[X] My Assistant

I actually had that exact thought- why don't I make some sort of doubker for the tach line.
Some research found that the tach signal is 5V open collector. There's two pulses per revolution, and the output (with a pull diwn resistor) is roughly a 50% duty cycle squarewave.
Some quick number crunching for reasonable limits:
200RPM (minium I expect to double) = 6-2/3Hz tach signal = ~75ms long pulses
300RPM = 10Hz tach = ~50ms pulses
2500RPM = 83-1/3Hz = ~6ms pulses
3000RPM (max to double) = 100Hz tach = ~5ms pulses
10K RPM (entirely plausible for server boards) = 333-1/3Hz tach = ~1.5ms pulses
Most of the getups I saw for multiplying a square wave were based on a delay made from a comparator, a buffer, or RC filter, then they fed the phase shifter signal and the original into an XOR gate. That approach isn't good for a terribly wide range, like an order of magnitude. If I really wanted to make that work, I suppose a workable setup would be to trigger a 2ms-ish pulse on each edge of the input signal, and run that through an open collector buffer.

The basic bitch way would be to intercept the i led wires, and block short positive pulses.

More basic: put everything on a single header with a splitter, use the rear panel fan for tach.

Most basic: Lift the tach wires, and hope the board doesn't do current sensing.

Neat idea, but I have too many open projects as is.

Another idea: Just use the power lines to engage an NE555 timer powered; if the fan seizes I think(?) (been a long time since i actually tried this, like a decade or more, so I've likely forgotten something basic) that'd prevent current from flowing and turn the NE555 off.

I do not know how much current a DC motor in a fan draws, but an NE555 should work on up to 16V DC, so if the current isn't too high that should be fine. If not, power it via a MOSFET and use a second MOSFET to cut the timer's output when the fan seizes.

Might not even need a second mosfet, but I have to run out now so I can't spend more time thinking about it.

This post has been edited by dragontamer8740: Oct 28 2021, 19:27

Oh, I'm not trying the control the fan's speed. The motherboard is driving the fans properly- but the fans respond to their PWM input as a speed % rather than a direct speed. My aftermarket fans are working properly, and producing enough airflow- but crucially, their maximum speed is a bit slow for the board, ergo their low speed at system idle, while producing enough airflow, is causing a flase alarm.
That said, the PWM signal is also a 5V one, but is roughly 25KHz. High impeadance defaults the fan to full speed.

On a related note, I just ended up leaving the new raid fan untouched. Noiseblocker XM2 40x10mm, at 3800RPM and a staggering 4CFM.

Oh, I'm not trying the control the fan's speed

I know that. But maybe I did a terrible job articulating what I meant, or changed what my goal was halfway through articulating it.
Let me draw what I meant since it's kinda hacky.

Edit: do not use this first picture; it outputs a 12V tach signal. I added another version that might work better.
Click images for full sizes.
[i.imgur.com] (IMG:[i.imgur.com] https://i.imgur.com/KLODZ7es.png)
[i.imgur.com] (IMG:[i.imgur.com] https://i.imgur.com/t1n00jlg.png)
I'd move the voltage divider to after the timer, but then we have the whole current drain vs. performance issue, and you said that the tach signal is open collector input so that might be too much load on it.
I think NE555 should be capable of handling this. When the fan works there should certainly be enough current to run a P-channel MOSFET at 12 volts, I'd think.

That 74HC08 thing might also be handy if the 555's output isn't push-pull (totem pole). But I think it's unnecessary. And it's not an especially great "protection" method either.

This post has been edited by dragontamer8740: Oct 28 2021, 22:29

I'm not sure what the mosfet in series with the fan is supposed to do?
The motherboard is correctly controlling the fan via the PWM input already.
Just for clarification, this is strictly dealing with 4 wire fans- the PWM signal originates in the mobo, and is working as intended- save for the fact that the BIOS isn't trying to target it's minimum speed via a closed loop system, nor does it have a learning procedure for the fans.

The only Chinese engineered fan in my comp has a two pin on the cpu fan header to report speed and is powered by a two pin molex adapter.

Probably precisely to avoid low RPM issues on older boards, it has an inline pot resister for fan speed control.

Is Whatsapp more popular globally than Telegram and Signal?

Is Whatsapp more popular globally than Telegram and Signal?

I have to imagine both are infinity more popular than Signal if only because I'm a cynic. Telegram vs Whatsapp would be completly anecdotal but I'd also guess Whatsapp.

Is Whatsapp more popular globally than Telegram and Signal?

haven't they nerfed the features of all three?

I think Whatsapp is more popular. Signal was gaining popularity but from memory platforms then temp removed them in response. Not sure how many people use Telegram.

I'm not sure what the mosfet in series with the fan is supposed to do?
The motherboard is correctly controlling the fan via the PWM input already.
Just for clarification, this is strictly dealing with 4 wire fans- the PWM signal originates in the mobo, and is working as intended- save for the fact that the BIOS isn't trying to target it's minimum speed via a closed loop system, nor does it have a learning procedure for the fans.

This is to send a fake higher speed signal. And the mosfet should, if the fan seizes, stop powering the timer.

Sure, you don't know the real speed it's running at, but you know it's spinning because the timer is reporting some non zero speed.
You ignore the PWM signal, that doesn't matter at all so I didn't draw it in the diagram.
The motherboard control shouldn't change if you just set it to a static rpm, either.

Basically all this was supposed to do is make the fan failure "idiot light" turn off until the fan actually fails. The mosfet supplies power to the timer, and this power is disconnected when the motor doesn't spin up.

I chose a P-channel mosfet because I don't like doing transistor math.

You may need to put some caps on it for while the fan spins up, actually. But eh, I knew this was a half baked idea anyway.

If you want the motherboard controlling the speed and changing it dynamically, this won't work, of course.
Although maybe if you used the fan tach to trigger the 555 you could do something interesting.

As another option, you could probably do actual, smart fan control if you slid an arduino or something in there and used it as frequency counter, pulsing an output at a multiple of the measured rate. I just don't like microcontroller programming much so I don't like thinking too hard about it.

Oh, I'm not trying the control the fan's speed. The motherboard is driving the fans properly- but the fans respond to their PWM input as a speed % rather than a direct speed. My aftermarket fans are working properly, and producing enough airflow- but crucially, their maximum speed is a bit slow for the board, ergo their low speed at system idle, while producing enough airflow, is causing a flase alarm.
That said, the PWM signal is also a 5V one, but is roughly 25KHz. High impeadance defaults the fan to full speed.

On a related note, I just ended up leaving the new raid fan untouched. Noiseblocker XM2 40x10mm, at 3800RPM and a staggering 4CFM.

4CFM is significantly more than zero. I have made a power supply with a 7805 (linear regulator) for a famicom that overheated until I put a dinky little 12V ~4cm fan running at ~5.5V on it. couldn't even tell it was running, but my system stopped "browning out" when using the disk system addon (which draws more power than an average cart and also somewhat blocks natural convection). The clearance was too tight for a larger heat sink.

This post has been edited by PrincessKaguya: Oct 29 2021, 06:00

eww gross a load of Central banks are planning to implement digital currencies



100% track-able, easy to 'print' money. They even talk about linking it with social policies and credit scores.

*vom*

They will also have different digital currencies for different groups with different rules.

Rules for thee but not for me taken to the extreme.

Its like crypto currency without benefits and evil government intervention.


You have a social credit violation, all of your digital currency has been frozen.


It will also crush innovation as financial intermediaries will need approval of the government.

"don't worry we won't abuse your data because we have no commercial reason to do so"

.... what about political reasons to do so...

Banks: No comment.

This post has been edited by EsotericSatire: Oct 29 2021, 09:24

Yet another clickbait video thumbnail

This is why i stopped watching youtube videos.

Police here are getting roasted because current laws allow them to declare people that are critical of politicians as terrorists.

This is because the laws are so vague.

Police: "children and people commit suicide due to 'online intimidation' so police require these powers"

I think we should have a rule where "for the children" is never a sole valid reason to make a law.

Things like trafficking, etc. are bad because they are human beings, being "a child" doesn't matter there (although maybe it's worse because of that?).

Nine times out of ten "for the children" is just the moral police trying to make things worse. Parents should be responsible for their kids and tell them "no" when they want to use social networking.

"If you have sensitive children, maybe you should tuck them into bed early tonight instead of writing us angry letters tomorrow."

This post has been edited by dragontamer8740: Oct 30 2021, 07:25

Apparently the social network that trump is attaching his name to might be violating (A)GPL.

Uses Mastodon underneath, but has scrubbed all references to mastodon and doesn't share sources.
[mashable.com] https://mashable.com/article/trump-truth-so...cense-violation

I for one am shocked that something with trump's name attached to it could be anything less than totally above water.

Note: I don't think trump knew about or cares about this, I am certain he did not know about mastodon or explicitly tell anyone to rip off a GPL project (I doubt he knows what GPL is). I don't think he had anything to do with it personally other than attaching his name. Someone lower down made this decision and people are quick to attribute it higher up the ladder because it's donald.

Reminds me, though. i forget what mastodon server(s) I registered on.

It's also a shame that at least half of mastodon servers seem to be for people who got kicked off of twitter (some pedos, political radicals, etc.) rather than people seeking a more open alternative by choice.

It's funny that it's called the TRUTH network though. Coming from the guy who said Obama was born in Kenya even after seeing the full birth certificate.
(IMG:[i.imgur.com] https://i.imgur.com/KihLjwn.gif)

Stating that any single political ideology is objective truth while holding public office tells me someone is unfit for said office. Anyone saying that while running for public office is similarly unfit. But I guess it's easier to do that than have a nuanced discussion of values.

This post has been edited by dragontamer8740: Oct 30 2021, 07:24

Apparently the social network that trump is attaching his name to might be violating (A)GPL.

I for one am shocked that something with trump's name attached to it could be anything less than totally above water.

All of Trump's projects are cubic zirconia upon close inspection.

He probably paid a cheap third party dev who 'art of the deal'd' him.

By patching a short crossover cable between the IMPI's dedicated NIC and one of the two normal NICs, then manually asigning an IP and mask to either, I can now access the small computer's IMPI interface.

Plugging the IMPI port's IP into a brower gives a simple java based web interface. That interface bitches my java is out of date because I'm using using OpenJDK instead of Oracle java. It still works though.
The linux version of the standalone program is also java based, but more feature rich. Curiously, I can view data in both, but neither says my fans are at failure RPM. I can set the fan control mode- but it doesn't really have any effect. I may try looping it back into the W10 box to see if the windows management software running on a different board than the one I'm trying to fiddle with gives better results.

Firefox 78 ESR will end soon, must get ready for the impact of 91 ESR (IMG:[invalid] style_emoticons/default/ph34r.gif)

I've officially used my gpu long enough for it to reach legacy status \o/

[ 142.769] (WW) NVIDIA(0): The NVIDIA GeForce GTX 660 Ti GPU installed in this system is
[ 142.769] (WW) NVIDIA(0): supported through the NVIDIA 470.xx Legacy drivers. Please
[ 142.769] (WW) NVIDIA(0): visit [www.nvidia.com] http://www.nvidia.com/object/unix.html for more
[ 142.769] (WW) NVIDIA(0): information. The 495.44 NVIDIA driver will ignore this
[ 142.769] (WW) NVIDIA(0): GPU. Continuing probe...
[ 142.769] (EE) No devices detected.
[ 142.769] (EE)
Fatal server error:
[ 142.769] (EE) no screens found(EE)

I've officially used my gpu long enough for it to reach legacy status \o/

Shit, that's a Kepler card. Oh, boy. My generation card comes next. Going to be so exciting when Nvidia no longer makes a compatible driver and the newer ones are paperweights with Nouveau.

Mine (maxwell 1, 750 ti) should be mostly okay on Nouveau, although I'll lose CUDA waifu2x and have to re-clock it via debugfs.

My Kepler GT630 is probably right there with yours.
Wish I knew when nvidia was planning to drop support for kepler cards in the "legacy" driver.
Also, fortunately for you, nouveau works pretty well on kepler, last I tried (in 2016).

Edit: Another thought:

I think I know why the 6502 CPU never really evolved beyond the 16-bit stage (65C816) - mainly because it was really better suited for hand-assembly than for running call stack based compiled languages. Especially since where in memory you put things determined how fast they ran (so you'd want to put your tightest loops in the first page of memory).

Still, it's fun to imagine what could have been. So I've been doing that. Only two things I know of that used a 65C816 for sure were the SNES and Apple IIgs, both of which chose it because of backward compatibility (although in Nintendo's case they abandoned that goal sometime in the development process; only a few things like the controller reading memory-mapped addresses are the same). It seems like no one really wanted to use the 6502 in newer systems... probably because of that compiled language problem again.

This post has been edited by dragontamer8740: Nov 2 2021, 06:28