• Please review our updated Terms and Rules here

GSETUP regarding expanded memory for IBM 5170

alejack12001

Veteran Member
Joined
Jul 26, 2020
Messages
528
Location
Fredericksburg, VA
I have been using GSETUP for setting up my IBM 5170. I have everything pretty much setup except for an issue regarding the expanded memory. For some unknown reason the GSETUP will not accept that, for now, no expanded memory cards have been installed. So, the lowest setting on GSETUP for expanded memory is either 113-kB or -15-kB, which must trim off 15-kB from the machines RAM of 512-kB. How do I tell this setup routine that for now, I don’t have expanded memory for this machine? I plan to add a Cheetah combo/memory card after, I am sure that the machine hard drive is setup and can boot from that drive.
 
First, you are confusing the terms, 'expanded' memory and 'expansion' memory. Easy to do. They sound very similar but are different. Expansion memory is IBM's early term for what later became known in the industry as extended memory. By the content of your post, you are referring to expansion/extended memory.

As for the symptom, that sounds to me like it is due to one of the flaws in GSETUP. See point #2 in the 'Problems/Issues' section of [here] - an inability of GSETUP to deal with a particular situation that the 5162/5170's CMOS/SETUP configuration can get itself into. As you will read, one way to deal with the situation is to enter in some particular Cassette BASIC statements, statements that will clear the affected area of the 5162/5170's CMOS/SETUP configuration.
 
Thank you for your response. I was hoping that you, Modem7, would respond since you seem to have the most knowledge about the IBM 5170.

I used the process described from the minuszerodegreeqs website for the GSETUP to place the image file on a 720-kB disk and then boot using the 1.44-MB floppy drive as drive A. Of course, I had to use the GSETUP to inform the system that the 1.44 was the A drive with the 1.2 basically disconnected as was the hard drive. Unfortunately, that part of the procedure didn’t quite work as the machine seemed to cause 1.44 to constantly illuminate and any execution of the programs from the image disk was displaying CDC errors. I am not saying that the image had these CRC errors; I believe that the errors were coming from the 1.44 drive. My drive might have been defective.

I had to get a little creative and make a 360-kB floppy bootable then load the GSETUP and the memsize programs on that media. I booted the machine from the 1.2-MB floppy drive using the newly created 360-kB boot disk. I was able first to trigger the memsize program and learn that the expansion was infact zero. I then setup the rest of the media including the 1.44 but still have problems there. So for now I have removed the 1.44 drive. I have configured my hard drive using the speed store tool, which identified both my drive’s manufacturer and size. I still had to put in a partition because the drive is greater than 32 MB.

So, I have two more questions: the BIO’s on the IBM Motherboard is dated with copy write from IBM for 1981, 85. So, assuming that the BIO’s is inline with the chart see on the following web page from minuszerodegrees: https://minuszerodegrees.net/5170/floppy_diskette/5170_floppy_drive_support.htm then a 1.44 drive should work with this machine, provided that I can get one that doesn’t give off errors? Would a GoTek drive configured as a 1.44 work?
 
So, I have two more questions: the BIO’s on the IBM Motherboard is dated with copy write from IBM for 1981, 85. So, assuming that the BIO’s is inline with the chart see on the following web page from minuszerodegrees: https://minuszerodegrees.net/5170/floppy_diskette/5170_floppy_drive_support.htm then a 1.44 drive should work with this machine, provided that I can get one that doesn’t give off errors? Would a GoTek drive configured as a 1.44 work?
I trust FlashFloppy firmware quite a bit. If a real 1.44m drive would work, so will the gotek.
 
I still had to put in a partition because the drive is greater than 32 MB.
Added a second partition (because you wanted to use all of the drive's space).

So, I have two more questions: the BIO’s on the IBM Motherboard is dated with copy write from IBM for 1981, 85. So, assuming that the BIO’s is inline with the chart see on the following web page from minuszerodegrees: https://minuszerodegrees.net/5170/floppy_diskette/5170_floppy_drive_support.htm then a 1.44 drive should work with this machine, provided that I can get one that doesn’t give off errors?
Note the distinction between drives and floppies/diskettes.

01/10/84 BIOS: Supports 5.25" drives only. However, if a 1.44M drive is connected (and CMOS SETUP is in a bank state), it will be found that 720K diskettes can be read and written to.
06/10/85 BIOS: Introduced support for 720K drives and diskettes. If a 1.44M drive is connected, it will be found that 720K diskettes can be read and written to.
11/15/85 BIOS: Introduced support for 1.44M drives and diskettes.

There are two 1985 dated BIOS revisions.

So, to answer your question, for all three BIOS revisions, a 1.44M drive can be connected up, but whether or not that drive can use 1.44M diskettes, or is restricted to 720K diskettes, depends on the BIOS revision.

( Beware of GSETUP. It will allow you to set a floppy/diskette drive type that the BIOS revision does not support. )
 
Added a second partition (because you wanted to use all of the drive's space).


Note the distinction between drives and floppies/diskettes.

01/10/84 BIOS: Supports 5.25" drives only. However, if a 1.44M drive is connected (and CMOS SETUP is in a bank state), it will be found that 720K diskettes can be read and written to.
06/10/85 BIOS: Introduced support for 720K drives and diskettes. If a 1.44M drive is connected, it will be found that 720K diskettes can be read and written to.
11/15/85 BIOS: Introduced support for 1.44M drives and diskettes.

There are two 1985 dated BIOS revisions.

So, to answer your question, for all three BIOS revisions, a 1.44M drive can be connected up, but whether or not that drive can use 1.44M diskettes, or is restricted to 720K diskettes, depends on the BIOS revision.

( Beware of GSETUP. It will allow you to set a floppy/diskette drive type that the BIOS revision does not support. )
In your initial statement, did i want the entire hard drive, yes. I did mention i was using Speed Store and I believe that the program informed me that 32-MB was the limit for FAT 16. The Speed Store that I was using was labeled SSTOR_65 and dated July 14, 1991. This date was seen in the ini file.

I have included a picture of the BIO's from the motherboard. As to the date in which the 1985 BIO's matches the listed BIO's dates from the table, I don't know.

I saw the notice regarding the GSETUP not correctly supporting the 1.44 setup. I do have the original IBM diagnostic diskette found in the IBM Operations Manual. I assume that diskette wouldn't help with the floppy 1.44 question either. I watched Adrian's Digital Basement where he was trying ro setup an IBM 5170. His solutions regarding the hard drive and floppy drives setup were to use Award/Phoenix BIO's, which were much, much newer than the original BIO's. My goal has been to make this machine as original as possible with a few minor exceptions such as the GoTek. Am I going to have to upgrade the BIO's to get the 1.44 recognized?
 

Attachments

  • IBM MB BIOS.jpeg
    IBM MB BIOS.jpeg
    3.1 MB · Views: 12
Last edited:
I have included a picture of the BIO's from the motherboard. As to the date in which the 1985 BIO's matches the listed BIO's dates from the table, I don't know.
At minuszerodegrees.net, in the IBM 5170 section, is a link named 'BIOS revisions'. In that, the 61X9266/61X9265 set are shown to be of the 11/15/85 BIOS revision.

Am I going to have to upgrade the BIO's to get the 1.44 recognized?
No. The 11/15/85 BIOS revision fully supports 1.44M drives and diskettes, which includes booting from 1.44M diskettes.

Rhetorical: So, is the problem cause the controller, cable, or 1.44M drive ?
 
At minuszerodegrees.net, in the IBM 5170 section, is a link named 'BIOS revisions'. In that, the 61X9266/61X9265 set are shown to be of the 11/15/85 BIOS revision.


No. The 11/15/85 BIOS revision fully supports 1.44M drives and diskettes, which includes booting from 1.44M diskettes.

Rhetorical: So, is the problem cause the controller, cable, or 1.44M drive ?
Thank you about the BIO's. I was hoping not to have to change that part of the machine. I think it might be the 1.44 drive. I have another drive and will set that one to test today.
 
I trust FlashFloppy firmware quite a bit. If a real 1.44m drive would work, so will the gotek.
Now that I have the 1.44 floppy working the next step is to see if the GoTek will work.

I tried the TexLec 8-bit CF card and received errors. I first set this CF card with the built on ROM enabled and the system didn't like it giving me an error at address CF000X and DA000X regardless of the jumpers set at D800 or C800. I can disable the ROM and see if the GSETUP will see the CF card. Is there another way? The CF card I am using is a 2-GB and I do have a 128-MB card as substitute if that is the problem.
 
I tried the TexLec 8-bit CF card
This sounds like you have moved on to trying an XT-CF card. I suspect that by "TexLec", you mean "texELEC". I know that texELEC sells XT-CF cards from Lo-tech.
In a different thread, you wrote, "James Pearse's 8-bit XT-CF card". But maybe you have more than one XT-CF card.
Which of the many cards shown at [here] are you referring to? That be relevant later.

I first set this CF card
Call it an XT-CF card, otherwise people will think that you are referring to a CF card (compact flash card).

giving me an error at address CF000X and DA000X
What is the exact error message shown?

I can disable the ROM and see if the GSETUP will see the CF card.
The XT-CF card. That will not work, because the XT-CF (and XT-IDE) card is not a controller type that the 5170 motherboard's BIOS knows how to communicate with.
 
This sounds like you have moved on to trying an XT-CF card. I suspect that by "TexLec", you mean "texELEC". I know that texELEC sells XT-CF cards from Lo-tech.
In a different thread, you wrote, "James Pearse's 8-bit XT-CF card". But maybe you have more than one XT-CF card.
Which of the many cards shown at [here] are you referring to? That be relevant later.


Call it an XT-CF card, otherwise people will think that you are referring to a CF card (compact flash card).


What is the exact error message shown?


The XT-CF card. That will not work, because the XT-CF (and XT-IDE) card is not a controller type that the 5170 motherboard's BIOS knows how to communicate with.
<The XT-CF card. That will not work, because the XT-CF (and XT-IDE) card is not a controller type that the 5170 motherboard's BIOS knows how to communicate with.> Okay, thanks for the reply, Well I guess all that I have written about below is irreverent.

The card in question is the Lo-tech XT-CF-lite rev.2 from the minuszerodegrees website. However, techELEC must have bought the rights to sell the product. All the card's acquired were from the same supplier. I always view them as James Pearse's card as was described in an article that I read on the first one developed.

I set all three jumpers labeled 1-3 on their respective posts. This configuration has worked well for most of my XT machines, however, this AT as you pointed out addresses things differently. The message received was an error that the address CF000X was not available. I took off the first two jumpers but left the third in play and the message repeated but with a different address DA000X was not available. Using the 320h the port address or changing the port address to 300h didn't seem to make any difference as far as the error was concerned.

I guess I'll have to find a different way to get more storage.
 
XT-IDE/XT-CF cards do work in the IBM 5170, but not if the 5170 motherboard has the first BIOS revision. Your motherboard does not have the first BIOS revision, and so no problem there.

Adding an XT-IDE/XT-CF card to an IBM 5170 is akin to adding a SCSI card.

The XT-IDE/XT-CF card will coexist with the 5170's stock floppy/hard controller. That is discussed, with diagram, in the second section of [here].

I set all three jumpers labeled 1-3 on their respective posts. This configuration has worked well for most of my XT machines, however, this AT as you pointed out addresses things differently. The message received was an error that the address CF000X was not available.
The default settings, which equate to: {Base address of ROM at C8000, Base I/O address of 300h, ROM enabled}

In relation to the card's ROM, the lo-tech web page contains, "Regardless of the chip used, the first 32KB only will be mapped into the PC address space."
So, with a starting address of C8000 (a.k.a. C800:0)(a.k.a. C800), the range of addresses used by the card is C8000 through CFFFF.

That will conflict with anything starting at CF000, and explains an error message (generated by ???) along the lines that CF000 is unavailable.

I took off the first two jumpers but left the third in play and the message repeated but with a different address DA000X was not available.
{Base address of ROM at D8000, Base I/O address of 320h, ROM enabled}
The range of addresses used by the ROM becomes D8000 through DFFFF, which explains an error message (generated by ???) along the lines that DA000 is unavailable.

So there is a conflict between the XT-CF-lite's ROM and something else.

I did a search for "unavailable" in the source code of the ROM content, which is the XTIDE Universal BIOS (XUB), and "unavailable" does not appear.

Are you aware of anything in your 5170 that might be causing the conflict?
 
XT-IDE/XT-CF cards do work in the IBM 5170, but not if the 5170 motherboard has the first BIOS revision. Your motherboard does not have the first BIOS revision, and so no problem there.

Adding an XT-IDE/XT-CF card to an IBM 5170 is akin to adding a SCSI card.

The XT-IDE/XT-CF card will coexist with the 5170's stock floppy/hard controller. That is discussed, with diagram, in the second section of [here].


The default settings, which equate to: {Base address of ROM at C8000, Base I/O address of 300h, ROM enabled}

In relation to the card's ROM, the lo-tech web page contains, "Regardless of the chip used, the first 32KB only will be mapped into the PC address space."
So, with a starting address of C8000 (a.k.a. C800:0)(a.k.a. C800), the range of addresses used by the card is C8000 through CFFFF.

That will conflict with anything starting at CF000, and explains an error message (generated by ???) along the lines that CF000 is unavailable.


{Base address of ROM at D8000, Base I/O address of 320h, ROM enabled}
The range of addresses used by the ROM becomes D8000 through DFFFF, which explains an error message (generated by ???) along the lines that DA000 is unavailable.

So there is a conflict between the XT-CF-lite's ROM and something else.

I did a search for "unavailable" in the source code of the ROM content, which is the XTIDE Universal BIOS (XUB), and "unavailable" does not appear.

Are you aware of anything in your 5170 that might be causing the conflict?
I think some clarification needs to be done here. This IBM 5170 has an IBM Type 3 Motherboard with the standard 512-KB RAM and you have seen the ROM's. The floppy and hard drive controller that I bought was for the IBM type 1 motherboard and it just didn't work well with the setup of my hard drive, which is a full height MFM Micropollis 1304, 42-MB drive. I noticed that there were other controllers that worked with the IBM type 3 motherboard but they are expensive and rather hard to get. I opted for a Western Digital 1003-WA2 controller and with the use of Speed Store, I was able to setup the hard drive even though it has a partition with two 21-BM spaces for C and D drive. The GSETUP recognizes the drive as type 3 even though the type 3 is a 31-MB drive. Someone explain that one to me.

I suspect that the motherboard although operational has some sort of flaw. I don't know if it's in the ROM or on the chip's; it's hard to say. When I first received the board, I put a NPS on it as I have done with all my motherboards. The NPS was an Intel 80287-8. I seated it correctly in the socket near the keyboard connector. As long as the chip was in the NPS socket, the motherboard would not power on. As soon as I removed the NPS, the motherboard powered on. I don't have a clue as to why. There isn't any wires on the reverse side that would prevent the insertion of the NPS from halting the powering up the computer motherboard. I have the IBM diagnostics but their limited in discovery of potential problems that might exist on this motherboard.

From your drawings, if I use I/O Port address 300 and Memory address D800 then see if the built in ROM will work. Worth a try.
 
The GSETUP recognizes the drive as type 3 even though the type 3 is a 31-MB drive. Someone explain that one to me.
GSETUP does not 'recognise' anything. Unlike IDE drives, MFM/RLL drives cannot be interrogated for drive information. If GSETUP was the source of the 'type 3' in the CMOS SETUP, then the type 3 had to have been selected by someone using GSETUP. Otherwise, GSETUP was just reporting the type that it found already in the CMOS SETUP (and the source for that might have been random RAM contents in the CMOS/RTC chip).

The NPS was an Intel 80287-8 ... I don't have a clue as to why.
Was that NPU verified as functional elsewhere ?

From your drawings, if I use I/O Port address 300 and Memory address D800 then see if the built in ROM will work. Worth a try.
That diagram assumes no hardware conflicts in the computer.

The I/O port range doesn't come into play just yet. It will later when the XUB displays its splash/banner text and then the XUB tries to communicate with the attached CF.
( Getting the XUB to display does not involve any I/O ports. )

The apparent conflict being experienced presently is between 'standard' addresses, ones used by the XT-CF's EEPROM/ROM circuitry, and something else.

Does the XUB appear if:
- The only fitted cards are video and the 1003-WA2 hard/floppy controller; and
- No CONFIG.SYS and no AUTOEXEC.BAT
 
GSETUP does not 'recognise' anything. Unlike IDE drives, MFM/RLL drives cannot be interrogated for drive information. If GSETUP was the source of the 'type 3' in the CMOS SETUP, then the type 3 had to have been selected by someone using GSETUP. Otherwise, GSETUP was just reporting the type that it found already in the CMOS SETUP (and the source for that might have been random RAM contents in the CMOS/RTC chip).


Was that NPU verified as functional elsewhere ?


That diagram assumes no hardware conflicts in the computer.

The I/O port range doesn't come into play just yet. It will later when the XUB displays its splash/banner text and then the XUB tries to communicate with the attached CF.
( Getting the XUB to display does not involve any I/O ports. )

The apparent conflict being experienced presently is between 'standard' addresses, ones used by the XT-CF's EEPROM/ROM circuitry, and something else.

Does the XUB appear if:
- The only fitted cards are video and the 1003-WA2 hard/floppy controller; and
- No CONFIG.SYS and no AUTOEXEC.BAT
You're right the GSETUP just reported the type 3 as the drive size. Speed Store setup the Micropolis drive, which had the following parameters of 830 cylinders, 6 heads and 17 sectors per track, which defined the drive to be 42.4 MB. They type 3 is a 615 cylinder, 6 head and 17 sector track drive with 31.6 MB. It will be interesting to see what happens when I exceed the 31.6 MB.

I stripped the AT machine down to just video and controller then configured the config.sys to buffers of 10, files of 20 with no autoexec.bat. On the XT-CF, I set the port to 300 and the memory to D800. I enabled the ROM setting then rebooted the AT. The computer didn't hang but the display didn't show with the different function keys with the C>C drive or A>A drive being the startup. When I checked the system using Checkit the memory test hung, which told me that there is a conflict. When I removed the XT-CF card and used Checkit again everything was back to normal. I now have a workaround with using NetDrive that Mr. Brutman stood up recently on his mTCP website, where the network gives me the same space as would have been provided by the XT-CF card.

Regarding the NPS, I have 3 of the 80287-8 co-processors or NPS's. Do they work, i suspect that they are pulls from other machines; you guess is as good as mine. There isn't a way to test these except to populate another motherboard and see if it's recognized.
 
You're right the GSETUP just reported the type 3 as the drive size. Speed Store setup the Micropolis drive, which had the following parameters of 830 cylinders, 6 heads and 17 sectors per track, which defined the drive to be 42.4 MB. They type 3 is a 615 cylinder, 6 head and 17 sector track drive with 31.6 MB. It will be interesting to see what happens when I exceed the 31.6 MB.
Re hard drives:

The IBM hard drive type number in CMOS SETUP, which is a mapping/pointer to a particular CHS geometry, is only used in particular situations.
E.g. When FDISK is creating partitions, FDISK needs to know the hard drive's CHS geometry, i.e the maximum cylinder count, maximum head count, maximum sector count.
E.g. Hard drive checking by the POST in the IBM BIOS. It does a test seek to the final track (track is a cylinder+head combination).
E.g. Information reporting software.

When DOS boots, it has no need to revisit that. It is going to believe what went into the partition table.

In your particular situation, I do not believe that you will have a problem asking DOS to use hard drive space from cylinders 615 (about) onwards.

In other situations, there can be problems.
E.g. Drive fitted has less cylinders than what the IBM type number in CMOS SETUP indicates. Expected is the IBM POST issuing a 1790 error, because the POST's test read of the final track would fail.
E.g. Drive fitted has less heads than what the IBM type number in CMOS SETUP indicates. Expected is the IBM POST issuing a 1790 error, because the POST's test read of the final track would fail.

And a head count mismatch is expected to impact on the cluster-to-CHS calculation that DOS makes.
 
I stripped the AT machine down to just video and controller then configured the config.sys to buffers of 10, files of 20 with no autoexec.bat. On the XT-CF, I set the port to 300 and the memory to D800. I enabled the ROM setting then rebooted the AT. The computer didn't hang but the display didn't show with the different function keys with the C>C drive or A>A drive being the startup. When I checked the system using Checkit the memory test hung, which told me that there is a conflict. When I removed the XT-CF card and used Checkit again everything was back to normal. I now have a workaround with using NetDrive that Mr. Brutman stood up recently on his mTCP website, where the network gives me the same space as would have been provided by the XT-CF card.
Per [here], there are some VGA cards that have a BIOS ROM that extends up to address CA7F0, but, everything else being functional, that is not going to be a problem with your XT-CF-lite's ROM set to start at D8000.

Do you have another computer that you can try the XT-CF-lite card in?
 
And in addition to that question, does the DOS based RAYXTIDE tool, available at [here], find the XTIDE Universal BIOS (XUB) ?
I have IBM 5150 that I can try the XT-CF in. The 5150 already has a XT-CF card in it to act as program/data repository. I downloaded the RAYXTIDE software; until just now, I didn't know that this tool existed.
 
Back
Top