jmacz Mac IIfx Project

[jmacz]

They are group regulated as well. For the most part the SS250 in my testbench SE/30 is fine but the 5V line sags a little bit below 5V with PDS card and accelerator in place.
It does get problematic for me with 2 PDS cards and an accelerator to the point that the machine stays in reset on power-on.

Hmm, ok. @zigzagjoe you’re using a Seasonic SSP-300SUB in your SE/30 right? I was looking to pick up an SE/30 also and was eyeballing that one. But curious if it has the same risk.

EDIT: cross posted - just saw your recent post. Ok, guess I am still going to have to look for another PSU. But yeah, thanks for your post! Great to know!

 

[zigzagjoe]

Hmm, ok. @zigzagjoe you’re using a Seasonic SSP-300SUB in your SE/30 right? I was looking to pick up an SE/30 also and was eyeballing that one. But curious if it has the same risk.

I am. I'm loading it as heavily as it ever would be in a SE/30, and it's fine, albeit with the aforementioned voltage sag. I expect it will continue to be fine, though I wouldn't buy another if I were intending to load the SE/30 to the gills as I ended up doing. For all other cases, it's perfectly fine.

On the group regulation, to clarify a little more... it's not an automatic death knell, good group regulated designs exist that can tolerate cross-loading without throwing everything out of whack (the Seasonic as an example of this). However, cheapie PSUs however don't tend to be well designed or /good/ anything, though, hence the risk of things going out of whack with the group regulated designs.

 

[jmacz]

However, cheapie PSUs however don't tend to be well designed or /good/ anything, though, hence the risk of things going out of whack with the group regulated designs.

I haven’t put as much on this budget PSU as you have, but it has handled multiple video cards at the same time, with ethernet card, and multiple scsi devices at the same time going through high load. Another one I have in my accelerated 50Mhz IIci with two additional cards is also doing fine.

But to your point, don’t want to risk it, so will look for replacements.

 

[jmacz]

Well, similar to the last time I searched for power supplies that provide enough amps on the -12V rail and are small enough to fit within the stock PSU enclosure, there isn't much out there. And then further reducing the search to PSUs that have either independent regulation or DC-DC, I found only one that was close enough?

ASUS ROG LOKI 750W SFX-L PSU
- https://rog.asus.com/us/power-supply-units/rog-loki/rog-loki-750p-sfx-l-gaming-model/
- has 0.8A on the -12V rail, DC-DC, and SFX form factor so it's small
- Pricey: $170 on Amazon (10% off right now)

There's also this one but it might not be enough on -12V? (0.5A):

EVGA SuperNova 650 G7
- https://www.evga.com/products/product.aspx?pn=220-G7-0650-X1
- has only 0.5A on the -12V rail, DC-DC, ATX size but it's a compact ATX
- $105 on Amazon (30% off right now)

For the ones without group regulation, I think that's it? Went 12 pages deep in Amazon so maybe there's more but not that I saw.

Then there's also a sketchy brand... JULONGFENGBAO... neverheard of them although searching seems to imply various crypo miners use these supplies from China. These have 0.8A on the -12V rail but I can't tell whether they are group regulated or not. One of the items (the flex atx form factor one) has a picture of the insides and from the number of coils, it might be independently regulated... but that item also has minimal reviews and one of them said it burned out... not good. They are cheaper, around $65 on Amazon.

Amazon.com

Amazon.com

Amazon.com

Again, haven't had an issue with my various Macs under load. Although I prize all of them, I think I'm willing to deal with any issues that may come up on the IIci's. But not willing to risk it with my Quadra 700 and IIfx. Will try recapping the IIfx first but if I feel like I need to replace the thing, maybe I'll splurge on the ASUS or wait to see if it drops price further during black friday.

 

[zigzagjoe]

Given there is a certain amount of DIY going into refitting a modern PSU into these, I'd personally find a PSU that meets all requirements except -12v and then put an -12v regulator board to derive a -12v rail with sufficient current capacity. Rather than trying to find a PSU that meets every single requirement.

Or, stick an ammeter on the -12v rail and figure out how much current is /actually/ required; -12v isn't usually used for all that much.

 

[jmacz]

Given there is a certain amount of DIY going into refitting a modern PSU into these, I'd personally find a PSU that meets all requirements except -12v and then put an -12v regulator board to derive a -12v rail with sufficient current capacity. Rather than trying to find a PSU that meets every single requirement.

I'll have to look into this. The rest of the DIY is pretty basic.

Or, stick an ammeter on the -12v rail and figure out how much current is /actually/ required; -12v isn't usually used for all that much.

From what I believe I saw before, it wasn't much, but might as well keep it close to or at spec if I'm going to do the rest. Thanks for the pointers.

 

[bigmessowires]

We all have our personal sense for what's normal or reasonable for vintage computers. Perhaps I'm living with my head in the sand, but I would sleep fine at night with a common ATX power supply in my vintage Mac. In the unlikely event that it crapped out and destroyed my logic board, that would be a shame, but I don't treat my old Macs as priceless museum pieces. Unless this is a one-of-a-kind showpiece or a Mac with extremely high sentimental value, I can't see paying $$$ for premium power supplies.

 

[s_pupp]

… but I don't treat my old Macs as priceless museum pieces...

Shame on you!

 

[jmacz]

I recapped the IIfx power supply, and unfortunately I still have the same issue. The display keeps flickering on/off with any ADB keyboard or mouse input.

Keeping everything the same, if I swap the stock IIfx power supply out for my ATX converted Mac II power supply, everything works perfect, including the external ZuluSCSI device without aux power.

I will need to decide whether to gamble with a cheaper PSU (which is what I have in my Mac II power supply and is working fine) or a much more expensive one with DC-to-DC.

I don't want to splurge on that ASUS PSU which meets all the requirements (it's now $190 as the sale ended on Amazon). Reasonable brand/quality SFX-size PSUs seem to be in the $100 range (which will fit the Mac IIfx enclosure) but none have a -12V rail that is in spec. So I guess I will have to read up on creating a -12V rail from the +12V rail, looks like maybe using a inverting buck/boost converter, if I want to go this route.

 

[bigmessowires]

Maybe you don't need to meet the -12V rating of the original PSU. Does anything other than floppy drives use it?

You can also purchase ready-made modules to give -12V out from a positive DC input, but I they don't look very high quality. https://www.amazon.com/Power-Module-Input-Output-Converter/dp/B07YY6PP94

 

[jmacz]

Maybe you don't need to meet the -12V rating of the original PSU. Does anything other than floppy drives use it?

You can also purchase ready-made modules to give -12V out from a positive DC input, but I they don't look very high quality. https://www.amazon.com/Power-Module-Input-Output-Converter/dp/B07YY6PP94

Not sure about what uses -12v. Can anyone confirm? Although I do use the floppy drive from time to time, but not sure it needs 1 amp? Problem is most power supplies these days only have 0.3A on that rail.

There seems to be many different inverter options, especially on eBay, I just don’t know which ones work and I need to understand what the gotchas are if any. I guess I only need it to support 1A but is it just as simple as feeding a 12v (or whatever it needs) input and I get -12v from it?

BTW, @bigmessowires we discussed briefly a better ATX adapter before. I think the ideal one for me would take an ATX 20 pin input, use an inverter IC for the power on circuit, also have on board a derived -12A rail supporting 1A (fed from the 12V rail), have a set of screw terminals for output (12V, -12V, 5V, 5V trickle, gnd, PFW) so that you can attach whatever plug you want (IIci/IIcx/IIsi/IIvx/Q700, II/IIx/IIfx, etc), and be small enough to fit within the stock enclosure alongside a donor PSU board, as I don’t want the PSU sitting outside like all the existing options. Only issue is that those ATX cables are pretty long so instead of a 20pin ATX jack, I may want input screw terminals instead. Solder pads instead of the screw terminals would be fine too.

 

[jmacz]

There seems to be many different inverter options, especially on eBay, I just don’t know which ones work and I need to understand what the gotchas are if any. I guess I only need it to support 1A but is it just as simple as feeding a 12v (or whatever it needs) input and I get -12v from it?

Actually most of the ones I am seeing are limited to 50-200mA only.

 

[bigmessowires]

I haven't looked at the IIfx schematics, but on the IIcx that I reviewed recently, -12v is only used by the floppy drives. It is also provided to the NuBus slots, but it's doubtful that any common NuBus cards actually use it.

Two years ago I did a lot of floppy drive testing related to my Yellowstone drive controller for Apple II (which also uses Mac floppy drives) - see the data here: https://www.bigmessowires.com/2021/08/13/floppy-drive-pin-9-is-not-my-friend/ None of the Mac 1.44MB floppy drives I tested actually use -12V at all. Possibly it was used in some of the earliest Mac floppy drives like the 400K, but -12V is mainly an Apple II floppy thing. For the Apple II drives where -12V is used, I think it's part of a potentiometer circuit that sets some kind of bias for read/write, but it's not powering anything hungry. I measured the -12V current consumption of a Disk II drive at 2.4 mA.

Nevertheless I personally wouldn't feel comfortable using a PSU with a 50mA -12V rating if the orginal PSU had a 1A rating. 0.8A is probably OK.

If your existing ATX-converted Macintosh II power supply is working and stable in your IIfx, then I would go with that. You can take some supply measurements while the computer is running, to make sure they're all close enough to spec. Set up the most extreme cross-loading test you'll realistically ever encounter, with tons of stuff plugged into 5V and nothing on 12V, and see how the supply regulation holds up. If it looks OK then you have your solution.

 

[bigmessowires]

I was just thinking more about your ADB-related video glitches. I think this could be explained by contention or a weak short circuit on the ADB Data line. ADB is a bidirectional open collector bus, where both the host and the devices pull the bus low as needed, or let it go high with a passive pull-up resistor. In a working system it should be impossible to get electrical contention with this design. But if you ADB Data is weakly shorted to +5V or to another data signal that's high, then you'll see a glitch and a current surge whenever a device pulls the line low in order to send data.

Now why wouldn't you have the same problem when the host sends data? Maybe you do, but the host messages tend to be brief and perhaps brief glitches don't cause visible problems. Device response messages tend to be longer. Or maybe there's a problem with the ADB controller chip on the IIfx where it's actually driving an active high voltage instead of letting it float high, so it can communicate OK but devices can't respond without causing a glitch.

A glitch on the 5V rail will cause glitches in any output signals that are currently high, including video sync signals.

OK, so why does it still work with your ATX conversion PSU but not the original stock PSU? The ATX has a higher 5V rating, so it will be more resistant to a partial short on 5V without the regulated voltage dipping too much.

This is a lot of conjecture but I think it fits your symptoms. Try using your scope to watch 5V and ADB Data while you move the mouse. I suspect you'll find a problem there.

 

[jmacz]

Thanks! This is super helpful and I think very plausible. I will dig into it to further.

 

[jmacz]

@bigmessowires you were right. Watched pin 1 (data) and pin 3 (vcc/5v) on my ADB port while moving it and I see movement on the 5v line. Also unplugged the LCD monitor and watched vsync, hsync, red, green, blue, and all have some movement when moving the mouse. Definitely was barking up the wrong tree. Looking further now to figure out what's going on with ADB. Thanks for the pointer.

EDIT: Watched the same thing with my ATX MacII PSU and there's movement there too... but it looks like with the stock PSU, mouse movement causes a slight drop in voltage whereas with the ATX MacII PSU, mouse movement causes a slight increase in voltage.

 

[bigmessowires]

How much change are you seeing on 5V and on the video sync signals, during the ADB activity? A little bit of cross-coupling wouldn't be unusual... maybe like 100 or 200 mV. That level of bouncing wouldn't be great, but probably wouldn't cause the desync you've seen.

Maybe check that the +5V and GND supply pins on your ADB transceiver chip are solid low-ohm connections to the PSU supplies. Also check the +5V and GND pins on the external ADB connectors for the same.

 

[jmacz]

How much change are you seeing on 5V and on the video sync signals, during the ADB activity? A little bit of cross-coupling wouldn't be unusual... maybe like 100 or 200 mV. That level of bouncing wouldn't be great, but probably wouldn't cause the desync you've seen.

I am seeing about plus or minus 200mV of change. When the monitor blacks out, if it stays black for long enough, I do see the LCD error dialog pop up and it's complaining about the sync being out of range. All the lines, red, green, blue, vsync, hsync are fluctuating about +/- 200mV but that's the only thing I can see. I've checked all the other pins (just in case) and don't see any change during mouse movements. It's just that +/- 200mV... it's about -200mV on the stock PSU and +200mV on the ATX PSU.

Will check the ADB chip tomorrow.

 

[bigmessowires]

Here's what you're dealing with on the IIfx:



Pin 1 on the ADB connector is data. There is an inductor involved (UB15) and maybe if that's damaged, or even if it's not, maybe that's acting like an antenna somehow? Just speculation... I have a Ham radio license but I'm still not great at that stuff. Or there could just be a weak short inside the UB15 component between the data line and another ADB supply line. I don't quite follow that schematic symbol. It's some type of quad-inductor?

There's also a 470 pF capacitor connected to the data line: C60. I would definitely check that one. Make sure it's actually connected to ADB connector pin 1.

Make sure there is a high amount of resistance between pin 1 on the ADB connectors and each of the other pins, especially pin 3.

Q1 at location 1A is used to pull the data line low, when the Mac is sending ADB commands. If it were damaged, you'd have trouble with ADB communication but I don't think it would cause glitching like you're seeing.

<speculation>Your symptoms might be explained by a weak short between ADB data and ADB +5V on the ADB connector side of those UB15 inductors. This might not cause a huge problem when the Mac is sending data, because the short to +5V is on the other side of the inductor from the Q1 transistor that's trying to pull low. But when devices tried to send data, it would be a more serious problem, C62 would be partially discharged and a surge of current would need to flow through those inductors to replenish it even after the data bit ended.</speculation>

If you still can't find the cause, I would start replacing all the components in the red box one by one.

When you see voltage bouncing on other signals during ADB transmissions, does it happen during the whole transmission (command and response), or only during the response part? See here for a diagram: https://www.bigmessowires.com/2016/03/30/understanding-the-adb-service-request-signal/ Everything from the beginning up to the stop bit comes from the Mac host, and everything after that (if there is anything) is from the answering device.

 

[jmacz]

When you see voltage bouncing on other signals during ADB transmissions, does it happen during the whole transmission (command and response), or only during the response part? See here for a diagram: https://www.bigmessowires.com/2016/03/30/understanding-the-adb-service-request-signal/ Everything from the beginning up to the stop bit comes from the Mac host, and everything after that (if there is anything) is from the answering device.

That Saleae you have is super nice but super expensive.

Will take a look further on this as well. I guess I have my next debugging project.