[peteeng]

Hi Craig - My search for small harmonic drives continues. The planatary I'm looking at has 30arc min backlash (from stepperonline, a couple of years ago they used to be 10 arc min). I'm going to use this on Lanky's Z as a test as I have them in stock. I have had 5:1 and 10:1 on a small router X&Y belt driven and was quite happy. As soon as I find suitable harmonics then that's the future. Working on Lanky's Z axis at the moment.. Trying to use up some old parts!! Peter

[joeavaerage]

Hi,
very wise to test them out, I suspect that 10 arc min lash will put you off. There are better brands but the cost goes up bigtime.

The two gear reducers I have used for my trunnion and fifth axis are <2 arc min and <1 arc min respectively. I've never actually succeeded in measuring any lash in either of them, the limits are manufacturers
published figures. To meet what I call reasonable accuracy for my machine I require better than 10 arc sec resolution. If the gear reducers actually had 1 and 2 arc min lash surely I would see it?
Any way I suspect you'll find much as I have that any angular lash is very detrimental to five axis operation.

Craig

[joeavaerage]

Hi,
just as a sanity check imagine a fifth axis platter with 1 arc min lash. At 100mm radius (from the centre of rotation) 1 arc min of lash translates to 0.029mm linear lash
at 100mm.....and that is to my way of thinking not good enough.

If I tighten the angular resolution to 10 arc sec at the same 100mm radius then the linear lash is 4.9um..... and that is the kind of resolution that I'm trying to obtain with my machine.

Craig

[peteeng]

Hi Craig your machine is a mill and requires higher tolerances then this machine. Mine is a Maker level router for wood and plastics, and being #1 its at preproduction level. I'm sure the tolerance of around 0.1mm is fine. Better to have a machine working then one in the hard drive. Its a machine to learn about 5 axis operation and sort & learn the electronics, logistics and programming. I'm sure in production I'll get a harmonic some where! Peter

Interesting - I looked up the data sheet for the 10:1 in my small Brevis machine. The data sheet is dated 2018 and its backlash is <15arcmin. SOL have two gearbox series one has 30 the other <15arcmin... I'll dig around for another brand... at the time I put a crank on the axle and could not get any lash but it was new....Peter

[peteeng]

Hi all - I have been looking at these harmonics. But I'm unsure which is the output holes or surfaces. I have sent a note to the supplier to explain. Currently my guess is that the D14 is the input shaft from a motor. The D90 is either a hole fit or use the outer hole ring for the static mount and the inner ring or holes for the output? Any thoughts? Peter

[joeavaerage]

Hi peteeng,
that diagram looks incomplete?? I'd swear that the diagram show the outer ring gear, the strain gear and the output bearing ........but where is the wave generator?

It's common that the wave generator be installed on the end of the motor shaft and rely on the motor bearings. That's what this unit looks like.

Craig

[peteeng]

Hi all - Looking at my rotary say D400mm then the planetary at 15arcmin (0.25deg) gives 0.87mm backlash and the harmonic at 30arcsec (0.0083deg) gives 0.028mm backlash so sorting the harmonic for the rotary seems a good idea... I'll pester the supplier for more mounting info. Peter

https://hfg3d.com/produto/harmonic-d...umanoid-robot/

[joeavaerage]

Hi,
I have got my new Delta VFD this morning from the US, its a second hand 11kw unit. I paid as much for it as I would have a new 5.5kW unit....so I thought why not?. If I ever get an even bigger spindle it will
not go astray.

I have just been programming it, it seems strange to be programming it for less than one third of its output....but my new spindle is rated at 3.5kW, and the VFD is capable of 11kW.

The new spindle runs beautifully, and so quietly. I've only run it at up to about 12000 rpm so far....I really want to stick to the run-in procedure and give it at least half an hour at rated speed before
taking it up to full speed. I don't have any water cooling yet so cannot run it for extended periods yet, it just gets too warm. I do not even have a monitor on the PTC temp sensor so I refuse to run it
for any length of time without cooling....it would be asking for trouble.....not that that has stopped me in the past, but a don't want/cannot afford to blow this thing up!

Using the multi-function display it consumes about 7A (at modest speeds, say 8000rpm) no load.....so it gets warm. Its leading me to think that careful and continuous temperature monitoring is
going to be required.

So still a few things to do:
1) Machine the spindle mount and fit to the headstock of the mill.
2) Devise a water cooling unit for the spindle.
3) Design and build an electronic monitoring unit for the PTC temp sensor and the three Tool-in-Place switches and wire to exclude VFD operation in unsafe/overheat condition. Will use the same module to mount
and monitor coolant flow, seal air, taper cleaning air and ATC air.
4) Install a 20A D curve three phase breaker block in the switchboard and a 20A three phase (plus neutral) socket to plug my machine into.

I've about blown my budget (for the next seven years any way), and would ordinarily contemplate selling off another child.....but I don't have any left!!! (Joke.....just to be clear)

Craig

[joeavaerage]

Hi,
found a local company to sell me an oscillating pump. Up to 55psi and 14l/min. I suspect and hope that at say 15psi I'll get 5l/min. The pump itself is industrially rated, and for my application
I can see it running for many hours at a time....so it will need to be industrially rated.

Back to 'as poor as a church mouse' again!

Craig

[joeavaerage]

Hi,
I rather think I've made a mistake......I've made one once before and I swore I'd never do it again!!!

The oscillating pump pictured above is 14 l/min......but I have reason to doubt that.

I have an oscillating pump that I took out of a welder years ago. It is very similar in style and size to the one I just ordered. I tried it out over the weekend and found its open flow rate is about 800ml/min.
That is consistent with advertised flow rates of other similar oscillating pumps.

I now feel like the advertising I saw, namely 14 l/min is wrong, in fact I'd believe 1.4 l/min. I think I have been guilty of believing something I saw WITHOUT question and bought on that basis. I may of course be wrong,
but as its now on its way from Auckland I'm stuck with it no matter what.

A friend pointed out another pump that I missed in my haste to buy the other day. Its a magnetically coupled stainless steel gear pump. Its a genuine quality unit. The motor alone is by Oriental Motors, the makers
of Vexta, and they 'put a horn on a jelly fish'. Its second hand at only $46.00NZD plus shipping. I would expect a pump like this of this quality to be several hundred USD new....so I got one of them.

So I now have two pumps on the way. I think this last one is the real deal.....not sure what to do with the first one.

Craig

[peteeng]

Hi Craig - Garden fountains are a useful distraction. Peter

[phomann]

Hi,

I seen a number of water cooled spindle systems where they use a closed loop gaming PC CPU water cooling system.
They seem to get mounted to the gantry of routers quite often.

Cheers

Peter


Sent from my iPhone using Tapatalk Pro

[joeavaerage]

Hi, Peters both,
they are good ideas.

I set up the oscillating pump I already have on hand and ran my spindle with it over the weekend. The flow rate I measured was about 800ml/min. I found that the temp increase at that flowrate
to be very small.

The specs for the spindle indicate that the flow rate should be >3 l/min and that I might expect a back pressure of 0.25 to 0.4MPa or 35psi to 60psi. I'm not sure how good these figures are, I rather suspect the back pressure
figures are overly high, but they are the OEM specs so I should keep them front of mind.

The second issue is how much flow is required? As I say even only 800ml/min results in scarcely discernible temperature increase. After half an hour the five litres (of my reservoir) of water had increased temp by 5C. I believe therefore that
even modest flow rates would be adequate.

Last thing is how much heat the spindle generates. What surprised me, at least initially, is that at low speeds, say any thing under 20,000 rpm the current was 7A/phase at no load. At 40,000rpm the current had dropped to 3A/phase.
The more I thought about it I realised that at low frequencies the inductive impedance of the stator windings is low and therefore the magnetizing current is high. Indeed this is exactly why we program a V/F curve to prevent
excessive current at low frequencies. At high frequencies the natural inductive impedance of the stator is high and therefore the magnetizing current is low.

These measurements were taken at no load, ie not cutting loads. Induction motors are reasonably efficient, I would guess over 80%. Thus if I exerted a 3kW load on the spindle then I would expect something
like 600W of heat to be generated over and above the magnetizing losses. The magnetizing current at 10,000rpm is 7A/phase and the applied AC voltage is 200VAC (at 10k) thus the magnetizing losses are
something in the order of 4kW. The upshot here is that the losses ( ie heat build up) is dominated by the magnetizing loses not the mechanical load.

The important conclusion I have drawn is that the magnetizing losses are in fact very high, and without continuous water cooling I could very reasonably expect the spindle to overheat and in short order. The spindle
weighs around 20kg, but if my calculation is correct, potentially 4kW being dissipated in there will cause a rapid temp rise! This rather means that I need a flow rate detector, and possibly a coolant temp alarm as well.
There is a well characterised PTC temp sensor in the spindle so maybe coolant temp monitoring is over the top, but I think a flow detector is pretty much mandatory.

Craig

PS:
I realise my calculation suggesting 4kW magnetizing loss is in error. What I have done is assume that all that reactive power turns up as heat, when in fact only a portion will depending on the power factor.
I would guess that the power factor will be low and no load as essentially the impedance is purely inductive, say PF=0.5 That would in turn suggest about half the apparent magnetizing current is
actually causing heat loss, so not 4kW but 2kW. Either way the magnetizing losses are high relative to load losses, so my conclusion remains. What this new appreciation does do is point towards
a more realistic estimate of the thermal mass of coolant required.

[peteeng]

Hi All- Great news! I have found a seller of flanged 50:1 harmonic drives. Saves making motor mounts and surface mounts. 20arcsec backlash very good. <$100USD per unit so cheaper than making the extra flanges and mounts... Now to redesign the B axis and the platter drive!! Peter

[joeavaerage]

Hi peteeng,
yep, they look like a cost effective way to do things.

Craig

[joeavaerage]

Hi,
my pump turned up, the Oriental Motor/Stainless Gear pump one. It's superb. I paid $46NZD plus $10NZD shipping from Napier, and yet I'd swear I'd pay $400USD or more for this thing new,
its genuine quality.

Pumps a little over 5l/min and I don't know what pressure, I cannot stall the flow by hand, my guess is well north of 100psi. I shoved it up my bum and bloody near popped!

I'm now thinking I might buy some more of these things, I suspect they would make a GREAT high pressure coolant jet.

Any Kiwis or Aussies reading this who want a coolant pump look on TradeMe....really good value for a superb quality unit. Any overseas readers whom are interested I'll get one for you. Apparently the NZ supplier has 50 or more of them.

Craig

[phomann]

Hi Craig,
I tried to send you a message but alas your inbox is full.

Could I ask you to snag one of the pumps for me?

Cheers

Peter


Sent from my iPhone using Tapatalk Pro

[joeavaerage]

Hi Peter,
I've just cleared my inbox.

Most certainly I'll get one for you, in fact I'm going to get another myself. They are good.

They are a capacitor run motor so you do have to wire them up. There is good documentation about Oriental Motors on-line, that is where I got the wiring info, took less than a minute thereafter.
I'll PM you when it arrives.

Craig

[phomann]

Hi Craig,

Thanks, I DM I tried to sent went through.

Cheers,

Peter.


Sent from my iPhone using Tapatalk Pro

[joeavaerage]

Hi,
given the rapidity of the flow through the spindle I don't think that I really need a flow rate detector....but just something to confirm the flow.

I'm thinking of a wee cheat, cheap and simple.

You may be aware that low fuel level sensors in cars are often just an NTC thermistor. When the thermistor is immersed in fuel (and cold) it is a high resistance state, when however the level drops the applied voltage
causes the thermistor to self-heat and threreby reduce the resistance. The circuit could not be simpler: Ignition supply to one side of a low wattage bulb (low fuel warning light in the dash), through the bulb, through the NTC to earth.
If the NTC goes low resistance (ie low fuel) the bulb lights.

I'm thinking I could do something similar. Place an NTC, which are about the size of a grain of rice into the return flow of the spindle coolant pump circuit. If there is flow the NTC will be high resistance
and in absence of flow the NTC will heat going to a low resistance state. Simple and cheap. I'm already going to have a small circuit board in order to monitor the PTC in the spindle and have lock-out logic
to monitor the Tool-in-Place switches so its not like I have to make a special board for it, or rather I'm already building a board for it, it would add next to nothing to monitor and extra thermistor.

Craig