After doing some very tedious testing (if I didn't control the CPU temp, baseline fans off draw would increase with CPU temp increase) involving constantly switching back and forth between zero fan draw and each test (while maxing out the lamptron fan between tests/baselines to keep the CPU temp bottomed out and baseline power draw stable)...
What follows is a large amount of stuff that turns out to be a relatively small difference in my situation. But, I'm satisfied that I've done the testing to settle this subject/decision in my mind, and now I have a big frame of reference for future cooling/fan operations.
TLDR: PWM is more efficient, but not enough to make a difference for me. Only read on if you like nerding out over details.
This is not the most precise testing because I think the Lamptron in particular just rounds the RPM in large chunks like 60, but whatever. It's close enough soundwise etc. I tried to match up the RPMs as close as I could.
There are 5 tests I did for each RPM. 2 tests for the HHE 3-pin fan, 3 tests for the HHE PWM fan.
Key:
HHE LAMPTRON = 3-pin fan on Lamptron controller
HHE MOBO = 3-pin fan on mobo header controlled with DC mode
HHE PWM MOBO DC MODE = 4-pin PWM fan on mobo header controlled with DC mode
HHE PWM LAMPTRON = 4-pin PWM fan on Lamptron controller
HHE PWM MOBO PWM = 4-pin PWM fan on mobo header controlled by PWM
Motherboard PWM mode adds about 0.25w at all times even at 0 RPM, compared to DC mode & no control (which just runs at 12v)
-[ ~1740 RPM ]-
Max's Version of "Quiet" Potential "Low Noise" Sweetspot
HHE LAMPTRON: 2.4w
HHE MOBO: 3.6w
HHE PWM MOBO DC MODE: 2.95w
HHE PWM LAMPTRON: 1.9w
HHE PWM MOBO PWM: 1.6w + 0.25w = 1.85w
-[ ~2000 RPM ]-
Potential Sound Sweetspot
HHE LAMPTRON: 3.3w
HHE MOBO: 4.4w
HHE PWM MOBO DC MODE: 3.35w
HHE PWM LAMPTRON: 2.45w
HHE PWM: 2.1w + 0.25w = 2.35w
-[ ~2220 RPM ]-
Potential Loud Sweetspot
HHE LAMPTRON: 4.2w
HHE MOBO: 4.9w
HHE PWM MOBO DC MODE: 3.7w
HHE PWM LAMPTRON: 3.05w
HHE PWM: 2.6w + 0.25w = 2.85w
-[ ~2520 RPM ]-
Loud But Still Sounds Good
HHE LAMPTRON: 5.5w
HHE MOBO: 6w
HHE PWM MOBO DC MODE: 4.45w
HHE PWM LAMPTRON: 4w
HHE PWM: 3.7w + 0.25w = 3.95w
-[ Maximum, ~2900 RPM ]-
Screamer Junior Status + Garbage Efficiency
HHE LAMPTRON: 7.4w (had a bit less power, couldn't quite maintain 2900, more like 2880 or something, mobo fan would always do around 2900 to 3000, situation was same when fans were swapped)
HHE MOBO: 7.8w
HHE PWM MOBO DC MODE: 5.3w
HHE PWM LAMPTRON: 5.2w
HHE PWM: 5w + 0.25w = 5.25w
--
Some interesting things to note.
PWM Is Indeed The Most Power Efficient
Big shocker there.
The PWM Fan Plugged Into The Lamptron Is The 2nd Most Efficient
This may suggest the PWM model fan is just more efficient than the 3-pin fan model. It's also striking how running that fan in PWM off the mobo header is only SLIGHTLY more efficient than running it in voltage control mode on the Lamptron. (Ignore the maximum speed test, as the Lamptron has less volts and couldn't maintain quite the same RPM at maximum.)
Both Fans Are More Efficient in Undervolting On The Lamptron Than The Mobo Header, But The Difference Narrows As The Power Draw Increases
It would appear for lower power fan applications, at least in the ranges I tested, the Lamptron is more power efficient for undervolting.
Comments On 3-pin Fan More Efficient On Lamptron Than Mobo
Swapping the 3-pin fan from the Lamptron to mobo, we see a consistent rise in power draw which is most striking at low speeds and smaller in % at max speed. One might be tempted to believe it's because of the lower voltage on the Lamptron, but with speeds about matched (I set the voltage setting higher on the Lamptron than on the DC mobo control to achieve the ~same RPMs) this doesn't seem to be the case.
This is hard for me to wrap my head around. It might seem like maybe, just maybe, due to imprecise RPM readings on the Lamptron, it was consistently being run slightly slower in RPM than off the mobo header, but I find it hard to believe the results would be so consistent this way. Also, the difference in power draw, % wise, is striking at the lowest speed tests. This suggests these numbers may be legit.
The % Difference Between Lamptron/Mobo in wattage goes down as the overall draw gets higher. This might suggest eventually the mobo header may overtake the Lamptron in efficiency at a high enough wattage. Perhaps the Lamptron's efficiency curve is skewed toward max efficiency at low wattage. Not a problem for me!
Thoughts/Conclusions
For a PWM fan, PWM control is the most efficient. However, running them in the 1.5w to 5.25w range from the wall, using undervolting control with the Lamptron results in a wholly insignificant loss of efficiency. Even with 5 of these fans, the difference between PWM and Lamptron undervolting would at best amount to a 1 watt total difference from the wall.
One thing I should avoid doing completely is undervolting any fan with the motherboard. At higher wattages this might be different, but clearly the Lamptron is more efficient in that regard.
As much as I love the the sound of the conventional 3-pin fan, the cost of getting to hear that motor buzz is an increase in power consumption compared to the PWM fan, regardless of whether the PWM fan is undervolted on the Lamptron controller, or controlled with PWM on the motherboard. Maybe the difference in efficiency even in undervolting is peculiar to these two nearly identical Delta models only, but it would certainly also be true in any case where two otherwise same fans are running with undervolting or PWM, though maybe the difference wouldn't necessarily be as striking. It's possible in this case the Delta PWM version of the fan was just built to a higher standard for efficiency.
If I were to run 5x of these PWM fans on the Lamptron vs 5x of the 3-pin version, at my likely target 24/7 sweet spot speeds (there's almost no chance I'd be compelled to change it unless I decided I like a different noise better), the difference in total would be 2.5 watts from the wall on the "quiet" 1740 rpm setting, 4.25 watts from the wall at my likely sweetspot setting of 2000 RPM, and 5.75 watts at the high end of my possible sweet spot settings of 2220 RPM or so. Overall, not a huge difference, though it would be a little larger if I ran all those fans on PWM. At maximum speed it would be a 10.5 watt difference. But since I'll probably never run more than 2220 RPM or so, not sure if 6 watts is worth abandoning the electric buzz.
However, maybe if I had 5x of those buzzes it would drive me nuts, with the different buzzes combining into a possible weird harmonic effect, which might have potential to be good, but just as easily would sound very annoying, unless the harmonics were adjusted just-so. The problem is minor voltage fluctuations would continually alter the harmonics of multiple electric motor buzz sounds. Running the 3-pin 1.6a SHE and the 3-pin HHE together might give me some idea of what to expect in those regards. I tend to believe the electric motor buzzing sound may sound best if it's just one distinct buzz, rather than a symphony of them. In that case, my best scenario would be just running one of those specific fans, and then just going with all other Delta PWM fans. Of course, having not tested some other Delta models or even other brands of fans, it's possible I may have other sound factors to play with before settling on the final configuration. It's very likely I will just end up running the 3 fans I ordered + the triblade 24/7, and just adding one more fan. I am not going full hobbyist here. Just trying to make things extra nice for my new build which I expect to last me as a primary system for at least 5+ years. But I need to get a move on with this, as I have other things I need to do and this is not really a project comp.
The most efficient thing for me to do would be to just go with all PWM fans. Likely it would cut down power usage 24/7 by at least 5 watts, if not more. However, this difference is not great enough to be worth abandoning the beauty and function of the Lamptron, nor the delightful sounds of the non-PWM fan models. I would predict that speed-controlling much higher power fans would make a more significant difference in which PWM, or at least more efficient PWM fan models, would be worth going with for me. However, since I just don't require much fan power, that difference won't be enough to sway me.
In conclusion, I occupy an interesting middle-ground that prevents PWM from truly benefitting me. On one hand, I run a lot more fan power and noise than quiet-computing enthusiasts and noise-haters, rendering the potential noise reduction aspect of PWM useless for me. On the other hand, I run a lot less fan power than people willing to run fans very loudly or running huge amounts of fans. The only situation in which the difference would be big enough for me, was if I was running 15+ fans, and if I was running 15+ fans, that probably means I'm at a hobbyist tier I'd only be interested in if I had my own garage workshop, which means I'd already be wealthy and I'd no longer care about shaving heat watts off my setup in a heat-trap room scenario, as I'd probably have an entire SFH at my disposal, or at least the greater part of one.