tanasen wrote:I can't find any info on this in the manual.
Here is a picture of the pinout:
https://i.imgur.com/EOuiSbK.jpg […]
Show full quote
I can't find any info on this in the manual.
Here is a picture of the pinout:
That's just the pins, doesn't tell you much about the pinout.
Solution is quite simple really - if you have a Voltage meter. "+5V" is exactly what it sounds like. So if you check the pins one by one measuring between the pin and a known ground, you should find the two +5V pins easily enough. If they're at the same end of the block, you have the new pinout and can just plug in a regular 2x USB bracket. If they're at opposite ends of the block you have the old one and need to somehow turn around half the connector (exactly what options you have depend on the connector itself).
If you don't have a Voltage meter...
1) Get one. Invaluable for hardware diagnostics in any case.
2) Just plug it in and see what happens (not recommended, but +5V is generally not enough to fry anything)
3) use a 5V LED (i.e. HDD LED) and do the same as with the meter. Put the -/GND wire from the LED onto the known ground, use the other one to probe the pins one by one.
I 've read that you can install the nusb33 driver in OSR2 and have proper usb support. Am I wrong on this?
Last time I checked that was Win98SE only, but tbh it's been a very long time since I attempted anything with Win95 and USB, so I might well be wrong.
jheronimus wrote:[...]
However, if I'm not mistaken, it also doesn't mean that you can just use modern ATX power supplies. Supposedly, they don't provide enough power over the 5V rail that some older hardware needs. I've never tried a modern PSU though, just noticed a lot of discussion of 5v rails whenever the topic of ATX comes up.
Yes and no. Yes, there's discussion, but no, with a board with low requirements like this, almost any modern PSU could handle it easily. You just would need to go overkill on the PSU, which would be hopelessly inefficient.
The discussion is quite simple really: traditionally, the CPU and PCI cards drew their power from the 5V line. When AGP was added to the mix, it drew most of its power from the 3.3V line. So the 12V line was really only used for powering the motors in disk drives and some legacy ISA stuff. That means that in the late 1990s and early 2000s, the 5V line needed by far the most current (even if total power over 12V could be higher), with 3.3V in second place, particularly in the later AGP years. This was the case up to Intel's Pentium 3, but AMD continued down this path with the Socket A platform for a few years more. A typical PSU from this era (like my FSP350-60PN) could deliver 28A on 3.3V, 30A on 5V and only 18A on 12V (note that these currents don't add up to 350W, because no PSU can handle the max current on all lines simultaneously).
The problem here is that delivering a lot of power at low voltage needs lots of thick cables, or the losses mount. When Intel introduced the P4, it knew how much power they would need and realized that getting that out of the 5V line was not a good idea. So they introduced the ATX12V standard, which is the little 2x2 (or in high-end 2x4) yellow and black connector that plugs in to the motherboard next to the CPU. AMD followed suit with the Athlon64. Suddenly the biggest single power user moved from 5V to 12V. That means that power supplies got re-engineered to deliver more power over the 12V line, or more commonly over two separate 12V lines. Compare my FSP ATX-350PNR with its older brother - 20A on 3.3V, 16A on 5V and two 12V lines delivering max 23A together.
Then PCIe came along and also moved most of the heavy lifting to 12V. So the two biggest draws were now both on 12V, and 5V basically just serviced the legacy PCI bus and some I/O. A recent 350W PSU (say for comparison's sake the FSP350-60APN) delivers 21A on 3.3V, 15A on 5V and max 32A on two 12V lines (note again that these currents don't add up to 350W, because no PSU can handle the max current on all lines simultaneously).
Now, the problem is one of efficency. A PSU is most efficient at about 80% load. If you only load the 5V line on a modern PSU, even if you completely max it out, the total load will be far below that 80%. That's bad for your electricity bill, but as energy is never lost, only converted, that "lost" energy is converted to heat, and heat is evil, because it stresses compontents (those poor caps that have a tendency to die anyway) and causes noise when you try to get rid of it.
TLDR: it's about balance. An early ATX PSU is well-balanced for a So7-SoA system, but unsuited to anything newer. A recent ATX PSU is well-balanced for an ATX12V+PCIe system, but unsuited to much older systems.
But that said, you only really hit problems on later So370 and particularly SoA systems with CPUs (and GPUs) that consume many times the stuff you intend to use here.
tanasen wrote:[...]
How much power do I need over the 5V rail for such build? My little seasonic has 14A output on the +5V rail. Is this enough? Also, about the adapter, do I need one with a power switch and -5V conversion?
-5V is only relevant for a few old ISA cards. It's unlikely that ESS1868F needs it.
As for how much you need, if you assume that the CPU and PCI cards draw all their needs from 5V and round up decimals, you get 16W for the CPU, 4W for the Riva and 3W for the Voodoo1. That gives you 23W. As W=V*A, that means less than 5A on the 5V line. So 14A will do the trick esaily. But you'd be better off with a lower-rated (but still good quality) PSU, say a 1990's ATX (or even better: AT) 200W PSU. That would probably still offer you more on the 5V line than 14A, but with far less unused capacity on the 12V (all the more so as you're not using a real HDD), so it would be a lot more efficient = cooler. However your modern Seasonic won't spontaneously combust if you use that for now - just be aware that a period PSU - even one with a much lower rating - would be a better idea if you see one.