Tandy 1000SX "Big Blue" (U30) interchange

[seat safety switch]

Hello all,

I have a Tandy 1000SX that I worked on quite a bit in 2019. A corner had broken off the power supply PCB which seems to have made it lose 5V regulation and cook almost everything it could get its little hands on.

I started by replacing the DRAMs which got white hot, and then bought a scope to chase down why it still didn't work. I replaced the "Light Blue" at U51 because I had no clocks or reset coming out of those, and that seems to work now (I have a spare Light Blue if someone else needs one.) Unfortunately, it appeared that the 28MHz master clock was being stepped on by the Big Blue custom chip at U30. Although my notes are spotty, I did enough other work in the area that I was fairly sure the Big Blue was bad.

Since I couldn't find a Big Blue at the time, I've had the computer shelved for the last four years, but I managed to find a Big Blue on eBay. Thing is, the part number isn't exactly the same.


Both parts have a Tandy copyright date of 1986, and the same internal Tandy part number of 8040684. Obviously, my old one is a -0002, and the new one is a -0001, as you can see above.



The technical reference manual only refers to it by name or as "#2684," so I'm guessing that this is the same master part with different revs. I would assume that an incompatible part would get a different master number, 2685 or something else.

I didn't want to power it up just yet, mostly because I have to move some more stuff around to get a monitor or TV on it, but my experience with the power supply has made me a little jumpy about firing up this machine without having my ducks in a row.

So my questions are:

1. Will this work?
2. Does anyone have a 1000SX with a 2684-0001 part riding in it?
3. (Super-duper extra credit) Is there any change documentation that survived somewhere which explains the difference between these chip revs?

Thanks in advance to anyone who can give some advice.

 

[seat safety switch]

It seems to be working, although the computer seems to be getting hung up on something before it can display video, and doesn't finish booting.

I don't have any video output on either RGBI or composite, but H and V syncs are being generated on RGBI (although my Rigol scope's frequency counter thinks that the horizontal sync is only at around 7kHz) and the composite signal looks to have a decent dip in it although my (digital) TV doesn't think anything is there. I might dig out my Monitor III to try it with a dumber tuner and see if there is any output at all.

The 8088 CPU didn't have any address bus activity, so I replaced it with an NEC V20, and now we seem to be getting somewhere: the ROM chip select is being hit, RAS and CAS and OE on the video DRAMs are being hit (also work DRAM, but only in "bank 0.") No writes to RAM that I can detect, which means it's probably failing somewhere early.

Hopefully I figure out soon why it's not completing the booting process. The configuration on the system isn't unusual, except I've replaced some Micron MT1259s with KM4164s and the MT4067 video RAM with 41464s, and there's no floppy drive installed yet.

 

[mojorific]

Glad to see you are getting somewhere. I have an Tandy 1000 SX that is (thankfully) still working. If there is something you need to compare with let me know. I don't have an oscilloscope to validate anything though.

Best of luck!

 

[seat safety switch]

Thanks, but I might not need it. I swapped my new AliExpress V20 for another new AliExpress V20, removed and re-socketed the parallel port custom, and changed the position of two 245s while testing it, and somehow this made the video start working.

I wasn't watching the monitor while I was doing those last few tests, so I will have to retrace my steps and figure out which step actually breaks it again when I undo it.



Now that I had working video, the Tandy complained about broken video RAM – the chip at fault was the last of the original Micron parts I left in the video RAM sockets, which failed a RAM test (with a Simon Raybould RAM tester) instantly.



Now that I've replaced that bad Micron VRAM, the Tandy is instead complaining about having broken system RAM, which makes sense because an MT1259 is a 256K x 1 RAM, and the 4164 I replaced a few of them with is a 64K x 1 RAM. That's one fewer row and column, so I can see why it would be upset.

So now I'm waiting to get some 41256s in the mail - I thought these were a pretty common part but I don't have any on hand and can't find any vendors that can get them to me before August. That's AliExpress for you

Even after all that's done, I will also still need to find a floppy drive (I have a Shugart-pinout FD-55B-06-U from a PC-8801mkII which I might be able to re-jumper; not sure what's different between it and the 55BV or 55BR, maybe I'll just use a Gotek) and build a keyboard adapter.

I'm happy to have gotten as far as I have, though!

 

[Eudimorphodon]

So now I'm waiting to get some 41256s in the mail - I thought these were a pretty common part but I don't have any on hand and can't find any vendors that can get them to me before August. That's AliExpress for you

Did you try Jameco? They're usually decently reliable.

 

[seat safety switch]

Did you try Jameco? They're usually decently reliable.

I have used them in the past, but in this case they unfortunately want a minimum of $37 US just for the shipping alone for 20 ICs to Canada.

 

[Eudimorphodon]

Ugh. I guess I should have thought of that after what I got charged once for admitting the tiny package I was shipping to Canada had an unpopulated PCB and a couple chips in it.

 

[Chromedome45]

how many 41256 do you need? 20? I have about 50 or more that I tested with my retro chip tester and all are good. Just that all my RAM is by different manufacturers.
TI, NEC, OKI, SAMSUNG. if interested let me know.

 

[seat safety switch]

Thank you for the offer, but it did eventually turn up! I finally finished testing and switching out all the RAM today, which took a little while.

Happy to say that it got past the memory test finally and is booting DOS 3.2. That means this thread has been answered: the 2684-0002 "Big Blue" can be replaced by a 2684-0001. I haven't exactly given it any intense use (I was hoping to get a game of Sopwith going) but I feel confident in saying if it works this well, it will probably work fine from here on.

Thank you to everyone in the thread for all your help so far.

Now onto another problem...

I'm testing with a Tandy 2000 keyboard, because I found this old mailing list thread that indicated it should work.

Well.. the keyboard is working, but in a strange way. It keeps generating periods for every keypress, if I hammer a key repeatedly, I can sometimes get the right letter or number out of it for a little while until it goes back to periods. Eventually, it seems to fill up some buffer on the computer side internally and just makes upset beeps at me on subsequent key presses.

I'm guessing this is a timing problem because of the randomness, but can't decide yet if it's a keyboard controller issue (another custom chip that might've gotten zapped) or a keyboard issue.

I do have a 1000TX so I think I'll dig that out later in the week when I have time, and see if the keyboard works properly there. Good thing I didn't pack up the scope yet!

 

[Chartreuse]

I'm testing with a Tandy 2000 keyboard, because I found this old mailing list thread that indicated it should work.

Well.. the keyboard is working, but in a strange way. It keeps generating periods for every keypress, if I hammer a key repeatedly, I can sometimes get the right letter or number out of it for a little while until it goes back to periods. Eventually, it seems to fill up some buffer on the computer side internally and just makes upset beeps at me on subsequent key presses.

I was under the impression that the Tandy 2000 keyboard used a different scanset to the normal XT one used on the 1000s, unless I'm wrong about the 1000 series using an XT scancode set. The scancode set for the 2000 keyboard is listed on page 225 (233 in the pdf) of the Tandy 2000 Technical Reference Manual.

Though I think I may be wrong on my impression the codes there seem to line up with the standard scancode set 1 (XT scancodes) on the ones I've checked. I guess the 2000 was an XT keyboard all along and I never thought to check it before sending it off to you. Just with the difference of having a /BUSY line and a different DIN plug. Sounds like it should just work there then. I guess it could be the opposite though that the Tandy 1000 uses a non-standard keyboard protocol, but the scancodes I'm finding for the Tandy 1000 also seem to line up with the XT, at least at a first glance.

I do remember the keyboard working though when I wrote my little teensy 3 powered scan converted to USB a while back. Though that project was buggy I was able to type with it, so I believe it should be still working for you.

 

[seat safety switch]

It has a similar failure mode on the 1000TX, although by all means the two technical manuals indicate they should work. I can type for a little while accurately, and then eventually either the computer or keyboard lose synchronization, and just start putting out period garbage for every keypress. Resetting the computer doesn't fix it.



Keyboard pinouts are nearly identical (except for the Reset that the 2000 keyboard isn't expecting, and whatever the "multi" stuff is for,) and the timing diagram seems to have been photocopied from one manual to the other:



Unexpected for sure. I find it a little unlikely that both the 1000SX and 1000TX would have identical keyboard controller failures, so I guess I'm going to be building a keyboard adapter. I don't think the keyboard is busted, though, as when it works, it works very well.

For the purpose of documenting this for future Google searchers, Chartreuse did point out that some Tandy 2000 keyboards say "Tandy 2000/1000" on them, whereas this one just says "TRS-80 Model 2000." It was floated that perhaps there's a later revision of the keyboard's firmware or circuit board out there that's more compatible, although I'm not sure how that could be from the evidence collected thus far.

 

[seat safety switch]

I got lucky and found a Tandy 1000 keyboard locally.

As you might expect, it worked perfectly on both the SX and the TX. Once the SX is done getting buttoned up (somehow I managed to have damaged my Gotek's OLED preparing it for this,) I'll put both keyboards on the logic analyzer and look for a difference. That'll be in another thread.