• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

mAcTX - ATX to Classic Macs

Compgeke

Well-known member
So if we were to switch to a board using a transistor rather than the inverter (smart move really) here's something along the lines of what it'd come out like. With rearranging of the big stuff, I'm able to clear up even more space which adds even more free plane space between the power supply connectors. This means more current handling again. Any traces possible have been routed along the outside as well, rather than through the center splitting it up.

Notable usability changes:

I've moved the fan connectors over to the edge of the board closest to where a fan mounts in the case. During testing I'm getting tired of fighting with the connectors behind the power supply cable. I'd imagine you guys would feel the same. 

The LEDs are now a bit hidden when installed in the system. The good news is you'll still be able to see red or green reflecting off everything as long as we use a diffused LED.

V3R1_Top.png

So, current handling fun: 

Here's the narrowest areas +5V has to pass through. I can run a IIci w/ nubus video card and spinning HDD off just one of those so this will hopefully be plenty. In theory ~75 mil * 4 with thinnest possible copper would net you ~10.5A current capability. Going to 2oz copper (the logical thing to do for a PSU board) would get you closer to the 18A mark.  Ground has more area to pass through than +5V so that won't be an issue. Now, 10.5A is less than a factory IIci power supply however it's also more than a stock IIsi power supply. 
image.png

As far as something like +12V, it's a single 75 mil trace. With 1oz poverty copper that should be able to push 3.5A give or take. Good news is that is higher than a IIci's 2.5A or a IIsi's 0.8A on 12V. 

 
Last edited by a moderator:

Compgeke

Well-known member
A few days late, but the third version boards finally showed up. Built and working 100% no issues for me on my test systems, a IIci, IIcx and a Quadra 650.

As some bonus fun, I can now load up the systems and stay (barely) within tolerance on the 5V rail using only poverty spec 1oz copper. +/- 5% (so 4.75 -> 5.25v) are generally considered acceptable ranges. On final boards they will be thicker than 1oz copper - we're just cutting a good $20/run off of testing boards by using the cheapest option.

aRDftji.jpg.d22a76da2c1240e63d36029519badb1b.jpg


That's a 33 MHz full '040, a spinning hard drive and three 24 bit color cards. A PaintBoard Li, a Supermac Spectrum/24 PDQ and a Radius Precision Color Pro 24xk. Maybe tomorrow I'll dig up some more ram and see if I can up that draw to better match what you'd find inside a compact mac. Maybe a CD drive or something.

36524cf9e6665dc3f924bee0a2be973f.jpeg.c602b67cc752793a13014319ef79b553.jpeg


 

superjer2000

Well-known member
What software do you use to design the PCBs?  Where are you printing and what does it cost for a run?  

I've been thinking of running a couple PCBs for personal use but there is s lot of information to digest with regards to good printers and there seems to be a lot of variability in cost. 

Thanks!

 

blusnowkitty

Well-known member
What software do you use to design the PCBs?  Where are you printing and what does it cost for a run?  

Thanks!
Compgeke does the initial design in EAGLE and I follow up and do final tweaks in KiCad. Printing is done through PCBway who charges $22 for a run of 10 1oz copper boards and basically double that for 2oz copper.

 

Trash80toHP_Mini

NIGHT STALKER
Exciting developments, very neatly done board! I'm curious about deletion of the second mounting point. Constriction of the +5V plane on the solder side of the test boards? All you guys rock! :approve:

View attachment 30720

edit: if you leave a pathway or just a few solder points on your +5V plane free of solder mask, you can tack a wire onto your test boards, making up for that missing second mil of copper. Use an insulated wire with headers hot glued to the board for a jumper on the bodge wire so you can test with current in either configuration?

 
Last edited by a moderator:

Compgeke

Well-known member
Compgeke does the initial design in EAGLE
Of course, if you're working with other people I'd recommend using KiKad There's Eagle can be a bit special when exporting to work with other programs and break silkscreen text placement and those kind of goodies. That, and you'll lose all the device footprints and just get holes in general.

Exciting developments, very neatly done board! I'm curious about deletion of the second mounting point? All you guys rock! :approve:
Not really something heavily needed and the exclusion enables it to fit in a IIsi easier. The 10 pin connector has a fairly snug fit (and really, you could get away with no standoffs) so as a good measure one's thrown on the other side. I ran some traces through there on the last (that should be final - pending IIsi testing) revision to allow for a cleaner layout and more space for the 5V and ground planes. I
IIs_ISizing.jpg

Here's the board we have for testing at the moment (first version with planes). It does work (I can pull a fair amount of current) but I'd like better.
image.png

Here's the current setup, the one that cut out that top standoff hole. A slight modification on the brown board gives significantly better area for the rails.

image.png

And ground:
image.png

 

Trash80toHP_Mini

NIGHT STALKER
Not really something heavily needed and the exclusion enables it to fit in a IIsi easier. The 10 pin connector has a fairly snug fit (and really, you could get away with no standoffs) so as a good measure one's thrown on the other side. I ran some traces through there on the last (that should be final - pending IIsi testing) revision to allow for a cleaner layout and more space for the 5V and ground planes.

View attachment 30765
I missed the deletion there.

Here's the board we have for testing at the moment (first version with planes). It does work (I can pull a fair amount of current) but I'd like better.

View attachment 30764


Here's the current setup, the one that cut out that top standoff hole. A slight modification on the brown board gives significantly better area for the rails.

View attachment 30767

And ground:
View attachment 30766


I figured that's what it was about, with just one mounting point to secure the board to a printed carrier's clips, it can be mounted anywhere inside a hacked PSU can.

Speaking of which: will your PicoPSU unit be adaptable to the Q605/LC series PSU cans? Replace the A/C connector with a socket for a 12V wall wart wired in line with the power switch and wire the necessary 10-Pin thru-hole connections to the 6-Pin LC cable harness? What mods might be necessary to activate the PICO-PSU directly from the power switch?

LC_PSU.JPG

Buck step-down conversion of a higher voltage Laptop or LCD lump-onna-rope A/C adapter might give a lot more oomph to any configuration you're already planning? I've gotten generic Laptop power supplies to scatter about here or there very inexpensively to keep one out of my bag between work and home etc.

 
Last edited by a moderator:

blusnowkitty

Well-known member
We're not planning to support the LC series at this time as that requires -5vdc in order to operate correctly, and our board doesn't provide -5vdc.

 

Compgeke

Well-known member
We're not planning to support the LC series at this time as that requires -5vdc in order to operate correctly, and our board doesn't provide -5vdc.


To expand a bit, the LC is a different thing all together. Physically this particular board won't fit within the LC PSU shell. The LC also doesn't require a /pfw inversion (they don't support soft power) and it uses -5V rather than -12V. We've talked about it and even have some plans however that's an "After this one" project. 

 

Trash80toHP_Mini

NIGHT STALKER
Cool, just thought I'd ask the obvious. An adjustable Buck Converter will output the +12V needed at higher amperage when stepped down from +24V (whatever) less losses from the conversion, no? Maybe too complicated if 12V supplies are easy enough to find with the required power output?

 

cheesestraws

Well-known member
Powering an LC series machine (at least the ones I've tried, anyway) is fairly easy from anything that will give you 5V.  I've got one running perfectly happily off a pair of DC/DC converters on veroboard.  You really don't need a chunky "proper" PC PSU for an LC, a smallish wall transformer will work perfectly well.

(If you want to talk about this we should do it on another thread, though, don't want to threadjack this one)

 

Compgeke

Well-known member
I got the IIsi board today, got it recapped and got it working. Some fun discoveries though.

To start with, the IIsi is possibly special. By that I mean it's /pfw voltage for button press seems to be a bit different. Lemme make a list.

IIsi I have here: 3.4v

@blusnowkitty 's IIsi: 4.4v

IIci: 5.1v

IIcx: 5.1v

Q650: 5.1v.

Notice a slight difference? I do. With the good news - the transistor board is working fine for me. As is the 74HCT04 and zero resistor setup. The bad news - why are the IIsi s giving different readings. Why is the one with the lower readings working and the higher not. I need to probe more IIsi for "wtf is going on" - see which is the anomaly. Good news is there's someone else close to me with a IIsi who also needs some help with another machine. I'll see if I can get a reading off theirs. 

 

Compgeke

Well-known member
More update: For get the whole voltage thing for now as I've now got two IIsi s confirmed working. Got a 2nd recapped and it's all rainbows and sunshine.

DSC_0002.jpg

DSC_0015.jpg

And here's a video of it working:



 

blusnowkitty

Well-known member
While Compgeke has been busy testing out random IIsis, I've been working on something of my own - introducing the mAcTX LC! As the name suggests, it's an ATX adapter board for the Mac LC series, and it's being designed as a drop-in replacement for the LC's PSU when combined with a picoPSU. We're even generating -5v on the board using a MAX660 voltage inverter! I've been looking at several options and I feel the MAX660 is the best option out of what we were testing:

  • TC7660 is cheap, but can only output 60ma peak (LC PSU needs 75ma) and the voltage drop under load is unacceptable - we get -4.5v when the LC is running.
  • 7905 is also cheap and can put out a rock solid -4.9v even under load, but most picoPSUs only put out 50ma peak on their -12v rail.
  • There are Chinese inverter boards out there, but I could only get -2v out of mine.
  • MAX660 is the most expensive (at least in thru-hole form) but outputs a solid -4.8v under load and can maintain 100ma of current.

Unlike the mAcTX II Compact we've been testing for the last four pages, the mAcTX LC is almost ready to go right out of the gate! The rev1 board has the following issues:

  • Doesn't have discharge resistors on-board
  • Drill size on the ATX pads is slightly too small for our ATX connector

These issues have all been fixed in Rev2. I'm also looking into ways of making a breakout board for the SMT version of the MAX660 to try and bring costs down - the thru-hole variant costs $5 per chip!

IMG_4333.JPG

IMG_4334.JPG

IMG_4335.JPG

IMG_4356.JPG

 

cheesestraws

Well-known member
Oh cool, hopefully this will obsolete my worrying veroboard mashup :) .

I've been looking at several options and I feel the MAX660 is the best option out of what we were testing: 


This seems sane to me.  I ended up using an isolated DC/DC converter to do it, mostly because I had one hanging around, but that was even more eye-wateringly expensive per unit than the MAX660, and I didn't really need the isolated-ness of it at all.

By the way, it looks like you're using a reused board connector there.  In case you are unaware, a 7-pin Molex KK396 is a near-perfect replacement for this connector and is readily and cheaply available (I just got a big bag of "compatible" connectors off eBay and they're fine) so one could also use this to breathe new life into LC logic boards that have had their power supplies go entirely missing.  (Forgive me if I'm telling you something you already know but this doesn't seem to be particularly common knowledge).

 

blusnowkitty

Well-known member
Hey mods, can we get the thread renamed to "mAcTX - ATX to Classic Macs"? Thanks!

By the way, it looks like you're using a reused board connector there.
Yep; I've got the part number for a replacement on-hand, I just didn't buy any for initial testing to keep costs down. I do have a pending order for some though!

I've also got a breakout board for the SMT version of the MAX660 inverter IC put together. For basically $2 we can have a DIP-style inverter while using the much cheaper surface-mount parts - though if you still want to use the DIP MAX660 you can cause I'll still be providing a DIP package layout on the board. Because it's surface-mount, I'll be including these boards as a pre-assembled component.

breakout.png

 

blusnowkitty

Well-known member
Update time! We have new mAcTX prototypes to check out!

The mAcTX Compact R4T board respins the board to use a 2N3904 transistor instead of a 7404 inverter. The biggest issue with this board so far is that some of the silkscreen on the back isn't aligned right and gets cut off by the Mac's 10-pin connector. Easy to fix, I just have to do it.

Rev2 of the mAcTX LC also arrived today as well as my SMT to DIP breakout boards. The main issue with this revision is that the resistor footprint is too small. I'm also not fully sure what value to use as a discharge resistor - 100 ohm in this revision drags down -5v too much. I'll be experimenting with different values over the next couple of days to see if I can figure out what I should be using.

IMG_4420.JPG

IMG_4421.JPG

IMG_4422.JPG

IMG_4423.JPG

 
Top