Iyonix questions

I’m trying to resurrect my Iyonix, prior to maybe selling it, but it isn’t going entirely smoothly. It’ a classic case, with a podule backplane but with no podules. The ROM has RO 5.22 (13-Apr-15).

Which slot does the video card normally reside in? Mine has it at the far end of the motherboard, closest to an edge of the motherboard. The Iyo only wakes up reliably when it’s warm. I’ve seen it wake up, clearly running the OS, but with no video output.

It ran for a long time, before I pensioned it off, with a SanDisk 64GB SSD. A couple of days ago I began to delete files from it, only to see “Filecore in use” partway through deleting NetSurf’s cache as it happens (lots and lots of folders and small files). Since then, despite reformatting (reinitialising, really) with HForm 2.75 on my BBxM, when it’s back in the Iyo the machine startup doesn’t complete with “Disc not understood – has it been formatted?” Needless to say, the disc checks out fine with DiscKnight on the BBxM, and is properly dismounted before removing it from the BBxM. Any idea why this should be, or what I can do to fix it or to diagnose it further? (Of course I have to use it with different interfaces; the BBxM needs SATA to USB; the Iyo needs SATA to PATA.)

It just occurred to me to try booting without the SATA-PATA interface, and plugging the SSD in to USB using the same SATA-USB interface as when I connect it to the BBxM. The result is that it boots to the same basic desktop but with the message “Disc drive not known”. There’s no hard drive icon. When I plug the SSD in, a HDD icon for drive 4 appears, but clicking on it gives “Error when reading SCSI::4.$ – Disc not understood – has it been formatted?”

Not sure about the SSD issues (did HForm start producing a newer format?), but the PCI slots on my Iyonix are (top to bottom in tower configuration): spare, USB, video, video fan. For machines without a video fan, some old pics suggest that it’s spare, spare, USB, video.

I don’t understand why HForm never even mentions LBA; it only talks about CHS. Is that realistic for an SSD?

It ran for a long time, before I pensioned it off, with a SanDisk 64GB SSD. A couple of days ago I began to delete files from it, only to see “Filecore in use” partway through deleting NetSurf’s cache as it happens (lots and lots of folders and small files). Since then, despite reformatting (reinitialising, really) with HForm 2.75 on my BBxM, when it’s back in the Iyo the machine startup doesn’t complete with “Disc not understood – has it been formatted?” Needless to say, the disc checks out fine with DiscKnight on the BBxM, and is properly dismounted before removing it from the BBxM. Any idea why this should be, or what I can do to fix it or to diagnose it further?

The Iyonix appears to be very susceptible to power supply variations and aging, because it appears to require voltages within a subset of the ATX specification (based on the information that John Ballance posted on the Iyonix list years ago). In particular, the kind of changes to the PSU’s output waveforms that you see as the output capacitors start to tire would seem to be the kind of things that it doesn’t seem to like.

The first things to start to fail are the network and disc interfaces, as I recall (again, from the Iyo list and first-hand experience). I seem to recall Martin Avison writing a small utility to hit the disc hard using lots of file reads or writes to test this problem; I could be thinking of someone else, though.

It’s a very hard thing to diagnose – and that’s speaking as someone who has designed SMPS for a living – because the waveforms on my “faulty” Iyo PSU, the two failed relacements (see below) and the “good” one that I eventually got to replace it all met the ATX specification. The differences were simply the absolute levels of supply rails (within the required limits), and the size of the ripple on the rails. And, of course, the voltages on the 3.3V, 5V and 12V rails in an ATX PSU are almost always interdependent, so that all three will vary as the loadings change on any one rail.

As I recall, John stated that the 5V rail on a “good” PSU should be above 5.06V. This, as any ATX PSU designer will know, is impossible to guarantee without insder knowledge of the design of an individual PSU, as the spec simply requires “between 4.75V and 5.25V” at all times¹ – and a good PSU design will utilise all of that leeway. After trying a couple of good quality PSUs from my local PC wholesaler, without success, I ended up buying a “replacement Iyonix PSU” from CJE Micros. Those other two supplies went on to do many years of sterling service in more conventional computers (one might still be powering the media PC sat behind the TV, in fact).

TL;DR: your symptoms as described exactly match the ones that I saw when my Iyonix decided that it didn’t like its PSU any more.

¹ With some limits on the shape of the ripple, as I recall, before the pedants arrive; I’ve not been in the ATX field for a few years, and don’t have time to look the spec up now.

as the spec simply requires “between 4.75V and 5.25V” at all times

That’s why I think it is “odd” powering computer devices from USB, as the spec requires 5V, with a leeway of +0.25/-0.60 (traditional) or +0.25/-0.55 (USB3). That means the actual conformant output can be 4.40V to 5.25V…

I run my Beagle xM from an old Iomega Zip power brick, as it’s the only thing I have that’s bang on 5V with enough current to successfully run the board.
My EeePC could power the Pi1 from a single USB port.
My desktop PC shuts down the port as running overcurrent. :-)
[thinking: this might explain why my USB DVD thingy never worked right in the PC?]

Now I run everything (except the Beagle) from a five-output power supply.

Point is, what devices require as power input and what the specs say is acceptable output can differ. The output can vary slightly as devices age. Better engineering would mitigate that, but what’s the point when those changes are still within spec? Unfortunately for us, the spec is aimed at a traditional PC, and the Iyonix is more sensitive to such changes.

A couple of days ago I began to delete files from it, only to see “Filecore in use” partway through deleting

I believe Jeffrey fixed this issue back in 2017, if you’re using ADFS. The thread that covers the issue/fix is here

That’s why I think it is “odd” powering computer devices from USB, as the spec requires 5V, with a leeway of +0.25/-0.60 (traditional) or +0.25/-0.55 (USB3). That means the actual conformant output can be 4.40V to 5.25V…

If something is designed to be powered from USB, it should be taking that into account. In just the same way that your “230Vac” supply in France is unlikely to be exactly 230Vac (so your Iomega power brick should be happy working from 198Vac to 264Vac at the extremes), “5V” is rarely 5.0V on the dot.

Point is, what devices require as power input and what the specs say is acceptable output can differ. The output can vary slightly as devices age. Better engineering would mitigate that, but what’s the point when those changes are still within spec? Unfortunately for us, the spec is aimed at a traditional PC, and the Iyonix is more sensitive to such changes.

Er, no. Anything that claims to be ATX compatible should be happy with what the ATX spec defines as “5V”: that is, something between 4.75V and 5.25V with some defined forms of ripple on it. If you need >5.06V, then you’re not ATX compatible, and if you need to run from an ATX supply then you’ll have to find a way to derive that >5.06V from what the supply is defined to produce.

I could also dispute the “better engineering” bit, too. A standard ATX supply is actually a very neat piece of engineering: it often manages to generate three supply rails at high current from a single power converter, using the relatively wide variations in the spec to give the voltages room to move around as the load shifts between them. You could easily get better regulation by throwing more money and components at it, but instead a cleverer and more pragmatic solution has been found.

Add to that the need to set up the new supply so that as the output caps start to age, the voltage ripple doesn’t fall outside spec until the unit has worked its design life, and it’s far from “simple”.

so your Iomega power brick should be happy working from 198Vac to 264Vac at the extremes

Until our electricity got changed (when the smart meter was fitted), it was a rare day when it would top the old 220V. Turning on the kettle would often make it drop below the 198V you’ve mentioned. Indeed, I used to know when the kettle (or immersion heater) was done by noting when the lights got a little brighter.
More reliable electricity and LED bulbs mean I know have to get off my fat ass and go look. ;-)

Anything that claims to be ATX compatible should be happy with what the ATX spec defines as “5V”

Is the Iyonix “ATX compatible” or does it just use an ATX power supply? The two aren’t quite the same thing.

I could also dispute the “better engineering” bit, too. A standard ATX supply is actually a very neat piece of engineering:

You could easily get better regulation by throwing more money and components at it,

That’s what I meant by it being better engineered. Something designed to military spec may well be able to hold a steady 5.06V through an apocalypse, but if it’d cost more than the computer it is running……
…the thing is, given the variations permitted by the spec, it allows for a power supply to be engineered to an acceptable price point. This does mean that something connected to it must be able to cope with, say, 4.75V on the 5V rail.

TL;DR: your symptoms as described exactly match the ones that I saw when my Iyonix decided that it didn’t like its PSU any more.

The first things to start to fail are the network and disc interfaces, as I recall

The disc interface (PATA) is not in use; I’m plugging USB devices in (I’ve tried the SSD and an SD card with reader). Same result with both. And of course the network interface is not in use because the thing won’t boot, therefore I can’t Configure anything and thus I can’t bring the network up.

The two main PSU mlines read 5.06V and 12.06V on one meter, measured at one of the Molex HDD connectors. I’ll check with my other two meters tomorrow, as none of them has been calibrated since manufacture. I’ll also check the 3.3V line and I’ll have a look at the ripple.

This PSU is the third one the Iyonix has had (i.e. the second replacement), but is not an “official” one. It should have had nowhere near enough use to make any capacitors dry out. But we’ll see when I get a ’scope on the lines tomorrow.