View Full Version : water cooling options for spindle

30-08-2010, 09:28 AM
I bought this water cooled spindle some time ago and am about to buy a water cooling system to go with it. When I asked about the radiator and pump heat rejection requirements etc. the seller helpfully suggested 'any kind of pump' to go with this system (didn't sell anything suitable himself).

So with that less than helpful spec I'm looking for options to cool this, anyone own one or something similar? My research has shown people using PC CPU coolers, and this Thermaltake 850i seems to be working for someone:

I've heard these spindles have a high flow resistance, and blowing into the inlet port last night confirmed this subjectively. My suspicion is that I need something which can create alot of pressure, but probably not need much in the way of flow.

I'm tempted to go for the Thermaltake option at about 90, but anyone used anything else? Looking to buy one ASAP to get the spindle running at last!


30-08-2010, 09:40 AM
Fishtank pump? :wink:

30-08-2010, 09:44 AM
Or even better....

A cheapish? brushless power jet washer? although this is pressurised...might not coolit down quick enough?

/I'll get me coat:whistling:

30-08-2010, 10:20 AM
There is a load of stuff to choose from in the PC world and certainly for cooling a CPU the best thing to be said about the Thermaltake is, it's green.

This may be of use..

... I'd guess something like this

... and this ...

would out perform the Thermaltake, as would almost any of the radiators on that site + that pump.

Add a 'T' junction and a blocked off length of tube for filling and off you go. Or you may want to add something like http://www.specialtech.co.uk/spshop/customer/product.php?productid=8159&cat=1630&page=1 and a little bit of silver for luck ;) http://www.xtremesystems.org/forums/showpost.php?p=3767420&postcount=18.

... should get you stated anyway :)

I wouldn't worry too much about flow rate, so long as there is some flow you just need to spec the radiator + fans to match your heat load at a comfortable temperature - say 30 to 35degC ?

30-08-2010, 12:32 PM
I used to use a windscreen washer motor and a large container ( sat on the cold floor of my garage (floor kept the water cold enough). This was to bench test and run in my model boat engines, on the same note you could use a small pond pump as you get decent pressure for long runs

30-08-2010, 01:11 PM
I've found out that these spindles are roughly 80% efficient, so clearly you need to dissipate 300W.
Lets say you have a 10L container of water at 20C and you run it for an hour...that will rise to 46C if the container was perfectly insulated.
I'm not at all certain about this calculation, but I'm pretty sure that that volume of water in an aluminium box would dissipate the heat fast enough.

I've got one of these pumps and it seems good:
Don't see why you couldn't use that with the setup I described.

30-08-2010, 01:29 PM
A bit elaborate but look here for an idea to put in line


30-08-2010, 09:19 PM
Thanks Guys, some useful stuff there.

I had considered fish tank pumps, but not sure if they would be up to it. If anyone has seen one used for this application, or seen anything suggested above actually used on a spindle, I'd be interested. Same goes for fountain pumps, high flow rate but would probably get hot with the restriction in the spindle - they are not very free flowing. Some people have tried windscreen washer pumps, but they did run very hot connected to spindles. I did consider running several together, but it feels like a bodge. Probably not as quiet as a PC cooler pump either.

Hi ecat, great stuff. Love the spreadsheet, would be great if there was a spindle option for the CPU block! I'd seen the silver strip idea to stop things growing in the water - was probably going to use a proprietary fluid of 25% proplyene glycol with anti-corrosion and anti-bacterial, but no harm to add that as well.

I've seen alot of the Thermaltake stuff around and thought it was respected, but perhaps there are better things out there, as you suggest. Thought this post might bring out some overclockers :smile: ! Thanks for the other options, I'd also considered a simple pump, radiator and T filling junction, but also like the idea of a reservoir to give more water volume and a bit of extra cooling. More water should allow me to run longer if the radiator and pump is a bit underspec.

So it's all a bit of a guess at the moment. These are the figures I think I'm looking at:
1. Don't want to spend more than 90 if possible
2. Want something to fit and forget - don't want to spend hours fiddling
3. Probably looking to dissipate about 200W, although I'll go with Jonathan's 300W
4. Prefer to run with a litre or so max because I'd like it all self contained near the spindle and water weighs ~1kg per litre!
5. Don't want the pump or fan to make too much noise - otherwise I'd have been better off with the aircooled version and this project is about a quiet overall system to replace that noisy router.

I'll do some more digging and if I take the plunge on something and it works I'll post it out in my build log.

30-08-2010, 09:39 PM
How come you never went for the air-cooled type?

30-08-2010, 09:53 PM
Hi George,

Two reasons. The built in fan creates extra noise, bit like a muted hair drier, and there's nothing you can do about it. Obviously nothing like a router in level, but I'm after peace and quiet. The water cooled spindles are probably about the quietest available, unless you know different. Since I'll be cutting alot of balsa (plus some liteply) I'm hoping that the overall levels during cutting will be fairly low, and I'll be able to hear the radio again. Also, for long runs I don't want to upset the neighbours.

Secondly the exhaust air flow blows down onto the cutting area (although it could be ducted away, but you don't want to restrict the flow). This blows all the fine dust (much of which you can't see) out through your collection shoe before the vac has a chance to grab it. With water cooled there is none of this.

John S
30-08-2010, 10:00 PM
I'm watching this post with interest to see what options are mentioned as i have an application that requires cooling.

I have a TiG torch that's water cooled, it's rater puny to say just what power it's supposed to handle and it requires constant cooling flow or the braided lead to the torch burns out.

The Tig welded is the new generation inverter model and is very portable, you can carry it in one hand and I'd like the cooler to be the same hence no header tanks.

30-08-2010, 11:39 PM
How do RCNC

The motors are identicle except one has air running through the channel and water through the other,to be honest even if its water cooled I don't think this will dampen the sound of it in both cases...water cooled is just that it keeps the motor cool for long periods of time.

31-08-2010, 01:13 AM
How quiet is quiet?

Yes I overclock the pc and yes it is very quiet, but I use 2 radiators with 6 fans running very slow. This is just for the CPU, say, 150W on load.


It's a balance between air flow and surface area. Don't suppose the base of your router is made of metal? Nice bit of Ally, even steel? Could pump the water through the section or fix a ring of copper pipe to the frame. It'll kill your flow rate but if you don't have much to start with you may never notice :)

The tube fittings on the spindle look quite small, any harm running the tube up and along the gantry?

One more thing, is any part of the spindle that contacts the water made from aluminium? I know you can use inhibitors, but life is more restful if you don't need too. Many pc water parts are made of copper, it does not play well with Ally.

CNC meets water, see pic 32 and 33...

31-08-2010, 09:41 AM
Was late last night...

The Tig welded is the new generation inverter model and is very portable, you can carry it in one hand and I'd like the cooler to be the same hence no header tanks.

Separate box: Sounds easy with these fangled push to fit, no leak, quick connectors.
No header tank: Sounds easy but only up temperature X. No header also means no expansion room.

Do you have any pics or spec?

Here's some RadBox action from the pc world for your amusement.


31-08-2010, 01:08 PM
Hi George,

The spindle motor itself is pretty quiet, aircooled or water cooled. The noise comes from the cooling system which is either a fan on the end of the shaft (20,000rpm) or a water cooling system to a proper radiator with a fan running at 2000rpm.

Hi Ecat,
Interesting project there! The ali/copper question is a good point, and is steering me back to the less efficient ali ones to be on the safe side.
Would prefer to use the internal cooling channels, and not mess about with extra cooling circuits, although its a nice idea.

I'm contemplating working out the flow resistance of the spindle cooling channel:
Cut the bottom off a plastic bottle, invert it, hang it out the 1st floor window
Run a pipe down to the spindle on the ground, with a pipe on the spindle outlet into a nearby bucket
By timing how long it takes to put say 10 litres into the bucket, whilst topping up the bottle to keep the same head, I can work out the flow rate and pressure characteristics.
Pressure is mass (of head of water) x gravity x head (of water), all of which are known and constant for this test.
Flow rate is the time is takes to fill 10 litres, converted back into some suitable units.

If I then repeat this test but without the spindle it would give me the flow characteristics of just the long pipe, and I'm guessing I could subtract that out.

If I then select a pump, look at the flow rate pump data at that pressure, it will tell me the flow rate through the spindle if I were to attach just that pump. I might then be able to plug this into the spreadsheet and see what would work in terms of radiator and fan spec.

I know water has a specific heat capacity value, the amount of energy required to heat a gram (or kilogram?) of water one degree. This might factor in somewhere, but it would be nice not to go down to first principles if possible.

I have read someone trying something similar but just holding the bottle at ceiling height, with the spindle on the machine on a desk (so head was about 1m). The water either did not flow out at all, or barely flowed, added to my suspicion that there is a lot of resistance in the cooling channels.

Whatever I end up do, I'm sold on the idea of using PC cooling bits, just a matter of getting something which works and is not too expensive.

31-08-2010, 02:59 PM
If anyone is looking for PC cooling bits to use such as pumps, radiators etc then try these people - http://www.overclockers.co.uk/productlist.php?groupid=962

Used them a few times for water cooling products and always got a good service.

I have stuck mainly to the Swiftech stuff and had a couple of their pumps/radiator systems running for over a year now with no issues at all.

31-08-2010, 05:59 PM
Hi routercnc

another good source for PC water cooling components is http://www.thecoolingshop.com/


31-08-2010, 06:52 PM
A few more links :)


Heh, found this, my design for a compact 6 fan cooler...


... you lose about 10% fan speed irc due to the fans fighting each other, the baffle didn't help at all. I'll have to get a case for this PC some day.

01-09-2010, 01:33 PM
Big thanks for all those links, am working through all the options now. The Swiftech MCP350 certainly is a good pump when compare to the Thermaltake one, although price / performance ratio is similar so it's a matter of trying to work out how much is required (and to spend!). Be glad when I've got past this bit and can get the machine cutting.

01-09-2010, 03:00 PM
Most of the Thermaltake and CoolerMatser stuff is so so - doesn't do the job all that well and I don't know if I'd trust it to see the year out.

I started with a 50 aquarium pump submerged in a bucket, yellow of course. No fancy fans or radiators just good old fashioned radiate heat to ambient and evaporation, lol. The open bucket works quite well in the short term but is prone to collecting all sorts of airborne dross, dust is bad enough but flora and fauna too :(

The Swiftech MCP350 is a badged Laing DCC. Solid, reliable and there are a number of bolt on tops that give a wider range of connection options, some of these tops include a small reservoir. There was a simple solder trick to convert the 10w model into an 18w model but I think Laing 'fixed' this.

Thinking about the all in one kits, Corsair and CoolIT have packages that receive good reviews. The down side is that the pump is built into the the 'cooling head', some plumbing would be required, I'm unsure of the metal mix and I don't know what kind of flow you'd end up with. The up side is possibly cost and you can always mount the 'cooling head' to your frame gaining some passive radiation :) or even drill the head out! The kits vary, but this looks sweet http://www.overclockers.co.uk/showproduct.php?prodid=HS-002-CI&groupid=701&catid=57&subcat=1395

01-09-2010, 08:03 PM
Hi ecat,

Thanks for the info, like the kit at the end but trying to spend less than that if possible. I've done some calcs to get used to the sort of numbers involved in all this, and came up with this quick spreadsheet to work out how long it would take to heat up a certain quantity of water (assuming no heat loss from the reservoir).

I also had another look at the inlet and outlet barbs screwed into the end of the spindle. Given that 3/8" (~10mm) ID pipe is quite popular for CPU cooling, I was disappointed to find the following rough dimensions on these barbs:
The thread into the spindle end itself is around 7-8mm (I've emailed the supplier for the exact size)
The barb which screws into this thread has a hole around 4.5mm at one end, and a hole about 2.5mm where the pipe connects.

This is quite a reduction! I repeated the 'blowing into it test' with and without these barbs connected and it made a big difference. When I find out what size thread these barbs are they are going to be replaced! If they are G1/4 (popular CPU cooling size), I'm hoping I can put some proper cooling mod parts in instead and increase the hole size from 2.5mm to 10mm at the pipe end.

Failing all that I guess the spindle endplate could be removed and the holes re-tapped. Feeling much better about this now since the flow will be so much better if I can open these restrictions up.

By the way, I've asked the supplier if the internal channels are aluminium.

01-09-2010, 10:07 PM
Impressive spreadsheet!

If I were the pedantic sort I'd remind you to add in the heat load from the pump, but I'm not so I wont ;)

The entry holes do look a bit piddly. Thin tube is more bendy (showing my full engineering acumen here) which may be a nice feature when connected to the moving head. I wonder how 12mm ID 19mm OD tube will fair after 2 weeks of constant bending and twisting?

The reduction in diameter sounds significant and may well be a design feature used to increase the velocity of the water over the cooling part:

1) Higher velocity means less laminar flow ie it minimises the occurrence of slower moving 'layers' of water near the boundaries which effectively 'insulate' the metal from the main cooling flow.

2) The common sense of many tell them that the longer the water is in contact with the metal, the more heat the water will take up, or give up if in a radiator. This is of course pants, in and out asap is the way to maximise the amount of energy transferred..... Unless achieving a satisfactory velocity requires a 200w pump ;)

Jet Plates!! Sounds good, lol http://www.thewatercoolingshop.co.uk/EK-Supreme-HF-Plexi.html

Hehe, extra points to anyone who can tell me what the 'Blank' jet plate is for :)

02-09-2010, 08:38 AM
Hi ecat,

Yes, didn't bother with the heat from the pump itself as it's quite small and this is a rough calc. By the way, I'm assuming you can ratio specific heat capacity values of water and Propylene Glycol in the proportion they are added to the mix.

Since I'm going to mount all this on the router plate there is no pipe bending and movement between the spindle and water cooling system. If you see my build log you'll see the router plate is long enough with space above the spindle to mount the cooling bits. It's all been worked out - well, package protected for anyway!.

I follow your point about narrow inlet/outlet increasing flow velocity, which it will, but immediately after the barbs (i.e. just inside the spindle) it opens out again which will slow it back down again. So I think this is effectively just a local bottleneck. My blow tests with and without the barbs back this up.

Thinking about your maximum heat transfer point, I guess this is like the wind chill effect where a windy day will cool you down more than a calm day. The rate of transfer of heat is proportional to the difference in temperature between the 2 parts. Hot stuff (relative to the surroundings) cools quicker. As it cools, it cools slower and slower. Therefore you want to keep putting fresh cold water next to the hot bits to maximise the transfer rate. Possibly.

I'm sure there are many factors involved, but to be honest I don't want to go down that long road - just want it to work so can get back to my hobby!

Blank jet plates - CNC machine your own?!

02-09-2010, 10:12 AM
Blank jet plates - CNC machine your own?!

To smart you be, yes. Indeed, drill your own holes :)

12-02-2012, 10:22 PM
Hello friends!
Not bother with containers, radiator, and pump! I connected a hose cold water, and the sewage will water the flowers!