Two Freddie Chips Question...

[DavidMil]

I have two working 130XE's with different Freddie chips in them. One Freddie is C061922, and the other is C061991. The C061991 works in either computer but the C061922

only works in one computer. If I plug it into the other computer I get a blue screen with garbage all over it. Comparing the rest of the chip numbers on the two 130XE's,

I see that there are different part numbers on the OS chips. The 130XE that the C061922 works in has a OS chip with part number C101687-001, Date Code 1987. The

130XE that does not work with the C061922 has an OS chip part number is C061598B date code 1984. Anyone have any ideas? Also, what's the difference between

the two Freddie Chips?

 

Thanks for any help,

DavidMil

[TXG/MNX]

Interesting could you dump both roms ? Maybe a diff from these roms would give extra info

 

Verstuurd vanaf mijn Le X829 met Tapatalk

[Rybags]

Freddie isn't a ROM, it'd probably be classed as an ASIC which performs Ram multiplexing, and divides down the master clock to provide the slower clocks used by the system.

 

Info I've found seems to indicate the numbers just indicate 2 different manufacturing sources but maybe it goes deeper. Supposedly Freddie was developed to be used in the 1400/1450XL but didn't see use until late in the 800XL run.

 

Maybe there's some difference that causes problems in some 130XEs. Do your 130XEs have the same number of Ram chips, early had 16x1 bit, later had 4x4 bit ICs.

[_The Doctor__]

a picture of both pcb's could shed some light

[DavidMil]

Freddie isn't a ROM, it'd probably be classed as an ASIC which performs Ram multiplexing, and divides down the master clock to provide the slower clocks used by the system.

 

Info I've found seems to indicate the numbers just indicate 2 different manufacturing sources but maybe it goes deeper. Supposedly Freddie was developed to be used in the 1400/1450XL but didn't see use until late in the 800XL run.

 

Maybe there's some difference that causes problems in some 130XEs. Do your 130XEs have the same number of Ram chips, early had 16x1 bit, later had 4x4 bit ICs.

 

I think that TXG/MNX meant the OS chips, not the Freddies. Info on the systemboards is:

Both are 16 x 1's and both have the same part number and Rev code: C070067 Rev. 3.

On the board with the older Atari OS (C061598B) there is a 1K resister soldered to pin 8 (ground) of the 74LS138 Chip (U2 between the Pokey and the PIA chips),

the other end of the resistor is stuck into a via hole between pin 7 of the Pokey Chip and pin 25 of the PIA chip.

[DavidMil]

Freddie isn't a ROM, it'd probably be classed as an ASIC which performs Ram multiplexing, and divides down the master clock to provide the slower clocks used by the system.

 

Info I've found seems to indicate the numbers just indicate 2 different manufacturing sources but maybe it goes deeper. Supposedly Freddie was developed to be used in the 1400/1450XL but didn't see use until late in the 800XL run.

 

Maybe there's some difference that causes problems in some 130XEs. Do your 130XEs have the same number of Ram chips, early had 16x1 bit, later had 4x4 bit ICs.

 

I just noticed one other difference, however I don't think it has any bearing but...

The board with the new OS ROM and the C061922 Freddie has TMS4164-12NL (120 nanosecond) RAM chips.

The other board has M5K4164ANP-15 (150 nanosecond) RAM chips.

 

David

[_The Doctor__]

the numbers he listed are Freddies

C061922 FREDDIE HA1922 3/21/1984 DRAM control chip, used in the 1450 8-bit computer (DLB) and maybe handles the whole 1200/1400 OS addressing and dram timing

C061991 FREDDIE (ASG) H1991 12/12/1983 is normal XE

Edited May 17, 2017 by _The Doctor__

[Rybags]

Supposedly the Atari can work with DRam with latency up to 250ns which sounds about right given that's still less than half a machine cycle.

 

But maybe with the faster DRam chips there was a change in Freddie? It'd need someone with a storage oscilloscope to look at - given Freddie operates at 8x the system clock speed that gives multiple possibilities of where to do Ras/Cas transtition. Maybe they found that a change was necessary to be more reliable with faster chips, and that the "fast Freddie" becomes incompatible with the longer latency chips.

[_The Doctor__]

two ways to make 1200/1400 freddie fail experiment... change the os like you would to add 32 in one or xe os, what about pbi upgrade.... change the memory chips to different latency....

what do you observe?

Edited May 18, 2017 by _The Doctor__

[DavidMil]

two ways to make 1200/1400 freddie fail experiment... change the os like you would to add 32 in one or xe os, what about pbi upgrade.... change the memory chips to different latency....

what do you observe?

 

This is getting curiouser and curiouser. Evidently I have never run the internal Memory test all the way through on the board with the different OS and Freddie

chips. When I let it finish, I noticed that there are four lines of RAM blocks and the fourth line has four blocks that are the size of four standard size blocks all

touching side by side with a gap between the four blocks. The total of the fourth line (space wise) comes out to 16 regular sized RAM blocks (4 blocks of 4).

I am wondering, does the 1400 have an internal Memory test and does it look like this (3 standard lines of 16 RAM blocks and one line of 4 large blocks)?

The computer runs fine (as far as I know). At least it boots to Basic, will boot to DOS, and runs Defender perfectly.

 

David

[Rybags]

Whatever OS is active should have no bearing on if a machine works or not with a particular Freddie. The chip itself isn't addressable, modifyable etc. although like the MMU it takes part in every memory access.

 

The memory test with the extra large bars for the Ram Test - Rev 3 and Rev 4 XE OSes as used in XEGS and late XE will also test the 130XE type extended 64K Ram if it's present and represents each 16K bank with a large bar each.

[_The Doctor__]

change the os out on the 1200/1400 and you must change the mmu and possibly freddie... it's a long time ago for me.... someone wanna hack apart the proto boards to see if it's the case?

Edited May 18, 2017 by _The Doctor__

[Rybags]

What I meant was in the case of an XE or production XL.

 

With the 1400/1450, who knows? There was extra hardware which never made it to other machines.

[DavidMil]

The memory test with the extra large bars for the Ram Test - Rev 3 and Rev 4 XE OSes as used in XEGS and late XE will also test the 130XE type extended 64K Ram if it's present and represents each 16K bank with a large bar each.

 

I guess that the few 130XE's that I've seen never had the Rev. 3 or 4 OS, because this is the first time I've seen those longer bars. Is it possible that the

C061922 is designed to work with the Rev 3 or 4 OS chip? I'm going to swap Freddies and try the older Freddie with the new OS chip and see what

happens when I run a memory test.

 

David

Edited May 18, 2017 by DavidMil

[Rybags]

Like I said before, I don't think the OS and Freddie combination would matter. Maybe with the 1400 and 1450. Or maybe with certain types of ECI devices present.

What would be good is the knowledge of any differences by design/intent or just circumstance between the 2 Freddie chips.

[Kyle22]

change the os out on the 1200/1400 and you must change the mmu and possibly freddie... it's a long time ago for me.... someone wanna hack apart the proto boards to see if it's the case?

You only need to change the MMU if you are using a single chip OS and BASIC instead of the OS being split on 2 chips.

[DavidMil]

 

I guess that the few 130XE's that I've seen never had the Rev. 3 or 4 OS, because this is the first time I've seen those longer bars. Is it possible that the

C061922 is designed to work with the Rev 3 or 4 OS chip? I'm going to swap Freddies and try the older Freddie with the new OS chip and see what

happens when I run a memory test.

 

David

 

Changing the Freddie's didn't make any difference. I did time the difference in the tests, and it could be human error but it seems like the

new Freddie ran a couple of seconds faster. So I guess all the difference is in the newer OS Chips.

By the way, For some unknown reason both Freddie's started working in both system boards.

 

Thanks to everyone for the really good info and feedback!

David Milsop

Kingwood, Texas

[Joey Z]

 

Changing the Freddie's didn't make any difference. I did time the difference in the tests, and it could be human error but it seems like the

new Freddie ran a couple of seconds faster. So I guess all the difference is in the newer OS Chips.

By the way, For some unknown reason both Freddie's started working in both system boards.

 

Thanks to everyone for the really good info and feedback!

David Milsop

Kingwood, Texas

Quite a bump here, but are you using the original atari sockets by chance? I bet they went bad, if you are. I've have several go bad on me during troubleshooting of machines. They truly are rated for several insertion/removal cycles apparently, and no more than that.

 

First thing I do when I start getting weird issues that weren't there before is to replace the atari sockets with machine pin sockets and try again.

[_The Doctor__]

Lots of time all I had to do was touch up the socket solder joint...with no chip in it after it cooled re insert and it was working again

[DavidMil]

Quite a bump here, but are you using the original Atari sockets by chance? I bet they went bad, if you are. I've have several go bad on me during troubleshooting of machines. They truly are rated for several insertion/removal cycles apparently, and no more than that.

 

First thing I do when I start getting weird issues that weren't there before is to replace the Atari sockets with machine pin sockets and try again.

 

Neither of the 130XE's had sockets on the Freddie's. I had to desolder them both and then install sockets for the chips. This is something I do on almost any XL/XE

computer I work on. But I agree with you, the sockets are cheap and allow a lot of corrosion to build up around the pins of the IC's.

 

David Milsop

[DavidMil]

Lots of time all I had to do was touch up the socket solder joint...with no chip in it after it cooled re insert and it was working again

 

I've seen a lot of Atari's ( as well as many other companies PC's) with cold solder joints. That's another one of the things I look for on the back side of the boards; any joints that are not smooth and

shiny, I resolder them! You'll especially see cold solder joints around anything with large or thick leads. I remember one HP color printer that had so many cold solder joints that the fix was to take

the controller board out of the printer, take it home, preheat your oven to 450 degrees, and put the board in the oven on a metal cookie sheet for 15 to 20 minutes. It worked! I fixed two printers that

way. HP got a little upset when they saw all the comments about this fix on the web. But it worked.

 

David Milsop

[MEtalGuy66]

There is a difference in the EXTSEL circuit (pin 3 of Freddie) on various XE generation machines.. Some, there is a +5v pullup here, others, it is left floating, and still others, it is hard-wired to +5v (eg. machines that do not have an ECI interface).

 

Do both boards treat this signal exactly the same?

[guus.assmann]

Hello DavidMil.

 

Most of the above is not right.

The difference is in the multiplexing for the DRAM chips.

In the oldest Freddy chip, this uses is only 8 bits.

And for the newer one it's 9 bit. If memory serves right.

The same was with the Antic, that also has 2 versions, where one has 8 bit refresh-counter and the other one has 9 bit counter.

So the newest Freddy would work in both machines, the older one just in the machine with the oldest (an probably slowest) Dram chips.

 

BR/

Guus

[+kheller2]

Antic was 7bit vs 8bit.

I never heard of Fred having a similar design .. but that would be interesting. Any sources for this information?

[Rybags]

Is it known how the Refresh works? Does Antic use A8-A15 for the row addressing here? And I suppose A7 might be in on the act for 9-bit addressing.

 

In any case, what has me confused is how the Refresh is routed to main or extended - the EMMU receives the CASINH signal which I assume would be used in refresh, but it's only receiving FA14, FA15 which the EMMU uses in conjunction with it's PB2-PB4 inputs to decide where normal Ram accesses should go. I can't see that both banks would be refreshed at once, otherwise why would 9-bit refresh be needed?

Edited May 26, 2017 by Rybags