VOGONS


Reply 121 of 1156, by feipoa

User metadata
Rank l33t++
Rank
l33t++

There are certainly some inter-wire parasitics at play. I think cb88 is going to get the design layed out on a PCB with minimal trace lengths and proper spacing for the operational frequency. As shown with another company's interposer, this should fix correct the problem. My wire-wrap prototype mess only worked at up to around 8 MHz.

Plan your life wisely, you'll be dead before you know it.

Reply 122 of 1156, by cb88

User metadata
Rank Newbie
Rank
Newbie
feipoa wrote:

There are certainly some inter-wire parasitics at play. I think cb88 is going to get the design layed out on a PCB with minimal trace lengths and proper spacing for the operational frequency. As shown with another company's interposer, this should fix correct the problem. My wire-wrap prototype mess only worked at up to around 8 MHz.

Yeah I just gotta get around to it... I think I have the circuit schematic correct at least now.

Reply 124 of 1156, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I just found a motherboard (EXP4349) which specifically mentions support and jumper settings for the PGA-168 version of the SXL2. Most likely, it is only for the 5 V version because I don't see any 3.6 V VRM settings. There is a 3.3 V jumper setting, however, I don't see a voltage regulator anywhere on the motherboard. I see what looks to be solder pads for a VRM though. Anyway, I'm looking forward to seeing if the PGA-168 version of the SXL2-66 3.6 V can hit 80 MHz.

EXP4349.jpg
Filename
EXP4349.jpg
File size
1.98 MiB
Views
1417 views
File license
Fair use/fair dealing exception

Plan your life wisely, you'll be dead before you know it.

Reply 125 of 1156, by Krey

User metadata
Rank Newbie
Rank
Newbie
feipoa wrote:

Most likely, it is only for the 5 V version because I don't see any 3.6 V VRM settings. There is a 3.3 V jumper setting, however, I don't see a voltage regulator anywhere on the motherboard.

do you know that yours TI SXL2-G66 is 5V-tolerant CPU? (according to 210 page of TI manual)

Reply 126 of 1156, by Paralel

User metadata
Rank Member
Rank
Member
Krey wrote:
feipoa wrote:

Most likely, it is only for the 5 V version because I don't see any 3.6 V VRM settings. There is a 3.3 V jumper setting, however, I don't see a voltage regulator anywhere on the motherboard.

do you know that yours TI SXL2-G66 is 5V-tolerant CPU? (according to 210 page of TI manual)

Good point. That is what the G is for.

Reply 127 of 1156, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie

I went throught the TI docs for SXL(C) family lately and I think the PGA168 version of the 486SXL will not work in any standard 486 mobo, unless the manufacturer added a proper support in both hardware and software (BIOS).

The reason is the SXL has a 386-like bus. Even in PGA168 package. It does not understand a true 486 cache flushing protocol and I'm pretty sure it can't do burst transfers either. Internal cache line is 4-bytes long, just like external cache on a 386 mobo would be. The HW mod for a 486 mobo wouldn't be that difficult but the BIOS is another story. It might be trying to test the internal cache using TR4/TR5 registers and those are not the same as on true 486.

BTW I've also found a Make-it 486 on ebay that is actually 486SXL in PGA168 package soldered to PGA132 interposer with just 2 extra chips, one is a counter (not sure what for) and the other is NAND gates for what I assume to be the external circuitry needed to generate the #FLUSH signal. Could be the setup is meant to not trigger on short HOLD cycles (this is when BIOS does RAM refresh and doesn't have a hidden refresh option).

Reply 128 of 1156, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I have only run across one motherboard with specific support for the 486SXL in PGA-168 format.

Plan your life wisely, you'll be dead before you know it.

Reply 129 of 1156, by maxtherabbit

User metadata
Rank l33t
Rank
l33t

feipoa asked for my help routing a board for this in kicad - I'd be happy to do it but it would be a while since I'll be doing it in my spare time

I must admit I don't really understand the use case here... so there is a PGA168 chip that doesn't work in 486 motherboards but needs a custom interposer to work in 386 boards???

Why would such a chip even exist, and why would anyone want to use it??

Reply 130 of 1156, by feipoa

User metadata
Rank l33t++
Rank
l33t++

That is a common curiousity - why was such an oddball chip produced? I don't really have a suitable answer, but some customer may have wanted it for a custom design and when it became available it was too late to be used? No idea. But it seems that industrial suppliers have quantity of it as new old stock. I've only seen one 386/486 hybrid board which had support for the 5 V variant of the SXL2-50 PGA168, but never for the 3.6 V version. And that board was extremely problematic and ultimately stopped working.

I've never seen a proper 486 board with jumpers to support the SXL pinout, although the SXL manual mentions the pin changes required. I guess there is quantity of these chips because nobody could use them in non-custom designs. They'd have probably seen more use in a PGA-132 format w/built-in VRM, similar to how POD's were produced.

Plan your life wisely, you'll be dead before you know it.

Reply 132 of 1156, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Well, this reseller has 37 immediately available: https://www.westfloridacomponents.com/IC490PE … ocessor+TI.html . $16.50 each.

And when I looked as of ~6 months ago, I had found other sellers (I think one was on eBay) who also had over 100 chips. So, yes, I think this is [more than] enough to satisfy the demand.
EDIT: Here's the other seller, https://www.questcomp.com/part/4/ti486sxl2-g66-ga/405969552 - direct link doesn't work, but go to https://www.questcomp.com/ and type in TI486SXL2-G66-GA . This seller has 300 at $19.50 each. No photo though.

Last edited by feipoa on 2019-08-14, 20:43. Edited 1 time in total.

Plan your life wisely, you'll be dead before you know it.

Reply 133 of 1156, by maxtherabbit

User metadata
Rank l33t
Rank
l33t

If I could get my hands on those schematic symbols that cb88 mapped out that would really make this about 1000 times easier for me. Other than that it's just a matter of matching trace lengths for address and data busses, letting the Float pin float, breaking out MEMW#, and handling voltage regulation, yes?

Reply 134 of 1156, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Sounds about right. You will probably want to re-read the thread starting from this sub-post, Re: Custom interposer module for TI486SXL2-66 PGA168 to PGA132 - HELP! to look for any caveats. I'm not free to re-read the thread right now. If cb88 didn't attach the desired files to the thread, then I don't have them either. I have sent him a PM.

As people like to overclock, I'd design the PCB for up to 100 MHz going into the CLK2 pin. We know the QFP version of the chip works at 80 MHz, so that's where 100 MHz came from. If 100 Mhz is problematic, then shoot for 80 MHz.

Plan your life wisely, you'll be dead before you know it.

Reply 135 of 1156, by maxtherabbit

User metadata
Rank l33t
Rank
l33t
feipoa wrote:
I recently acquired what looks like a commercially produced version of the SXL2 interposer I had been working on. I had hoped t […]
Show full quote

I recently acquired what looks like a commercially produced version of the SXL2 interposer I had been working on. I had hoped that it contained the voltage regulator for 3.6 V CPUs, however it did not. This is a PGA132 to PGA168 interposer for the 5 V version of the TI 486SXL2.

PGA132_to_PGA168_486SXL2_Interposer.jpg

Ideally, I'd like to run the SXL2-66 at 3.6 V and 80 MHz.

TISXL2-66_PGA168.jpg

To do this, I had to add a voltage regulator interposer module. I modified an extra Gainbery PGA168 interposer module to have a variable voltage output and set it for 3.6 V. I also removed the 3 IC's on the Gainbery module. Unfortunately, a few pins on the Gainbery aren't passed through, so I had to wire BUSY# and PEREQ externally, otherwise any FPU calls result in errors. Vcc5 also isn't connected on the PGA132-to-PGA168 interposer, so I wired it to 5 V.

Gainbery_VRM_mod_for_SXL2.jpg

This is what the package looks like in sandwich form.

SXL2-66_Interposer_with_Gainbery_VRM.jpg

This combination appears to run well at 66 MHz in DOS and Windows 3.11, however I could not get it to function at 80 MHz. I tried two different CPUs. I also tried it at 70 MHz, and while it boots, there are artifacts in Windows 3.11. I am surprised that an extra 3.3 MHz causes instability. Anyone have any ideas on how to get 80 MHz running? Perhaps the extra trace lengths associated with the Gainbery VRM is causing the problem. Or maybe something to do with the IC's on the PGA132-to-PGA168 interposer. It contains 4xNAND gates, a counter, and one unknown. I could remove them, but I don't want to modify this rare item. One other idea might be to run the 3.6V SXL2-66 at 5 V, that is, without the Gainbery interposer. This is probably too risky for me.

On the contrary, the QFP144 version of the SXL2-66 has run just fine for me at 80 MHz.

SXL2-66_QFP144_Upgrade.jpg

I'm still willing to give this project a shot, but I seriously doubt anything I can design will surpass the performance of this stacked config you've tried here.

For the simple reason that it's going to be impossible to hand assemble anything that places two through hole PGA sockets on top on one another. It would have to be a side-by-side arrangement, which is going to mean longer traces and more parasitics.

Reply 136 of 1156, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Why does it need to be a side-by-side arrangement?

"this stacked configuration"? We don't need the Gainbery module, it would be built into the design.

Basically, we are copying the Evergreen QFP144 interposer, but without the [unnecessary] flush logic circuits, using PGA instead of QFP, and installing a VRM that can output more current. I wonder if the QFP144 SXL2-66 chips are still available...

We need a top-down approach for the ability to fit into more motherboards. How many layers do we need to accomplish this while limiting the parasites to an acceptable limit? To me, this is the part where past and proficient experience is needed with PCB design.

Did you see the drawing supplied by cb88? Have you been able to read thru the whole thread?

Plan your life wisely, you'll be dead before you know it.

Reply 137 of 1156, by maxtherabbit

User metadata
Rank l33t
Rank
l33t
feipoa wrote:

Why does it need to be a side-by-side arrangement?
*snip*

Did you see the drawing supplied by cb88? Have you been able to read thru the whole thread?

yes, I saw the rendering - care to explain to me how exactly someone is going to hand solder 2 PGA sockets on opposite sides of a PCB??

Reply 138 of 1156, by feipoa

User metadata
Rank l33t++
Rank
l33t++

It will be a little cumbersome, but I think you'd first solder on the male-to-female PGA socket, and cut the male leads just to the end of the PCB such that they do not pultrude. The other end is a male-to-male machine pins. The male-to-male machine pins cannot be in a full grid, otherwise you'd not be able to solder the centre rows. The male-to-male ends would need to be single rows only so that you can access the ends to solder. Then you solder on row by row. Ideally, you'd want to reach to the top end of the PCB to get the row started and such that it is normal to the PCB. Once two pins are soldered from the top side (not all top pins will be visiable obviously), then you can remove the black sleeve and solder the remaining pins from the bottom side. The black sleeve is left on initially to help keep the non-soldered pins normal and aligned.

Plan your life wisely, you'll be dead before you know it.

Reply 139 of 1156, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Here is a photo of a PGA168 socket over a PGA132 socket to give you some idea as to how many PGA132 pins should be accessable from the top of the PCB, even with the PGA168 socket in place. For the non-top-accessable pins, I've attached an image of the male-to-male machine pins. It may be possible to solder the single rows in place without removing the black sleeve. It might make it a little easier if there are 45 or 90 degree bends on solder tips.

PGA168_over_PGA132.jpg
Filename
PGA168_over_PGA132.jpg
File size
162.39 KiB
Views
1170 views
File license
Fair use/fair dealing exception
male-to-male_machine_pins.jpg
Filename
male-to-male_machine_pins.jpg
File size
76.57 KiB
Views
1170 views
File license
Fair use/fair dealing exception

EDIT:
This method is just the first thought which came to my mind. There may be other methods to accomplish this without the male-to-male machine pins reaching the top of the PCB.

The following photo is posted earlier in this tread, but I'll post it again here in case you missed it. Here is another interposer design, which is essentially what I am after, but without that extra logic, and with a VRM. They are using PGA168 with with a PGA132 male array. How do you suppose they did this? Perhaps filled the PGA132 vias with solder then used hot air while lowering the PGA132 pins into place?

PGA132_to_PGA168_486SXL2_Interposer.jpg
Filename
PGA132_to_PGA168_486SXL2_Interposer.jpg
File size
421.54 KiB
Views
1165 views
File license
Fair use/fair dealing exception

Plan your life wisely, you'll be dead before you know it.