MechMate CNC Router Forum

MechMate CNC Router Forum (
-   50. Toolheads (
-   -   Water cooling a spindle - KISS Method? (

Radishworks Thu 03 September 2009 22:23

Water cooling a spindle - KISS Method?
I'm going to start another thread on water cooling the Chinese Spindle. Gerald, feel free to merge with another thread if you think that's best.

I've committed myself to the Chinese Water cooled Spindle and ordered the 2.2kw one today. I have to say the quiet-ness (fan-less) of what people report is really appealing to me :-). It hasn't arrived yet, but I'm already thinking about the cooling system.

After reading all the threads on this topic, I'm really thinking a pretty small pump will do the trick. The pump that comes to mind is a automotive fuel pump. Yeah, its 12v, but it should fill a beer glass in a few minutes, fuel lines are ~1/4" and are designed to pump long distances (back of car to front). Maybe even a submersible one in the tank. Thoughts anyone? Anyone tried a fuel pump?

I'm also thinking about the tank, and I'm wondering if a few gallon aluminum tank mounted under the table will provide enough cooling (i.e. no fan, or radiator). If I really get motivated I'm may try to tig one up and add some aluminum cooling fins on the bottom. Again, any feedback?

Thanks in advance.


Gerald D Thu 03 September 2009 23:35

A fuel pump will not like plain water - you would have to use a non-corrosive liquid.

My first port of call for a pump would be the gardening center to look for a fountain pump.

Quick google gave this:

javeria Fri 04 September 2009 00:33

Mike did you get a chance to look at my video on youtube?

Gerald D Fri 04 September 2009 01:31

Here is a test that someone can do in the interest of science:

- Put a full bucket of water on a step ladder, 1 meter [3ft] above the spindle.

- Put one of the spindle hoses into the full bucket, below the water surface.

- Put an empty bucket on the floor, 1 meter below the spindle.

- Suck on the second pipe till the water siphons down by itself, then drop the second pipe into the empty bucket on the floor.

- Measure how long takes to drain the top bucket.

Good old school stuff involving buckets of water.....who could resist? :)

Anyway, this test will tell us the flow resistance of the spindle and piping and then we can select a pump based on real facts. (we need to know the difference in water level height between the two buckets, as well as how long it takes a gallon or liter to flow)

bradm Fri 04 September 2009 06:55

Mike, I'm using a simple, plastic, 5 gallon pail with about 4 gallons of water in it. With small, submersible, fountain type pumps it worked, but not with a flow rate I was comfortable with; I'd describe it as about two drops per second coming back through the hose. With a 1/6 HP sump pump, I'm getting a continuous trickle; enough to fill that beer glass in a minute. Unfortunately, the sump pump is heating the water. My next stop is with an external pump similar to the one Irfan shows.

Radishworks Fri 04 September 2009 08:09

Thanks for all the feedback and the video link.

A fuel pump will not like plain water - you would have to use a non-corrosive liquid.
I was thinking about using automotive anti-freeze / coolant. At least that way I can use my MM when the temp is below zero ;-).

submersible fountain type pumps
Ahh, good idea. I wasn't making that connection with the "fish pumps" on some threads. I may even have a couple of these around to play with. Good point about the pump itself heating up the water, I'll watch for that.

Good old school stuff involving buckets of water.....who could resist?
OK, I'm up for that! As soon as my spindle arrives I'll give that a try.

BTW, on a side note. I got a message back from the Spindle people, that they don't offer the 110v spindle anymore, only 220v. Not a big deal, this just means my MM will need a 220v outlet rather than a 110v one.

Gerald D Fri 04 September 2009 08:51

A 1/6th HP pump is about 120 Watts. Even if that pump was totally inefficient, and converted all its power to heat, that is only 120 Watt of heating. Sure, it will raise the water temp a bit, but most of the heat is expected to come from the spindle. If a 2kW spindle is 90% efficient, it will make 200 Watt of heating worst case. Which mean means we need a bucket/drum/tank of water that can dissipate 300 to 400 Watt with the water temp being say below 50 C [120 F]

Taking the pump out of the bucket will not give you cold water - the pump is still putting energy into the water.

Gerald D Fri 04 September 2009 08:54

Brad, are you holding the return pipe under the water level in the bucket? (to keep the siphon effect going). If you want to do a flow rate test without putting the hose under the water, you have to lift the bucket to the same height as the spindle.

Radishworks Sat 05 September 2009 14:36

4 Attachment(s)
OK, what could be more fun than running experiments with pumps, water, and electricity! :-).

I looked about and I did have an old pond pump in the shed. Itís a 115v .8Amp pump, see pictures below. I decided to run a couple tests.

Test #1: I filled two buckets with 4 gallons of water each, one is my control and the other has the running pump inside of it. This little submersible pump raised the temperature of the water 8 degrees over 3 hours, or about 2.5 degrees per hour. So I can confirm the pump raises the temperature of the water, not a huge amount, but it does none the less.

Test #2: I wanted to see what the flow of the pump would be through all the tubes needed to get though the cable chains and to the height of the spindle. I ran some tubes through my MM and joined the ends together where the spindle will be. When the pump is turned on it pumps the water no problem to the height of the spindle and back into the bucket. The end result is 1 gallon pumped in 42 seconds, or about 1.4 GPM.

Conclusion: This particular pump will work, but might be a bit overkill. I'll run the same tests when I get the spindle to see what kind of resistance is in the pump - Gerald's ladder experiment. It looks like Beckett (the pump maker) has several other sizes, a 60 and 90 GPH (about $20-30) one that might be a better choice for less flow and less heat into the water.

I'll keep you posted as I progress looking for the right combo.

Gerald D Sat 05 September 2009 20:11

Brad told us that there is a lot of resistance in the spindle's water channels, so I wouldn't consider going smaller just yet. Here is your Beckett pump's sales sheet, it is rated about 350 gallon per hour.

larry1larry Sun 06 September 2009 19:28

HSD,Colombo etc,spindles are rated for temperature of the bearings.They must be warmed up to operating temperature before use.10to 15 minute no load warmup is advised.Are the Chinese spindles temperature rated?Too high a flow rate may"Overcool"the bearings.

domino11 Mon 07 September 2009 20:01

The other aspect I have also wondered about is those of use in a cold climate. In an unheated garage, with temps at or close to 0C, can you still run a spindle without damage to the bearings? Would you need a heater in water then to maintain the proper temp for the spindle? :)

larry1larry Tue 08 September 2009 12:21

Hay Heath,
You would not need to heat the water at 0C.Pro spindles,Colombo,HSD etc. rate the bearings at a max of 140F.May be you don't need cooling in the winter.A good temperature guage is in order.Bearing warmup is necessary at low RPM before heavy cutting.

Radishworks Sat 12 September 2009 09:23

1 Attachment(s)
Time for more cooling pump test results.

My spindle arrived a couple days ago. (Very well packaged and a nice looking piece of equipment BTW). But for those wondering why the instructions that come with the spindle don't answer these questions, I will scan and attach the sum total to all the documentation that came with the spindle. See below.

The water connections on the spindle are for 1/4" OD tubing - i.e. very small. I hooked up my existing pond pump (the G325AG20 discussed above). Using this pump the flow rate dropped through the spindle from 1.4 GPM to 0.3 GPM (or 1 gallon in 3:10). So yes, as discussed, there is a pretty high resistance through the spindle.

Just for fun I purchased a much smaller pond pump from Lowes (the other Home Depot) for $20 and some 1/4" OD tubing. This one:

Unfortunately, this little pump wasn't strong enough to pump through the 1/4" lines through the machine much less the spindle. Oh well, I had to try.

Conclusion: I think I'm going to go back to the fuel pump idea. Pond / fountain pumps are great (and cheap) but these pumps are designed for high volume / low pressure. I think fuel pumps are designed for a higher pressure.

I'll let you know what happens.

Gerald D Sat 12 September 2009 09:52

If there is a lot of resistance then you need a more powerful pump, especially one with a higher "head" or "shutoff height" (ie. pressure when you hold it closed with your finger).

Your Beckett was only getting 0.3GPM through the pump, which is 18 GPH which means it was experiencing about a 12ft head or shutoff height. 30 GPH would probably be a very healthy cooling rate through the spindle - so we want a pump that can do 30 GPH at 12 ft head. [150 liter/hour at 4 meter head]

If I look at the Dragonfli pump curves at then I see that the DF5050 pump will easily handle that. None of the other Dragonfli pumps will cope.

Radishworks Sat 12 September 2009 10:20

Thanks for the feedback!

Just as your post came in I found (and ordered) this little pump:
This seems to be right in the flow rate (a little higher) than you suggested. I found it on sale for $49 at, that should be in the price range of the DF5050.

I also found this:

How cool with that display be on the door of my control box. For $19 who could resist! :)

sailfl Sat 12 September 2009 10:22


That will be a nice feature to have.

Gerald D Sat 12 September 2009 10:34

$19 . . . . that is what I am expecting to pay for dinner for two tonight! :D

If that pump can run off a PC's power supply, only drawing 18 Watt, then it will only give a small trickle of water. Sorry, but it is going to fail the beer-glass-in-a-minute test. :(

Reading that spec:

Nominal head (@ 12 V) 20.2 ft (6.1 m) This happens with zero GPH
Nominal discharge (@ 12 V) ~ 120 GPH (454 LPH) This happens at zero head

You want 30 GPH at 12 ft head, simultaneously, and you can only decide that if they give you a pump graph.
Nominal head (@ 12 V)20.2 ft (6.1 m)Nominal discharge (@ 12 V)~ 120 GPH (454 LPH)
Nominal head (@ 12 V)20.2 ft (6.1 m)Nominal discharge (@ 12 V)~ 120 GPH (454 LPH)

Realise that the Dragonfli pump draws 150 Watt . . . . . there is no way that a 18 Watt pump will give a similar performance.

bradm Sun 13 September 2009 11:17

Finally getting back to this.

Gerald, with the spindle in it's normal vertical position, the amount of water that flowed through using the ladder test was ... zero. It is possible that there were some air bubbles in the system, hiding inside the spindle preventing a proper siphon effect.

Also, when I originally reported heating of the water by the sump pump, it was without the spindle running, so the only source of heat was the sump pump. I suspect having such a restricted output flow caused it to operate far enough out of it's design specifications that it became more of a heater than a pump.

This response was delayed, because I waiting for a chance to test out a wretched overkill approach, using a 1/2hp 330 GPH 115Ft head pump ( harborfreight #1479 )

I have this in place with a reduction from 1" to 3/8", a tee into two 3/8" lines, one of which runs unrestricted right back into the supply bucket; the other up to the spindle and back. I now have a solid flow capable of repriming itself (*), which moves about 1/4 gallon per minute through the spindle.

(*) Note, when working with more reasonable sized pumps, I often found it helpful to completely clear the lines with compressed air before making an attempt with a new pump.

Now I do have water heating, but it's from the spindle. So the last step is to replace the plastic bucket with something that will dissipate heat, or add a radiator.

liaoh75 Wed 16 September 2009 23:40

I am finally putting in my Chinese Spindle and have been reading this with intense interest. My 3KW spindle arrived yesterday and I have already aquired a radiator out of a compact car from a junk yard which cost me next to nothing.

A few questions though, would too big a pump be a problem? I'll post pics after getting all parts together. Is there anyone using a 3KW water cooled spindle? I've seen a lot of 2.2KW spindles being used. Is 3KW overkill?

javeria Thu 17 September 2009 00:25

I dont think its a overkill just that it will be more expensive - off late people are experiencing with the Hyunang VFD, think going for a standard delta is better, I think you need 3 phase supply for your 3kw spindle?


Radishworks Thu 17 September 2009 08:39

2 Attachment(s)
Hi Irfan,

I'm determined to find the right pump here. I'm convinced the pump needed is all about presure not volume.

I have the 2.2KW spindle, but I suspect the cooling will be similar to the 3KW one (but that's only a guess).

I'm still waiting for this pump to arrive:

But in the meantime, I did find this little Delco 12v pump at a local surplus store - $10. I think itís a windshield washer pump for a car. (pictures below) I hooked this little pump up to the 1/4" lines that run through all my cable chains and through the spindle. Flow rate was 0.33 GPM, almost exactly the same flow rate the big G325AG20 pond pump was! I think this still passes the beer glass test :-). This Delco pump is only about 6 Watts (12v X 0.5A), yet the larger pond pump is 96 Watts (120v X .8A). I don't think this little Delco pump is the right one to use, I don't think it will hold up to continuous use and itís a little noisy.

I'm not sure a pond pump is the right pump for the job here (sorry Gerald :-), please forgive me!), I think pond pumps are better at volume, not pressure and that's why it takes such a large pump to get the needed flow.

I do agree with Gerald that a radiator may not be needed at all, but I'm trying to find the right pump first.

I will report again when the next pump arrives.

WTI Thu 17 September 2009 09:28

1 Attachment(s)
If you look at the Excitech CNC routers with factory installed 2kw spindles, they have a rather small radiator and motor that rides along right behind the spindle itself.

So a huge radiator is not needed.

WTI Thu 17 September 2009 09:37

This site has a ton of pumps, radiators and fittings that are small enough to ride along with the spindle itself.

WTI Thu 17 September 2009 09:46

Using long coolant lines with lots of twists and turns adds way too much resistance (head pressure) for a pump. It is often said that every 90* turn adds 3 feet to your head pressure.

A short hose run, with the least amount of tubing bends will let a small pump motor be sufficient.

Gerald D Thu 17 September 2009 12:06

I wouldn't agree with the 3 ft for every turn saying, when applied to this particular system. Our problem seems to be the resistance of the spindle itself - the tubing seems to be of little consequence. See test #2 in this post.

WTI Thu 17 September 2009 15:30

1 Attachment(s)

the above kit has been used successfully by several people to cool the 2.2kw spindles.

Yes, the kit cost $150, but at least you are using a proven system. You will have the CPU water block left over to sell on eBay.

Everything is 12v, the pump draws 600mA and pushes 130 gallons per hour.

Greolt Thu 17 September 2009 15:57

Originally Posted by WTI View Post
Using long coolant lines with lots of twists and turns adds way too much resistance (head pressure) for a pump. It is often said that every 90* turn adds 3 feet to your head pressure.
Head pressure is a static thing. Twists and curves will not effect that at all.

Those type of restrictions, along with the diameter and length of the tubing, come into play when you consider flow.

With the "Beer glass in a minute" level of flow that Gerald has talked about these restrictions have minimal impact.

However if we expect the coolant to flow like a fire hose then that is a different story.


WTI Thu 17 September 2009 16:33

Lets assume that the flow rate is actually 120 GPH. This represents the worst-case scenario because head loss in a 90 degree elbow is highest for the greatest flow rate.

Flowrate, Q = 120 GPH = ... = 7.7 []
Pipe area, A = 1/4*pi*D^2 = 0.1104 [in^2]
Nominal flow speed, V = Q/A = 5.8 [ft/s]

Dynamic pressure = 1/2*rho*V^2 = 0.5*1.938[slug/ft^3]*(5.8[ft/s])^2
Pd = 32.6 lb/ft^2.

The loss from one 90 degree bend is approximately 0.5*Pd, so the head los PER bend is going to be about 16.3 [lb/ft^2], or 0.26 ft of head. This means that for each 90 degree bend, you can expect to lose about 1% of flow through 3/8" tubing at 120gph.

I just got schooled by the king of overclocking...

Greolt Thu 17 September 2009 17:49

Originally Posted by WTI View Post
Lets assume that the flow rate is actually 120 GPH.
That is my point. We don't need anything like 120 gph.

So therefore the rest is irrelevant.


All times are GMT -6. The time now is 06:14.

Powered by vBulletin® Version 3.8.3
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.