Danner Mag 3 disassembly & sealing info and impeller/magnet direct coupling attempt

[krazy]

I ordered a Danner Mag 3 from www.reefgeek.com (the lowest price I could find considering shipping) a few weeks ago and have been playing with it.

From my research before I bought it (info from here and elsewhere on the web), the only main problem I heard about was the O-ring around the impeller housing leaking intermittently. Some folks have complained about clattering noise from the shaft, but my attempted solution for this caused the pump to be too difficult to use properly, although it did work.

Basic Disassembly


The basic disassembly of the danner is to simply remove the four stainless steel screws that hold the impeller housing onto the main chassis. Wiggle and rock the impeller housing gently and carefully pull upwards to get it off without damaging anything. The O ring is stretched slightly to fit over the round protrusion on the main motor chassis. be careful not to make any scratches or dings in either mating surface for this O ring to minimize the chances for trouble later.

The impeller and magnet ride on a white ceramic shaft to prevent potential corrosion and alleviate the need for oil or grease in the bushing. Slide this whole assembly out all at once by pulling on the impeller.



Notice how the impeller and shaft are wiggly when the assembly is in place with the impeller housing removed: The support for the impeller-end of the shaft, resembling a thinble with three supporting spokes, is built into the intake hole of the impeller housing. Someone asked what would happen if they cut this support off, hoping to improve flow or something. It would be a bad idea, as your shaft would then have nothing to ride on at one end and the pump would not work for beans.





Properly sealing the impeller housing O-ring:
People have complained that the O-ring around the impeller housing sometimes leaks. My pump did not leak when I first had it out of the box, but I didn't test run it long enough to be absolutely sure. I disassembled the impeller housing and decided that the way the O-ring seals is somewhat spooky. The two parts do not solidly clamp down vertically on the O-ring, but rather make a sliding seal that does not seem to compress the O-ring very much. Slow leaks coming from this seal are probably a result of the O ring being twisted a bit or possibly small scratches or grooves in the mating surfaces.



The easy and effective fix for both of these potential issues is silicone lubricant. By simply coating the O-ring with a thin (but not patchy) layer of silicone lubricant, I was able to slide the O-ring down over the raised part of the pump chassis without it trying to twist at all. The lubricant also obviously fills any sort of gaps and hairline leak paths, taking care of common potential leaks.










Magnet/impeller lock attempt (and why it can't work)
The magnet and the impeller in these pumps are not solidly connected together. There is not possibility of slip, but the design incorporates about 3/4 turn of slop before the tabs on the impeller and magnet hit eachother. This makes some clattering/rattling noise when the pump is started or stopped, as well as when it is passing lots of air bubbles. I noticed a little bit of clicking and popping while the pump ran smoothly, and decided to try locking the magnet and impeller together like I had seen done successfully with similar pumps (the Hydor I believe?).

Common theory is that the slop joint is to allow pebbles or other solid things to pass through the pump without damaging it. I thought this seemed odd because the thimble-shaped shaft support inside the intake is too small to allow any potential jamming objects into the impeller housing in the first place, and the pump should be pulling up tightly to the tabs when running, making the only way to slip by overspeeding the impeller temporarily. This doesn't seem like it would help pass rocks at all.





[my camera's batteries died when I wanted to take a picture of the silicone smoothed out, so just imagine it.]



I disassembled the impeller/magnet/shaft assembly by twisting and wiggling one of the rubber caps off the end of the ceramic shaft. The impeller is coupled to the magnet by a single plastic tab, allowing the magnet and impeller to rotate separately for the better part of a complete rotation before catching. I put four small blobs of blue silicone sealant (much like caulk) on the magnet's coupler part. I used a razor blade like a mini putty knife to smear the blobs around into a more even surface that was slightly larger than the inside diameter of the mating hole on the impeller. I was very careful to make sure that the sealant couldn't ooze out and get on the shaft for obvious reasons. After leaving the part to cure overnight with the shaft in place to hold things true, I reassembled the pump and tried it out.

Initial results were frigtening: the pump didn't work anymore. It would make noise and seemed to be spinning, but it couldn't move water more than about three inches vertically. I could blow in the end of the output hose and push air right through the pump with ease. What surprised me was what happened when the water flowed back into the submerged housing: the pump began pumping. Once it had cleared all of the air, it was running like new, and maybe a little quieter because of the mod. I tried it many times, and concluded that I could only get the pump going by letting it spin up full of air and then suck in water.

Obviously, this is no good for a watercooling loop that will run full of water all the time. After a little research and consulting with a mechanical engineer (my old man ), we realized why the "improvement" caused the problems it did. The pump is magnetically coupled, but instead of having a spinning magnet both outside and inside the water seal, the only moving magnet is the inner one. The magnet in this pump is the stator of the motor (the motor is technically inside-out, but this is not an ordinary motor either). It is directly driven by two electromagnets on the outside of the watertight plastic housing. This makes it a two-pole motor that uses 60hz line , and it has to surge faster and slower as it goes past each pole to run well. The slop joint allows the motor to bounce the impeller back and forth a few times to get spinning and allow the water to come up to speed a little more gradually, even if it is in lurches. This is why the pump clicks a few times when you plug it in.

By locking the impeller onto the magnet, I had forced it to instantaneously come up to speed and spin a mouthful of water up while doing so. It simply could not get going, and the magnet/impeller jerked back and forth between two fields of the motor without getting enough momentum up to coast to the next field. The rapid back-and-forth motion of the impeller was enough to generate a few measley inches of head, but the mass of the water prevented the pump from starting properly. Once a fair amount of air was introduced, the pump could easily get up to speed, and then was able to take on water somewhat gradually by having to pass the air as the water flowed in.



I carefully separated the two parts and scraped all the silicon out of the mating areas and reassembled the pump. It worked like a charm. The impeller/magnet lock trick might work for some other pumps, but the Danner (at least the mag3) does not have a motor capable of starting under these conditions. I think I will try adding small silicon pads on either side of the magnet's drive tab to dampen the clicking noise. This should still allow the impeller to bounce around while the water gets up to speed.

[UberBlue]

Nice work!

It's always nice to see something like this accompanied with pics and good explanations for phenomenon.

Good thing you didn't use something more permanent to glue the pieces together.

[AntiBling]

you said..
Initial results were frigtening: the pump didn't work anymore. It would make noise and seemed to be spinning, but it couldn't move water more than about three inches vertically. I could blow in the end of the output hose and push air right through the pump with ease. What surprised me was what happened when the water flowed back into the submerged housing: the pump began pumping. Once it had cleared all of the air, it was running like new, and maybe a little quieter because of the mod. I tried it many times, and concluded that I could only get the pump going by letting it spin up full of air and then suck in water.



Thats EXACTLY what happened when I tried it with my mag3. It wouldnt spin up unless it was started without water inside it. I used JB Weld to lock up the impeller/magnet. I was fastidious in making sure I didnt glue them together off balance ( at least to a visual inspection anyway) I tried running it in a bucket with the front part off to see what was happening, and it would spin when it was plugged in but not very fast. I had to scrape all that crap off to separate the two pieces to get it to work again . Once I quit screwing with it and just plugged it in it worked fine. After my JB weld experience, strangely enough the minor clattering that I was attempting to cure went away. Could be they just need some time to wear in? BTW mine has never leaked from the front as others have reported. Thanks for the excellent documentation! I'm quite pleased with my Mag 3. I got mine from reefgeek as well, and also ordered two spare impellers. My experience with reekgeek was great, I would order from them again.

[Jabo]

Pristine work, m8!
YOu certainly know jow to make pictures dude (it's quite unusual to see so pro-quality pics on any PC-type forum these days).
It read very well, and generally made an impression of professionaly put article with lots of effort behind it. I'd suggest a sticky

Btw, there is a similiar tyoe of mod for eheims since they are almost of the same construction.

[lolito_fr]

A little more about loose impellers:

When I was testing a little Hydor, I found that the max head was nowhere near what it should be (1.35m). In fact, it was fluctuating wildly about the 1m mark, and the rattling seemed to start whenever the head would reach a certain value. So it seemed like the pump wanted to, but the rattling was somehow stopping it.

Anyway, I fixed the loose impeller with a tiny blob of Blu-tack, and miraculously the head then maxed out to 1.33-1.35m and the pump was completely rattle free. Luckily, it also seemed to make no difference to starting performance.

I like your explanation about play allowing the rotor to spin up before it has to do any work, I also wonder if maybe these aquarium pumps are intentionnally designed to rattle, to alert you to a blocked filter or something like that? (aquariums have filters, right?)

[krazy]

Originally Posted by Jabo

Pristine work, m8!
You certainly know how to make pictures dude (it's quite unusual to see so pro-quality pics on any PC-type forum these days).
It read very well, and generally made an impression of professionaly put article with lots of effort behind it. I'd suggest a sticky

Btw, there is a similiar tyoe of mod for eheims since they are almost of the same construction.

Thanks for the compliments. I decided that since I was going to go through the trouble of figuring all this info out for myself, I should at least share it with you guys.

I used a Canon A70 for all of those pictures. Some were taken with just the camera's own optics, and some were taken using a macro lens and polarizing filter for the uber-closeness and glare-free effect.

I personally was a little disappointed with the shadowing of some of the pictures taken on my bathroom counter. Next time I'm doing this kind of photography I will be using more light sources and get a little more anal about shadowing.

If anyone wants some good detailed pictures of anything else inside this pump or any of the other stuff I'm working with, I'd be happy to take some pictures for you. Just let me know what you have in mind.

-----------------------------------------------

I intended to have this information in the original post, but I forgot it. :P

Reliably sealing the hose barb fittings

The Mag3, like many similar pumps, has a plastic impeller housing. People have (very understandably) reported cracking the threaded pipe connections on the housing by overtightening fittings on them. I decided right away that I wanted to avoid cracking my brand new pump, but I also didn't want it to leak because the barbs were too loose. A little looking into how pipe threads are supposed to work should provide some valueable insight into how to get around this problem in the most professional (non-ghetto) way possible.

warning! The following paragrah contains very basic information. I'm not trying to insult anyone's intellegence, but rather step back and start at the beginning so people that don't know this kind of stuff can understand the way this stuff works better.

Pipe threads are different from the threads you'll see on bolts and bench vise mechanisms. Pipe threads have a taper to them; the far end has a smaller diameter than the end at the middle of the fitting. The mating part also has this same taper. When a male pipe thread is threaded into a femal pipe fitting, the fit is initally very loose. As it is tightened down, the fit eventually becomes tighter and tighter. This is caused by the threads' tapers biting into eachother, much like a wedge. Note than a normal bolt and nut do not behave in this manner when tightened.

Pipe threads like to leak. Common sense would suggest that tightening the threads down on eachother further would tighten the seal and stop the leak. This is not true. No matter how tight you make a pipe threaded part, it can never be expected to seal completely. This is due to the shape of the threads. They are not a perfect mesh together regardless of how tight they are threaded together. There will always be a very small helix-shaped leak path going out around the seam in the threads. Tightening the fitting down more just stretches the female fitting further apart, but can not eliminate this leak path. The only thing you can do by tighening the fitting more is crack the female part by applying too much force. Once you have a cracked impeller housing, you're going to have a real fun time trying to reseal it properly, so just avoid this step in the first place.

The proper way to seal this up is to put something soft inside the leak path. Old wooden ships with rows of planks were sealed by hammering tar-soaked rope into the gaps between the planks. Several products are made that will do this job, all with varying applications and weak points.

Teflon tape:
Thin, smooth, & stretchy plastic film based on some kind of teflon. Usually comes on a small spool. It is quick to install and less messy than a comparably paste. It's main flaw is that thin hair-like strands of teflon are extruded out of the joint as the fittings are tightened down. These little teflon fibers flow through whatever is supposed to be on the inside of the piping and can clog pumps and precision valves. This is most commonly seen when working with hydraulic systems, but it should be considered when working with watercooling equipment too. These strands could plug up a nozzle in blocks like the Cascade, and might cause trouble in some pumps. For these reasons, I recommend:

Teflon pipe dope/sealant stuff:
A thick, gooey, somewhat-chunky white paste that comes in a can with a brush lid (a lot like those PVC adhesives and rubber cement are sold in). It's a bit less convenient to work with because of the potential mess, but the finished product can be a lot better. Dope does the same job as teflon tape, and is made of the same stuff, but it goes on wet, which makes it easier to avoid the teflon fiber problems. Note that dope can cause a whole new set of problems. If you aren't careful about how sparingly you apply it or get it on the inside of the fittings, it will be washed into the flow and circulate with the water in the form of little blobs. Dope never really dries, but it does get sort of a surface skin to it after a while. I recommend doping fittings carfully, then letting them sit for a while before running water through the system to let the dope skin over a bit.

Here are the barbs on my pump, sealed with a thin (but even) layer of pipe dope and torqued with just my hand to avoid over-tightening and cracking the housing. The dope takes care of sealing up the leak paths in the threads, and these fittings do not leak in the slightest.

The first time I doped the fittings, I put too much on and it started coming through the water in little blobs. I disassembled everything, wiped up all the dope I could, and reapplied a much thinner layer, putting it only on the male threads and using a Q-tip to apply the dope.

[JamesAvery22]

I just got a mag3 in the mail today and IM quite disappointed with the vibrating and the noise =( I read about gluing the impellar but then I saw this thread and a bunch like it.

Is there anyone that has glued their impellar and not have a problem with the mag3? My mcp600 was much quieter than this. If there is nothing I can do to quiet it down Id rather just get another pump...

If anyone has some tips to quiet this thing down could you please post?

[JWFokker]

It will make a tiny bit of clicking, though it does get better with time, and as for the vibration, I just put a mousepad under it and everything was fine after that. I think McMaster sells vibration deadening pads just for this purpose. Well, not for in PCs but for aquarium pumps and whatnot.

[AntiBling]

How odd that someone would reply to this thread today.... my Danner has run 24/7 this entire year,and I havent given it a second thought until today. I am about to purchase another one for a new gaming PC Im building. My Mag3 has made a Danner fan out of me. $45 dollars and its worked so well I forgot all about it. I give it two thumbs up. You're right on the money about it getting quieter. Mine has a 60Hz hum, but to hear it you have to kneel down and put your head next to the case. There is some vibration. I nested mine in a foam cube. That has to be dealt with or it will rattle against whatever its resting on.

[JamesAvery22]

Originally Posted by AntiBling

How odd that someone would reply to this thread today.... my Danner has run 24/7 this entire year,and I havent given it a second thought until today. I am about to purchase another one for a new gaming PC Im building. My Mag3 has made a Danner fan out of me. $45 dollars and its worked so well I forgot all about it. I give it two thumbs up. You're right on the money about it getting quieter. Mine has a 60Hz hum, but to hear it you have to kneel down and put your head next to the case. There is some vibration. I nested mine in a foam cube. That has to be dealt with or it will rattle against whatever its resting on.

Mine is pretty loud... I have it on some foam so its not vibrating the sink its on right now but I can hear the hum when its 10' away in the bathroom... I really think mine is louder than it should be from what Ive read from others.. If its not on the foam I can hear it like 3 rooms over...

[Razor6]

A possible reason that wasn't mentioned about why the o-ring leaks sometimes is that the 4 screw hole arms on the cover touch the face of the pump before the rest of the cover can come down all the way to compress the o-ring. At least that's how it is on my Mag5, it worked for over a year before developing an extremely slow leak. Maybe the reason that only some report no problems is that the degree to which the screw hole arms extend below the face of the cover has some variance.

[JamesAvery22]

Originally Posted by Razor6

A possible reason that wasn't mentioned about why the o-ring leaks sometimes is that the 4 screw hole arms on the cover touch the face of the pump before the rest of the cover can come down all the way to compress the o-ring. At least that's how it is on my Mag5, it worked for over a year before developing an extremely slow leak. Maybe the reason that only some report no problems is that the degree to which the screw hole arms extend below the face of the cover has some variance.

my o-ring is leaking barely, and I mean barely. But its still leaking. The vibration is so bad Im emailing marine depot and going to see if I can exchange it for a eheim 1048(with some extra $ of course =\ ). The noise the pump makes is pretty important to me. I still think there is something pretty wrong with mine.

[Kobuchi]

Great posts krazy!

Teflon tape is preferable to pipe dope where the joint might seize up from corrosion (It works on mechanical fasteners too!). That's not an issue setting barbs in a pump. But note there is a right direction and a wrong direction to wind teflon tape.

Pipe dope also comes in crayon form. This is a stiff paste and is rubbed into the threads.

[JWFokker]

I prefer RTV sealant. It worked very well for me.

[gruntledweasel]

Forget a sticky, I think this is front page material. Very good writeup, with nice detail (and pretty pictures). There's a whole lot of people with Mag 3s who could make use of this info. Good work.

[the jew (raven)]

What fittings are you guys using on your Mag 3? I've heard that just getting a 1/2" NPT to 1/2 OD barb is too restricitve for the pump. So if you're not supposed to use those fittings, what would one use? I'm using 1/2" ID tubing (obviously).


EDIT:


I pressuming the answer of female 1/2 - 5/8 barb (outlet) and a male 1/2 3/4 od from this thread is still true?

http://forums.procooling.com/vbb/sho...light=mag+pump

[JWFokker]

Ideally, you could use 3/4" ID tubing at least for the inlet if you have a reservoir.

[the jew (raven)]

Originally Posted by JWFokker

Ideally, you could use 3/4" ID tubing at least for the inlet if you have a reservoir.

Well, yeah, but I've got 20 ft of Clearflex 1/2" ID hose in front of me so I'd rather just stretch it over. Wouldn't the rest of the hose being 1/2" ID negate the benefits of the 3/4" section?

[Cranky]

nice tips tbh. i have had 2 mag 5's leak from that exact same place..

[killernoodle]

My new mag 3 has developed a rattle of some kind, I think it might be the impellor being loose. I'm going to take it apart and goop the impellor together and see if that might fix the problem, because it is starting to get irritating. I'll post my results tonight.

[JamesAvery22]

Originally Posted by killernoodle

My new mag 3 has developed a rattle of some kind, I think it might be the impellor being loose. I'm going to take it apart and goop the impellor together and see if that might fix the problem, because it is starting to get irritating. I'll post my results tonight.

I wouldnt do it... I searched a lot on this before I just went with a 1048. Everyone that did it had to hit their pump to get it to start when they glued the impellar...

And as an FYI on a single block system my 1048 gave me better temps. Too bad I bought that mag3 before I read Cathar's thread on pumps =(



As for what fittings I used a 1/2" pipe threaded nylon T. The actual ID of the threaded fittings is > 3/4". I still used 1/2" ID tubing and fittings at the ends of the T though.

[killernoodle]

Ok, I did the mod only this time using thin walled tubing to isolate the impellor. The pump does have problems starting, it takes about ten seconds and makes a helluva lot of noise, but because I run it 24/7 I dont have to worry about it starting. There is definitely no more rattle and all I can hear now are the numerous air bubbles which should disappear in a few days. Here is a simple pic of what I did, it doesnt completely isolate movement, but definitely helps with the rattle. Plus it is instant and I didnt have to wait for any glue to dry, and I dont have to worry about chunks of glue in the system.

[JWFokker]

Originally Posted by the jew (raven)

Well, yeah, but I've got 20 ft of Clearflex 1/2" ID hose in front of me so I'd rather just stretch it over. Wouldn't the rest of the hose being 1/2" ID negate the benefits of the 3/4" section?

In a closed loop, yes, but with a reservoir you want as large an inlet as possible. Ideally you'd mount the pump flush with the reservoir so the inlet is actually inside the reservoir. I just switched to a t-line, it made no difference in temps, and it was a pain in the ass to fill. I'm going to switch back to a reservoir when I get a better one. If I have to, I'll tap a huge outlet on it so I can use 3/4" ID tubing for my pump inlet.

[killernoodle]

Ok, all the air bubbles are settled and I have to say, this mod actually does work. The only downside so far is the problem starting, but everything else is a definite benefit. There is less vibration, less rattling, less noise in general from the pump (considerably less I might add), the only noise coming from my computer now being the 4 hard drives.

I highly recommend this mod for anyone who leaves their pump on 24/7.

[krazy]

Some brief updates on the pumpology:

The pump has been running more or less nonstop since I bought it now with little trouble. I've still never had a single leak from anything on or directly attached to the pump. I actually just disassembled it tonight to have a look at the impeller. If you look at the pictures I posted initially (or your own mag-drive impeller), you will see that there is a small washer made of some kind of synthetic material that rides on the shaft between the impeller and one of the rubber end bushings. A few months ago, mine simply wore out. The pump started to make a rattling sound that was initially faint, but grew worse and worse by the day. I eventually shut the machine down, drained the loop, and tore open the pump. Everything looked fine, but I noticed that when I stuck the impeller assembly into the (removed) impeller housing, I could get the impeller to scrape against the inside face of the housing without much trouble at all. I popped one of the busings off of the shaft and took the pieces off of the shaft. The washer was worn right down to nothing for some reason. It seems that either the suction of the pump or pressure from the magnets pulls or pushes (respectively) the impeller assembly's magnet towards the inlet and was putting enough pressure on this washer that it simply ate right through it. All that was left was a paper-thin wafer with part of a rim around it that escaped being worn to nothing.

Since my computer couldn't run without a pump and I don't exactly have another workstation handy, I decided that I would have to fix the impeller pronto. I scrounged in the workshop, hoping to find something synthetic and the same size as the former washer, but all I could come up with was an ordinary steel #6 hardware store washer. It fit onto the ceramic shaft PERFECTLY and was also the same outer diameter and approximate thickness as the original deceased washer. I worried about rust and corrosion for about 30 seconds and then slipped the washer onto the shaft in place of the original one and reassembled the pump.

When I broke down the pump tonight, several months later, the washer was still there and still doing it's job well and wasnt even very rusty or corroded. The water I drained had a very fine black sediment that settled into the bottom of the dishpan I used after a few minutes of sitting still on the floor, but I don't know if it is related to the steel washer or something else in the loop. (the loop is aluminum-free and just has a few copper blocks, acrylic tops, clearflex tubing, acrylic reservoir, and a soldered copper/brass heatercore and the pump, so...)

Now the impeller assembly is rattling again (the reason for my disassembling it) but it is not scraping on teh impeller housing face. It seems that the actual individual impeller component no longer has tight tolerances on the ceramic shaft and is rattling around like a dying bushing would. If I stop and restart the pump a few times, I can pretty much get it to quiet down on a hit-or-miss basis, but I am ready to order a spare impeller assembly (or two). If anyone has a recommended place to get one for a good price (or with cheap/free shipping!!!), feel free to speak up. It hurts to pay $7 to ship an $8 part. I guess it's just the principle ot it...