Inexpensive but good quality fans

[Lurking]

1 hour ago, RevGAM said:

Hmmm...The article I read on one manufacturer's site (Renkeer) said that they can be adversely impacted by temperature and humidity variation.

 

I also found an anemometer from Testo that uses a TINY propeller to measure. I think it'd probably be the best bet for testing fan airspeed and flow.

That is a really weird statement they make. The hot wire anemometer basically has 2 temperature sensors. One measures incoming air (this also has a dual purpose if you need to measure temperature for other purposes), the other measure the heating probe. Based on the dT, it derives the velocity. If the air comes in at 0°, or 80° doesn't matter since the dT will be the same. Heat capacity of air changes a bit with temperature, but the meter should have that built in to compensate. Without a dewpoint sensor, it may not be able to compensate for varying humidity. But humidity has small impact only. A car air intake is the definition of a very wide range of temperatures. Hot wire anemometers are some of the best  air flow stations in HVAC systems (others are vortex meters). Pressure dependent readings are outdated and not accurate since the pressure sensors aren't accurate enough (at least not at an acceptable price point). Sadly, they still get used in HVAC with very questionable results. 

 

Either way, those meters only give you the velocity. You will need flowrate. You need a defined duct and traverse the duct  and take the velocity measurements at the positions outlined in the table. The reason you need to do is that there is a velocity profile. Velocity isn't the same at all locations and the table values take that into account. This will be impossible to do with the mechanical meters since you can't get them into the duct, and their cross-ection is too large to measure velocity at  a specific point. A hot wire anemometer can be lead into the duct via a drilled hole. It is really hard to be accurate and I've never seen a field person do it by the book. 

 

A pre-defined venturi seems to be easier to use and it looks like that is what the fan testers use. 

 

I would look into understanding fancurves. A fan curve should be the result of fan testing. Sadly, PC fan manufacturers don't publish fan curves, but just give a flow and pressure value with no indication where on the curve that is (if both values are even from the same point on the curve). 

 

Check out Bernoulli's equation and understand dynamic and static pressure and how they interact. 

 

 

[RevGAM]

8 hours ago, Lurking said:

That is a really weird statement they make. The hot wire anemometer basically has 2 temperature sensors. One measures incoming air (this also has a dual purpose if you need to measure temperature for other purposes), the other measure the heating probe. Based on the dT, it derives the velocity. If the air comes in at 0°, or 80° doesn't matter since the dT will be the same. Heat capacity of air changes a bit with temperature, but the meter should have that built in to compensate. Without a dewpoint sensor, it may not be able to compensate for varying humidity. But humidity has small impact only. A car air intake is the definition of a very wide range of temperatures. Hot wire anemometers are some of the best  air flow stations in HVAC systems (others are vortex meters). Pressure dependent readings are outdated and not accurate since the pressure sensors aren't accurate enough (at least not at an acceptable price point). Sadly, they still get used in HVAC with very questionable results. 

 

Either way, those meters only give you the velocity. You will need flowrate. You need a defined duct and traverse the duct  and take the velocity measurements at the positions outlined in the table. The reason you need to do is that there is a velocity profile. Velocity isn't the same at all locations and the table values take that into account. This will be impossible to do with the mechanical meters since you can't get them into the duct, and their cross-ection is too large to measure velocity at  a specific point. A hot wire anemometer can be lead into the duct via a drilled hole. It is really hard to be accurate and I've never seen a field person do it by the book. 

 

A pre-defined venturi seems to be easier to use and it looks like that is what the fan testers use. 

 

I would look into understanding fancurves. A fan curve should be the result of fan testing. Sadly, PC fan manufacturers don't publish fan curves, but just give a flow and pressure value with no indication where on the curve that is (if both values are even from the same point on the curve). 

 

Check out Bernoulli's equation and understand dynamic and static pressure and how they interact. 

Oh boy, this is such a can of worms. I gather you work on HVAC?

I am absolutely confused as to how to traverse a wind tunnel. Best case would be a wind tunnel with a slot in it that has a sliding hole for the probe or fan, because I would think that if you put the probe in on an angle, the top of the probe would block at least a bit of the wind, resulting in a (slight?) change in the results.

 

There's the Testo 416 (this is one of several variants), which has a tiny prop in it, and now I'm looking at several hundred dollars to over $1,000, and that still doesn't give me static pressure. The Testo 405-i hotwire is reasonably priced and has a second sensor that isn't explained on the website, so I hope it's WB or DP.

What are reasonable yet accurate choices for flow and pressure?

I'll have to read those links. I'm not good at maths, so I probably won't understand Bernoulli's equation.

[Lurking]

35 minutes ago, RevGAM said:

Oh boy, this is such a can of worms. I gather you work on HVAC?

I am absolutely confused as to how to traverse a wind tunnel. Best case would be a wind tunnel with a slot in it that has a sliding hole for the probe or fan, because I would think that if you put the probe in on an angle, the top of the probe would block at least a bit of the wind, resulting in a (slight?) change in the results.

 

There's the Testo 416 (this is one of several variants), which has a tiny prop in it, and now I'm looking at several hundred dollars to over $1,000, and that still doesn't give me static pressure. The Testo 405-i hotwire is reasonably priced and has a second sensor that isn't explained on the website, so I hope it's WB or DP.

What are reasonable yet accurate choices for flow and pressure?

I'll have to read those links. I'm not good at maths, so I probably won't understand Bernoulli's equation.

You basically have to drill as many holes as you need probe positions on one side. And you need to mark up on your probe how deep to insert. And you have to do it straight, and in the correct angle. With other words, it never happens to be done correctly. Like on a large duct the balancer will claim they measured the flow, but I only see 3 drilled holes while 6 are required. Also not at the correct location. You see the same incompetence on the hydronic side.

 

Yes, I design and commission HVAC systems. i would never eve be able to select a fan if there isn't a full fan curve provided. I came to the conclusion it is better to use as much as possible factory calibrated equipment instead of field-calibrated devices. There is a very wide gulf between what that balancing text books say how measurements are to be taken, and how they are taken in the field. Maybe there is one somewhere, but i nave never met a professional balancer where i was under the impression they know what they are doing. They measure something, and think the result is something. All the flow measuring devices (air or hydronic) i use have their own design flow conditions and don't need field calibration. Older flow stations (like the VAV pitot tube devices) need to be field calibrated.... oh boy are those wrong. 

 

The one meter you have has a 16mm probe, so you need large holes. I would stay away from mechanical devices with turbines or propellers. 

[RevGAM]

15 minutes ago, Lurking said:

You basically have to drill as many holes as you need probe positions on one side. And you need to mark up on your probe how deep to insert. And you have to do it straight, and in the correct angle. With other words, it never happens to be done correctly. Like on a large duct the balancer will claim they measured the flow, but I only see 3 drilled holes while 6 are required. Also not at the correct location. You see the same incompetence on the hydronic side.

 

Yes, I design and commission HVAC systems. i would never eve be able to select a fan if there isn't a full fan curve provided. I came to the conclusion it is better to use as much as possible factory calibrated equipment instead of field-calibrated devices. There is a very wide gulf between what that balancing text books say how measurements are to be taken, and how they are taken in the field. Maybe there is one somewhere, but i nave never met a professional balancer where i was under the impression they know what they are doing. They measure something, and think the result is something. All the flow measuring devices (air or hydronic) i use have their own design flow conditions and don't need field calibration. Older flow stations (like the VAV pitot tube devices) need to be field calibrated.... oh boy are those wrong. 

 

The one meter you have has a 16mm probe, so you need large holes. I would stay away from mechanical devices with turbines or propellers. 

Ok, I understand, but which device types should I be looking for? What should I avoid? I certainly can't afford anything close to what HW Cooling has and I know my wind tunnel is not perfect but there is no such thing as a perfect wind tunnel, especially since fans aren't used in wind tunnels IRL.  

[Lurking]

17 minutes ago, RevGAM said:

Ok, I understand, but which device types should I be looking for? What should I avoid? I certainly can't afford anything close to what HW Cooling has and I know my wind tunnel is not perfect but there is no such thing as a perfect wind tunnel, especially since fans aren't used in wind tunnels IRL.  

You have to to build a defined condition. To measure flow, you need to have something with a defined cross section.  I'm sure there is some ANSI or ASTM or DIN standard how to measure a PC fan. Keep in mind, the measuring device (the tunnel etc.) will impact flow compared to just open flow. So if you do it your own way, your results aren't useful comparing to a standardized test. What you attempt at best will allow you to rank the fans you have. But you can't measure noise well. So without noise-normalizing this all is meaningless since anyone can make a powerful fan if noise isn't an issue. Measuring noise will be equally, if not more, challenging than measuring pressure and flow. 

 

If you want to measure velocity with a probe, a hot wire anemometer is better. Testo is a reputable company, but they may have different models. Note that those devices should be factory-calibrated annually or some other interval. This isn't an issue the first year since it comes calibrated. At work we send them in to calibration. Every once a while this leads to us needing to buy a new device since at some point it can't be made to measure correct anymore after some years. When I was in school I learned that any test report that doesn't include the latest calibration dates of measuring instruments is useless. 

[RevGAM]

1 hour ago, Lurking said:

You have to to build a defined condition. To measure flow, you need to have something with a defined cross section.  I'm sure there is some ANSI or ASTM or DIN standard how to measure a PC fan. Keep in mind, the measuring device (the tunnel etc.) will impact flow compared to just open flow. So if you do it your own way, your results aren't useful comparing to a standardized test. What you attempt at best will allow you to rank the fans you have. But you can't measure noise well. So without noise-normalizing this all is meaningless since anyone can make a powerful fan if noise isn't an issue. Measuring noise will be equally, if not more, challenging than measuring pressure and flow. 

 

If you want to measure velocity with a probe, a hot wire anemometer is better. Testo is a reputable company, but they may have different models. Note that those devices should be factory-calibrated annually or some other interval. This isn't an issue the first year since it comes calibrated. At work we send them in to calibration. Every once a while this leads to us needing to buy a new device since at some point it can't be made to measure correct anymore after some years. When I was in school I learned that any test report that doesn't include the latest calibration dates of measuring instruments is useless. 

Thank you for your guidance. I am testing noise, but I need to improve my methodology. I'm currently getting results that simulate the noise you'd hear if you blow it in your ear from 15 cm, which certainly isn't representative of reality, so I'm considering alternatives.

I tried to find an old computer at the charity store, but they've started to sell things online because of bargain hunters buying things cheap and selling them high online - charity stores are realizing they can get a lot more money if they auction off or otherwise sell online valuable items. Since even a used computer can net them hundreds, I think I won't have much luck getting a testing computer. Thus, I'm stuck using my own PC.

Am I correct that the hotwire's probe needs to be replaced periodically as it wears out? Do you have any suggestions on which hotwire tools from Testo or others would be best? I gather you're not suggesting I use the mini-vane 416, or a large vane, or a Pitot? How, then, can I get static pressure?

As for sensors, given that my tunnel is 8"x2', how many should I put in, and at what points?

I got an email back from Degree Controls and they certainly have good equipment, but it's outside my budget. I'm still going to consult with them for the future, though. I left a message for Testo but they haven't called me back yet.

I realize that my results aren't going to be ideal but that they will let me rank fans, and they will also provide enough data for end-users to make informed choices when they are confused by all the reviewers and false reviewers out there. At least, that is my intention. I'm very ethical and will not sell out just to get rich.

Again, I appreciate your guidance, and I hope you won't give up on me!

[Lurking]

37 minutes ago, RevGAM said:

Thank you for your guidance. I am testing noise, but I need to improve my methodology. I'm currently getting results that simulate the noise you'd hear if you blow it in your ear from 15 cm, which certainly isn't representative of reality, so I'm considering alternatives.

I tried to find an old computer at the charity store, but they've started to sell things online because of bargain hunters buying things cheap and selling them high online - charity stores are realizing they can get a lot more money if they auction off or otherwise sell online valuable items. Since even a used computer can net them hundreds, I think I won't have much luck getting a testing computer. Thus, I'm stuck using my own PC.

Am I correct that the hotwire's probe needs to be replaced periodically as it wears out? Do you have any suggestions on which hotwire tools from Testo or others would be best? I gather you're not suggesting I use the mini-vane 416, or a large vane, or a Pitot? How, then, can I get static pressure?

As for sensors, given that my tunnel is 8"x2', how many should I put in, and at what points?

I got an email back from Degree Controls and they certainly have good equipment, but it's outside my budget. I'm still going to consult with them for the future, though. I left a message for Testo but they haven't called me back yet.

I realize that my results aren't going to be ideal but that they will let me rank fans, and they will also provide enough data for end-users to make informed choices when they are confused by all the reviewers and false reviewers out there. At least, that is my intention. I'm very ethical and will not sell out just to get rich.

Again, I appreciate your guidance, and I hope you won't give up on me!

Every sensor will drift over time and electronic elements also will change characteristics. This has many reasons, many of them on the atomic level. Like some atoms are displacing other atoms and so on. The hot wire anemometer relies on accuracy of the 2 independent sensors, and the control circuit that meters the electricity going into the heater. So if one sensor now shows you 0.1°C more, the other shows 0.1°C less, now you have an 0.2°C more OR less dT at the very same airflow. Add some minute change of electric power to heat, and you can have quite different flow readings. in addition, any dirt on the heater will change heat transfer coefficient (so never touch it!). Similar things are true for all measuring devices. 

 

A rectangular tunnel, especially one with such extreme aspect ratio will be really bad. Round is much better. The velocity profile in a round duct is even due to no corners and all dimensions the same. Even in a square duct, you have the corners. just not great for flow. the rule of thumb in duct design is to never exceed 4:1, even that isn't great and if that ever is needed, oval duct is better. so i suspect for measuring flow, round also is best.  The above link about traversing shows that for under 30"duct to use 5 probes on each side. It is silent about what to do with your short side, but really, suhc extreme differences aren't great. Positioning should be in those ratios: 0.074, 0.288, 0.500, 0.712, 0.926 based on the duct size. So multiply 30"with those ratios. It sounded like Testo has an app that also may give you those values. So you see, between drilling, and holding the probe accurately, there is a lot of variation. 

 

For noise measuring the first challenge is to have a room isolated from ambient and with no reflections. Like you have to remodel your home AND live in a very quiet area to get that even remotely done. I read some online fan tests that talked about a floor of 30dB and measuring 32 dB, which is ridiculous. You probably need to be at 20dB or under. You also need to remove yourself  (breathing noise etc.). I don't know if 1m is the standard for PC fans, it is for many louder devices. But I suspect PC fans are so quiet that they need to measure closer to those and then re-calculate for 1m. How to calculate that depends on the type of distribution (360° vs. 180° etc.). 

 

You'll need to repeat tests multiple times. Even with the best conditions you will see variations between tests. 

 

There is a reason why there are only very very few places that actually test fans. You probably have figured out by now that this is very complex

[RevGAM]

4 minutes ago, Lurking said:

Every sensor will drift over time and electronic elements also will change characteristics. This has many reasons, many of them on the atomic level. Like some atoms are displacing other atoms and so on. The hot wire anemometer relies on accuracy of the 2 independent sensors, and the control circuit that meters the electricity going into the heater. So if one sensor now shows you 0.1°C more, the other shows 0.1°C less, now you have an 0.2°C more OR less dT at the very same airflow. Add some minute change of electric power to heat, and you can have quite different flow readings. in addition, any dirt on the heater will change heat transfer coefficient (so never touch it!). Similar things are true for all measuring devices. 

 

A rectangular tunnel, especially one with such extreme aspect ratio will be really bad. Round is much better. The velocity profile in a round duct is even due to no corners and all dimensions the same. Even in a square duct, you have the corners. just not great for flow. the rule of thumb in duct design is to never exceed 4:1, even that isn't great and if that ever is needed, oval duct is better. so i suspect for measuring flow, round also is best.  The above link about traversing shows that for under 30"duct to use 5 probes on each side. It is silent about what to do with your short side, but really, suhc extreme differences aren't great. Positioning should be in those ratios: 0.074, 0.288, 0.500, 0.712, 0.926 based on the duct size. So multiply 30"with those ratios. It sounded like Testo has an app that also may give you those values. So you see, between drilling, and holding the probe accurately, there is a lot of variation. 

 

For noise measuring the first challenge is to have a room isolated from ambient and with no reflections. Like you have to remodel your home AND live in a very quiet area to get that even remotely done. I read some online fan tests that talked about a floor of 30dB and measuring 32 dB, which is ridiculous. You probably need to be at 20dB or under. You also need to remove yourself  (breathing noise etc.). I don't know if 1m is the standard for PC fans, it is for many louder devices. But I suspect PC fans are so quiet that they need to measure closer to those and then re-calculate for 1m. How to calculate that depends on the type of distribution (360° vs. 180° etc.). 

 

You'll need to repeat tests multiple times. Even with the best conditions you will see variations between tests. 

 

There is a reason why there are only very very few places that actually test fans. You probably have figured out by now that this is very complex

Well, I guess I made the right choice going for a round tunnel from the start. I figured it made the most sense, especially since the ones sent out by fan makers to some reviewers are round. I will try to understand the traversing article but I'm afraid I'll need equations (and hope I can solve them - after so many years, I've forgotten most of what I learned about maths). Yeah, Testo and some others have apps that'll do it. Can I afford it? Ummm...If possible, I want to semi-permanently install the sensors so I don't have to worry about placement and such.

 

I think that all I need to do is make a small chamber which is lined with sound-deadening material to test sound. I don't agree with HWC's testing because they're trying to pick up the noise using a funnel on the mic, which is not going to be representative of what users will hear except people like Zakius who are super-sensitive to noise. That makes it at least as bad as my current test. Ideally, there would need to be 2 chambers in the box - one to simulate the fan being in a case, and the other for the mic, facing the fan for both intake and exhaust. Once the chamber is closed, virtually no outside noises will interfere as long as it's made right. I would place it on top of, say, a rubber mat or something else to absorb vibrations from outside.

 

Yes, repetition. Such fun. And people wonder why reviewers don't pump out videos more often.  

 

It's all complex, but my feeling is that fans are potentially easier than some other things, like PSUs, CPUs and mobos. have a multimeter but I don't remember much of anything from my circuitry classes to test things. Honestly, I think that there's a point beyond which testing is unnecessarily complex and/or isolated, producing results that are so pure that they are divorced from reality. That is the disadvantage of testing in a lab because there is a massive number of variables when we consider all of the components inside a computer, and then add in all the external variables, which results in a different user experience than the data suggests. One simple example is the number of fans, which are intake or exhaust, and the case's airflow design (if any). I think it's a bit OCD to go to such extremes as lab testing, but I applaud the effort anyways.

 

Looking for specific suggestions about meters I should consider, if you don't mind.  

[Lurking]

A standardized test is about repeatability. A user installs in a variety of cases along with other equipment and the sound goes through and bounces off all surfaces. So making a PC case, of rexample, isn't practical for testing fans. A lab test basically attempts to reduce or control most variables. So you basically test the fan on its own. 

 

HVAC or industrial fans get tests for varying resistances (pressuredrop) and flowrates and you get a fan curve. I think the HWcooling test did a good job to make this practical since they had tests of nylon filter, radiators etc. This doesn't give us a fan curve with pressure values. But something the user or builder of a PC can use. There also is fan-effect if you have the resistance up or downstream the fan too close (before flow can develop). 

 

Like if you measure light output of a lightbulb that will be without a shade even if a user puts the lightbulb in some lamp or puts a shade around it.  

[RevGAM]

3 hours ago, Lurking said:

A standardized test is about repeatability. A user installs in a variety of cases along with other equipment and the sound goes through and bounces off all surfaces. So making a PC case, of rexample, isn't practical for testing fans. A lab test basically attempts to reduce or control most variables. So you basically test the fan on its own. 

I know. Simulating a partial case in the box is repeatable,  or even an entire case which would be better, but a complete computer to get a sound profile would be best even if it's turned off... although having it on would be best.

 

3 hours ago, Lurking said:

I think the HWcooling test did a good job to make this practical since they had tests of nylon filter, radiators etc.

Yes,  although I'd like to see their data that supports their conclusion that putting barriers opposite where they are in a case works the same. 

3 hours ago, Lurking said:

This doesn't give us a fan curve with pressure values. But something the user or builder of a PC can use.

Do you mean my testing or HWC's?

3 hours ago, Lurking said:

There also is fan-effect if you have the resistance up or downstream the fan too close (before flow can develop).

Yeah, this is what I'm talking about, since they do it differently. As for fan-effect, what about all the cases that do just that - putting a fan against 1 or 2 barriers?

[Lurking]

3 minutes ago, RevGAM said:

I know. Simulating a partial case in the box is repeatable,  or even an entire case which would be better, but a complete computer to get a sound profile would be best even if it's turned off... although having it on would be best.

 

Yes,  although I'd like to see their data that supports their conclusion that putting barriers opposite where they are in a case works the same. 

Do you mean my testing or HWC's?

Yeah, this is what I'm talking about, since they do it differently. As for fan-effect, what about all the cases that do just that - putting a fan against 1 or 2 barriers?

But you are buying a fan, not a specific fan/case combo. And the case (or any obstruction) also creates noise from air going over it. And what case setup would everybody agree to use for testing? And if 20 years ago we decided on one case that was typical back then, how do we now deal with airflow cases, radiators etc.? 

 

A PC case kind of has to violate the rule to not allow fan-effect. For an industrial or HVAC fan the rule is to have 3-5 fan diameters of unobstructed flow up and downstream of the fan. That just isn't possible in a PC case. Like a 140mm fan would require 420mm up and downstream. The HWCooling test also took that into account since they mounted the fan to the radiator etc. They basically did a real world (or close to it) test according to how one would use it in a case or cooler. They did varying radiator thicknesses, filters, hexagonal holes etc. I think they covered most cases a PC builder would need. 

 

I think what HWcooling tested is very useful for selecting a fan. If you have an exhaust fan, you look at the hexagonal results, if you have intake fan through filter, you look at the filter test, same for radiator. etc. You will notice that the good fans typically always were on top, but not necessarily in the same order for all applications. I don't think I could come up with a better idea for a fan test for PCs. 

 

All that HWCooling are missing are more of the fans I want to buy

[RevGAM]

14 hours ago, Lurking said:

But you are buying a fan, not a specific fan/case combo. And the case (or any obstruction) also creates noise from air going over it. And what case setup would everybody agree to use for testing? And if 20 years ago we decided on one case that was typical back then, how do we now deal with airflow cases, radiators etc.? 

 

A PC case kind of has to violate the rule to not allow fan-effect. For an industrial or HVAC fan the rule is to have 3-5 fan diameters of unobstructed flow up and downstream of the fan. That just isn't possible in a PC case. Like a 140mm fan would require 420mm up and downstream. The HWCooling test also took that into account since they mounted the fan to the radiator etc. They basically did a real world (or close to it) test according to how one would use it in a case or cooler. They did varying radiator thicknesses, filters, hexagonal holes etc. I think they covered most cases a PC builder would need. 

 

I think what HWcooling tested is very useful for selecting a fan. If you have an exhaust fan, you look at the hexagonal results, if you have intake fan through filter, you look at the filter test, same for radiator. etc. You will notice that the good fans typically always were on top, but not necessarily in the same order for all applications. I don't think I could come up with a better idea for a fan test for PCs. 

 

All that HWCooling are missing are more of the fans I want to buy

And therein lies the rub. Please bear with me as I try to explain.  I'm not disrespecting your input, so I hope you will not take offense. 

First, the question of which case and setup to use. As you pointed out, it's about repeatability and repetition to acquire a valid average and/or mean. Unless I intend to test every case, sub-category of case or just every major category, which I don't because I can't due to finances, time, knowledge, tools, and space constraints, it doesn't matter at all. I'm looking for repeatability that is relevant to the real world, not results from a lab that don't reflect reality. As for obstructions, that is the normal state of a PC, so the point is moot. Having obstructions inside the noise chamber's case or case simulation is critical to providing a successful simulation, even if it cannot possibly represent all cases. To be able to represent both intake and exhaust can be achieved either by turning the case around, or by having a bisected case, but the two halves would need to be separated so they don't impact each other's results through reflectivity. Frustratingly, using sound deadening materials on the walls of the noise box has a negative impact on the results because it actually decreases the amount of noise emanating from the case and thus a lower dBA score will be achieved. Even this, however, is preferable to the way that HWC (and, I believe, many others) is doing it.

Yes, 3'-5' is relevant for HVAC but it is NEVER relevant to computing because I doubt that there's ever more than 1 foot of EMPTY space behind a fan, much less at 3-5, which won't even exist at 3' for the vast majority of consumer and commercial cases. That is, unless the case is empty, in which the results will be skewed at least as badly as in a wind tunnel. Testing HVAC is very different in that regard from PCs because HVAC MUST by definition have only clear space, and PCs cannot do that. So, I don't need to have even a tube as long as I've created in order to get accurate results because the airflow and pressure from any fan will never go that distance inside a case.

Let's not forget about other fans in the case that create cross-currents and turbulence, or the fact that laminar flow is a major factor in a tube and a minor one in a case. There are many variables that exist in a complete PC that are absent in all of the high-quality tests. They are absolute measurements under sterile conditions. There is absolutely no possibility of providing end-users (or anyone else) with consistent data from one PC to another. Some PCs will amplify the noise of the fans while others will muffle it. Some will slow down airflow while others will allow close to the maximum. Some cases will decrease static pressure, and so on. There is literally no way, without millions of people-hours and money, tools and special rooms, to approximate even a small percentage of all the possible variances. All I can do is choose one, just like the labs, Aris, Ľubomír, LTT, GN and everyone else. In other words, NO tester will provide data with such a high level of accuracy that it will be valid for everyone. However, if different testers all use different repeatable, repetitious procedures, anyone who cares to know will be able to put together a decent understanding of how the options will behave and perform in different PCs. There are far fewer variables in an HVAC system, wouldn't you agree? Do companies take an HVAC system to a lab to have it tested every time a customer calls for help? Certainly not - that would be ludicrously expensive and time-consuming, as well as very inconveniencing for the customers, right?

 

I agree that they did a great job with the variety of barriers that they chose to test, but do you know why they mount things opposite the way they would be IRL?

 

Aris has a "better" setup for fan testing as he actually bought the scientific-grade equipment like that seen in GN's lab and more. I cannot even guess how many tens of thousands of dollars he's spent to get it all, between financing research for HWBusters and Cybenetic Labs. I would absolutely rely on HWC's results to choose a fan, too, but I also wish they could test more fans - an endless and thankless task for a man who apparently sleeps 4 hours a night.  

Questions:

1. Since both HVAC and PCs are semi-closed systems (the air has to go somewhere, after all), it seems to me that I should cap the other end of my tunnel with some sort of material to encourage pressure. What would you think would be the best option? I don't think that a vent or solid cap is a good idea, but I have nylon mesh. I could also, I'm sure, procure something like the large-bore stuff often found at the rear of a case. I guess I could cut up a lot of straws like ThermalLeft.
2. Which type of pressure anemometer+anemometer would be the best bet, or should I purchase them separately? And, again, which?

Can you recommend specific brands and/or models that are reliable but affordable for me?

Ok, I'll bite - which fans do you want to have reviewed?

Thank you, as always, for your continued guidance.  

[RevGAM]

You may find ThermalTake's work interesting. He also has a YT channel. https://quasarzone.com/bbs/qf_cool/views/365052

[Lurking]

2 hours ago, RevGAM said:

And therein lies the rub. Please bear with me as I try to explain.  I'm not disrespecting your input, so I hope you will not take offense. 

First, the question of which case and setup to use. As you pointed out, it's about repeatability and repetition to acquire a valid average and/or mean. Unless I intend to test every case, sub-category of case or just every major category, which I don't because I can't due to finances, time, knowledge, tools, and space constraints, it doesn't matter at all. I'm looking for repeatability that is relevant to the real world, not results from a lab that don't reflect reality. As for obstructions, that is the normal state of a PC, so the point is moot. Having obstructions inside the noise chamber's case or case simulation is critical to providing a successful simulation, even if it cannot possibly represent all cases. To be able to represent both intake and exhaust can be achieved either by turning the case around, or by having a bisected case, but the two halves would need to be separated so they don't impact each other's results through reflectivity. Frustratingly, using sound deadening materials on the walls of the noise box has a negative impact on the results because it actually decreases the amount of noise emanating from the case and thus a lower dBA score will be achieved. Even this, however, is preferable to the way that HWC (and, I believe, many others) is doing it.

Yes, 3'-5' is relevant for HVAC but it is NEVER relevant to computing because I doubt that there's ever more than 1 foot of EMPTY space behind a fan, much less at 3-5, which won't even exist at 3' for the vast majority of consumer and commercial cases. That is, unless the case is empty, in which the results will be skewed at least as badly as in a wind tunnel. Testing HVAC is very different in that regard from PCs because HVAC MUST by definition have only clear space, and PCs cannot do that. So, I don't need to have even a tube as long as I've created in order to get accurate results because the airflow and pressure from any fan will never go that distance inside a case.

Let's not forget about other fans in the case that create cross-currents and turbulence, or the fact that laminar flow is a major factor in a tube and a minor one in a case. There are many variables that exist in a complete PC that are absent in all of the high-quality tests. They are absolute measurements under sterile conditions. There is absolutely no possibility of providing end-users (or anyone else) with consistent data from one PC to another. Some PCs will amplify the noise of the fans while others will muffle it. Some will slow down airflow while others will allow close to the maximum. Some cases will decrease static pressure, and so on. There is literally no way, without millions of people-hours and money, tools and special rooms, to approximate even a small percentage of all the possible variances. All I can do is choose one, just like the labs, Aris, Ľubomír, LTT, GN and everyone else. In other words, NO tester will provide data with such a high level of accuracy that it will be valid for everyone. However, if different testers all use different repeatable, repetitious procedures, anyone who cares to know will be able to put together a decent understanding of how the options will behave and perform in different PCs. There are far fewer variables in an HVAC system, wouldn't you agree? Do companies take an HVAC system to a lab to have it tested every time a customer calls for help? Certainly not - that would be ludicrously expensive and time-consuming, as well as very inconveniencing for the customers, right?

 

I agree that they did a great job with the variety of barriers that they chose to test, but do you know why they mount things opposite the way they would be IRL?

 

Aris has a "better" setup for fan testing as he actually bought the scientific-grade equipment like that seen in GN's lab and more. I cannot even guess how many tens of thousands of dollars he's spent to get it all, between financing research for HWC and Cybenetic Labs. I would absolutely rely on HWC's results to choose a fan, too, but I also wish they could test more fans - an endless and thankless task for a man who apparently sleeps 4 hours a night.  

Questions:

1. Since both HVAC and PCs are semi-closed systems (the air has to go somewhere, after all), it seems to me that I should cap the other end of my tunnel with some sort of material to encourage pressure. What would you think would be the best option? I don't think that a vent or solid cap is a good idea, but I have nylon mesh. I could also, I'm sure, procure something like the large-bore stuff often found at the rear of a case. I guess I could cut up a lot of straws like ThermalLeft.
2. Which type of pressure anemometer+anemometer would be the best bet, or should I purchase them separately? And, again, which?

Can you recommend specific brands and/or models that are reliable but affordable for me?

Ok, I'll bite - which fans do you want to have reviewed?

Thank you, as always, for your continued guidance.  

While the fundamentals are the same, there are significant differences between HVAC/industrial fans and fans in equipment, like PCs. In HVAC the required airflow typically is a known number (based on heating/cooling load and/or ventilation needs). And the whole system pressure drop is relatively well known since most parts come with manufacturer data (tables, curves, selection software etc.) that show pressure drop at specific flowrates. We basically evaluate the entire system inc. all branches  (all ducts, all elbows, all junctions, diffusers, filters and so on) Based on required flow and pressuredrop a fan is selected and that fan manufacturer provides flow and pressure data (curve, table, software). In an equipment, like a PC, it is more eye-balling and based on what typically works and we typically ignore theoretically best fan location etc. where  fan is and what space we have is determined by other factors and we can't change it. 

 

For fan testing and providing a fan curve a manufacturer would use a flow-station (basically an array of hot wire anemometers) or a Vortex flowstation. You also can use a Venturi. No one would continuously move a probe around in a permanent setup. Those flowstations are calibrated, have an accurate location of probes and measure the entire cross section simultaneously. then they connect a fan (with enough straight duct to avoid fan-effect) and a damper to throttle. They measure pressure up and downstream of the fan and the flow. the damper will vary the flow, and then at different flows the pressure gain over the fan will be measured. The same then at different fan speeds. You then get a fan curve for each speed. Example:

this then can be used for applications where flow and pressure (system curve) is known. Like in HVAC. I suspect a PC fan manufacturer does something similar, except they don't publish those data and only give one flow and one pressure number. Look at the curve, and you see a single value is pointless if you don't know ehre the pressure and the flow number is taken from. Maybe the flow number is taken at zero pressure, and the pressure number taken at the max. Totally pointless if they are not from the same part of the curve. 

 

But since no one in PC building does all that and also doesn't know the system pressuredrop (case), The HWCooling approach is a much more practical method. They basically skip the measuring of pressure gained and only measure the flow, but over a standardized system (i.e. the radiator, or filter).

 

To me, it is the one or the other. Really depends on your objective. For someone to select a fan for a radiator, or case, the HWCooling approach is better. I appreciate you making the effort. But really, I would just use a really good case with good and as large as possible fans and call it a day. I recently bought a fractal Torrent. I don't need to measure that 180mm fans and unobstructed flows will be inherently well performing at low noise. 

[Dogzilla07]

@RevGAMYou should probably give up on manometers, the ones that have enough resolution/fidelity are probably too expensive.

And you don't really need the pressure data, if u slap different obstruction on the front of your tunnel.

[RevGAM]

29 minutes ago, Dogzilla07 said:

@RevGAMYou should probably give up on manometers, the ones that have enough resolution/fidelity are probably too expensive.

And you don't really need the pressure data, if u slap different obstruction on the front of your tunnel.

Which manometers are you thinking of?

Can you elaborate, please, about your second comment?

I realize that using obstructions allows me to infer which fans have higher SP. Is that what you mean?

[RevGAM]

7 hours ago, Lurking said:

While the fundamentals are the same, there are significant differences between HVAC/industrial fans and fans in equipment, like PCs. In HVAC the required airflow typically is a known number (based on heating/cooling load and/or ventilation needs). And the whole system pressure drop is relatively well known since most parts come with manufacturer data (tables, curves, selection software etc.) that show pressure drop at specific flowrates. We basically evaluate the entire system inc. all branches  (all ducts, all elbows, all junctions, diffusers, filters and so on) Based on required flow and pressuredrop a fan is selected and that fan manufacturer provides flow and pressure data (curve, table, software). In an equipment, like a PC, it is more eye-balling and based on what typically works and we typically ignore theoretically best fan location etc. where  fan is and what space we have is determined by other factors and we can't change it. 

 

For fan testing and providing a fan curve a manufacturer would use a flow-station (basically an array of hot wire anemometers) or a Vortex flowstation. You also can use a Venturi. No one would continuously move a probe around in a permanent setup. Those flowstations are calibrated, have an accurate location of probes and measure the entire cross section simultaneously. then they connect a fan (with enough straight duct to avoid fan-effect) and a damper to throttle. They measure pressure up and downstream of the fan and the flow. the damper will vary the flow, and then at different flows the pressure gain over the fan will be measured. The same then at different fan speeds. You then get a fan curve for each speed. Example:

this then can be used for applications where flow and pressure (system curve) is known. Like in HVAC. I suspect a PC fan manufacturer does something similar, except they don't publish those data and only give one flow and one pressure number. Look at the curve, and you see a single value is pointless if you don't know ehre the pressure and the flow number is taken from. Maybe the flow number is taken at zero pressure, and the pressure number taken at the max. Totally pointless if they are not from the same part of the curve. 

 

But since no one in PC building does all that and also doesn't know the system pressuredrop (case), The HWCooling approach is a much more practical method. They basically skip the measuring of pressure gained and only measure the flow, but over a standardized system (i.e. the radiator, or filter).

 

To me, it is the one or the other. Really depends on your objective. For someone to select a fan for a radiator, or case, the HWCooling approach is better. I appreciate you making the effort. But really, I would just use a really good case with good and as large as possible fans and call it a day. I recently bought a fractal Torrent. I don't need to measure that 180mm fans and unobstructed flows will be inherently well performing at low noise. 

That is very  interesting! It sounds as if, once you get sufficient experience, it almost becomes an art to designing HVAC, but I don't mean that in a bad way. How long does it take to design and commission, on average, in a home? Would you provide examples of mistakes in designs? Does ductwork from the unit to branches have to be consistent in size?

What are "in. wg" and "Pa"? Is KW kilowatts? So, to provide "valid" info, I would select, say, the intersection of the 1.5 KW line with 600 Pa just under 5000 m3/hr, right?

I think I might have mis-communicated regarding highly technical methodologies like HWCooling's,  ThermaLeft's and Aris's. I'm not saying the data is invalid, nor did I mean to imply that they're doing it wrong. I was really just talking about the difference between real-world data and what comes out of standardized testing. One of the things I quickly realized as a teacher is that standardized testing has weaknesses. Now, fans aren't pupils, of course, but any type of standardized testing has strengths and weaknesses. Honestly, the problems with testing in education are far more severe than, I think, in testing fans. It is not my goal to disprove their work but, rather, to corroborate the work of others and be able to tell which reviewers are just sell-outs or selectively promoting brands with skewed or false data, with the ultimate goal of providing useful info to end-users.

For you, it is easy to make certain assertions about one fan over another, and to suggest which fan(s) in a group of options are best in a given task, but most people don't know even the basics about fans. They don't understand that there is both AF and SP, they just know "low" means quiet and slow, "medium" means noisier and faster, and "high" means very noisy and very fast, but it varies from one fan model to the next. Some don't even understand RPMs, as flabbergasting as that is. Thus, it is surprising how many people get overwhelmed just trying to choose computer fans because choosing a box fan to keep you cool is a comparatively simple task that doesn't require understanding anything more than "how noisy" and "how well" a fan operates in a general sense.

 

I'm thinking about cutting out a viewing window in my tunnel, replacing that with hard plastic, and having some pieces of perhaps crepe paper hanging there to visualize the airflow for me.

Would you please answer the questions I posed in my last post?

[Dogzilla07]

1 hour ago, RevGAM said:

Which manometers are you thinking of?

Can you elaborate, please, about your second comment?

No clue, just experience other people setting up their testing had before, best would be to send a message, talk to doyll at overclock . net 

He was building a testing setup that's not too expensive and always complaining that adequate manometers were too expensive to get in the UK (plus he worked in the industry if I'm not miss-remembering).

 

1 hour ago, RevGAM said:

I realize that using obstructions allows me to infer which fans have higher SP. Is that what you mean?

Yeah pretty much, SP on it's own is not really useful. When u put a filter/radiator/mesh block infront or behind the fan, the new airflow values u get that are different from free airflow tell you what fan has higher SP.

 

It's redundant to have both unless you're gonna be doing 3-rd party testing and validation for companies like HWBusters intends to. Or if you want to compare to data from other reviews xD

[Dogzilla07]

11 hours ago, RevGAM said:

In other words, NO tester will provide data with such a high level of accuracy that it will be valid for everyone. However, if different testers all use different repeatable, repetitious procedures, anyone who cares to know will be able to put together a decent understanding of how the options will behave and perform in different PCs

Well yeah, you've got 4 separate general things to tell you if a fan will do better in your case than others (as an end user):

1. standardized testing to tell you which fan has best raw noise (dBA) to airflow (CFM/m3^h) numbers

 

2. noise spectrum frequency analysis and vibrational analysis to tell you if a fan has design issues or QC/QA issues (X+Y-axis harmonic resonance noise especially)

 

3. the quality of intake filter/mesh, combined meshfilter of the case you're using

 

4. everything else about case you're using and all the little quirks it has and all the fan quirks related to it like micro-turbulences/turbulent noise, air vortices/hot pockets, Z-axis harmonic resonance noise due to blade lift/fan design, inadequate fan placement, number of fans, type of fans, etc, ...

and it's almost impossible to test/show no.4 to people who don't follow tech, don't research their own case/fan behaviour in depth (like you mentioned). Those who know don't need no.4 testing info. Those who need no.4 info can't get it in a understandable and simple enough manner.

Though No.2 shows you a ton of stuff that falls into No.4 category, that you can do pay attention if it applies to your case (if you know what you're looking for/at )

 

so far for no.1 the top on the internet are VSG (Techpowerup, Thermal Bench), HWcooling, HWBusters, Quasarzone, Expreview. and there's more that are good. for no.2 there's only really HWcooling (but fortunately their findings fit with user experience almost 1:1, I've been following user experience for weird frequency and vibrational noise for about ~7 years now).

for no.3 the best is GamersNexus

 

and then for no.4 some youtubers give you bits and pieces but mostly just Machines & More for a few small form factor cases.

 

doyll who I mentioned, also made a really good thread back in 2014, combining all the usual questions people have / all the standard fan related stuff:

https://www.overclock.net/threads/ways-to-better-cooling-airflow-cooler-fan-data.1491876/

Not all of it applies nowadays, but it is the most comprehensive, useful starting point to understand fan basics

[RevGAM]

@Dogzilla07Thank you for the information! Right now, I'm tired and frustrated because my spreadsheet keeps on doing erroneous things when I sort - it doesn't matter which program I use. It's been driving me nuts for over a week now, so my brain is now full of cotton and I'm stupider than a rock.  

Do you know of any that should be avoided?

[Dogzilla07]

8 minutes ago, RevGAM said:

Do you know of any that should be avoided?

Any what ? (I'm not sure what part you mean specifically)

[RevGAM]

10 minutes ago, Dogzilla07 said:

Any what ? (I'm not sure what part you mean specifically)

Sorry, like I said - I'm a rock now. The opposite end of the spectrum from the best reviewers you've listed, both text and videos.

[Dogzilla07]

25 minutes ago, RevGAM said:

The opposite end of the spectrum from the best reviewers you've listed, both text and videos.

Ah, you meant reviews to compare to, hm, I've only really kept track of the the ones that stand out, and I kept track of tier II reviewers in the past.

 

In general, you could look for ideas for graphs and presentation, etc, from anyone. But not the accuracy of the numbers itself. Usually if they don't show their methodology, don't have anywhere explained in detail how they do it, that's a bad sign.

[RevGAM]

49 minutes ago, Dogzilla07 said:

Ah, you meant reviews to compare to, hm, I've only really kept track of the the ones that stand out, and I kept track of tier II reviewers in the past.

 

In general, you could look for ideas for graphs and presentation, etc, from anyone. But not the accuracy of the numbers itself. Usually if they don't show their methodology, don't have anywhere explained in detail how they do it, that's a bad sign.

Yeah, I automatically bypass any reviewers who don't have charts of data, or whose methodology is flawed (like Tech Notice using an open test bench). If I don't know how they're making their decisions and where the data is from, it's not worth looking at. There are a lot like that, as I'm sure you know!

[Lurking]

6 hours ago, RevGAM said:

That is very  interesting! It sounds as if, once you get sufficient experience, it almost becomes an art to designing HVAC, but I don't mean that in a bad way. How long does it take to design and commission, on average, in a home? Would you provide examples of mistakes in designs? Does ductwork from the unit to branches have to be consistent in size?

What are "in. wg" and "Pa"? Is KW kilowatts? So, to provide "valid" info, I would select, say, the intersection of the 1.5 KW line with 600 Pa just under 5000 m3/hr, right?

I think I might have mis-communicated regarding highly technical methodologies like HWCooling's,  ThermaLeft's and Aris's. I'm not saying the data is invalid, nor did I mean to imply that they're doing it wrong. I was really just talking about the difference between real-world data and what comes out of standardized testing. One of the things I quickly realized as a teacher is that standardized testing has weaknesses. Now, fans aren't pupils, of course, but any type of standardized testing has strengths and weaknesses. Honestly, the problems with testing in education are far more severe than, I think, in testing fans. It is not my goal to disprove their work but, rather, to corroborate the work of others and be able to tell which reviewers are just sell-outs or selectively promoting brands with skewed or false data, with the ultimate goal of providing useful info to end-users.

For you, it is easy to make certain assertions about one fan over another, and to suggest which fan(s) in a group of options are best in a given task, but most people don't know even the basics about fans. They don't understand that there is both AF and SP, they just know "low" means quiet and slow, "medium" means noisier and faster, and "high" means very noisy and very fast, but it varies from one fan model to the next. Some don't even understand RPMs, as flabbergasting as that is. Thus, it is surprising how many people get overwhelmed just trying to choose computer fans because choosing a box fan to keep you cool is a comparatively simple task that doesn't require understanding anything more than "how noisy" and "how well" a fan operates in a general sense.

 

I'm thinking about cutting out a viewing window in my tunnel, replacing that with hard plastic, and having some pieces of perhaps crepe paper hanging there to visualize the airflow for me.

Would you please answer the questions I posed in my last post?

Most residential home HVAC systems in the US aren't really designed and perform poorly. Contractor just uses a 2-ton unit for a specific house size and runs 6"duct to each room. if the run is longer, you just get less flow in that room.  For larger homes they get a 3-ton unit...  Yes, they should get designed properly, but aren't. Use of flex duct is another mistake. I only design commercial systems and we use sophisticated controls and actual design etc. 

 

If you move 1 m^3/s at 1 Pa pressuredrop, you need 1W of air power. If we assume fan and motor efficiency multiply to 0.5, this means you need 2W electric power. It is a bit more complicated in IP units (and you learn quick why metric system is superior). "in-wc" is 1 inch of water column, which is one of the multiple IP units for pressure.  You can look up conversions, but all are pressure. mm-Hg is mm of mercury, psi= pounds/square-inch, psf- pound/squarefoot. You all can convert them to Pa or other units.  

 

Before worrying about fans, I'd optimize the case. If you don't have a case that allows good flow and large fans, all the time and money spent on fans is kind of wasted. I stumbled across this thread because I considered upgrading fans in my Fractal R5 case, which actually is a relatively well ventilated case compared to some of the abominations of cases. But it has 4 intake fans and one exhaust fan (assuming i don't use the top fan locations). So 5 good 140mm fans... Even if i just get the cheap Arctic fans, that would have been $50, or over $100 for Noctua or over $150 for Bequiet fans. It sure would have helped a bit. But I ultimately decided to get an airflow case with large fans. One option was the Lancool 216 with 160mm fans for $100. I ended up with a $90 used Torrent Compact with 180mm fans and now have a superior and cheaper solution (better cooling at low noise) compared to buying new fancy fans. and I now have the (still OK) R5 for other uses. What I'm trying to say is, before finding the best fan for a bad case, get a better case (and if it comes with good fans, it is a bargain). 

 

I hope this large fan (i.e. 180mm) idea is catching on and the next step is to get rid of the small 120/140mm fans the same way we got rid of 80 mm fans. It is just simple physics that larger fans are more efficient and larger cross sectional areas (filters etc.) allow much less pressure drop at a given flowrate if the larger fans are used. There is just no magic you can apply to a 120 or 140mm fan that will make it as good as a 180mm fan. Large fan airflow cases are modern Dreadnoughts and made every small fan case obsolete at once.