[ignore]
That missing tranny is actually marked "SS". Looking it up is is an N-channel mosfet.

https://www.mouser.com/Semiconductors/Discret … =BSS138&FS=True

As for the caps, I'd snag a cheap mlcc capacitor assortment and just sort of eyeball it. Just choose the one closest in physical size at 16v
[/ignore]

luckybob wrote:

That missing tranny is actually marked "SS". Looking it up is is an N-channel mosfet.

https://www.mouser.com/Semiconductors/Discret … =BSS138&FS=True

As for the caps, I'd snag a cheap mlcc capacitor assortment and just sort of eyeball it. Just choose the one closest in physical size at 16v

Wrong thread?

Also you mentioned earlier using MELF components I personally tend to use 805 or 1206 any reason why you'd use melf over those?

how the happy hell...

anyway, i just like the look of Melfs.

And in the business of vintage computer hardware, looks account for a large bulk of the lust.

feipoa wrote:

And in the business of vintage computer hardware, looks account for a large bulk of the lust.

Point taken... I have no preference personally as long as they aren't larger than the laternative. Looks kike they tend to be color coded (personally not a fan of color coding).

I'm not quite sure what I'm looking at. The PGA168 side should be female, while the PGA132 side is male. Are you intending the user to place male-to-male pins into that PGA132? I just as well assume use the male-to-male pins right onto the PCB to reduce height, increase beauty, and reduce touch/pressure-based connections.

I assume you placed the VRM on the bottom so that the DIP block could be on the top?

What is the function of the two pin header? 5V and GND? MEMW#? My thought was to turn the header another 90 degrees to ensure that it is not in the way of some motherboard component.

feipoa wrote:

I'm not quite sure what I'm looking at. The PGA168 side should be female, while the PGA132 side is male. Are you intending the user to place male-to-male pins into that PGA132? I just as well assume use the male-to-male pins right onto the PCB to reduce height, increase beauty, and reduce touch/pressure-based connections.

I assume you placed the VRM on the bottom so that the DIP block could be on the top?

What is the function of the two pin header? 5V and GND? MEMW#? My thought was to turn the header another 90 degrees to ensure that it is not in the way of some motherboard component.

I don't happen to have a male version of the 3d part.. sooo.... yeah ignore that 🤣 It's just ther so I can double check that I don't place something in a place where it would block the socket etc...

see the image description anything beyond that is irrelevant! 😜 ... and yeah... I kinda figured the PCB would deflect a little heat too...instead of it cooking the underside of the CPU a little. The 2 pin header is just chillin there untill I can bother to make it a 4-5 pin header with GND 5V 5V REG_VIN EX_REG_VOUT EXT_REG_VOUT MEMW so yeah 6 pins... you'd put a jumper on REG_VIN to select the desired voltage input.

I am fairly confident that nobody here is going to add an external power source to this. My recommendation is to keep it simple. I haven't, for example, seen anyone else on the forum, besides myself, who would replace a resistor with a trimmer to increase voltage to their CPU. Same for modifying a POD83 with a diode to get it to run at 4V and 100 MHz. The user base here seems a bit more focused on use. What generally happens is people move onto something else when they need higher specs. During prototyping, I tested about a half dozen regulators for noise, reliability, and heat at 735 mA. The linear regulator was surprisingly cool after x minutes compared to the switching. I am not particularly concerned about needing more voltage for a SXL2 chip. These are essentially 386 chips with benefits.

I liked your idea of having two models. Have you abandoned this?

We'll it's not like its even done yet but once it is modifying it to use a different regulator should be trivial... for anyone at least anyone willing to put a couple hours effort into modifying the design.

Also going to have to rethink the 132pin package footprint... I accidentally mirrored it. The footprint in the datasheet is obviously for a CPU socket... but this is on the bottom layer of a PCB so when I put the part on there send it to the other side it gets mirrored.

Is there an option to put a male-male pin? And just put 132 such pins spaced at 2.54 mm, or whatever it is. That should eliminate the mirroring issue as well.

feipoa wrote:

Is there an option to put a male-male pin? And just put 132 such pins spaced at 2.54 mm, or whatever it is. That should eliminate the mirroring issue as well.

No the pinout has to match the schematic symbol... it's actually less work that way. You technically could individually wire male pins yes but that'd be alot harder in the end since you'd be ignoreing where the cad tool can actually help even though it's getting in the way in this case because it doesn't allow putting a part on the other side without flipping it.

What software are you more familiar with? I seem to recall something called Protel being the thing back in the day. Maybe it still is.

Protel became Altium.

gdjacobs wrote:

Protel became Altium.

Yep, still the go to PCB software in the industry as far as I know, never used it myself but I know people that have.

If you don't have a couple thousand to burn on Altium though or even if you do you'd probably just settle for KiCad as it does most of what you need and its getting a lot better especially since CERN has been paying for development work to be done on it (yeah the particle accelerator guys).

And I quote "Altium Designer 14 pricing starts at $7,245 USD, including a one year subscription."

Another free but not open source PCB cad is DipTrace... I can't really compare it to kicad but I seems reasonable at first glance, I didn't even know where to start with Eagle though I haven't tried it again recently since I know a bit more about how PCB design works... no point as it doesn't have the features I want at the price point I want (free to under $100) Also note that KiCad 5.0 can now largely import Eagle projects its a new feature.

A few months ago, I started some Eagle tutorials, then noticed that only 2 layer boards were included in the freeware part of the software. That was the end of my Eagle experience.

feipoa wrote:

A few months ago, I started some Eagle tutorials, then noticed that only 2 layer boards were included in the freeware part of the software. That was the end of my Eagle experience.

Yep, btw you can download the final KiCad 5.0 release now... if you already installed just install it over top of the previous 5.0 nightly. It's roughly eqivalent to Eagle... just without the license limitations.

KiCad can do 32 copper layers + 14 technical and 4 auxiliary layers and doesn't have any real size limitaitons either. I was watching a KiCad presentation awhile back and someone was asking for greater than 32+ copper layer support I think it is apparently a technical challenge but they were open to the idea.

A good place to start is to just do a microcontroller project, (pick a MEMS osc instead of a crystal as it is a bit more fool proof) https://datasheets.maximintegrated.com/en/ds/MAX7375.pdf if you pick a microcontroller like atmega328, atmega32u4, pic32mx270 or pic32mx274 etc... you can use the Arduino toolchains with your board or the offical vendor toolchains if you want to go all in on it. I've been playing with the pic32mx270 it's amazing how much faster a 40Mhz 32bit micro is than a 16Mhz 8bitter. Probably fast enough to emulate an IBM PC full speed.

cb88 wrote:

gdjacobs wrote:

Protel became Altium.

Yep, still the go to PCB software in the industry as far as I know, never used it myself but I know people that have.

Altium and Orcad/Allegro are the two giants in the ring that I'm aware of.

I'm sure this will make me sound like I'm a little touched in the head, but have you tried contacting TI and asking them if they would be willing to give you a PDF copy of the 1995 "TI486SXLC and TI486SXL Microprocessors" Reference Guide? (I'm assuming the 66 Mhz part came out in 1995, if not ask for the appropriate year instead, 1996, etc...) If there is one copy left in the world, I'm pretty sure that's where it would be. I mean, what reason would they have for not giving it to you? It's not like its confidential information. It was meant to be a reference for those that work with the various processors. They must have an intern somewhere they can send to try and find it somewhere in their extensive archives.

When it comes to persistent dead-ends, I always go to the source, do they usually say "No". Yes. But, every so often I get a "Yeah, sure, you can have it." and it is a glorious day.

I'm not sure what aid a newer reference manual will provide. It is easy enough to over spec. a voltage regulator for uncertainty concerning current draw.

I meant just so you can know everything about the processor inside and out