Cooling fan upgrade comparison: Explorer 11-blade fan and HD clutch versus SPAL 19" 850-watt electric brushless fan
- Thread starter Steel City 06
- Start date
Figure I would post an update after a few months of running the SPAL fan.
Overall, I am very pleased and impressed with the performance of the SPAL 19" fan in comparison to both the OEM fan/clutch assembly and the Explorer fan/clutch assembly. A summary of things I've noticed:
As for bugs, I really only have one small one I'm still working to address. On engine start-up, the fan is obviously shut off, since the coolant temperature starts at ambient, and stays at ambient until the thermostat has opened to a significant degree, usually ~5 minutes or so after startup. This is not a problem (and actually mostly a good thing) in most scenarios, since it shortens warm-up time, and also doesn't consume unneeded power.
The issue presents if you also turn on your A/C right at startup, and then don't immediately start driving or turn your fan override on. With the A/C going and no airflow (since the coolant is still ambient), the A/C condenser can't reject enough heat, causing the condenser pressure to rise until the high pressure switch trips and disengages the compressor. This is very hard on the compressor, and risks causing damage if repeated many times.
The basic solution is just to have the user turn the "override on" switch on at startup, and then turn it off once they are driving or the coolant is up to temperature. However, this solution depends on the user remembering to do this, and generally speaking, human input is the most unreliable factor in a control scheme. Thus, an automated solution would be better. In addition, the fan doesn't need to be maxed out just for the A/C, and can be run at just 25% of the power to get nearly identical A/C performance with less fuel consumption. My goal is to develop a "low override on" capability that turns the fan on at a minimum state when the A/C compressor is on, thus eliminating the risk of damaging the compressor. I've had a few ideas:
In summary: Highly recommend the SPAL 19" fan and controller.
Overall, I am very pleased and impressed with the performance of the SPAL 19" fan in comparison to both the OEM fan/clutch assembly and the Explorer fan/clutch assembly. A summary of things I've noticed:
- The sound made by the SPAL fan assembly is very different than the Explorer fan assembly. Imagine a dump truck taking off from a traffic light going up a hill. That's what the Explorer fan sounds like when the clutch is engaged. Meanwhile, the SPAL fan is fairly quiet at lower speeds, but as it ramps up to max it can also be quite loud, though not nearly as loud as the Explorer fan. Standing in front of the car, it sounds like an air raid siren spinning up. However, unless you flip the switch, you are rarely running max power.
- The power consumption of the fan is normally very minimal (unless you engage the override on switch), with some peaks during high load situations and while idling with the A/C on. I would estimate that >90% of the time, the fan is drawing between 0-12A.
- The A/C is certainly noticeably cooler when the fan is on max and the vehicle is sitting still. Above about ~35 mph, the fan doesn't seem to make a difference in the A/C temperature, which is expected based on the speed of the air entering the radiator.
- Coolant temps seem to be fairly stable (almost always staying between about 195F and 205F) with the current control scheme. Warm-up time is slightly reduced, and the system usually reacts within about 5 seconds to a change in load or road speed.
- For example, during normal driving, the fan is usually off, as there is more than adequate airflow to drop the radiator return temperature below the 130F setpoint. Once you approach a stoplight, usually the fan comes on low within about 10 seconds or so.
- I have never had an overheat condition or anything approaching it with this setup. That said, the highest ambient temperatures I've tested it in are about 95F (per the forecast), though that was with the A/C on maximum and a wide variety of speeds and fairly steep/long hills.
- Coolant temps vary even less if the override on switch is used, since the fan no longer reacts to temperature, but allows the thermostat to do 100% of the control. Not really useful for normal driving, but might be useful if you're abusing the car off-road with many rapid changes in power demand and road speed.
- I suspect there is still quite a bit more headroom in overall cooling capacity, so I would not hesitate to try this setup in harsher conditions.
- Haven't gotten any real good data on the difference in fuel economy, but there is a noticeable reduction in drag on the engine especially at higher RPMs for the SPAL fan in lieu of the Explorer fan.
- Warm-up times are reduced a bit, but not a whole lot. (Faster warm-up generally improves fuel economy and emissions.)
- There actually is a fairly noticeable increase in engine drag when the override on switch is turned on. I would estimate about half to 2/3rds that consumed by the A/C compressor. With the A/C on high and the fan maxed out, the drag is VERY noticeable compared to with the A/C off and fan off (or at low speed), especially when decelerating.
- The engine drag created by the SPAL fan (when maxed out) is probably comparable to the Explorer fan (with the clutch engaged) at idle to ~1,500 RPM. However, unlike the Explorer fan, the drag does not increase with engine RPM, and is not very noticeable at higher RPMs. Theoretically, the engine drag gets even more interesting:
- The clutch fan, when engaged, operates at a speed roughly proportional to the speed of the engine, usually 80-90%. Using the fan affinity laws, if we 2X RPM, airflow increases 2X, static pressure and torque increase 4X, and shaft power increases 8X.
- The electronic fan, when maxed out, operates at a constant shaft power/speed/airflow, independent of engine RPM. The PCM compensates for engine RPM by adjusting alternator field current to maintain constant power, which results in the system having constant power drag on the engine, and an inversely proportional torque drag on the engine.
- So if you double engine RPM, on the clutch fan, you have four times the torque drag on the engine. Meanwhile, on the electronic fan, torque drag is actually halved in the same scenario! Alternatively, doubling engine RPM causes the clutch fan to use 8X the power, while the electronic fan still only uses 1X!
- Fuel economy has seemed to be noticeably better, but I haven't had a chance to get a very good apples-to-apples comparison yet. I suspect I'm getting about 17-18 MPG, whereas before, I was closer to 16-17 MPG. (Note I am the typical heavy Rubicon on 35s and all steel everything.)
- The engine does seem to be able to rev up a bit more easily, especially on the high end.
As for bugs, I really only have one small one I'm still working to address. On engine start-up, the fan is obviously shut off, since the coolant temperature starts at ambient, and stays at ambient until the thermostat has opened to a significant degree, usually ~5 minutes or so after startup. This is not a problem (and actually mostly a good thing) in most scenarios, since it shortens warm-up time, and also doesn't consume unneeded power.
The issue presents if you also turn on your A/C right at startup, and then don't immediately start driving or turn your fan override on. With the A/C going and no airflow (since the coolant is still ambient), the A/C condenser can't reject enough heat, causing the condenser pressure to rise until the high pressure switch trips and disengages the compressor. This is very hard on the compressor, and risks causing damage if repeated many times.
The basic solution is just to have the user turn the "override on" switch on at startup, and then turn it off once they are driving or the coolant is up to temperature. However, this solution depends on the user remembering to do this, and generally speaking, human input is the most unreliable factor in a control scheme. Thus, an automated solution would be better. In addition, the fan doesn't need to be maxed out just for the A/C, and can be run at just 25% of the power to get nearly identical A/C performance with less fuel consumption. My goal is to develop a "low override on" capability that turns the fan on at a minimum state when the A/C compressor is on, thus eliminating the risk of damaging the compressor. I've had a few ideas:
- Use a Zener diode to clamp the voltage across the thermistor to a maximum value when a ground is provided by the A/C on switch. This I did not have success with, as the tiny amount of current provided by the controller was not enough to trigger a Zener (reverse) breakdown. I may try again using a diode in forward orientation; the issue is that the forward voltage drop across a diode is not independent of current, and thus much harder to calibrate.
- Use a resistor instead of a diode to lower the sensed resistance of the thermistor when a ground is provided. This would certainly turn the fan on, but it would also have the effect of lowering the "max speed" temperature a significant amount, which would result in the fan running unnecessarily high whenever the A/C is turned on.
- Swap the controller out for an Arduino. This is probably the best option in terms of getting things to work as intended, but my C++ and Arduino skills are currently just enough to make me dangerous with a 3D printer, and building a program from scratch would be quite the undertaking for me.
- Wire the "override high" switch to the A/C switch in parallel, so the fan runs at max speed whenever the A/C is on. Certainly the simplest solution, but also would waste an awful lot of energy.
In summary: Highly recommend the SPAL 19" fan and controller.
Great write-up, Steel. Thanks for the follow-up data.
So while messing around with Zener diodes, I kept running into the problem of non-linearity in the breakdown voltage at the low currents that are used to read the coolant temperature sensor. I originally expected a clean Zener breakdown voltage; i.e., I expected a 770 mV diode to have a Zener voltage at very close to 770 mV regardless of the current flowing through it. But this was not the case. With the 2,200 ohm sense resistor inline, the actual Zener voltage fluctuated somewhat with the changing resistance of the coolant temperature sensor (which is simply a NTC thermistor). This is why I originally gave up on the Zener diode.
But, I decided to try a circuit anyways, and actually found a completely new control strategy I like even better than the original. Essentially, the Zener diode wired in parallel to the NTC thermistor is used to ensure the fan always is running at least it's minimum speed, and then as the NTC thermistor dops resistance (rising radiator outlet coolant temperature), eventually it takes over, ramping the fan to a higher speed. The issue I expected was that it would be a harsh cutoff, given that I would have to set the minimum coolant temperature to turn on the fan to 160F just to get it to run at minimum state most of the time. However, what I actually found is that the non-linear response of the Zener diode allows the fan to ramp up non-linearly as the coolant temperature increases.
The setup includes the following:
Edit: Turns out the forum software doesn't like copy-pasted Excel, so the below is provided as a screenshot.
Effectively, this system generates an output in which the fan is always on at the lowest speed, and non-linearly ramps up as radiator outlet coolant temperature rises. At the lowest speed, this system only uses about 60W. The fan ramps up only minimally in response to coolant temperature until about 170F, at which point it ramps aggressively, maxing out at about 200F.
This setup, although it uses a small amount of power constantly, seems to be even more efficient than the previous linear setup. 1st, the fan, when running at minimum speed, only uses about 60W, or about that of your cabin blower at its lowest speed. This comes out to about 0.2 gallons every 1,000 miles in a worst-case scenario (versus a fan that doesn't run at all). 2nd, the fan responds gently to moderate increases in temperature occurring during normal driving conditions, and even at 170F, the fan is still consuming less power than your cabin blower takes to run at it's highest speed. Ramping to high power only occurs when coolant outlet temperature begins to spike above 170F, which would be indicative that the radiator is nearing its rejection capacity, and only running max power in a near-overheat condition.
With this setup, there is no risk of damaging the A/C compressor either, and overall, the fan runs much quieter, since it stays at low RPM the vast majority of the time. It may also present a benefit by lowering underhood temperatures, though this I have not tested quantitatively. I have not noticed any fan speed cycling as a result of this setup.
Edited as I messed up the Zener diode info. I used a fairly high voltage Zener diode in the forward configuration, not in Zener configuration. Updated part number and link.
But, I decided to try a circuit anyways, and actually found a completely new control strategy I like even better than the original. Essentially, the Zener diode wired in parallel to the NTC thermistor is used to ensure the fan always is running at least it's minimum speed, and then as the NTC thermistor dops resistance (rising radiator outlet coolant temperature), eventually it takes over, ramping the fan to a higher speed. The issue I expected was that it would be a harsh cutoff, given that I would have to set the minimum coolant temperature to turn on the fan to 160F just to get it to run at minimum state most of the time. However, what I actually found is that the non-linear response of the Zener diode allows the fan to ramp up non-linearly as the coolant temperature increases.
The setup includes the following:
- Set "fan on" temperature to 160F
- Set "fan max" temperature to 200F
- Coolant temperature sensor in radiator outlet (same as previous)
- 1N6006B Zener diode wired in parallel with GM coolant temperature sensor (NTC thermistor), wired in forward direction (stripe towards sensor ground).
Edit: Turns out the forum software doesn't like copy-pasted Excel, so the below is provided as a screenshot.
Effectively, this system generates an output in which the fan is always on at the lowest speed, and non-linearly ramps up as radiator outlet coolant temperature rises. At the lowest speed, this system only uses about 60W. The fan ramps up only minimally in response to coolant temperature until about 170F, at which point it ramps aggressively, maxing out at about 200F.
This setup, although it uses a small amount of power constantly, seems to be even more efficient than the previous linear setup. 1st, the fan, when running at minimum speed, only uses about 60W, or about that of your cabin blower at its lowest speed. This comes out to about 0.2 gallons every 1,000 miles in a worst-case scenario (versus a fan that doesn't run at all). 2nd, the fan responds gently to moderate increases in temperature occurring during normal driving conditions, and even at 170F, the fan is still consuming less power than your cabin blower takes to run at it's highest speed. Ramping to high power only occurs when coolant outlet temperature begins to spike above 170F, which would be indicative that the radiator is nearing its rejection capacity, and only running max power in a near-overheat condition.
With this setup, there is no risk of damaging the A/C compressor either, and overall, the fan runs much quieter, since it stays at low RPM the vast majority of the time. It may also present a benefit by lowering underhood temperatures, though this I have not tested quantitatively. I have not noticed any fan speed cycling as a result of this setup.
Edited as I messed up the Zener diode info. I used a fairly high voltage Zener diode in the forward configuration, not in Zener configuration. Updated part number and link.
Last edited:
Also, worth noting I am now up to 17.4 MPG as measured on a 150 mile stretch of mountainous highway (US 30). True highway mileage is probably closer to 18, and city is probably closer to 16, but I don't have hard data to back that up yet. Again, note I am the usual 35s on 5" of lift and heavy steel bumpers/skid plates.
So to sum this up for idiots like me you are sending an alternate ground(?) signal to the lingenfelter controller with a diode in addition to the coolant sensor?
This is for the purpose of ac performance and pump protection before the motor is warmed up?
This was done because you couldn't find a good way to take a signal from the ac controls on the dash or pcm and send it to the lingenfelter?
how does that effect the temp sensor ramping up the fan speed? Does the lingenfelter need to be adjusted to take the total added ohms ?
This is for the purpose of ac performance and pump protection before the motor is warmed up?
This was done because you couldn't find a good way to take a signal from the ac controls on the dash or pcm and send it to the lingenfelter?
how does that effect the temp sensor ramping up the fan speed? Does the lingenfelter need to be adjusted to take the total added ohms ?
That was the original idea. Basically the diode would be wired to the A/C enable wire, which provides a ground when the A/C is on. However, for this, I actually chose to simply wire the diode in parallel with the temperature coolant sensor, so the fan is always on (when the ignition is on and the system enable switch is on as well). Basically you're faking out the Lingenfelter controller by making it think that the coolant temperature is almost always about 166F (or otherwise higher).
The Lingenfelter controller is adjusted by changing the setpoints for the "fan on" and "fan max" temperature dials. It isn't actually directly reading the temperature; instead, it is reading the resistance of the engine coolant sensor, which negatively correlates with temperature. So if you fake out a different apparent resistance to the controller, you can get it to do anything you want. The diode is a way of introducing a resistance "floor" into the system so it never sees a "resistance" below about 400 ohms.
In reality, the Lingenfelter isn't even directly measuring the resistance of the temperature sensor, but actually the voltage drop across a fixed internal resistor wired in series with the temperature sensor. This resistor is 2,200 ohms, and is fed with 5V. So, for example, if your engine coolant sensor is also 2,200 ohms, the controller sees an effective voltage of 2.5V, which it then uses to calculate a temperature of 86.3F, which it can then use to determine what the fan speed should be (in our case, zero). What the diode is doing is limiting the maximum voltage drop across the coolant temperature sensor to 0.77V, or 166.0F. If the diode were perfectly linear, this would mean the controller would think there is always a temperature of 166F, except in the case it was higher (less than 0.77V), in which it would know the actual temperature in the system. (In reality, the lines are blurred due to the non-linearity of the diode.)
In layman's terms, we're simply feeding the controller fake temperature signals part of the time. Sort of like when you have a cheap landlord who locks the thermostat, so you put an ice pack on top of it to get it to crank the heat. Or perhaps more accurately, putting a hot lamp under it to get it to crank the AC.
Last edited:
Stock 4.10s. Originally planned to regear to 4.88 or maybe 4.56, but now have no intention of regearing. The combination of engine modifications and tuning I've done has way more than made up the loss of torque caused by switching from 31s to 35s, and I would definitely miss my current 6th gear way more than I would like having a lower 1st/2nd.
Did not expect that. I'm on 4.10s and metric 33s, getting 13.5 mpg on a good tank, less if i spend time on the highway. Granted I drive it like I stole it.
How much of that mpg gain do you attribute to the SPAL fan?
I can add that my 05 with the 6 speed gets into the 18’s with 33’s and 3:73’s . Cai, exhaust, Redline oil in the entire driveline(except the engine). I run a 2 speed electric fan off a ford Taurus. Like Steel city I have enough incremental gains that I pretty happy with the current configuration. I will say in my observations not one thing move the needle significantly- I have a “put in the bucket” approach. You don’t feel 1,2,3,4% increases but when you add a couple of those up and put it in the bucket- you start looking like 10-12% gains and that is something.
Hard to say. If I had to guess, maybe on the order of half a MPG. I've done a lot of things that have improved fuel economy, and unfortunately I didn't do them all in a vacuum, so I can't attribute any specific amount to any specific mod. That said, here is my general best-guess rank:
- Tuning - A lot of things to be gained here
- Advance timing where possible. The TJ has a very conservative timing map, and you can extract a lot more power from the fuel you are already burning. (Note I will not share my map as it is based on a FRP canned tune.)
- Dial back some of the excessive protective features, such as component overtemp protection. These will aggressively pull timing, but since the user reacts by adding more throttle or more RPMs, aren't really effective and often counterproductive.
- Set Power Enrichment to come on only in the last 5% or so of your pedal travel. You should only be in PE if your foot is to the floor. (Note you can also zero the delay and the min RPM for a cheap torque boost.) You should be in closed loop operation for >95% of your driving.
- Make Deceleration Fuel Cut-Off (DFCO) more aggressive. (Note I'm still having a stalling quirk with this, so not 100% de-bugged yet.)
- Increase idle RPM spark advance. Ever notice how if you just barely touch the throttle while idling, it quickly jumps in torque? This is because the spark advance is jumping from about 10 degrees to more like 32 degrees. If you set the idle spark advance to more like 20-24 degrees, idle fuel consumption is significantly reduced. The only downside is that the closer you go to the max advance (32 degrees BTDC), the less room your engine has to adjust idle torque, so there is a bit more propensity to stall if you let out the clutch too fast at idle. Idle also becomes a little less stable at the very high end.
- Windstar cowl air intake, with insulated intake tube - Massively lowers intake air temperatures. Note that most people will claim that the power to be gained here is mainly by the increased density of air. It is true that will help, but it isn't the biggest reason it works. What it actually does is cause the computer to see lower IATs, so it doesn't pull timing as aggressively. Meaning you can extract more power from the same amount of fuel burned. There is also a very slight ram air effect from the position of the intake, but probably nowhere near enough to be noticeable.
- Ceramic coated intake header, exhaust header, and catalytic converter assembly - reduce heat transfer from exhaust gas to intake gases. Not a huge deal in its own. but allows for more aggressive timing across the board in your tune. Basically, increases the detonation threshold, so you can be more aggressive without pinging. Same concept as the Windstar intake, although you have to tune for the gain. Note there may also be a small torque benefit from the higher density intake air, and possibly better exhaust scavenging due to the hotter exhaust flow.
If you do this, I highly recommend Cerakote in one of the low-emissivity variants, such as "piston coat". The low-emissivity variants are even better at trapping heat and reflecting heat than their standard counterparts. - SPAL fan - will gloss over since it's already discussed here.
- 12-hole injectors - These don't make a huge difference, but they make the burn more consistent, so you can be a bit more aggressive with timing and other things. Like many other items below here, they allow for a better tune, and together with the tune make a big difference. There may also be additional gain with these in combination with the insulated cowl intake and the ceramic coated headers/exhaust, since the colder air will not vaporize fuel as quickly unless droplet size is reduced.
- EcoBoost oil heater/cooler - discussed here: https://wranglertjforum.com/threads...ost-oil-cooler-heater-in-a-4-0-and-why.73481/
- Tire pressure - experiment with ~40 PSI. Definitely a significant effect on rolling resistance, though beyond a certain point you may begin to see increased tire wear. One of the side gains of higher tire pressure is less tire sideslip, meaning steering will feel tighter. Note higher tire pressure will magnify the effects of any slop you have in your steering or suspension, so these need to be 100% slop-free, or the increased pressure may make steering feel even worse.
- 5W-30 oil in lieu of 10W-30. No difference after warmup, but when cold, it has a lower viscosity than 10W-30, so will flow easier and lubricate better in that time. Also may reduce engine wear and tear. (The only advantage of 10W-30 over 5W-30 is that it is cheaper.)
- Flowkooler water pump - More flow at low and mid-RPMs means more stable coolant temperatures, which leads to the PCM pulling timing less. It may also mean lower risk of detonation during heavy load, allowing you to keep the timing more aggressive. Not sure if there are any gains in isolation, but it seems to help stabilize short term coolant temperature fluctuations.
In addition to these things, there may be a few additional means of improving fuel economy I have not tried:
- Tune for higher octane - allows for even more aggressive timing
- Reduce vehicle weight - generally only has a small effect based on my research, but noticeable for city driving
- Improve aerodynamics - This is really the elephant in the room for the TJ at freeway speeds
- Higher temperature thermostat - Fuel economy generally improves with higher engine temperature, at least until the point that timing has to be pulled because of pinging
- Higher engine compression ratio - more expansion means more extractable power
- Turbocharger - If done right, can significantly improve fuel economy, since it acts as an on-demand extension of the engine's compression ratio
- Drive like a Boomer - be "that guy" that does only 65 when everyone else is doing 80+
- Hybrid conversion kit - very expensive but effective
Guys running boost know this all too well.
What are your IATs (above ambient) when fully warmed up and in the throttle for minutes at a time?
I just installed a ceramic coated JBA exhaust manifold last week. It was great to no longer have exhaust leaks. I’ll get my gauges back soon to find out if my IATs change at all.
I use Redline 5w-30, which has an even better pour point…but I’m running it mainly for the turbo.
I can’t go for this one. My new Gates water pump (didn’t want to pay for Mopar this time) has a cast impeller and my system runs cooler now than it ever has. The thermostat opening and shutting allowing for a properly working radiator to cool the fluid seems to matter the most. I don’t have much fluctuation at all.
My MPGs are unchanged running a 180* thermostat since it warms the coolant up to that point just as fast as a 195* does…so I see little benefit with going higher than a 195*. Too much risk for little, if any, practical gain.
It also acts as an oil heater much like your oil cooler does when the oil is cold. Boost definitely helps efficiency in stop and go traffic.
I’d rather enjoy my life or do that in a Prius and to get 50 mph lol.
Didn’t know this existed…where would the hardware go? At a certain point, things simply get too impractical.
Please do this lol.
- -
Good man on not sharing FRP’s tune. I don’t use it anymore, but it would be unethical to share it.
I really want to try out your 2018 Camaro SPAL fan experiment. I think you found an amazing option there.
Ok so it's a 5v signal. That's nice in that it won't add to the total once the gm sensor starts to do it's thing.
My transit 250 van kicks the fan into high gear as soon as I turn on the ac. This seems like an elegant solution for us without needing to remember to turn switches on or off.
Lexus has a transmission with all that stuff integrated. I would probably start with one of those and 10 years of computer classes to make it work
This is good news as far as I'm concerned. This is not a small list of quick hits. The gap between your mpg and mine makes much more sense. I appreciate the detailed list, this puts it in perspective and makes it tangible.
I'm going to apologize in advance for what will be a long post, but because I'm a nerd for this stuff, I put some pen to paper on how I would start with a program for my own fan controller (Arduino) and now I'm gonna nerd out about it on you guys.
It's no doubt more complex than most fan controls probably are, but since my concern is primarily noise, I want to run as little fan speed as possible at all times, and I do far more complex things in modern HVACR controls as my day job so don't have even the slightest concern with executing it. In my day job it's usually about energy consumption which is a nice bonus here as well...and it works here since the the goals of energy and noise are both served by only spinning the fan as fast as you need to.
Proportional Control:
The Lingenfelter control is all proportional - the output is linearly proportional to the error between the control variable and the setpoint.
Inputs:
Radiator Outlet/Leaving Fluid Temp (LFT)
Lingenfelter: Fan Speed = (LFT - LFTlo) / (LFThi - LFTlo)
LFThi = 190 (temp for 100% speed)
LFTlo = 130 (temp for 5% speed)
I take the same proportional control but I introduce ambient into the mix, because if it's 50 degrees outside you don't need to run as much fan to achieve the same LFT and your output adjustments don't have to be as big to effect the same change in the control variable. Not only does this run the fan slower in colder ambients, being less aggressive will also make it less prone to hunting and oscillations. On the other hand, if it's above "base", it will be more aggressive. It'll still be limited at 100%, it'll just get there a few degrees earlier. All of the parameters would be stored as variables for easy adjustment, including the 90° baseline.
Added Inputs:
Ambient (Tamb)
Fan speed: (Tstat - base) / (Tstat - Tamb) * (LFT - LFTlo) / (LFThi - LFTlo)
Tstat = temperature where thermostat is fully open
base = baseline ambient temp - this is where my equation has a result identical to what the Lingenfelter controller would have, no ambient scaling.
LFThi = LFT that results in fan at 100% speed
LFTlo = LFT that results in fan at 0% speed.
For the table below I'm using 90 as a baseline ambient and the numbers @Steel City 06 recommendation as a starting point, in hopes that the ambient scaling would give me more flexibility to tighten up the band without causing instability. I would perform some tests to find their final values:
Test 1 - Idling with AC on and fan at 100%. Find base ambient that results in ECT = Tstat. This is minimum engine load and should therefore be minimum engine dT (ECT - radiator LFT), so the LFThi parameter should be set to the LFT found during this test.
Test 2 - Low load cruise at speed that provides same airflow as 100% fan speed (measure air velocity in front of condenser using anemometer at several points arrange in a grid and calculate mean). LFT during this test should determine LFTlo, so that the fan shuts off when you don't need it. I think this will come out higher than 130 but you can't set it that high without the ambient compensation or it'll be too touchy when it's cooler outside.
It assumes that airflow volume is linear to fan speed, which it is if the fan static was constant, but it's not, because the radiator and condenser will have a pressure drop curve which will result in 50% fan speed providing more than 50% airflow. I'm ok with erring on the side of a little excess airflow vs trying to determine the APD curve AND the fan curve and plotting them against one another.
Integral Control:
This is more of a "cool to have" but completely unnecessary feature for this application, but because I can, I think it would be fun to implement what basically would work like a fuel trim based on O2 sensor readings. Proportional control always results in what's called a "steady state error", which means that for midrange conditions, it will settle with the control variable and the output somewhere midrange, rather than controlling the variable to a specific value. With proportional only, you are pretty certain go end up somewhere between 130 and 190. With integral, you can pick a constant temperature you want and hold it in variable conditions - and in this case, I think the ideal temperature would be an ECT just high enough that the thermostat is fully open so no fan speed is wasted.
For this exercise I want to base it on ECT instead of radiator leaving temperature.
Inputs:
ECT
Tamb
Fan Speed = (ECT-Tamb)/(Tstat-Tamb)*(Previous Fan Speed)
So what this does is if the ECT isn't exactly 205 (or 210 or whatever I set Tstat to), it will trim up or down. The ambient temperature being included is for the same reasons as above - it's less aggressive with lower ambients to prevent overshooting.
To keep it from being fiddly and constantly moving around, I might implement a deadband, call it +/- 2 degrees, so it doesn't make any adjustment within 203-207.
For example I'm idling at a stoplight and it's 90 degrees out and my fan is at 80%, but ECT breaches 207, the integral term will
(207.1-90)/(205-90) * 80 = 81.5%. On the next calculation cycle, the fan speed will increase by 1.5%, and it will repeat that operation on a regular interval which can be adjusted to modulate the intensity...I'd probably start around 5 seconds and expect to do some tuning. Say my fan speed ends up creeping up to 86% before it pulls it back below 207 and then the light turns green. I take off and vehicle speed takes over so my ECT starts dropping and once it pulls down to 203 I start dropping the fan about 1.5% every 5 seconds until it gets back over 203. Or say the ECT really drops to the point that the thermostat completely closes, the fan speed will drop 10% (of the previous fan speed, not absolute fan speed) every 5 seconds until it shuts off.
There's some issues and common pitfalls that I know how to deal with but there's a couple that I haven't nailed down yet:
1. If it's hot and I'm on the highway and have plenty of airflow but the ECT just wants to be something above what I've calculated Tstat full open temp, the integral will speed up the fan and run it unnecessarily if I don't do something to prevent that. Setting Tstat somewhere up around 215 where it only hits in the most extreme conditions would prevent that but I don't want it running there all the time, either. I may do something like monitor the LFT and override the fan off if its low enough to indicate the vehicle is at speed.
2. If I screw up on the time interval and it can't react fast enough, the ECT could spike before the fan can catch up. I'd probably put a secondary safety in there that just puts the fan on full blast if an ECT threshold is exceeded, but still be impacted by point #1 so it doesn't turn the fan on at 75mph because the ect hit 220.
It's common to use both proportional and integral control together, but since I would want each of them pointed to different control variables they'd end up fighting each other so I need to pick one or the other.
The final piece is the making sure the AC condenser gets air when its in use before the engine is calling for cooling.
Input:
Air Conditioning Liquid Temperature (between the condenser and orifice)
I haven't quite worked out exactly what the threshold should be but I should be able to monitor that temperature to evaluate whether the AC is running or not, and if it is, I'll set a minimum fan speed like 20% to get air moving through the condenser. It'll run that way until the first time the coolant temp control kicks the fan on, which will "unlatch" it and let coolant temp control the fan for the rest of that drive cycle. I only need the AC control until the engine warms up enough to run the fan, and once that happens, it'll be harder to distinguish AC heat from engine heat anyway.
It's no doubt more complex than most fan controls probably are, but since my concern is primarily noise, I want to run as little fan speed as possible at all times, and I do far more complex things in modern HVACR controls as my day job so don't have even the slightest concern with executing it. In my day job it's usually about energy consumption which is a nice bonus here as well...and it works here since the the goals of energy and noise are both served by only spinning the fan as fast as you need to.
Proportional Control:
The Lingenfelter control is all proportional - the output is linearly proportional to the error between the control variable and the setpoint.
Inputs:
Radiator Outlet/Leaving Fluid Temp (LFT)
Lingenfelter: Fan Speed = (LFT - LFTlo) / (LFThi - LFTlo)
LFThi = 190 (temp for 100% speed)
LFTlo = 130 (temp for 5% speed)
I take the same proportional control but I introduce ambient into the mix, because if it's 50 degrees outside you don't need to run as much fan to achieve the same LFT and your output adjustments don't have to be as big to effect the same change in the control variable. Not only does this run the fan slower in colder ambients, being less aggressive will also make it less prone to hunting and oscillations. On the other hand, if it's above "base", it will be more aggressive. It'll still be limited at 100%, it'll just get there a few degrees earlier. All of the parameters would be stored as variables for easy adjustment, including the 90° baseline.
Added Inputs:
Ambient (Tamb)
Fan speed: (Tstat - base) / (Tstat - Tamb) * (LFT - LFTlo) / (LFThi - LFTlo)
Tstat = temperature where thermostat is fully open
base = baseline ambient temp - this is where my equation has a result identical to what the Lingenfelter controller would have, no ambient scaling.
LFThi = LFT that results in fan at 100% speed
LFTlo = LFT that results in fan at 0% speed.
For the table below I'm using 90 as a baseline ambient and the numbers @Steel City 06 recommendation as a starting point, in hopes that the ambient scaling would give me more flexibility to tighten up the band without causing instability. I would perform some tests to find their final values:
Test 1 - Idling with AC on and fan at 100%. Find base ambient that results in ECT = Tstat. This is minimum engine load and should therefore be minimum engine dT (ECT - radiator LFT), so the LFThi parameter should be set to the LFT found during this test.
Test 2 - Low load cruise at speed that provides same airflow as 100% fan speed (measure air velocity in front of condenser using anemometer at several points arrange in a grid and calculate mean). LFT during this test should determine LFTlo, so that the fan shuts off when you don't need it. I think this will come out higher than 130 but you can't set it that high without the ambient compensation or it'll be too touchy when it's cooler outside.
It assumes that airflow volume is linear to fan speed, which it is if the fan static was constant, but it's not, because the radiator and condenser will have a pressure drop curve which will result in 50% fan speed providing more than 50% airflow. I'm ok with erring on the side of a little excess airflow vs trying to determine the APD curve AND the fan curve and plotting them against one another.
Integral Control:
This is more of a "cool to have" but completely unnecessary feature for this application, but because I can, I think it would be fun to implement what basically would work like a fuel trim based on O2 sensor readings. Proportional control always results in what's called a "steady state error", which means that for midrange conditions, it will settle with the control variable and the output somewhere midrange, rather than controlling the variable to a specific value. With proportional only, you are pretty certain go end up somewhere between 130 and 190. With integral, you can pick a constant temperature you want and hold it in variable conditions - and in this case, I think the ideal temperature would be an ECT just high enough that the thermostat is fully open so no fan speed is wasted.
For this exercise I want to base it on ECT instead of radiator leaving temperature.
Inputs:
ECT
Tamb
Fan Speed = (ECT-Tamb)/(Tstat-Tamb)*(Previous Fan Speed)
So what this does is if the ECT isn't exactly 205 (or 210 or whatever I set Tstat to), it will trim up or down. The ambient temperature being included is for the same reasons as above - it's less aggressive with lower ambients to prevent overshooting.
To keep it from being fiddly and constantly moving around, I might implement a deadband, call it +/- 2 degrees, so it doesn't make any adjustment within 203-207.
For example I'm idling at a stoplight and it's 90 degrees out and my fan is at 80%, but ECT breaches 207, the integral term will
(207.1-90)/(205-90) * 80 = 81.5%. On the next calculation cycle, the fan speed will increase by 1.5%, and it will repeat that operation on a regular interval which can be adjusted to modulate the intensity...I'd probably start around 5 seconds and expect to do some tuning. Say my fan speed ends up creeping up to 86% before it pulls it back below 207 and then the light turns green. I take off and vehicle speed takes over so my ECT starts dropping and once it pulls down to 203 I start dropping the fan about 1.5% every 5 seconds until it gets back over 203. Or say the ECT really drops to the point that the thermostat completely closes, the fan speed will drop 10% (of the previous fan speed, not absolute fan speed) every 5 seconds until it shuts off.
There's some issues and common pitfalls that I know how to deal with but there's a couple that I haven't nailed down yet:
1. If it's hot and I'm on the highway and have plenty of airflow but the ECT just wants to be something above what I've calculated Tstat full open temp, the integral will speed up the fan and run it unnecessarily if I don't do something to prevent that. Setting Tstat somewhere up around 215 where it only hits in the most extreme conditions would prevent that but I don't want it running there all the time, either. I may do something like monitor the LFT and override the fan off if its low enough to indicate the vehicle is at speed.
2. If I screw up on the time interval and it can't react fast enough, the ECT could spike before the fan can catch up. I'd probably put a secondary safety in there that just puts the fan on full blast if an ECT threshold is exceeded, but still be impacted by point #1 so it doesn't turn the fan on at 75mph because the ect hit 220.
It's common to use both proportional and integral control together, but since I would want each of them pointed to different control variables they'd end up fighting each other so I need to pick one or the other.
The final piece is the making sure the AC condenser gets air when its in use before the engine is calling for cooling.
Input:
Air Conditioning Liquid Temperature (between the condenser and orifice)
I haven't quite worked out exactly what the threshold should be but I should be able to monitor that temperature to evaluate whether the AC is running or not, and if it is, I'll set a minimum fan speed like 20% to get air moving through the condenser. It'll run that way until the first time the coolant temp control kicks the fan on, which will "unlatch" it and let coolant temp control the fan for the rest of that drive cycle. I only need the AC control until the engine warms up enough to run the fan, and once that happens, it'll be harder to distinguish AC heat from engine heat anyway.
Cool stuff. So long as it could grab 5v signals from the pcm for this,what about using tps,rpm,map and mph to modify the ramp curve? Predicting load would be useful
Last edited:
This is very interesting and I would definitely be interested to follow this and even try it if you can come up with an even better solution.
I do have a Rugged Circuits Arduino Mega (basically a US-made Arduino designed for extreme temperatures) on hand that I originally planned to drive the fan with. But my C++ programming skills are just barely enough to be dangerous with a 3D printer...
There are few ways to get an Arduino to interface with the PCM so you could poll PIDs. My simplest idea was to just enable the PWM fan option in the tuner/PCM, and then use the Arduino to just poll that and run the fan at that speed. The program in the PCM is fairly sophisticated, from what I've learned; it's biggest flaw is that the 4.0 doesn't have the A/C pressure transducer like the 2.4s and other electric fan models of it's era. Alternatively, poll various PIDs so you can monitor things like A/C status, engine RPM, road speed, ambient air temp, coolant temp sensor readings, etc. to allow the Arduino to calculate a basic PWM value.
There are definitely Arduino programs out there that do full Proportional-Integral-Derivative (PID) control of temperatures that would probably be easy to adapt to a fan as well. The derivative part of the control might also be very useful in addition to proportional and integral, given that it would allow the fan to react aggressively to fast changes in temperature.
While I couldn't even start to build one from scratch, I know the Arduino powering my 3D printer has several PID control functions, notably for the extruder heater cartridge and the heated bed. The source code for this is free and open-source (for example, Prusa MK3S+), so someone with more programming experience could certainly figure out exactly what to rip. (Temperature.cpp or temperature.h I think are the sub-files that have the PID control, but it's been a while since I've messed with it.)
Edit: temperature.cpp in the Marlin/MK3S firmware has the PID program.
Last edited:
one of the things I enjoyed about swapping to electric was being quieter…so I was sucked in for the read. However by the end of the read, I believe i would be driving around radio off, just to listen for a fan responding to all that mental effort put into practice.
Similar threads
