This is apparently another method to lower only medium’s speed, discovered by error. I had thought these two fans were identical, but the speed difference was found to be due to a speed switch I had replaced in that particular fan several years ago which routed power through the fan’s capacitors slightly differently, and which I had forgotten about having installed.

The replaced switch fan, when the speed switch was set on medium, routed power through only two capacitors, Med Δ = f(C1,C2), instead of the factory’s design of all three, Med Δ = f(C1,C2,C3), and my spreadsheet formulas hadn’t been intended for that logistic. I recall that after the switch’s installation some years ago, the wiring going to it had to be reordered so high was on chain pull one, medium on pull two, and low on pull three. I also recall simply being happy enough that its prior broken switch was replaced and that the fan seemed to work again with three reasonably acceptable speeds, even though I didn’t understand why medium speed had slowed down.

Here’s how the original and replacement switches work, presented in semi-schematic form that is accurate for the two fan switches we have according to continuity testing. Note the differences, especially with regard to the column under “Medium”. The innermost circle and four dots represent potential connection points, lines between those dots indicate connections or continuity. Moving outward, the next circle with numbers indicates printing on the case of the switch housing near each wire insertion point, and the area just outside any circle may indicate the color of the factory wire connected to it.

If the fan has different speed switch-to-capacitors-to-motor coils’ logistics, then the coil difference formulas won’t necessarily work without some rather minor modifications, nor would these switches necessarily work.

Ultimately, this project has also taught me that it doesn’t matter a great deal what kind of switch the fan has provided it’s all wired logically. I’ve even conceived of how to adapt our fans’ inner wiring to the two-capacitor, three-speed ceiling fan wiring scheme, but I don’t have one of those kinds of switches to play with, so I think I’ll let that conception pass, for now, but trying and testing it could shed further understanding upon the spreadsheet issue explained below. Why buy a switch if one isn’t needed? But it’s nice to know that if it was the only kind of switch available, with sufficient forethought and rewiring time, perhaps additional capacitor microfarad changes, it could probably be made to work pretty much the same.

Discussion
I used the spreadsheet by first inputting the values of the factory capacitor (which should be printed on each capacitor case), noting the calculated answers, and then comparing those answers to other possible capacitor combinations’ answers. I then used simple if-then formulas, such as testing for medium values being lower than low values, then sorting the answers and making appropriate row deletions, as well as coloring and cut & paste features quite a bit to pare down the vast number of potential choices. However, there is an even simpler way to make these decisions, but I didn’t know what it was until after writing out, then revising, what I had learned, an iterative process which seemed to provide further insights.

One issue to be aware of is the need to get each capacitor’s respective value into the appropriate C1, C2, or C3 spreadsheet cell: wire colors, where those wires go, and schematic diagrams are all useful for this, but the single best way to locate and identify them seems to be following the wires to and from the chain-pull speed-switch wire-insertion number (switch terminal).

The spreadsheet numbers claim I have reduced High’s Δ by 25%, medium’s Δ by 35%, and low’s Δ by 18%: I don’t know how well Coils’ Δ correlates to actual fan speeds, but it seems to roughly agree judging by visually observed speeds. Since high was reduced more than low, I’ve narrowed the high-to-low speed range; also, since I’ve lowered medium the most, I’ve brought it closer to low relative to high. This is about what I wanted to achieve.

Skip all the rigamarole and jump to the next table entry.

One identified spreadsheet issue is seen in the light green column, which represents high. This column’s answers cannot be directly compared to the answers in the light blue columns: a lower value in High’s Δ may result in a faster fan speed than a higher value for Med’s Δ, or the next one which represents Low’s Δ. For example, using the data for the factory module, the “4” is faster in the high-speed circuit than “6.23” is in the medium-speed circuit. However, it seems the two light blue columns can be compared to each other, these represent “medium” and “low”, respectively. It is not clear whether medium and low values associated with one high Δ (in the same row) can be compared to other medium and low values with a numerically different high Δ existing in their respective rows, though that is a presumption I made both at first and in the table above.

C1 entered values are included in the function or calculation of medium and low Δs of the same row, as well as simply echoed for the high circuit. Since a shorted or solid wire apparently approaches infinite capacitancesee prior post for reference link, it’s possible it’s in error, then if one coil is supplied with infinite capacitance, and the other with C1’s value, then high Δ is the difference between these two coils. I’m not sure how to mathematically deal with the infinity concept correctly, there are several possibilities that I can conceive, two of which are ∞-C1 or C1-∞, but how does one convert that into a practical number unless infinity and zero can be substituted for each other on the number line?

One potential rationale for the error is that C1 feeds one coil in the high circuit, in the other two circuits, it appears to feed the other coil, judging from the schematic. Perhaps these two coils have different angles with respect to each other and or the motor’s magnets. Anyway . . . if high’s value could be compared to the medium and low circuits, then the high to low range could easily be calculated, and a spreadsheet formula such as (med-low)/(high-low) could give an accurate abstraction of the relationship of medium to high and low, and which could also help to filter the many possible capacitor combinations down to a smaller subset of preferred values.

It seems the best way to use the spreadsheet to select capacitor values, for the particular wiring schematic under discussion and using the factory speed switch, is to choose the high value that results in an acceptable high speed either by noting the spreadsheet value’s percentage change, installing that value capacitor and testing the high speed, or a combination of both. Then, keep only those rows that include that particular high column’s capacitor value, deleting all the other rows that have different values in that same column; then select low speed values only from those remaining rows, and thirdly or lastly, select medium speed capacitor values. My reasoning for this is due to the fact that the high speed circuit uses only this one single capacitor (C1): the other two slower circuits also use this same capacitor in addition to others, therefore, all circuits are dependent upon C1 to one degree or another. Following this reasoning, the slow circuit uses two capacitors, C1 and C3, so it seems selection of C3 is the next logical choice. Finally, medium speed uses all three capacitors.

Another way of stating this is: if medium speed is unsatisfactory, changing capacitors C2, C3, and/or C1 will change medium’s speed, but changing any capacitor other than C2 will also change other speed settings; if low speed is unsatisfactory, changing C3 and/or C1 will change low’s speed, but changing C3 will also affect medium speed while changing C1 will affect all three speeds; and finally, if high speed is unsatisfactory, changing C1 will change high’s speed, but changing it also affects all other speeds. Thinking of it this way is somewhat more complicated than as-simple-as-it-can-be construed.

This is also, apparently, determined from the formula functions:

High Δ = f(C1)
Low Δ = f(C1,C3)
Med Δ = f(C1,C3,C2)
In the search for the perfect combination of fan speeds, it may be necessary to compromise to some degree, due to space limitations and capacitor values that are available, though if the capacitors are small enough and there is enough space for more than three, then by using multiple capacitors in series and or parallel, and substituted for each single capacitor in the basic schematic, speed choices would seem quite numerous and adaptable.
With one set of medium and low Δs, I found that multiplying them by “4” resulted in a close prediction of the fan’s respective speeds over a 10-second time period, the speeds that in some cases I could count visually by timing with a stopwatch. However, that relationship did not hold with some other capacitor values I checked, so I concluded I either made an observation error or the relationship of coils’ Δ to fan’s RPM, if one exists (it certainly seems to), is not linear.

Installation

A removed and fully-functional module is seen next to the new and, except for the 4uF in the front, lower-value single capacitors on the right (I bought these capacitors from one of the suppliers listed in my previous post which was titled: Ceiling Fan Capacitor Woes). The reason the module is both still functional and removed is the fans I wanted to slow were not the fans with blown capacitors. It was my intent to move these good modules to the broken fans, to minimize the number of capacitors purchased. However, there was another problem, the factory module had five wires, these replacement capacitors had six wires.

When I wired the capacitors into the fans, I used male and female insulated crimp connectors. I thought this would make the process of switching their ordering, or future replacements, that much easier. The harness simulates the schematic printed on the factory’s modules. There may have been a cleaner or simpler way to do this, but this was what I happened to think of at the time.

With respect to the mini-harness, I was worried about the wire’s insulation, so I additionally wrapped it with electrical tape.

This is a photo of the factory switch fan’s control housing or case with the new capacitors installed. Some of the model numbers on the speed switch can be seen. The reverse switch is also partly visible.

Final Summary

Pulling it all together, and in spite of some initial confusion, spreadsheet issues, seemingly needless complications, and errors, this is the final installation outcome for the fan with the factory switch. Does it look familiar?

To summarize the ordered and incremental steps to alter the fan’s speeds, with respect only to the simplified three-capacitor ceiling fan schematic, using the factory switch to identify each particular capacitor (C2 and C3, C1 is the sole remainder), then:

Set high speed first by altering C1’s value if high speed is unsatisfactory. Changing C1’s value changes all speeds.
Set low speed second by altering C3’s value if low speed is unsatisfactory. Changing C3’s value also changes medium’s speed.
Set medium speed last by altering C2’s value if medium speed is unsatisfactory. Changing C2’s value only affects medium speed.
How much does changing the microfarad values affect other speeds?

Writing this post out sure helped me to understand how the capacitors affect different speeds. Remember, I do not advise you to repair or alter your own fan!