[jdelaney44]

Thanks for the compliments!

I have not cut anything in a couple of months. Been tied up with getting a new day job and starting some .NET classes, helping out a couple folks with some CNC stuff and tiling my floor. Whew!

Accurate, yes. Reliable, not sure yet. Need more hours. With respect to the accuracy I probably had very reasonable expectations after months of on line research. I went over the machine with a dial indicator before I bought it so I knew what I was getting into. The electromechanics are scary accurate, well unless I run the IPM too high. The base iron is the real issue for this rig.

As always glad to hear the post is helping people.

School and work are settling down. I hope to start making chips in a few weeks. The floor can wait a little.....Don't tell my wife.;-) Kidding. She's been a terrific supporter of the project.

Best,
John

[jdelaney44]

Well. Got the next power supply version built and load tested. The 5 ohm NTC Thermistors work great. They do get hot but not blazing hot.

This one is for a friend. Will retro fit mine soon.

My machine continues to work nicely. I have not used it a lot, but when I need to use it, it does it's thing.

More details on the new supply design in a while.

Update - uploaded one pic. More to follow. Lost my mini USB cable.

This is with a low power bench test using a 300W regular light bulb. The final test rig was seven 300W bulbs and one $20.00 heat gun. The heat gun worked nicely as it had a DC motor and a bridge rectifier built right on the back of the motor. So it didn't matter how I plugged it in to the DC side of the supply.

The bulbs drew about 2 amps each. The heat gun pulled 6 amps. All tolled it hit 20 amps. I rewired after the first 20 amp test. Went with bigger wire and more runs.

Things that get hot:

The bridge rectifier
The NTC Thermistors
The wires at the recitifier
The trannies start to warm up too

This was about 1300 - 1400 VA. The trannies are rated at 800VA each. If you wanted to run it at 20A continuous you would need really good cooling. This will require decent cooling for the 15A I fused it at. That last 5A makes a big difference.

The trannies are rated for 55 deg. C temp rise at full power. So they need cooling to get that under control.

The caps stayed cool really. I was expecting some heat off them at full power. The wires to the caps got a little warm.

My concept going in on these is to over build so we don't end up running hot. I think I've gotten there. We'll see once this gets installed and running.

Best,
-jd

[jdelaney44]

Here are some testing / power supply Pics.

First is the full load test with the seven 300W bulbs and the heat gun which is out of view on the floor.

As I think I mentioned this is pulling 20A total. 72V max. At full load it dropped to 63 Volts.

2nd is the NTC Thermistors. BTW, these make interesting fuses. I blew up a couple in testing. Blew up is not an exageration either. $6.00 each. They are not much more expensive than a good fuse. Oh well. $12.00 education, priceless. ;-)

3rd. is the fuse block. Notice that I doubled up on the 12 GA wire. I can't get 10 ga or 8 ga in small amounts with high temp insulations locally. This is THHN building wire. Stranded. Courtsey of Home Depot of course.

4th is the top of the caps. I went with my triangle mount again. This is really stable. Note the bleeder resistor and the LED. I like the LED so I know the caps are still hot. The bleeder resistor was selected to be slightly higher than the Geckos this will eventually be hooked up to, at idle. That ensures the bleeder does not see current until the Geckos are totally off line.

Had an interesting learning experience.

So, if you charge up a cap and discharge it by putting a resistor across the terminals you'd think it was really discharged right? Checking it with the volt meter shows nothing. Discharged right? Not! Try hooking it up to another cap that has never been charged or has been sitting around for a while. I got a little spark and backed off quick!

Measured the potential between them and I got 8 volts. Not a big deal except that these are 20,000 uf caps. So there's a bit of juice in there.

So I put (+) to earth ground on all of them (across a resistor, 100 ohms I think.) to get them equalized before I hooked them up again. All is good. No sparks!

As always, be careful, this stuff can kill you. Seriously.

Best,
John

[jdelaney44]

See attached.

Regarding the costs.

Don't forget shipping and tax..... Shipping on the transformers is a lot. They are heavy.

Also don't forget an allowance for hardware and some money for the mounting plate. This was built on a piece of surplus 6061 that's about 0.50 thick I think.

You also might want to allow for blowing up a few fuses and such.


Best,
John

PS -updated file on 2/25/07 with wiring diagram and a couple of corrections. I left the bleeder resistor off the schematic and did not have a spec listed for R3.

[jdelaney44]

Did some more figurin'
The formula for the transformer current is -

Ip = Is * winding ratio

winding ratio (WR) = Vs / Vp

WR = 50 / 115 = .435 (115 is from the spec sheet)

Ip = 16 * .435
Ip = 6.96 = +/- 7 amps for each trans.

I am using two so 14 amps. So at 20 amps we have 6 amps of safety or about 43%. A little much, but the next size down is 15 amps and that's too close.

Correction to above: The max continuos rated current of the four windings in parallel would bE 32 Amps. 8 A per winding. This is off the Avel Lindberg spec sheet.

But as I noted, running at 20A got things plenty warm and I think you would need to pay extra special attention to cooling at 32 amps. I'm certainly no cooling engineer, but it seems obvious after the bench testing.

Best,
John

[jdelaney44]

Quick update. Few weeks back I installed the new transformers on my power supply. I'm still experimenting with the acceleration settings in Mach 3 but it looks like I'm going to be able to get a stable max ipm of around 50 ipm.

So bringing the voltage up from ~56 to about ~72 made a difference.

Recall that my motors are 100V motors. So this confirms for me that making sure your motors are close to the max voltage of your drives and power supply is very important.

If I get some cash, I may switch out the motors for something around 70 - 80V.

I want to throw something out there too. The common concensus is that you only need linear or unregulated power supplies. I want to start making the case for regulated supplies. When I load tested my last power supply build, the voltage dropped significantly under full load. There's a lot going on there I know. And I have a ton to learn about power supplies still. But it seems as if regulation might not be such a bad idea. Just a theory at this point. Yes, it's a cost trade off too. Getting into a regulated supply opens a can of worms that might be just as well left shut.

There is also the very true point that the servo drive regulates the effective voltage, or maybe power is a better term, very well so further regulation is not required.

Best,
John

[mikep_95133]

Quick update. Few weeks back I installed the new transformers on my power supply. I'm still experimenting with the acceleration settings in Mach 3 but it looks like I'm going to be able to get a stable max ipm of around 50 ipm.

So bringing the voltage up from ~56 to about ~72 made a difference.

Recall that my motors are 100V motors. So this confirms for me that making sure your motors are close to the max voltage of your drives and power supply is very important.

If I get some cash, I may switch out the motors for something around 70 - 80V.

I want to throw something out there too. The common concensus is that you only need linear or unregulated power supplies. I want to start making the case for regulated supplies. When I load tested my last power supply build, the voltage dropped significantly under full load. There's a lot going on there I know. And I have a ton to learn about power supplies still. But it seems as if regulation might not be such a bad idea. Just a theory at this point. Yes, it's a cost trade off too. Getting into a regulated supply opens a can of worms that might be just as well left shut.

There is also the very true point that the servo drive regulates the effective voltage, or maybe power is a better term, very well so further regulation is not required.

Best,
John

Hi John,

Voltage sag amounts to losing torque due to lost watts. So you are on the right track. Not all servo systems regulate the effective voltage. So a regulated power supply is a good upgrade.

Some time ago I posted alot of questions about using regulated supplies. It was obvious that most thought that they were a waste of time. Even a commercial unregulated version that I found claimed a 10v drop at full load. I can't remember if I posted it here or in the Mach3 forum.

I agree with you that regulated power supplies are very important to a well regulated and accurate system. One way to do this was to find a rack mount supply already working that just needed to be wired in. That was much cheaper and easier than trying to design and build regulated supplies. My system uses 5 amps continuous and 10 amps peak dc output per axis. There are 3 axis. So that makes the AC side pull up to about 15 amps peak from 120vac. The peaks don't happen very often. The worse that can happen in a regulated supply is that the voltage droops a bit at peak loads. If you can run a rack mounted regulated supply from 240vac, you won't have any voltage sag.

Good luck on your quest.

Mike

[jdelaney44]

Mike, thanks for the reply. I was beginning to feel like I was on an island on this one...

[mikep_95133]

Mike, thanks for the reply. I was beginning to feel like I was on an island on this one...

No prob. I proved that it works on my system, no matter what the nay sayers say.

Mike

[jdelaney44]

Could I get a review on the write up from someone? I tested the circuit so I know it works. But I might have my terminology screwed up. See the attached post.

Best,
John

Strike this! I need to refigure the math!

[lerman]

John,

You don't really need the 1.5K resistor. The series resistor (shown as 10K), serves to limit the current. I haven't checked to see if 10K is appropriate for the optoisolator you are using, but it is probably about right.

On the subject of power supply regulation, I don't believe that it is necessary for this application. The voltage affects the servo speed; not the torque. That's controlled by the current. If you are worried about achieving a certain maximum speed at maximum load, just use a higher supply voltage. In practice, though, maximum speed will occur when traversing at no load.

I could imagine that poor regulation could affect the servo tuning, but in practice, tuning is not that precise, anyway. Other things like varying part mass, material, depth of cut and so on have a much more significant effect.

Ken

Ken

[jdelaney44]

Ken, Thanks for the reply.

I put the 1.5k in there to form a voltage divider as you probably figured. That gives me a voltage of 1.56 according to the math. The LED is rated at 2V max. So you think the voltage drop across the 10K resistor will cut it?

Thanks,
John

[lerman]

Optos consist of an LED which has a specified forward voltage drop -- usually around a volt. They also have a specified current to actuate them. (Sometimes they specify a current transfer ratio. It might be 10:1 for a single transistor type, or 100:1 for a darlington type.)

So, if the diode drop is D (say .7 volts) and the desired current is I in milliamps (say 1 ma.), then for a 12 volt power supply, the resistance R (in K ohms) is:

R = (12 -D)/I
= 11.3/1 = 11.3 Kohms.

In that case, an 11K resistor should be just fine.

Ken

[jdelaney44]

Took me long enough. I wanted to bread board this and test it. It works and by design seems to be right for the specific application on a Sound Logic Break Out Board.

Thanks for all the input and assistance.

Best,
John

[jdelaney44]

Update on raising the power supply voltage. It's working pretty well with the actual program. I am not getting 50 ipm running the program however. But I was able to increse my "F" settings to 15 from 9. Will be increasing it over time. I hopeful I can get to hang at 20.

I got agressive on the first try an took it up to 30 ipm, no good. It did experience some step loss.

This program took about 10 hours to run. It's now down to between 6 - 8.

There are something like 25,000 lines that were generated by Visual Mill. I think I need to work with that some more too.

Still looking at the motor swap however.

Best,
John

[jdelaney44]

So my wife named the machine Rosie. Presumable after Rosie the Riviter. Not the other Rosie. Anyway....

Attached is a short clip of Rosie doing a finish pass on a foam trial for what I hope will be a sand casting pattern eventually. This is part of three hours worth of video I am cutting down for an educational DVD for a younger cousin of mine. She wanted to stay and watch it all work, but had to go home to NY from CA. So I'm trying to get her the video.

While I was at it I thought I'd put something up here showing actual motion.

As you can see she is capable of rather nice rounds and curves.

The shop vac hose is mounted to a mag base not far out of the shot to suck up all the foam dust.

This program is around 25,000 lines. There are a ton of X,Y,Z moves all the time. I need to work on the CAM settings to dampen that. It's creating moves when it really doesn't need to.

Not bad for .002 or worse backlash, mis matched motors, and a kludge of other stuff that this thing is made of I guess.

So if you are thinking about doing one of these, the results are worth it. I think so anyway.

One of these days I'll post more video over at my own web site.

Sorry for the crappy video. Working on getting the file size up to the 500K limit.

Best,
John

[jdelaney44]

So, after pricing out having my Bridgeport moved to a shop I recently rented and getting back prices from $500.00 - $2,000.00, decided to give it a go with rented equipment. I also had some time constraints with my travel schedule so getting this done on a Sunday was worth some extra $$.

Sorry no pictures of this yet and none of the actual move day. :-(

Anyway here's the gear:

5,000 lb capacity warehouse forklift
JLG "Triple L" hydraulic trailer (for the forklift)
Stake side F-Superduty Ford truck
A very understanding Wife
My best working partner ever, dear old Dad

All rented from United Rentals.

In retrospect the JLG trailer could have been replaced by normal tilt bed equipment trailer.

Plan A was to drive the lift and the machine onto the trailer and tow the whole shootin' match across town. The JLG trailer had a capacity of 10,000 lbs. The forklift weighed 9,200. The machine is close to 3,000 with the CNC gear on it. No go.

The JLG trailer's bed drops flat on the ground. No tilt and a small ramp on the end. Very nice rig.

Plan B was to use the forklift as a crane to hoist it on to the back of the truck. That's what we went with.

Note, I was way to chicken to try lifting the machine from under with the forklift. It seems too top heavy for this. Maybe with a bigger lift.

The first challenge was getting the Bridgeport out of the garage. There was some big show at Disneyland (really) and all the lifts in the area were is short supply. So the short ones I could drive into the garage were all rented out.

I was able to get my hands on some skates at Wright Industrial Hardware over in Costa Mesa. Actually my wife was good enough to go get them for me since I was out of town. I'll post pics of those later.

She also tried helping me use a pry bar to work the machine up on blocks to get the skates under it. We got it up, but stopped and let it back down. Not especially stable or safe. Getting it on the skates was looking to be a dicey proposition. This was a week or two earlier when I moved some other gear with the same trailer as kind of a test run.

So after thinking about this and looking at the Bridgeport I decided to try using a floor jack to lift up the machine. But how?

On the front, my Y axis drive goes straight down. The bottom of the drive housing is low enough to where I can get the floor jack under it and lift the machine. I was a little worried about putting 1500 lbs. of load on it.

So I ran the table out until the Kurt vice was over the Y drive housing, braced the drive housing with wood between the Kurt vice and the top of the drive.

On the back decided to make a strut from the utility mount on the back of the ram to within 5 - 6 inches of the floor. Enough room for the jack. I had some thing wall 2x2 square tube from another project and some square 1/2 plates from my scrap pile.

I drilled a 1/2 inch hole in one plate to align with the hole in the utility mount. Then a 1/2 bolt was welded into the hole. This bolt was used to secure the strut to the utility mount point.

Then the plates were welded onto the ends of the tube. Got lucky on the tube length and did not have to cut it. Good thing since the band saw went with the first load to the shop.

So once the strut was made up and bolted on it was a simple matter of jacking up the machine. We jacked it up on one end about 2 inches at a time, blocked it, and jacked the other side. Did that twice on each end.

Then we put the skates under it and lowered it onto the skates. It's still on the skates so I can probably get a picture of that.

We used a ratchet cargo strap around the machine and under the skates to help keep them in place. This worked real nice when we picked up the machine.

Once on the skates we rolled it out of the garage by using some 18 ga. sheet metal strips for runners. The skate wheels are sealed bearing units and they get stuck really easily. It required a good shove, but we got it out. Had to help it with the jack at the slab joint between the driveway and the garage slabs.

Once out of the garage we could get the lift up to it. We took several measurements to make sure we had enough mast height on the lift to get the Bridgeport up on the truck. Oh my wife calls it Rosie.

The forks were put together in the center of the fork carriage. Mast up, carefully drive the lift into Rosie as close as possible to reduce the moment arm of Rosie's 3000 pounds.

Then two nylon tow straps were wrapped around the ram and the table and then the forks. Most of the load ended up on the ram strap. The table strap acted to stabilize the machine.

Did a very slow pick. Drove it to the stake side. Lifted it up, set it down on blocks to keep it off the skates, backed up the lift about 20 inches, picked it again and drove it further to the center of the truck. Done, on the truck.

Strapped it down, loaded the lift back on the trailer, headed to the shop. Reversed the process to unload and returned the truck, trailer, and lift to United. Done and dusted.

At present Rosie is sitting on her skates waiting to be placed in her final home.

Cost ~ $600.00 all tolled including the skates, straps and fuel. I could have had it done for that, but now I have the means to move her around the shop. I also have the confidence to move her again if need be. Next time it'll be faster and a little cheaper with the tilt trailer.

I can see where a beater used forklift can pay for itself quick.

Later,
John

[jdelaney]

More later. Got the X-axis problem above fixed. Put a 1uf MLCC cap on the HEDS encoder power terminals per a post by Mariss from Gecko. Re-tuned the G320 on the machine with a scope. Worked perfectly for two hours. Then I went to test more today and ... the motor shaft sheared, perfect. Well, I picked up a 3/4 hp motor a couple years ago in case I needed it. No time like the present!

https://youtu.be/5TLrlYtQer0

[jdelaney]

Here's the post where Mariss discussed the 1uf cap.
https://www.cnczone.com/forums/gecko...osition-3.html

[jdelaney]

After some thought I think the motor tuning on the machine caused the shaft failure.

I used a function generator to create an instant forward / reverse so I could see the wave form at the G320 test point on the scope. That's all fine when there is nothing being driven by the motor. But ... I left the belt hooked up so I was instantly reversing about 30 - 40 pounds of ball screw and 400 lbs. of bed. I'm thinking all that that inertia against the instantaneous torque of the motor, effectively at stall, was enough to start the failure. Didn't help that the 1/2" shaft was turned down to 3/8" and the inside of the shoulder is very sharp on these motors.

So ... when I go to tune the others I will remove the belts and sheaves. Found some replacement NEMA 34 motors @ CNC4PC and they are on order.

Live & learn I guess.