| Author | Message | ||
     Hildstrom |
Here is a link to a page detailing my diy stator rewind and rotor modification. I'll post a few photos here, but much more detail is on my page. I rewound with 16 AWG polyimide heavy build magnet wire, which should be good to 240C. I protected the windings from vibration with J-B Weld epoxy, thinned with heat, which should be good to 260C. I drilled six 3/16" holes in the rotor to force some air past the stator to help keep it cool. I upgraded my voltage regulator to the CE-605 SB series style regulator to reduce stator current. This was my first time trying to do something like this, so I was very happy when it worked. I got 21V AC at idle and 40V AC at 3000 rpm before hooking it up to the new regulator. I performed an hour-long hot-idle test yesterday and everything seems to be working well. I'll post updates to my page and this thread as I rack up some miles. Hopefully it will stand the test of time.         | ||
| Baf |
Sweet. How much did the wire end up costing you? I took a look at the supplier linked in the article, and I see wire in half gauge increments, and several different insulation sizes. Did you investigate bumping up to 15 or 15.5 gauge wire, or getting wire with thicker insulation at all? Also, is the JB Weld going to pose a problem (when removing) if you end up needing to do another re-wind? | ||
| Hildstrom |
Baf, The quotes were $0.84/ft for 150ft or $0.163/ft for 1000ft. I opted to get 1000 feet because I have some other magnet wire projects coming up and the cost difference was not too much. Shipping on the 1000ft was $14, so $0.84*150+14 = $140 for enough wire to do the stator. I paid $166 for the CE-605 SB series regulator kit. So, you could do the same for about $306 shipped. I considered using larger wire, but I did not want to sacrifice turns or idle voltage. I don't want to have to rev the bike up in a post-traffic parking lot just to make sure it will start when I return. Thicker insulation would only increase the heat problem, decrease the wire size, or decrease number of turns. The heavy build is more than tough enough and it can insulate much more voltage than we need in this application. I chose heavy build, which is thicker than single, but not as thick as triple or quad. Also, thicker insulation is not much use if its useful temperature is exceeded. The series regulator accomplishes the same thing as the "harness fix", but it is much better in just about every way. It reduces current in the entire stator, not just 1/3 of it. Also, the rotor modification should draw more air through the stator and help cool it. After the series regulator, the rotor modification, and the high-temperature polyimide insulation, I figured I would be alright using 16 AWG and the original number of turns. Time will tell. The J-B Weld is definitely tough stuff. However, if it doesn't come apart like the original epoxy, it's nothing that a propane torch and determination can't take care of. Hopefully I will not need to find out. | ||
| Reepicheep |
Great write up and thanks for taking the time to do it! Very cool project. The only thing I might disagree with so far is saying the series regulator is better than the harness fix in every regard. I'd be OK with "almost every regard" though.  The one way it is better is that it opens up an entire stator leg during a specific condition that is known to be problematic, even if its starving the bike to do so. The series regulator isn't that bright (though it could be), and doesn't target that specific condition as well, at least not deliberately. Series is better than shunt for sure, but the harness makes one leg even aggressively "series" as well. That being said, if the harness change worked, we wouldn't have all these smoking stators on the other ends of those harnesses... so I would argue your solution will be at least as good, and likely much better with some empirical information gathering over time. In a nutshell, "Bravo" for moving this real problem such a big step forward! | ||
| Baf |
With larger wire, I wasn't sure if there was some extra room in there that would allow for a slight increase in wire size. And as far as the thicker insulation, I thought it was insulation failure that was causing the stator issues (though it does make sense that it would trap more heat in as well). The CE605 was actually on my list of things to do, along with a drilled rotor from EBR, once they become available. I am definitely interested in the DIY stator re-wind though, and I might grab a fried stator to rewind and keep as a spare, ready to go. Oh yeah, and thanks for the excellent write-up. I'll definitely make use of it if necessary. | ||
| Hildstrom |
I forgot a few other things in the cost estimate that you may need: 150' 16 HML: $140 shipped CE-605 SB: $166 shipped 1"x36' F4 tape: $14 shipped ignition cover gasket: $17 w/ tax rotor nut: $33 w/ tax high-temperature threadlocker: $10 w/tax 2oz J-B Weld: $7 w/ tax total: $387 | ||
| Timebandit |
Rolling your own! What an interesting write-up! Thanks so much for posting. Too bad you didn't mention that you were going to do this before you actually did it -- I might have been able to hook you up with some embedded temperature sensors that you could have placed inside of the windings on a couple of key poles, which would have allowed you to take direct measurements of stator temperatures at a couple of key poles, like 1 (at 6:00) and 3e (2:00). That would have been a great opportunity to get direct stator core temps. One thing that's important to take home from all of this is that even if you DIY the stator rewind, it's going to cost you about $300. Given the time and effort, that makes a $600 over the counter price for the stator seem a bit less unreasonable. | ||
| Hildstrom |
Timebandit: Keep in mind that the $300 included a series regulator, so I still think $600 for just a new stator is unreasonable. If that price also included the rotor or regulator, maybe it would be more reasonable. Yep, some temperature sensors would have been a great idea, but only if they could deal with the rapidly changing magnetic fields. There is not a whole lot of extra room in there, so it would have to be pretty small like a thermocouple with the wires twisted together. I wonder if I can estimate the temperature by measuring before & after resistances of two stator wires. I realize this method will give average wire temperature and not detect hot spots, but it's better than nothing. The equation can be rearranged to solve for Thot. http://www.allaboutcircuits.com/vol_1/chpt_12/6.ht ml Rhot = Rcold[1+a(Thot-Tcold)] Thot = [Rhot/Rcold-1]/a+Tcold If I let the bike sit in the garage for a day or so, Tcold will be very close to the ignition case temperature. The trick is getting accurate small resistance measurements, which are probably too low for my cheap multimeters. The delta winding resistance across any two output wires is the resistance of one phase in parallel with the resistance of the other two phases. With around 4 milliohms/ft for 16 AWG copper wire, each phase is about 4mR/ft*11ft/pole*4poles/phase = 176 milliohms. The measured resistance across two delta wires is then about 1/Rd = 1/Rp + 1/(Rp*2) = 1/176 + 1/352 = 0.0085227 and Rd = 1/0.0085227 = 117 milliohms at room temperature. http://www.allaboutcircuits.com/vol_2/chpt_10/5.ht ml http://en.wikipedia.org/wiki/American_wire_gauge If I use my current-limited bench-top power supply, I can accurately measure voltage and current to derive resistance. I know my multimeters can accurately measure millivolts and milliamps so I should be able to accurately calculate resistance. A 5A test current would yield V = IR = 5*0.117 = 0.585. An increase of 1 milliohm would produce V = 5*0.118 = 0.590, which is a 5 millivolt increase and easily measurable with most digital multimeters. The one-phase 20C resistance of about 176 would be more at the engine's coolant temperature of around 100C and the stator will likely be even hotter. The above equation lets us estimate the resistance with a=0.004041 for copper. Rhot = 176[1+0.004041(100-20) = 233 milliohms. The measured delta resistance is then 1/Rd = 1/233 + 1/466 = 0.006437 and Rd = 155 milliohms. The minimum change in resistance is then 155-117 = 38 milliohms. With a 5A test current, that would cause at least a 190 millivolt change. I might give this a shot if I find some free time this week. | ||
| Parrick |
Very nice. I was considering the same project but the hold up for me was the coating. A local electric motor repair shop said they could resin-impregnate it for me for reasonable $$. I'm interested to see how the JB Weld holds up. QUESTION: During the de-wind did you find the failure point in the original windings? How did it look? | ||
| Nuts4mc |
1)what's the di-electric strength of the JB weld? does it conduct electricity? 2) back in the day I used to work at at transformer/auto former/hi tech transistor company (PTI Controls...we built our own IGBTs 400V @ 100A- very big IGBTs)- we use to mix our epoxy and then subject it to a vacuum to remove the air 3) all of our transformers were baked after coating to help take out the moisture and "set up" the epoxy 4) may want to look here for the "professional" grade stuff (Emerson & Cummings is a name I remember - but it is very $$$$s) http://www.ellsworth.com/ 5) possibly the local rewinder may help by doing the epoxy and bake good luck | ||
| Oldog |
There are several types of JB weld some are metal filled I hope that was not used... | ||
| Hildstrom |
Parrick, I did not look for the short. There are more disassembly photos on my page. Nuts4mc, The dielectric strength of JB Weld is not critical because the polyimide dielectric strength alone is more than sufficient for this application. JB Weld does not conduct electricity. Oldog, My page details what was used and why. I used normal JB Weld, which contains a small amount of very fine steel particles. It does not conduct electricity and the metal content is not high enough to significantly affect the magnetic fields; the proof is in the voltage numbers. (Message edited by hildstrom on March 26, 2012) | ||
| Nuts4mc |
not trying to attack your effort (I applaud it!....too bad the "factory"(HD) doesn't have the same sense of urgency when solving this problem)...just trying to suggest a possible better solution to JB-weld- there are many choices in the Mcmaster-Carr website that may help you (and others) to become even more successful with this re-wind...best of luck especially with the new riding season upon us!...nuts | ||
| Hildstrom |
Nuts4mc, "not trying to attack your effort". No doubt . I was just answering your two questions. My detailed page mentions Cotronics Duralco 4461 as one of my spare-no-expense choices for coating the windings. | ||
| Oldog |
Thanks for putting that link up That needs to be in the KV! Your post here did not indicate that the JB weld used was non conductive, I have machined jb weld before and knew it to be metal filled, Sorry if you took that as critical I hope that the job holds up well... the high grade magnet wire and smoothing the poll corners should help.. | ||
| Reepicheep |
Now y'all got me trying to think up clever hacks for pulling a vacuum using stuff I have in the garage. I wonder how much vacuum you need? And how long my shop vac could go duck taped to a box in a trash bag before the motor over heated? | ||
| Dannybuell |
LOL! | ||
| Kenm123t |
rent a refrigeration vacuum pump build a vacuum chamber so you dont get goo in the pump Think like a mightyvac brake bleeder . I have even used the intake side of a oiless type HDepot air compressor to a quick high volume but not a deep vacuum. A stroll through the Grainger catalog will have you set up with something. | ||
| Oldog |
Reep, JC whitney sells an air powered venturi probably pull enough vacuum. save your shop vac burn up the compressor I have a hand held plastic pump that will pull 26"Hg the chamber could be made from plastic pipe ( not cheap but doable )} | ||
| Nuts4mc |
an old pressure cooker you might be able to find one at the Salvation army store = vacuum chamber... refrigeration supply store = "yellow jacket lines" and some fittings like 123KenM sez - RENT a refer vacuum pump - pull 1 atm ( - 14.7 PSI) the pressure cooker should be able to hold that for a couple of hours, but you need to use sloooow epoxy - or you'll be building expensive (and hard) hockey pucks | ||
| Timebandit |
Greg, using the stator as a thermocouple is a great idea. Indeed, it would indeed limit you to average temperature readings, but accurate average temperature readings would in and of themselves be quite helpful. The problem is that accurate numbers aren't so easy to get using the theory you've cited. The theory is straightforward, but as you mentioned, the execution leaves a bit to be desired. Nightsky and I discussed doing this back in December or January, and he's already done the experiment you've proposed. (He lives in a place where you can ride year round, and my bikes were put up for the winter.) We have the precision meters and tried doing the experiments that you referenced on an 2009 bike that he bought just for a testbed. A bench meter didn't have sufficient resolution, so he tried using precision 4-wire meters. Unfortunately, the calculated results were so far off as to be unrealistic/unusable. We couldn't get the system to work without having an embedded RTD to provide a calibration standard, or a hole drilled into the case for an IR temp sensor; of course, needing to use these rendered the experiment meaningless. Just in case you weren't aware, there are precision thermocouples and RTDs that are designed specifically for embedded stator applications. I had been interested in trying to attach them to the outside of the coils on an existing stator; this is exactly the plan that a few of the forum members have been working on together. It would have been 100x more cool to embed them in a custom wound stator. Minco has a nice catalog if you're interested. If you proceed with the experiment, I'd be interested in hearing your results. | ||
| Timebandit |
Harbor freight sells vacuum pumps -- presumably aimed at those people who don't care that the EPA gets pissed when people self-service their car's AC. | ||
| Reepicheep |
What temps will thermistors survive up to? The one I am using for my heated grip controller reads very accurately and easily with a 10 bit A2D. Probably easier to figure out how to bury that in the stator epoxy and get the leads out. Thermocouples should be a joint between dissimilar metals, and it's a tiny voltage you have to pick from the weeds. Thermistors are a much easier analog problem to solve. Measuring the resistance change in the copper versus temperature could work, but that's going to be a tiny change. You would have to measure both voltage and current really accurately. And you would have to factor out connector drops. | ||
| Oldog |
As TB mentioned some where above, about using an RTD To measure temp. I use them in industrial applications, the temp / resistance curve is non linear... but they work well | ||
| Timebandit |
"You would have to measure both voltage and current really accurately." that's exactly the problem. your garden variety benchtop meter won't cut it. you need something with 6.5 digit precision that will resolve 100 uOhms or 100 uA, like the Fluke 8845A or 8846A. These are 4-wire "kelvin" meters. http://www.allaboutcircuits.com/vol_1/chpt_8/9.htm l | ||
| Hildstrom |
Well, I verified that my meters are not nearly accurate enough. I'm probably ok with 1mV accuracy, but at 5A, my best meter only goes down to 10mA. It would be better to have 0.001 or better accuracy for both measurements to accurately calculate resistance in milliohms. If I have any kind of problem and I need to rewind again, I'll embed a few sensors. For now, I just hope the series regulator and drilled rotor can keep it cool enough. | ||
| Timebandit |
props to you for coming up with such a good idea. it's good that there are lots of smart people working on these problems, and thinking of novel approaches that are based on good science. it's no surprise that more than one person thought of the same approach -- that confirms that the approach is theoretically valid. it turns out that from a practical standpoint, this is a complex problem that requires expensive gear to study. even if you have the right gear, the practical side of the project turns out to be more difficult than you'd expect based on the theory. it's easy enough to get accurate measurements with the right gear, but the problem that we ran into when we did the experiment was that the calculated temperature was far too low to be believable. the only recourse in that situation is to have a reference temperature measuring device to calibrate your stator resistance measurement; but once you've got that, you don't need to measure stator resistance any more.  it's really a great idea from the theoretical standpoint, but the practical application is tough. it seems like the best thing to do is to embed a probe when you do a rewind, and take measurements from that. | ||
| Timebandit |
"What temps will thermistors survive up to?" Hotter than our bikes will ever get -- I've seen thermistors rated up to 550*C and RTD rated up to 600*C. | ||
| Reepicheep |
Cool. I have a birthday coming up and was asked if there is anything I want... I asked for one of these: http://www.sparkfun.com/products/10244 I suspect it will either do logging out of the box, or be trivial to hack to do it and record to an SD card. One of these, and a simple regulated voltage divider going to a embedded (or just glued on) thermistor could give lots of really interesting data. If I am ever in the Uly primary, I'll put one on my stator. | ||
| Mako |
great work guys. I was going to try this route if my 3rd stator burns up. This really helps to eliminate lots of research and trial and error. It's great to have professionals doing this instead of someone like me who would just try to copy a tried and true system. Thanks again for all of the time and effort. |
