VOGONS


First post, by Telev

User metadata
Rank Newbie
Rank
Newbie

Hello,

It's my first post on this forum, and the english is not my primary language. 😀
I'm rebuild a PC like my childhood, so I get on ebay a motherboard with a CPU 486DX2 66.
file.php?mode=view&id=126646
I have no documentation about this motherboard and my reserch on internet do nothing.
When i be able to boot, i'm surprise about the poor performance of the CPU. After discution with the seller and many read of the PCB, i realize a misconfiguration on the turbo jumper. The seller put a jumper on the turbo led jumper ... and the turbo switch is not set.
file.php?mode=view&id=126645
JP9 -> turbo SW
JP8 -> turbo LED
So, now the performance is good.

My question :
What is JP10 Turbo polarity ?? I don't understand the utility of this jumper.
Where put the power LED ? I think on JP5 Keylock, but i'm not sure.... or JP15 JP16 ? These 2 jumpers are not labeled on the motherboard.

Can you help me ?

Best Regards,

Telev

Attachments

  • motherboard.jpg
    Filename
    motherboard.jpg
    File size
    507.26 KiB
    Views
    242 views
    File license
    Public domain
  • 20211121_143328.jpg
    Filename
    20211121_143328.jpg
    File size
    430.09 KiB
    Views
    242 views
    File license
    Public domain

Reply 2 of 11, by mkarcher

User metadata
Rank Oldbie
Rank
Oldbie

JP9 is meant to be connected to a turbo switch. In some cases, the switch is open when the turbo button is "active", in other cases, the switch is closed when the turbo button is active (and there are cases that have a three-pin turbo switch connector, and the switch switches whether 1-2 or 2-3 is connected). I wonder why JP10 has three positions and is not plugged at all. You only need to choose between "turbo is when the switch is open" and "turbo is when the switch is closed", which would usually be implemented as a 3-pin jumper to be toggled between 1-2 or 2-3, or 2-pin jumper that will be set or not. You can just try how the board reacts to JP9 in the three possible configurations of JP10.

Keylock/Power LED is usually combined into a single 5-pin connector, in your case it is JP5. I have no idea what JP15/JP16 are supposed to do on your board.

It seems a battery has been removed from that board. This is a good idea, because most of these batteries start to leak and corrode the board. But that also means the board won't be able to keep time and settings when it is powered off, as there is no replacement power source installed.

Furthermore your board doesn't have any L2 cache. I looked up the HTK340 chipset, consisting of the HTK321 ISA bridge and the HTK320 memory controller (the latter has a fan on your board), and the manufacturer claims that the memory interface is performance-optimized in a way that the system is "fast enough without L2 cache in most usage scenarios". I don't doubt the claim that the board has a high-performance memory interface, and the write-merging write buffer built into that chipset will (if designed correctly) indeed help a lot with typical memory access patterns by 16-bit software that otherwise would be handled by the L2 chace. The chipset can do bank interleaving on the DRAM. I strongly recommend you make use of that option, i.e. you plug in 8 identical 30-pin SIMMs, either 8*1MB = 8MB total or 8*4MB = 32MB total.

Reply 3 of 11, by mkarcher

User metadata
Rank Oldbie
Rank
Oldbie
vstrakh wrote on 2021-12-27, 13:01:

Is there a jumper shorting out the JP8, which is supposed to be a LED output?

Yes, there was:

Telev wrote on 2021-12-27, 11:58:

When i be able to boot, i'm surprise about the poor performance of the CPU. After discution with the seller and many read of the PCB, i realize a misconfiguration on the turbo jumper. The seller put a jumper on the turbo led jumper ... and the turbo switch is not set.

Reply 5 of 11, by Telev

User metadata
Rank Newbie
Rank
Newbie

Thanks you for your comments

mkarcher wrote on 2021-12-27, 13:20:

JP9 is meant to be connected to a turbo switch. In some cases, the switch is open when the turbo button is "active", in other cases, the switch is closed when the turbo button is active (and there are cases that have a three-pin turbo switch connector, and the switch switches whether 1-2 or 2-3 is connected). I wonder why JP10 has three positions and is not plugged at all. You only need to choose between "turbo is when the switch is open" and "turbo is when the switch is closed", which would usually be implemented as a 3-pin jumper to be toggled between 1-2 or 2-3, or 2-pin jumper that will be set or not. You can just try how the board reacts to JP9 in the three possible configurations of JP10.

So, for you the JP9 and JP10 do same thing but with a different pin out ?
Actually, i have put my button on JP9 and leave JP 10 totally open. In benchmark program, i can see the difference when i push the button or not. So with this configuration it's totally working.

mkarcher wrote on 2021-12-27, 13:20:

Keylock/Power LED is usually combined into a single 5-pin connector, in your case it is JP5. I have no idea what JP15/JP16 are supposed to do on your board.

In the case i get, i don't have the keylock. So i must just found the good pin on the JP5 for put my power LED ?

mkarcher wrote on 2021-12-27, 13:20:

It seems a battery has been removed from that board. This is a good idea, because most of these batteries start to leak and corrode the board. But that also means the board won't be able to keep time and settings when it is powered off, as there is no replacement power source installed.

Yes, I have make a battery holder with 3AAA batteries and a diode for protection. and connect to the header JP7 near the BIOS. It's working fine.

mkarcher wrote on 2021-12-27, 13:20:

Furthermore your board doesn't have any L2 cache. I looked up the HTK340 chipset, consisting of the HTK321 ISA bridge and the HTK320 memory controller (the latter has a fan on your board), and the manufacturer claims that the memory interface is performance-optimized in a way that the system is "fast enough without L2 cache in most usage scenarios". I don't doubt the claim that the board has a high-performance memory interface, and the write-merging write buffer built into that chipset will (if designed correctly) indeed help a lot with typical memory access patterns by 16-bit software that otherwise would be handled by the L2 chace. The chipset can do bank interleaving on the DRAM. I strongly recommend you make use of that option, i.e. you plug in 8 identical 30-pin SIMMs, either 8*1MB = 8MB total or 8*4MB = 32MB total.

Exactly, the seller of this board claims the same.
I have 8Mb (8 x 1Mb) of memory. 4Mb getted with the board and 4Mb buy on ebay. But, it's not the same RAM.
Some benchmark in attach

Regards

Attachments

Reply 6 of 11, by Deksor

User metadata
Rank l33t
Rank
l33t

Hello, we don't have this board in ultimateretro.net

Could you provide a bios dump using this DOS program http://cd.textfiles.com/microhaus/mhblackbox3 … MORY/GETROM.ZIP and a good picture without a watermark ?

With these infos we might be able to identify your motherboard and find documentation (or at least help someone in need with the same board later)

Trying to identify old hardware ? Visit Ultimate Retro - Project's thread The Ultimate Retro project - a stason.org/TH99 alternative

Reply 7 of 11, by mkarcher

User metadata
Rank Oldbie
Rank
Oldbie
Telev wrote on 2021-12-27, 13:52:

So, for you the JP9 and JP10 do same thing but with a different pin out ?
Actually, i have put my button on JP9 and leave JP 10 totally open. In benchmark program, i can see the difference when i push the button or not. So with this configuration it's totally working.

Exactly. That configuration fits your need. If you are happy with the way the turbo button works, and you don't want to have it inverted, leave it as is.

Telev wrote on 2021-12-27, 13:52:
mkarcher wrote on 2021-12-27, 13:20:

Keylock/Power LED is usually combined into a single 5-pin connector, in your case it is JP5. I have no idea what JP15/JP16 are supposed to do on your board.

In the case i get, i don't have the keylock. So i must just found the good pin on the JP5 for put my power LED ?

https://old.pinouts.ru/Motherboard/AtLedKeylock_pinout.shtml has the standard pinout of that connector. The two-pin connector of the power LED thus needs to be connected with the positive end to one end of the connector and the negative end to the next pin ("ground"). Typically, the positive wire of a power LED is green colored and the negative wire is white or black.

Telev wrote on 2021-12-27, 13:52:

Yes, I have make a battery holder with 3AAA batteries and a diode for protection. and connect to the header JP7 near the BIOS. It's working fine.

That's a good way to handle it.

Telev wrote on 2021-12-27, 13:52:
Exactly, the seller of this board claims the same. I have 8Mb (8 x 1Mb) of memory. 4Mb getted with the board and 4Mb buy on ebay […]
Show full quote
mkarcher wrote on 2021-12-27, 13:20:

you plug in 8 identical 30-pin SIMMs, either 8*1MB = 8MB total or 8*4MB = 32MB total.

Exactly, the seller of this board claims the same.
I have 8Mb (8 x 1Mb) of memory. 4Mb getted with the board and 4Mb buy on ebay. But, it's not the same RAM.
Some benchmark in attach

Regards

I was unclear when I said "identical". You don't need to use the same type of module. All modules being 1MB modules is good enough, so you are fine with your setup.

Compare Re: Doom runs slow on 486Dx2 machine for SpeedSys timings of a 486DX2 systems with L2 cache. In this kind of benchmark, your system shows a clear disadvantage compared to L2-equipped boards. This is because this kind of benchmark is specifically designed to be limited only by memory performance. To get a fair comparison between a "standard" system and a system optimized for L2-less operation, you would need application benchmarks. The HT chipset doesn't magically create L2 cache out of nothing, but it might be doing a good job of hiding most of the delays you get when writing to RAM that would have been caught be write-back cache due to its write buffer (write buffers with write combining were uncommon in 486 days, so on paper, this is a real advantage, and I guess the write buffer works out quite well for 16-bit software). Also, it might be able to get nice read performance due to bank interleaving, but in practice, the speed doesn't look impressive. Maybe you can improve the read speed by reducing the wait states using the BIOS setup. This is especially true if you use 60ns RAM.

Reply 8 of 11, by Telev

User metadata
Rank Newbie
Rank
Newbie
Deksor wrote on 2021-12-27, 14:17:

Hello, we don't have this board in ultimateretro.net

Could you provide a bios dump using this DOS program http://cd.textfiles.com/microhaus/mhblackbox3 … MORY/GETROM.ZIP and a good picture without a watermark ?

With these infos we might be able to identify your motherboard and find documentation (or at least help someone in need with the same board later)

The board is now installed in the case, but i have also this picture in attachment. Hope it's ok.
I will tried later to dump the bios with your software.

mkarcher wrote on 2021-12-27, 14:19:

https://old.pinouts.ru/Motherboard/AtLedKeylock_pinout.shtml has the standard pinout of that connector. The two-pin connector of the power LED thus needs to be connected with the positive end to one end of the connector and the negative end to the next pin ("ground"). Typically, the positive wire of a power LED is green colored and the negative wire is white or black.

Ok, i will try

mkarcher wrote on 2021-12-27, 14:19:

Maybe you can improve the read speed by reducing the wait states using the BIOS setup. This is especially true if you use 60ns RAM.

I don't know the timing of my RAM ... and my BIOS is set to default. The only modification i have made in the BIOS is the harddrive and the floppies drives.
In attach, the picture i have of the RAM

Attachments

  • memory2.jpg
    Filename
    memory2.jpg
    File size
    261.38 KiB
    Views
    169 views
    File license
    Public domain
  • memory2 back.jpg
    Filename
    memory2 back.jpg
    File size
    205.53 KiB
    Views
    169 views
    File license
    Public domain
  • memory1.jpg
    Filename
    memory1.jpg
    File size
    491.69 KiB
    Views
    169 views
    File license
    Public domain
  • motherboard (without watermark).jpg
    Filename
    motherboard (without watermark).jpg
    File size
    267.5 KiB
    Views
    169 views
    File license
    Public domain

Reply 9 of 11, by Telev

User metadata
Rank Newbie
Rank
Newbie
mkarcher wrote on 2021-12-27, 14:19:

https://old.pinouts.ru/Motherboard/AtLedKeylock_pinout.shtml has the standard pinout of that connector. The two-pin connector of the power LED thus needs to be connected with the positive end to one end of the connector and the negative end to the next pin ("ground"). Typically, the positive wire of a power LED is green colored and the negative wire is white or black.

It's working. I have found the ground with my multimeter. So with the pinout you provided, i put my LED on JP5.
Thanks you for your advise.

Deksor wrote on 2021-12-27, 14:17:

Could you provide a bios dump using this DOS program http://cd.textfiles.com/microhaus/mhblackbox3 … MORY/GETROM.ZIP

In attach the file resulting of the ROMSAVAT.EXE
When i tried to lanch the ROMSPLIT.EXE, it fails to open TEMP ...

Attachments

  • Filename
    ATROM.zip
    File size
    42.09 KiB
    Downloads
    8 downloads
    File license
    Public domain
  • 20211227_164004.jpg
    Filename
    20211227_164004.jpg
    File size
    477.33 KiB
    Views
    149 views
    File license
    Public domain

Reply 10 of 11, by mkarcher

User metadata
Rank Oldbie
Rank
Oldbie
Telev wrote on 2021-12-27, 14:45:

I don't know the timing of my RAM ... and my BIOS is set to default. The only modification i have made in the BIOS is the harddrive and the floppies drives.
In attach, the picture i have of the RAM

That RAM has a row access time of 70ns. You can easily identify it by looking at the model number both of the chips and the modules, as they end in -7 or -70. This is the typical memory speed of RAM modules used in 486 systems, and the default memory timing settings should work perfectly with them. Likely one step faster than the default configuration is also stable, but I won't recommend you to fiddle with the memory timings unless your goal is trying to max out benchmark scores.