up a white ECS to a PAL Intellivision (i.e. mismatched hardware), then you will be able record and playback data cassettes just fine with that combination of hardware. However, these tapes will not be compatible with the normal ECS+Intellivision combinations (white+NTSC, brown+PAL). Also, tapes from one mismatched combo will not be compatible with the opposite mismatch (ex: brown+NTSC can't handle tapes from white+PAL). Also, mismatched hardware combinations also affects hooking up an ECS to a printer (or other external hardware) to the AUX port. Most likely, the external hardware will simply not be able to communicate with the ECS

 

That all said, I personally like the look of a brown ECS with my woodgrain Intellivision I which is NTSC. This issue only affects the TAPE IN/OUT and AUX ports. All other hardware is just fine (as far as I have tested). I rarely use those ports normally.

 

Lastly, if you have recorded from TAPE IN/OUT into a WAV file on your computer from a mismatched ECS+Intellivision combo and want to use it on a matched ECS+Intellivision combo, you should be able use any free audio file editor to speed up or slow down the the playback (I have not actually tested this, yet).

 

Edit: updated the slow baud to 1075 after I was able to double-check some notes

[+Lathe26]

Also...

 

U52 (the 24-pin chip on the 4239 daughterboard) is confirmed to be the ROM. This is known from working out the chip's pinout and from simply yanking it out of its socket and scanning the Intellivision+ECS memory. Side note: underneath the socket are the pads for it to be repackaged as the 2nd glob-top chip (chip-on-board).

 

U42 is strongly suspected to be an MMI PAL chip. This is because MMI mostly invented PAL technology and the pinout matches a particular PAL chip model. This chip is yet to be desoldered and dumped. It's main purpose is to mask off the BDIR, BC1, and BC2 control signals that go back into the Intellivision.

 

Edit: originally, the part numbers were accidentally reversed in this post.

[intvnut]

Also...

 

U42 (the 24-pin chip on the 4239 daughterboard) is confirmed to be the ROM. This is known from working out the chip's pinout and from simply yanking it out of its socket and scanning the Intellivision+ECS memory. Side note: underneath the socket are the pads for it to be repackaged as the 2nd glob-top chip (chip-on-board).

 

U52 is strongly suspected to be an MMI PAL chip. This is because MMI mostly invented PAL technology and the pinout matches a particular PAL chip model. This chip is yet to be desoldered and dumped. It's main purpose is to mask off the BDIR, BC1, and BC2 control signals that go back into the Intellivision.

 

Just a minor correction: U52 (the big 24-pin chip) is the known-to-be ROM, and U42 (the smaller MMI chip) is the suspected PAL chip.

 

[+Lathe26]

 

Just a minor correction: U52 (the big 24-pin chip) is the known-to-be ROM, and U42 (the smaller MMI chip) is the suspected PAL chip.

 

 

Doh! That's what I get for looking at schematics on my phone.

 

I corrected my original post for future readers.

[intvnut]

 

Do we know when work started on the Intellivision III, even with the limited known history of it?

 

 

Unfortunately, the Intellivision III target specification (CCF10232011_00017.pdf at PapaIntellivision) doesn't have a date on it. There's a separate Sony engagement PDF (CCF10242011_00005.pdf) with an earliest date of Sept 2, 1982, that describes the Intellivision III and IV.

 

Lots of documents in the same short time window (Sept 1982). You mentioned the Colecovision coming out in August 1982, so I wonder if this was all part of a counteroffensive with potential business partners? I have a feeling the Intellivision III was in development for much of 1982, but probably got more attention when Coleco came out.

[mr_me]

Mattel marketing would have known about colecovision from at least february 1982, when it was introduced at new york toy fair.

 

Regarding the keyboard component, I wonder what dave chandler was thinking when he first saw the intellivision ii. There's a document from may 1982 saying the intellivision ii should be widened to match the 2609. The last KC related doc on papaintellivision.com might be dated june 2 1982.

Edited March 10, 2019 by mr_me