SAMS card troubles

[+jedimatt42]

4 hours ago, Vorticon said:

Is your card a Fetzner one or the old generation? I was chatting with Richard Bell about this and he had never seen an old generation card fail, until mine did after 25 years of nearly daily use...

It is a 2016 pcb from Jim, but assembled by myself. I also have a 256k AMS that had failed before I bought it in 2015. All the chips on that have been swapped or tested, but that uncovered nothing. 

 

The impatient child in me, thinks he can program a gal to replace the comparator, while he waits for USPS.

[+jedimatt42]

This WINCUPL code seems to be working as a replacement for the '688 ignoring that it is supposed to compare 2 sets of 8 bits... since in the SAMS, one of those sets is hardwired to 0x1E, I built that into the GAL.  Mostly cause WINCUPL drives me crazy... 

 

Name     74LS688 ;
PartNo   00 ;
Date     12/23/2020 ;
Revision 01 ;
Designer Engineer ;
Company  jedimatt42 ;
Assembly None ;
Location  ;
Device   G16V8MA;

/* *************** INPUT PINS *********************/
PIN 1   = !EN                     ; /*                                 */ 
PIN 2   = A0                      ; /*                                 */ 
PIN 4   = A1                      ; /*                                 */ 
PIN 6   = A2                      ; /*                                 */ 
PIN 8   = A3                      ; /*                                 */ 
PIN 11  = A4                      ; /*                                 */ 
PIN 13  = A5                      ; /*                                 */ 
PIN 15  = A6                      ; /*                                 */ 
PIN 17  = A7                      ; /*                                 */ 


/* *************** OUTPUT PINS *********************/
PIN 19  = !EQ                     ; /*                                 */ 

EQ = EN & !A0 & !A1 & !A2 & A3 & A4 & A5 & A6 & !A7;

 

One of my 74HCT244's seemed to be at fault. I'll have to put the '688 back in and see if it still works. 

 

Things degraded throughout the day. The behavior where simply removing the '688 allowed the 32K function to proceed, stopped working. The card was always driving the bus.

So, I replaced the '688 with a pull-up resister in pin 19 of the '688 socket, which should also turn it into a 32K card... asserting that the crubase is never matched. But that made no change.

So I started swapping other chips. 

 

Presently, it is passing my memory test with the GAL16V8 replacing the 74LS688, and all the 74HCT244's replaced.  

[+jedimatt42]

2 hours ago, jedimatt42 said:

This WINCUPL code seems to be working as a replacement for the '688 ignoring that it is supposed to compare 2 sets of 8 bits... since in the SAMS, one of those sets is hardwired to 0x1E, I built that into the GAL.  Mostly cause WINCUPL drives me crazy... 

 

Name     74LS688 ;
PartNo   00 ;
Date     12/23/2020 ;
Revision 01 ;
Designer Engineer ;
Company  jedimatt42 ;
Assembly None ;
Location  ;
Device   G16V8MA;

/* *************** INPUT PINS *********************/
PIN 1   = !EN                     ; /*                                 */ 
PIN 2   = A0                      ; /*                                 */ 
PIN 4   = A1                      ; /*                                 */ 
PIN 6   = A2                      ; /*                                 */ 
PIN 8   = A3                      ; /*                                 */ 
PIN 11  = A4                      ; /*                                 */ 
PIN 13  = A5                      ; /*                                 */ 
PIN 15  = A6                      ; /*                                 */ 
PIN 17  = A7                      ; /*                                 */ 


/* *************** OUTPUT PINS *********************/
PIN 19  = !EQ                     ; /*                                 */ 

EQ = EN & !A0 & !A1 & !A2 & A3 & A4 & A5 & A6 & !A7;

 

One of my 74HCT244's seemed to be at fault. I'll have to put the '688 back in and see if it still works. 

 

Things degraded throughout the day. The behavior where simply removing the '688 allowed the 32K function to proceed, stopped working. The card was always driving the bus.

So, I replaced the '688 with a pull-up resister in pin 19 of the '688 socket, which should also turn it into a 32K card... asserting that the crubase is never matched. But that made no change.

So I started swapping other chips. 

 

Presently, it is passing my memory test with the GAL16V8 replacing the 74LS688, and all the 74HCT244's replaced.  

 

I think this only kind-of works. I'm seeing a less than stellar behavior in Force Command which uses the SAMS if available for enhanced features...

 

I think this works 

Edited December 24, 2020 by jedimatt42

[+Vorticon]

In my case, my defective SAMS card seems to behave normally as a 32K expansion for all programs, but FC still misbehaves with it, probably seeing it as a fully functional card.

[+Ksarul]

41 minutes ago, Vorticon said:

In my case, my defective SAMS card seems to behave normally as a 32K expansion for all programs, but FC still misbehaves with it, probably seeing it as a fully functional card.

That sounds like the 612 is operating in pass-through mode only. . .I've had several 612s fail that way during initial testing.

[+jedimatt42]

2 hours ago, Vorticon said:

In my case, my defective SAMS card seems to behave normally as a 32K expansion for all programs, but FC still misbehaves with it, probably seeing it as a fully functional card.

Does my memtest detect it as a SAMS? 

 

I think on reset, the '612 is in pass-through mode. So non-SAMS software that doesn't go looking just functions.

If the SAMS is detected, then Force Command will use it in NOT pass through mode... I have recently discovered some variable initialization issues in Force Command, where my C runtime fills 32k expansion RAM with zeroes... before loading static data, but that code isn't SAMS aware. So, I'm seeing history act messed up on startup. Paging in upper memory expansion for history, or swapping it for free banks to load executables (like the bundled FTP) seems to be fine. (for the record, I tried other unreleased programs that use the same mechanisms, not FTP) 

 

What symptom are you seeing in FC? Knowing what FC feature is failing, can map back to the SAMS functionality that is suspect... 

[+Vorticon]

1 hour ago, jedimatt42 said:

Does my memtest detect it as a SAMS? 

 

I think on reset, the '612 is in pass-through mode. So non-SAMS software that doesn't go looking just functions.

If the SAMS is detected, then Force Command will use it in NOT pass through mode... I have recently discovered some variable initialization issues in Force Command, where my C runtime fills 32k expansion RAM with zeroes... before loading static data, but that code isn't SAMS aware. So, I'm seeing history act messed up on startup. Paging in upper memory expansion for history, or swapping it for free banks to load executables (like the bundled FTP) seems to be fine. (for the record, I tried other unreleased programs that use the same mechanisms, not FTP) 

 

What symptom are you seeing in FC? Knowing what FC feature is failing, can map back to the SAMS functionality that is suspect... 

Yes, memtest detects the card.

What I usually see is that the XB command fails to launch a program and I end up with a syntax error. Also LOAD acts flaky as well.

[+jedimatt42]

17 minutes ago, Vorticon said:

Yes, memtest detects the card.

What I usually see is that the XB command fails to launch a program and I end up with a syntax error. Also LOAD acts flaky as well.

Hmm... those both exercise SAMS, in that if Force Command see a SAMS, it maps page 0, and 1 into lower expansion, page 2 into 0xA000 for history, and leaves the rest of upper expansion unmapped. 

LOAD and XB reset the mapping so pages 2 & 3 go to lower expansion, and 10, 11, 12, 13, 14, and 15 go to upper expansion ( the same order as '612's pass through mode ) 

The mapper is left on. I didn't see a reason to set it back to pass through mode given that the layout is the same. 

 

If the 32k aspect works from other points of entry, then that seems like a '612 issue. Sounds like the SRAM is still functioning, and the '245, and the '244s, and the '138.

 

Although, I thought my '688 was bad, but it turned out to be one of the '244s transitioning from partial to full failure mode.

 

I don't feel like this was helpful. 

[+TheBF]

3 hours ago, Vorticon said:

Yes, memtest detects the card.

What I usually see is that the XB command fails to launch a program and I end up with a syntax error. Also LOAD acts flaky as well.

For what it's worth Vorticon my Samstester has all of Forth underneath so it can be used like a debugger on memory.

 

Since you are familiar with Forth it's a shorter learning curve.

 

When you start the program the card is initialized to pass through mapping.  It will error out if it can't read back what it writes to the registers.

But you can still play around.

 

COMMANDS:

 

• HEX         switch interpreter to BASE 16
• DECIMAL switch interpreter to BASE 10
• SAMS-ON  turn on the mapper
• SAMS-OFF turn off the mapper
   
• HEX <page#>  DMAP   maps the SAMS page into the window at expansion RAM address >3000
  
   Example:   SAMS-OFF  HEX 3000 ?  ( will show pass thru value I wrote on startup of >BFBF)
                   SAMS-ON   HEX 8 DMAP  ( maps in page 8 to >3000 expansion RAM)
                   3000 ?  will show some data from 1st cell of SAMS page 8
                   994A 3000 !  ( store >994A into SAMS)
                   3000 ?   ( NOW shows 994A)
                   (You can read any address from 3000..3FFF to see the SAMS page
• DUMP     ( addr len )
            Example:    HEX  1F DMAP
                              3000 100 DUMP   ( shows contents of SAMS page 1F, 1st >100 bytes)
   
• REGTEST  just re-inits SAMS to pass thru values
  
  
  That's might help you narrow down the problem. ?

[+Vorticon]

Any recommendations for a reliable source for the 612 mapper or is Ebay my only option?

[atrax27407]

Unicorn Electronics (Hodge16 on eBay). If they have them in stock, they are reliable. Unicorn shows them in their inventory. However, the minimum online order to avoid shipping is $30.00. You might want to email them and see what shipping would be on a couple of them. Of the Chinese sources, Funkward Tech has them available in various sized lots. They are usually fairly reliable.

Edited December 29, 2020 by atrax27407

[RickyDean]

3 hours ago, Vorticon said:

Any recommendations for a reliable source for the 612 mapper or is Ebay my only option?

Vorticon, if you're in the states, and can't get a 612 reasonable or in a timely fashion, pm me. I am in Knoxville for a few more days, but will be back in South Georgia eventually and can send you a good mapper, bought about 20 or so back when I was getting ready to build my Sams.

[+Ksarul]

I bought some of these (about 100 of them) when I was stocking up for the general SAMS card builds. Almost all of them worked (one or two will always fail in groups of 50 or more). The seller is a generally reliable source of surplus parts too, and I have bought a lot of stuff from them over the years.

[+Vorticon]

1 hour ago, RickyDean said:

Vorticon, if you're in the states, and can't get a 612 reasonable or in a timely fashion, pm me. I am in Knoxville for a few more days, but will be back in South Georgia eventually and can send you a good mapper, bought about 20 or so back when I was getting ready to build my Sams.

Thanks for the offer, but I have already bought some from ksarul's source. Appreciate it!

[+Vorticon]

45 minutes ago, Ksarul said:

I bought some of these (about 100 of them) when I was stocking up for the general SAMS card builds. Almost all of them worked (one or two will always fail in groups of 50 or more). The seller is a generally reliable source of surplus parts too, and I have bought a lot of stuff from them over the years.

That is a good deal for sure. I went and bought a set of 4. Thanks for the tip.

[+Vorticon]

Quick update: it turned out the 612 mapper was bad. Everything back to normal now

Incidentally, RXB 15's SIZE command does not seem to see the SAMS card, but FC does.

[RXB]

1 hour ago, Vorticon said:

Quick update: it turned out the 612 mapper was bad. Everything back to normal now

Incidentally, RXB 15's SIZE command does not seem to see the SAMS card, but FC does.

Try using CALL AMSINIT in RXB 2015 that seems to work.

I have fixed RXB 2020 but need to get it tested on real TI99/4A as soon as I can get my new TIPI to work, presently it does not!

[+Vorticon]

2 hours ago, RXB said:

Try using CALL AMSINIT in RXB 2015 that seems to work.

I have fixed RXB 2020 but need to get it tested on real TI99/4A as soon as I can get my new TIPI to work, presently it does not!

After CALL AMSINIT, SIZE reports only 240 banks of 4K pages. Shouldn't there be 256 pages? It does see a 1024K AMS however.

[RXB]

2 hours ago, Vorticon said:

After CALL AMSINIT, SIZE reports only 240 banks of 4K pages. Shouldn't there be 256 pages? It does see a 1024K AMS however.

That is absolutely correct!

If you read the RXB Docs you will find 64K is reserved with 32K

(8K Assembly and 24K upper 24K for future use, and 32K reserved for Pass mode i.e. 8K Assembly and 24K normal program space.)

 

16x4=64K and 240+16=256 

[+Vorticon]

Well now the card is not functioning again... I tried all of the 4 612 mapper chips I got from Jim's source and the results are all over the place. Not sure why it worked the first time around. I'm going to order replacements for all the chips on the board except the 612 and just go from there. I'm also going to replace the regulator in case it is getting flaky. Scorched earth approach

[+Ksarul]

Suggestion: replace the 74LS688 with a 25LS2521. They are functional equivalents and a lot more stable over time.

[+Vorticon]

15 hours ago, Ksarul said:

Suggestion: replace the 74LS688 with a 25LS2521. They are functional equivalents and a lot more stable over time.

I will, although I have a strong suspicion that the 612 mapper chips I got were bad as well. The card initially passed all the tests using AMSR4 and Matt's checker when I replaced the 612, but then later failed them, so I think the 612 failed after a short burn-in use and the other ones were bad from the outset. These chips have a TI logo on them, but I kind of doubt that this is genuine.

Can I buy one of your known working 612 chips by any chance? That way I can at least eliminate one potential failure source.

[+Ksarul]

16 minutes ago, Vorticon said:

I will, although I have a strong suspicion that the 612 mapper chips I got were bad as well. The card initially passed all the tests using AMSR4 and Matt's checker when I replaced the 612, but then later failed them, so I think the 612 failed after a short burn-in use and the other ones were bad from the outset. These chips have a TI logo on them, but I kind of doubt that this is genuine.

Can I buy one of your known working 612 chips by any chance? That way I can at least eliminate one potential failure source.

I'll do a test on a chip and send it your way this weekend, @Vorticon. I'll include a tested 25LS2521 as well, just in case.