Wind simulator fans

Edit - Just putting this in at the front now that all the answers seem to be in. It works great! The TerraBloom fans are quiet and provide plenty of air. So far they give very close to what the SeaFlo fans do for wind speed out of a 4" tube but at much lower sound level. I wasn't measuring identical setups and still think the TerraBlooms are putting out more air but that's not in the numbers. Makes flying and driving a joy. There is lag but for me it's not noticeable in an airplane but noticeable if I look for it in a race car, but not out of place or objectionable to me at all. Control is easy - PWM, DAC, or digital potentiometer - you just want to swing 0-10V which is why the digital pot might be the best solution. The fans have their own power supplies so you just need outlets. No Monster Moto needed either. My installation is just an Arduino Uno with 2 channels out of a 4 channel level shifter on the PWM outputs.



..............................................................................................

After wearing out a couple of sets of Sea-Flo 4" 12V 6A fans, I started looking for possible alternatives. I found these from Terra-Bloom. They are ball bearing instead of sleeve and have their own built-in electronic commutation - no brushes and no additional power supply needed. Plug them in and feed them the appropriate control signal for 0-100%. They take a PWM signal or a 0-10V level (DAC or digital pot) to control the speed. The Sea-Flo 4" units are rated at 270 cfm. The Terra-Bloom ECMF-150 is rated at 350 cfm. These are also apparently engineered to do better against a back pressure and may do better if ducting the air with flexible tubing since working against pressure is a feature they bullet on their information.

Down sides are cost and a fairly slow response time. They are about $90 each for the 6" model (similar footprint as the Sea-Flo) and by ear it sounds like 12 seconds for a 0-100% step function. Car drivers may notice that lag. I haven't tried them yet in my sim to be able to say for myself - waiting on the parts to interface my wind Arduino to run them under sim control using 21-25kHz PWM. I think it should generally be OK for flight, though. I asked customer support about the lag and the person I was talking with said it's basically a soft start to prevent stress on components. I would guess it's for keeping currents down closer to running currents so it doesn't need a beefier built-in power supply - the down side of getting away from brushes. At 40W it still draws less than bilge fans rated at 12V/6A/72W.

The ECMF-150 has 7 rotor blades, 7 stator blades, are much, much quieter than the Sea-Flo, and the bearings are rated for 70,000 hours (8 years) continuous. They have sound level videos where you can see/hear what they sound like.

Website is https://terra-bloom.com/

https://terra-bloom.com/collections...-and-hydroponics-airflow-boosting-350-cfm-40w

Edit 2/8/21 - TerraBloom has modified their pages for these fans to include information that they can be controlled by PWM from Arduino and Raspberry Pi microcontrollers. Apparently they are embracing this kind of application. If they are paying attention to us, there is more hope for faster response or maybe even a line of fast response fans sometime in the future.

Anyway, I'm going to give them a go and see how they work out. I'm tired of fans that squeal and squeak, tick, or whatever. The quiet will be nice. They still make noise - just not so much of it. They also have a glasspack muffler style that is supposedly extremely quiet (another $60) but it doesn't have the speed control options the ECMF series does. The ECMF-150 is noticeably quieter than the Sea-Flo, though. The extra wind is a bonus.

It also has a sculpted inlet and the rotor gets significantly bigger in diameter before the air hits the stators. The exhaust is just about the full 6" diameter so I'll get a bit of back pressure with a 6" to 4" adapter but they are supposed to work what is considered well with back pressure so time will tell. That should get me better velocity too. But one spin of that rotor and you instantly feel a difference from the bilge fans. It's got some weight to it (part of the lag issue) but it spins smooth. I can't even hear it at low speeds. It's silent.

The PWM control is new as of only a few months ago. If anyone else is interested in these with PWM, be aware that there are old and new versions and you want the new to be able to use it. There is a tamper sticker on the electronics bay hatch with manufacture date. (Just flexing that hatch up a little to take a peek breaks it, FYI. Guess how I know ) I’ll see what the cutoff is or if there are ways to visually verify old vs. new. Amazon has that it’s a new version in the product description and it’s probably safe any bought direct are new. I’m not sure yet how to set an Arduino to a 21-25kHz PWM frequency but other micros might not have that issue, if it is an issue. An I2C or SPI DAC or digital potentiometer can also be used.

thanks for your experiments and sharing!
nice PWM driven fan
maybe I'll have a try with the wind simulation one day, it's tempting

yes, the soft start is a pain in this case...

you can drive 20kHz PWM with arduino indeed, here is the code I shared https://www.xsimulator.net/community/marketplace/ultrasonic-pwm-with-arduino.102/ in order to prevent the humming of brushed motors...

Hey Mat, thanks for the heads up! I remember the code in the wind Arduino had a high speed silent option but it’s been so long and I hadn’t looked back at it. If it’s already there and a compatible PWM frequency, it’s possible everything is good to go already and I just need to level shift up to 0-10V. Adafruit has a level shifter that I think will work great and is capable up to a MHz or two. I’ve already got one so it will be easy to test. (https://www.adafruit.com/product/757)

The soft start is the biggest issue IMO but once moving, speed changes would be more incremental. Cars can rocket out from stopped though and that may be weird to have wind grow in after already largely at speed and a frequent break in “reality”. I mostly fly so I’m hoping it’s less of an issue there. I’ll know more once I get them wired in. I’ve got a few questions in to customer support and will update with their answers.

But wind is great in these sims. Highly recommended! It really changes things and for me in VR and it also increases comfort on long flights/races. The whine and noise from the Sea-Flo's was a negative, though, and immersion gets wrecked when a fan starts squeaking or ticking and you hear that above other sim ambient sounds. I’m really curious how quiet these are in actual use but so far are very promising. Unless the bearings act up, there really should be no mechanical noise and wind noise itself is low. May be able to quieten it up even more by trimming small flash bits still at mold seams.

Got this image from customer support showing the different control wirings. There’s also a tach output but no real use in this application I can see. The 6” fan has a 5000 RPM max speed.

Thanks again for the heads-up! Will be testing later today.

It works!

I heard back from customer support and stuff they sent shows the input is basically an RC filter so frequency isn’t critical and lower is fine. I’d bet the stock analogWrite at 980Hz would even be fine. No buzz with these.

Another surprise was the 10V line. It’s a 300mA, 10V budget designed to drive external circuitry. The Adafruit level shifter I used is being loaded slightly by the control input so need to sort that but it’s only off by 5% (9.46V) with pwm 255. Missing about 5% off the top end but there is plenty of power for whatever circuitry is needed.

So almost there but it’s good enough to use now. I need to wait for its twin to arrive, get them mounted, wired, and adapted to the existing 4” ducting or maybe go 6” later as the velocity is already really good. Turns out this is way more simple than it originally seemed. I just need to get that PWM signal to a full 10V.

The only question now is how noticeable the soft start lag is. Control is easy.

And the way to tell the new PWM-capable fans from the older style is just by the connectors used. The new style uses a 4 conductor miniplug. The old style was a three pin connector.

Minor update. Second fan arrived today, got the old fan unit off, new fans mounted, and printing the adapter cone parts now. It’s going to take all day tomorrow and into Wednesday to get the cones printed provided no troubles. I contoured the inside surface to hopefully help the flow. The black look is a bonus.

I didn't view the code but timer settings are specific to the microprocessor. So for clarity isn't this just for an Uno? I know for fact that a Mega for example needs different settings. Note I posted here because discussion isn't included on the download page.

You are right, BlazinH.

But at least to run these fans, the PWM frequency doesn’t need to be high frequency. I think the customer service rep was using old information or not sure how he came up with the original 21-25kHz frequency requirement, but it works great all the way to 4kHz that I’ve tried. I’m just using the stock code I was using with the Monster Moto shield now. 8kHz. Works great!

The final adapter cone, flow straightener, and fan tail cone. If anyone else uses these fans and the tail cone, it will void the warranty since the cover is what has the tamper tag on it. You should also be able to place this cone over the electronics bay cover without disturbing the tamper tag but may need longer screws. Haven't tried it yet. At any rate, I'm hoping the tail cone will lower noise even more and maybe even pick up some volume and velocity. It's modeled after a jet engine tail cone. Even if it adds nothing, it looks cool.

I'll post the .stl files as soon as I have one complete to know it all works together. These elbows I'm using are these: amazon.com/dp/B005VRJ6OI The flow straightener is a press fit into the elbow and the elbow is a press fit into the 6" to 4" adapter. The adapter is a press fit onto the exhaust of the fan and the fans are oriented vertically with intake down and exhaust up.



Old setup was clunky with flow straighteners taped on. Flow straighteners in the elbows is much cleaner.

New setup is kind of trick but the real test comes when checking out the lag. Control works (will post PWM and Arduino box details when that’s final), physically everything fits, what remains is if the extra wind is noticeable, and how noticeable the lag is.

They are an expensive option, though. But after wearing out two sets of SeaFlo fans, I needed to do something else. One of the second set fans was squeaky almost from the beginning. Maddening.

I made some minor changes to the adapter section to give a better mate to the elbow so I get to start over printing those. Couldn't completely test until I had the four segments to be able to bolt up. 8-32 x 3/8 black finish pan head screws to hold the adapter sections together. And it turns out the tail cone for the fan goes right over the stock cover. The stock screws hold it well so no need to void warranty.

I'm not going to put the tail cone in the other fan until after I get it up and running. Curious if it makes a difference but with two fans, I should be able to see a difference if it turns out to be significant. It looks cool, though.

Kind of tempted to go after the surfaces in the wind channel with filler to smooth them...




Edit - added a simple box for the Arduino and level shifter that has cutouts for the connectors that come with the fans in the little control boxes. Screw threads are M3.



Also, the tail cone adds about 2 mph at top speed. It's such a small amount that I don't feel any difference qualitatively comparing between the two fans but it is another 2 mph if you do put them in.

Me too!

Will post. Got the final bits to hook things up arriving soon. But the rest will take a couple of days to get done.

Just a random thought: if lag ends up being a problem and they're really quiet, another option might be to run the fans at full speed all the time and instead motorize a damper/diverter valve in the ductwork.

@noorbeast, I think @dododge might be referring to a servo-controlled damper/diverter where it would be varied via telemetry.

@dododge, when the fans are at full speed they do make noise even though what seems like significantly less than noise than the bilge fans. At low speeds they are almost silent. I got an app for my tablet that measures sound pressure levels and broke down and ordered an anemometer to be able to quantify the wind speeds so this isn’t all hand waving. I would rather put numbers to this, even unflattering ones if they are.

But I think the numbers will show these fans are a big difference by both metrics. The old fan rig is still on its mounting plate so it’s just plugging in power and hooking up a USB cable to bring it back to life.

I was running it using the pots the fans came with and am cutting over to Arduino control, but air volumes/wind is nuts. It seems way up and noise seems way down. But no idea on how the lag is just yet. Will post more soon...

Still don't have info on lag, sound levels, or wind speeds, but do have information on the interface. I am going to keep looking for a good digital pot or DAC breakout board but it's one of those things where it's do I really need to? Thing is that the level shifter doesn't seem to be able to get all the way to the +10V rail for full fan speed. It's close, though, and closer than it was. The stock Adafruit level shifter (https://www.adafruit.com/product/757) would get up to 9.46V at PWM value 255. The pullup resistors on the high and low sides are all 10k chips and between whatever drop there is in the BSS138 FETs, that high of a resistor doesn't allow but 1mA to feed the fan input circuit (10V/10k). I soldered another set of 10k chip resistors on top of the stock ones on the HV side to make the equivalent of 5k pullups for double the current. It improved things but not that much. High side voltage only comes up to about 9.65V now. At PWM 255 the output becomes a level so the fan speeds are at about 96.5% of full speed. I don't want to provide another voltage just so I can swing the full 0-10V, though. I can stand losing 3.5% of the wind.

Here are scope traces and you can see what is going on. At low duty cycle, the pulse gets rounded off pretty severe and probably makes the fan more non-linear at low RPM. They also don't start running until I hit a PWM value of 16 so I lose the very low speeds but starting force and stiction is also a thing with brushed motors so no big deal? There is also a built in step where it's easier to make sure you can send a zero RPM command even if your circuit doesn't go all the way to zero. There will always be some offset before the fans start.



Look close and you see the dead FET gap.



Those are PWM values of 0, 10, 20, and 200 with the 5k equivalent pullups on the high side of the level shifter. I'm losing some voltage to whatever is happening in the fan speed control input circuit but rise times are faster now thanks to the lower resistance pullups. Less rounding. The Arduino is hardly getting loaded (it has to sink most of the current to pull the pins low) so I probably could drop the resistance even more but it's a bother for another percent or two. The 5V I/O of the Arduino is banging next to 0 and 5V. So far it doesn't care about the level shifter or fans.

One other thing about the level shifter - you can pop them fairly easy. I slipped with my scope probe and popped a FET just that quick. Luckily there are 4 channels of level shifting and I only need two. Just had to move input and output on the dead FET to a good one. One left now before I have to swap out the little board for new.

Up and running full on the Arduino now for Game Dash control. Printing out the last section of the adapter cones now so will soon be able to put it together and test in sim to start seeing qualitatively about the lag.

Really curious about the numbers. So far it's working great.

5V from the Arduino goes to the LV input of the level shifter. Gnd from Arduino goes to Gnd and the two PWM outputs just go to level shifter inputs. Similarly, the level shifter outputs go to the two blue fan control input wires and Gnd from the fans get tied together on the level shifter Gnd. The oddball is how to hook up the 10V from the fan power supplies. Each cable and fan has one. I didn't want to tie them both to the HV terminal on the level shifter because not knowing about their power supply, it could kill one or both. I didn't want to OR them with diodes because that guarantees a diode drop down from the 10V lines so reduces the max voltage available to the speed controls and therefore fan speed. So I punted. I use one 10V supply and stuck the other 10V line into a drop of hot glue to insulate it. The consequence is whichever fan you use for the 10V supply has to be powered for the circuit to work. In practice it's no big deal but something to be aware of if anyone duplicates this. I also don't use the yellow tach lines. Those also got stuck into glue, but if anyone wanted to servo the wind fans, that would be the ticket.

More later.

Yeah I was definitely thinking of a variable damper, basically just using the same speed telemetry to control how much of the airflow is directed toward the driver. I don't know if this sort of damper-and-alternate-path arrangement might have major downsides for noise or turbulence, though, and perhaps the resistance difference in each path would require the actuator to use some special response curve.

Variable diverters with motors already exist for HVAC zoning, though I suspect the turnkey ones are for really large systems and you could do it much cheaper DIY. For example here's a 150mm tee with a shaft and mounting plate ready for mounting your own actuator, for £26:

Sound levels go way up at high fan speeds for both inlet and exhaust, though. Otherwise it would be a very responsive solution.

These fans have significantly higher CFM ratings and are already quiet. By working lower on the fan curve, I can get the same wind at lower noise but damn if they don’t push a lot of air at high speed. With my Index the wind noise itself isn’t bad but the fan/blade whine is.

I sure hope it works well. Had to stop. Almost there.

Just got it working where I can test it in actual use. It’s just qualitative, but this setup rocks for flight! In sound level testing so far, the user location is seeing up to 94 db at full speed where the wind tubes are aimed, but 59 db over by the inlets. Background without fans running was 39 db. (Measurements made using Audiobel on an iPad.) It sounds like the inlet noise is dominated more by blade noise and kind of higher frequency inlet turbulence, but in the pilot position it seems much more like just plain old wind turbulence. I can’t hear the fan motors at all.

I can really influence the inlet noise by putting my hand near it to disrupt the flow. @Avenga76 did some great work with inlet velocity stacks and I'll be adding those too. With the amount of noise that gets generated by moving my hand near the inlet, it could be velocity stacks would make the fans even quieter.

The fans not starting until a PWM value of 16 just takes a math + 15 statement in the dash setup in Game Dash. Pretty much as soon as I start rolling, the fans start coming in and that step is no longer an issue. Having the fans start spinning nearly immediately may also help with eliminating the perception of lag.

And at least in flying, for me the fan lag isn’t an issue at all. In closed cockpit stuff you aren’t out in the wind so all I’m trying to mimic are fresh air vents and speeds are math'ed out lower anyway. In open cockpit stuff the winds are definitely higher ( * 1 ) and louder but now all I hear are the sim sounds and very realistic turbulence. Tons of wind but the turbulence drowns out the fan noise.

Definitely a winner for flight. It’s like night and day.

Will also try some Project Cars and Assetto to try out open cockpit wind. Will be able to get some actual speed numbers later as well.

In a nutshell, 2 of these fans at $90 each plus a level shifter ($4) vs 2 bilge fans at $30 each + power supply ($10-20) + Monster Moto ($15). These definitely cost more but replace the bilge fans a few times and you’ve paid for these.

I can’t comment on longevity on these new fans. They use ball bearings rated for 70,000 hours so I think that issue is licked. No brushes since electronically commutated. Nothing really to make noise but don’t know how the control electronics will hold up long term. That is an unknown in this application with constantly varying speeds.

No idea how many are having issues with the bilge fans. Their noise really got to be an issue for me when I went to the Valve Index with the off ear headphones/speakers. They sound great but other sounds easily come in. The other half was that I've been through two sets of bilge fans already as the bearings squeal or brushes click (not certain it’s the brushes clicking but that doesn’t matter anyway), and one of the ones I was using until now also started clicking.

But for flying, I’m more than happy. I’ll go swap for driving configuration and see what that’s like...

FWIW from my brief experiment with making a velocity stack (I have the SRS kit that uses computer fans) it added a noticeable whooshing noise to the intake. I didn't have anything to measure the wind speed at the time to see if it made a difference, but it certainly sounded like something was going on.




This was just a quick test using sketchup so there's a lot of visible segments and edges in the print. I keep meaning to do it again using Fusion360 to tinker with the parameters and maybe get a smoother curve.