Horun wrote on 2025-02-15, 02:33:
I am going off his sig which puts max power about 110w assuming also 1 HD and 1 CDROM. Double that for safety so a decent 250watt psu would be more than good, a cheap 200w may not be IMHO.
Nah, I looked up the hardware in more detail. Even just about any "gutless wonder" PSU will run this rig with room to spare for the power draw. Of course, one may not want to do that for obvious reasons. 😉
That K6-III CPU is 29W TDP max... make it 30W wit VRM inefficiency. The Matrox GPU is less than 5W total for the entire card (I have one and measured its power draw many years ago.) The Voodoo cards are around the range of 10W each. HDD 20W during spinup, but less than 1/2 of that when running... but let's give it 10W for an easy number. CD-ROM fully spun up, about the same. So total comes to about 75W. Give another extra 10W for chipsets and other hardware, and we're up to 85W range absolute max (CD-ROM reading a CD, HDD writing, CPU @ 100% and both GPUs @ 100% load.)
A 100-Watt PSU will still cover this (unrealistic) scenario, though may stress a bit. A "true" / decently-built 150W PSU will run such a rig with absolutely no sweat (provided it's a proper era-appropriate PSU, expecting to provide more load on the 5V rail.)
Gennadios wrote on 2025-02-15, 02:08:
Since you 'had a look' at this cultist network, you would have naturally seen the 'PSU' tab which has detailed reviews of every model including the tested power loads and internals pictures, right?
I could not find many of the PSUs from that list.
Also, if they were reviewed elsewhere, why didn't they include a link to that?
You see, to me this is "lazy reporting". Basically, it's just a list of PSUs. How do you expect me to take that for granted / as 100% truth?
Horun wrote on 2025-02-15, 02:33:
There are only 26 PSU reviewed under that tab (counted them on all 5 pages). And all ATX case types are 650watt and above. No need for a 650 watt psu to power a Pentium 😀
Exactly!
Also, I looked up a few random reviews in the "D" & "F" tier list. I can sort of understand why some of these PSUs were placed there when it pertains to using these to power modern PCs. After all, modern PCs can have power draw spikes that are nearly double of what the hardware might be specified to nominally (particularly high-end GPUs.) But when it comes to retro/vintage hardware, this does not apply at all.
That being said, some of the group-regulated units in the D and F tier list should still probably be avoided for retro PCs, mainly because these PSUs may use an old topology that was "tuned" for the power demand of modern PCs (i.e. expect most of the load on the 12V rail.) So using any of these in a retro PC with 5V-heavy draw could make the 12V rail go above spec.
Horun wrote on 2025-02-15, 02:33:
And there is no mention of Delta branded PSU which historically have been extremely good.
Old Delta's and HiPro's (Chicony) are my go-to units for retro PCs and sometimes testing even not-so-retro stuff. 😉
Gennadios wrote on 2025-02-15, 02:08:
Secondly, the AT PSUs I've pulled out of my cases had cheap garbage steel clad wires.
Just because that was your "luck" with a few still doesn't mean that ALL AT PSUs are built like that.
I always like to evaluate my PSUs on a per-unit basis.
Speaking of which, that cultists network place puts all non-APFC PSUs (ones that have a voltage selector switch) in the "F" tier list... which I find funny. But again, I can see some point to it, at least when it comes to newly-made PSUs, as lack of APFC typically means it's likely going to be a cheapo unit that probably cuts corners in many other ways... and probably based on an old topology that may not be able to handle the harsh (IMO synonymous with trash) power transients of new PC hardware. So I can give that a pass. But in reality, the inclusion or exclusion of APFC does not automatically make a PSU good or bad. Rather, it's just a good "marker" to identify old and/or new cheapo units that may not be able to handle the latest modern PC hardware that has silly power spikes (ATX 3.0 spec.)
When it comes to retro hardware, though, those power spikes don't exist... so for such, this is a completely irrelevant exclusion.
Another exclusion they have there that would theoretically put a PSU in the "F" tier list is "Any units released more than 12 years ago, bought more than 10 years ago, or made for ATX spec v2.2 or earlier due to the age of design and components"... which again, I can understand when it comes to using these to power modern PC hardware. But again, this does NOT apply to retro hardware.
So again, this is why I cannot recommend for anyone to go off from such "PSU lists" blindly and think they are the ultimate truth. Rather, take the time to try to understand what they mean before making a decision about a PSU purchase.
In my case, most of the PSUs I use for my own stuff were specifically chosen to be ones *without* APFC. It's not that there's anything wrong with APFC. However, it does stress the main input filter cap quite a bit more. As such, the life expectancy of APFC PSUs will typically be lower compared to non-APFC PSUs... and my own experience with repairing these affirms this notion. A lot of times, the bulk input cap goes bust - even if it's a quality Japanese brand sometimes. This often results (about 50% of the cases) in many other components to go *bang* on the primary side, making the PSU not worth to repair. Meanwhile, non-APFC (PPFC or no PFC units at all) don't put this kind of stress on their primary cap(s), and even PSUs that use completely crap-cap brands almost never have their primary caps go bad. On the occasion that this failure occurs, the result is -at worst- a PSU that either shuts down / cannot stay powered up or just dumps lots of line frequency hum on the output, typically making the PC unstable, especially if something disrupts the AC line. So when it comes to reliability and life longevity when all other factors are equal, non-APFC units should be considerably more reliable and expected to have longer life time. My experience with decommissioning old industrial hardware reaffirms this notion.
Gennadios wrote on 2025-02-15, 02:08:
Good quality motherboards have better fault tolerance and are more likely to blow before your motherboard does.
??
Not sure what you mean by this.
Perhaps: "Good quality motherboards have better fault tolerance and are more likely to blow before your PSU does." ??
If that's the case, I'd like to see the source of this kind of info... or at least good detailed examples.
Again, I can tell you from industrial hardware (with both old AT and ATX PSUs) that the motherboards were never damaged by the PSU, so long as it was a decent brand and built well. One would think the dirty AC power in such environment would have had an effect... but not really. And it's worth noting that some of this "industrial hardware" was often just plain consumer-grade hardware, so not always anything special. That said, about all of the motherboard failures were really due to bad electrolytic caps from (now) known crap brands... or in rare cases, the mobo dying due to noise/spikes coming in from connected peripherals (i.e. COM port -connected hardware and similar with long unshielded cables running near noisy power-hungry machinery) - this is nothing even a "super-fancy" modern PSU will be able to do anything about.
Anyways, sorry for the wall of text... but it really irks me when I see generalized and "sub-par" info posted about retro hardware & PSUs. It's as if we've been brain-washed or completely reset about what retro hardware expects in terms of power supplies. It's not like using a modern top-tier "A"-list PSU from that list above would do anything bad to the hardware... but it's really a waste of money and a misnomer to think that this is the only way to power a retro/vintage box.
