Micro-controller controlled Pneumatic Power Draw Bar

[jcdammeyer]

This is what I was thinking might work? As the inside flats of the socket out of alignment by 30 degrees, will tend to slide down the curve in the direction of one of the edges. The angle can't be too shallow or I suspect it would then still hang up on the surface.

Comments?

 

[jcdammeyer]

This is what I was thinking might work? As the inside flats of the socket out of alignment by 30 degrees, will tend to slide down the curve in the direction of one of the edges. The angle can't be too shallow or I suspect it would then still hang up on the surface.

Comments?

Update on this. I first tested a 45 degree and then 60 degree angle on a 17mm bolt head. Seemed to make it slip on more easily. So I turned a 60 degree angle on the 19mm draw bar bolt head. It does slip on more easily but still when off center the socket comes down so hard with the pneumatics that it hangs and I've seen the hardened impact socket actually leave a mark on the softer draw bar bolt.

I've ordered a 12 point socket. And I'm going to take a dremel grinding tool to the inside sharp edges of the 6 point socket to round them so they won't hang up but will slide. Won't affect the overall socket operation. More when I've tried that.

 

[jcdammeyer]

The newest problem. I've ground the inside edges of the impact socket. Since it's a 3/4" on a 19mm head (really 18.6mm) certainly loose enough to slip over easily when lined up. On another group it was suggested perhaps giving the wrench a blip to spin it as it's pushed down that the spinning socket would self register. It does. And the few times it might not as soon as the power cycle starts the wrench, now with bevelled head on bolt and inside socket has no trouble going down.

Not tight to lift off either but, as an impact socket, is heavy. And the air cylinder really moves up very fast and that results in the socket being left behind occasionally. Worse is if as it goes up it pulls halfway off the wrench while the wrench is still turning. Been hit in the chest and shoulder with a flying socket. Not a good thing.

A 3D printed guide/guard will help that. But it looks like I will need some sort of better retention on the socket. Not sure if the 3/8" drive is hardened. Suspect it is.

 

[jcdammeyer]

Just a quick update. Turns out the replacement registration pin I made for the wrench was too short. The finishing nail I trimmed looked like it protruded high enough above the gasket but I hadn't actually gotten it into the bottom hole deep inside the wrench. So once the wrench was operated a few times the outer body, not held from turning by the pin, rotated enough to almost completely prevent CCW operation due to the blocked air passage. I kept adjusting software not realizing it was mechanical.

So I now have a new pin made from a 0.104" diameter 2.5" finishing nail turned down to the correct 0.084". The shorter nails were 0.071 and I didn't have a longer one that diameter. So I center drilled the big end, turned off the point end and center drilled that too. Then shaved it down on the little Unimat DB200.

Now I have full power back in both directions. Everything is now on hold until KMS Tools delivers the 12 point deep socket. I drilled a relief in the side of the 3/8" drive. Ultimately I'll do as one of the other members and use locktite to prevent the socket from dropping but not enough to prevent that little bit of wiggle for guiding onto the nut.

Now the other issue was the socket spinning too fast once the draw bar was loose. This is a standard operation of air tools. So the guide plate demonstrates that a bit of resistance can slow it down along with PWM on the air valve. Ultimately I think I'll add a brake of sorts to keep it from spinning too fast.

And it looks like I will need a wrench down switch before I start turning. I've found it does too much spinning on the top of the bar mangling it a bit. The 12 point will help but it may not be enough because if the socket doesn't go down then the magnet on the socket isn't picked up by the hall effect sensor and so it tends to over spin the wrench and I get way more than 2 turns on the draw bar. Sometimes the R8 even drops out.

More when I have it.

 

[Susquatch]

Dumb question - seems you are going to a lot of trouble trying to get a socket to mate up with the drawbar head.

Why not leave it engaged permanently and just spin the drive with the drawbar?

Alternatively, spin the socket with the drawbar but install a clutch between the socket and the drive? In fact, eliminate the socket and nut entirely and just put a clutch in. ???

 

[jcdammeyer]

Not a dumb question at all.
The bushings in the impact wrench are designed to be oiled with the supplied air. As such spinning them without that oil might damage them. Then there's also the tangs inside the wrench that under light load may well click back and forth making noise.

A clutch such as this: DC Clutch might well work but the size is prohibitive without a total frame redesign.

Eliminating the socket would then also require a redesign of the end of the draw bar. At this point, aside from the redesign, there still needs to be a way to manually remove it if something fails with it tightly in place.

Again the big issue is for TTS tooling I want no more than 2 turns in the CCW and probably 3 seconds in the CW so the impact wrench can reach 15 to 25 ft-lbs of torque. The other postings with various rotary power draw bar subjects all use R8 tooling and spin CCW until the R8 drops out. Watch the videos closely of the manually engaged versions and if the socket doesn't go down all the way there's a slight twist of the handle to blip the air wrench just enough to move it slightly without excess pressure on the nut. And in one thread the need for the 12 point socket was emphasized. So when a human is involved the slight behavior changes are handled without thinking.

But I also want the ability to swap R8 tooling so mine is a dual operation draw bar. So far I haven't seen any other threads or you tube videos that deal with that.

If I were to start over I'd look at a 300 oz-in stepper motor and probably a 20:1 planetary drive. A few days ago the drives on Aliexpress were all in the $39 range. Yesterday they were all in the $70 range. In either case, a change to a fully electric version still runs up to $200 with drive etc. And some major redesign.

I've found one of the bigger issues is when the socket doesn't descend all the way and it starts to spin the magnet isn't lined up with the hall sensor so it defaults to timed turning which polishes or marks up the top of the draw bar even more. I think a switch to detect fully down before turning might be the solution. Then if it isn't down I can move it up, blip it to turn a tad, and try again. With a 12 point socket likely it will drop into place quite quickly and then a 2 turn spin to undo it leaves the R8 in place and the TTS drops out.

 

[Susquatch]

Not a dumb question at all.
The bushings in the impact wrench are designed to be oiled with the supplied air. As such spinning them without that oil might damage them. Then there's also the tangs inside the wrench that under light load may well click back and forth making noise.

A clutch such as this: DC Clutch might well work but the size is prohibitive without a total frame redesign.

Eliminating the socket would then also require a redesign of the end of the draw bar. At this point, aside from the redesign, there still needs to be a way to manually remove it if something fails with it tightly in place.

Again the big issue is for TTS tooling I want no more than 2 turns in the CCW and probably 3 seconds in the CW so the impact wrench can reach 15 to 25 ft-lbs of torque. The other postings with various rotary power draw bar subjects all use R8 tooling and spin CCW until the R8 drops out. Watch the videos closely of the manually engaged versions and if the socket doesn't go down all the way there's a slight twist of the handle to blip the air wrench just enough to move it slightly without excess pressure on the nut. And in one thread the need for the 12 point socket was emphasized. So when a human is involved the slight behavior changes are handled without thinking.

But I also want the ability to swap R8 tooling so mine is a dual operation draw bar. So far I haven't seen any other threads or you tube videos that deal with that.

If I were to start over I'd look at a 300 oz-in stepper motor and probably a 20:1 planetary drive. A few days ago the drives on Aliexpress were all in the $39 range. Yesterday they were all in the $70 range. In either case, a change to a fully electric version still runs up to $200 with drive etc. And some major redesign.

I've found one of the bigger issues is when the socket doesn't descend all the way and it starts to spin the magnet isn't lined up with the hall sensor so it defaults to timed turning which polishes or marks up the top of the draw bar even more. I think a switch to detect fully down before turning might be the solution. Then if it isn't down I can move it up, blip it to turn a tad, and try again. With a 12 point socket likely it will drop into place quite quickly and then a 2 turn spin to undo it leaves the R8 in place and the TTS drops out.

I hear you. I'm just noodling the issues and asking myself what would I do.

Maybe that helps, maybe it doesn't. But at a minimum it provides other ways of looking at things. Starting over or switching to Gen II is always an option. "A wise man changes his mind often."

Here is a smaller clutch to consider. It's a riding lawn mower PTO clutch for the blade deck. Plenty strong enough and can be made much smaller than it is if you turf the pulley.

Amazon.ca

 

[jcdammeyer]

I hear you. I'm just noodling the issues and asking myself what would I do.

Maybe that helps, maybe it doesn't. But at a minimum it provides other ways of looking at things. Starting over or switching to Gen II is always an option. "A wise man changes his mind often."

Here is a smaller clutch to consider. It's a riding lawn mower PTO clutch for the blade deck. Plenty strong enough and can be made much smaller than it is if you turf the pulley.

Amazon.ca

Thanks. Brings back memories of equipment I worked on in the early nineties. It was a 30 Ton Flywheel press with a clutch and absolute encoder. The clutch was engaged and as the flywheel rotated the micro-processor was interrupted on absolute locations. It would then load the next location for an event until the revolution was completed. Depending on what it was doing it would release the clutch or do another rev.

These events did thinks like engage air valves or tell stepper motors to move. All to move IC lead frames under the press for trimming, bending and forming the leads of the ICs.

The issue of cost has to be factored. A reduction drive and stepper/servo motor can, in addition to controlled speed and position can also control torque. Assuming the curve says the motor has 288oz-in of torque then that's 1.5 ft-lb. If I want between 12 and 20 ft-lbs then a 10:1 planetary gear box would create 15 ft-lbs. If I set the current to create 12 ft-lbs and turn it slowly as it tightens then I get predictable torque. If I set to 15 ft-lbs for unloading then I can turn it slowly at full torque for 1/2 a turn and then quickly for 1.5 turns. But then it might be nice to close the loop and have an encoder on the back of the motor too.

That would allow fractional turns if the first try didn't move the socket all the way down.

Like you said, different ways of looking at things. Since I've done this kind of computer controlled hard for many decades it's the way I tend to look at it.

 

[Susquatch]

Thanks. Brings back memories of equipment I worked on in the early nineties. It was a 30 Ton Flywheel press with a clutch and absolute encoder. The clutch was engaged and as the flywheel rotated the micro-processor was interrupted on absolute locations. It would then load the next location for an event until the revolution was completed. Depending on what it was doing it would release the clutch or do another rev.

These events did thinks like engage air valves or tell stepper motors to move. All to move IC lead frames under the press for trimming, bending and forming the leads of the ICs.

The issue of cost has to be factored. A reduction drive and stepper/servo motor can, in addition to controlled speed and position can also control torque. Assuming the curve says the motor has 288oz-in of torque then that's 1.5 ft-lb. If I want between 12 and 20 ft-lbs then a 10:1 planetary gear box would create 15 ft-lbs. If I set the current to create 12 ft-lbs and turn it slowly as it tightens then I get predictable torque. If I set to 15 ft-lbs for unloading then I can turn it slowly at full torque for 1/2 a turn and then quickly for 1.5 turns. But then it might be nice to close the loop and have an encoder on the back of the motor too.

That would allow fractional turns if the first try didn't move the socket all the way down.

Like you said, different ways of looking at things. Since I've done this kind of computer controlled hard for many decades it's the way I tend to look at it.

When the gang first started doing automotive engine controls, they wanted to replicate the mechanical stuff in electronics. But how do you do things like timing advance (vac or speed) when the mechanical systems really are advanced ahead of the event? So some genius figured out we could delay from the previous point which would be the same as advanced from the current event. I argued that a lot could change in 90° of engine rotation. But the electronics guys won that round and we did it that way for a few years. It made their life easy when fast computer controllers only have to wait to act. Then the problems came home to roost and they started to listen a bit better. We switched to real time using position sensors and let the computers do what computers do best - a million calcs between counts. That works! Yup, those were the days.......

 

[jcdammeyer]

My ELS still does awesome threading with 1PPR under the restriction that the spindle speed as to be solid. Even then it did track variations on the smaller Sherline Lathe. Not to say the larger PPR isn't better but at the time I designed the ELS an encoder pulleys and belts were more expensive than the entire ELS kit which included a micro-stepping driver for the Z axis.

I designed a CDI multi-point fuel injection ECU for Honda VTEC that were used in home built aircraft and hovercraft. We only used two sensors: one at TDC and one at 20 degrees (I think) BTDC. One one rev an engine just doesn't change speed that fast.
Firewall Forward Engine Control

 

[Susquatch]

My ELS still does awesome threading with 1PPR under the restriction that the spindle speed as to be solid. Even then it did track variations on the smaller Sherline Lathe. Not to say the larger PPR isn't better but at the time I designed the ELS an encoder pulleys and belts were more expensive than the entire ELS kit which included a micro-stepping driver for the Z axis.

I designed a CDI multi-point fuel injection ECU for Honda VTEC that were used in home built aircraft and hovercraft. We only used two sensors: one at TDC and one at 20 degrees (I think) BTDC. One one rev an engine just doesn't change speed that fast.
Firewall Forward Engine Control

Awesome stuff. I do agree that engine parameters don't normally change that much in one rev. Trouble is that these were mass produced engines used by consumers all around the world. Sometimes things don't work so smoothly in the real, world. Low speed operation, cold starts, and misfires can change engine speed dramatically in just a few degrees of rotation. It happened often enough that we had to fix it with a much more accurate system that knew exact crank position in real time.

Funny that you mention lost spark. I was actually thinking about that when I wrote my previous note because it was another problem that came up for the same reasons. It was another one of those "it worked great in the lab" situations from the early days. Mechanical & vacuum advance systems moved the rotor plate as well as the spark timing signal via points, so rotor to cap position was compensated for almost automatically. But when they were replaced by a hall effect system (also before engine position control), that auto positioning advantage was lost and so was the spark! Unfortunately not lost well enough as it would sometimes go to a cylinder that wasn't at all ready for it! We found that one quite by accident looking for other things related to combustion emissions processes. I was on a cold trip up north looking at low temp performance of catalysts. I just happened to have a storage oscilloscope monitoring a lot of other functions and couldn't explain that one oddball signal that popped up at the oddest of times. Then the lightbulb went on! I wonder if.....

 

[jcdammeyer]

The only major problem I had, and our engineer was in the cockpit logging data with the pilot doing a dead stick landing, was subtle. Never showed up on the dyno so it may have been linked to a wiring/ground issue which of course we see a lot of in CNC systems too. We caught this one with the logged data. The system was doing A/D sensor acquisition on a periodic basis often enough so we could filter and get clean data. Seems that if a conversion was done at the same time as a spark event the electrical noise caused MAP values to change and therefore that changed the mixture and that stalled the engine.

A restart would have worked in this case and he could have flown it back under power. Changed the code to hold off A/D conversions during spark events and no more problems. The CD ignition was modeled after a SAAB patent and the coil I designed (in that one photo on the web page) could create as much as 300V even at 8V during very cold poor battery cranking.

The original ignition system (no injection) used dual PIC controllers and sparked the exhaust stroke as well as the compression stroke which is why the one photo shows 4 ignition coils. That was all a long time ago now when we still had to use UV Eraseable EPROMs.

 

[Susquatch]

That was all a long time ago now when we still had to use UV Eraseable EPROMs.

Hey, you say that like UV Eraseable EPROMS were a bad thing! They were AWESOME compared to the non-eraseable one-time ROMs and going back a bit further still the cold bootstraps that came before that!

Ah yes, the memories......

 

[jcdammeyer]

This was the precursor to what is now an Arduino. Monitor program in Mask ROM. Expansion bus with EEROM and 32K RAM added an RTOS I wrote and a Tiny PASCAL Interpreter written by the late Ian MacKay that could create tasks for the RTOS.

But the ability to flash code via serial/USB to a processor with more memory wasn't available for 20 years. We also had a 3 channel quadrature encoder card. 12 bit A/D card. But made in Canada means it would never compete with made in China.

 

[Susquatch]

This was the precursor to what is now an Arduino. Monitor program in Mask ROM. Expansion bus with EEROM and 32K RAM added an RTOS I wrote and a Tiny PASCAL Interpreter written by the late Ian MacKay that could create tasks for the RTOS.

I can't believe you kept all that stuff. I pitched all mine about 20 years ago in a big fit of anti-nostalgia. Then I pitched again about 10 years ago when we moved to the farm. I think what hurt the most was the s100 boards - especially all the a/d & d/a stuff. But then again, now you can get an add-on package for a tablet or laptop that will do all that better and faster for much less.

Out with the old, in with the new?!?!

 

[jcdammeyer]

I think what hurt the most was the s100 boards - especially all the a/d & d/a stuff. But then again, now you can get an add-on package for a tablet or laptop that will do all that better and faster for much less.

I still have all the old S100 boards and a couple of boxes along with my PERSCI 8" drive. An original 8" Turbo Pascal Disk etc. Probably about 10 years ago I tried to boot once of the machines and the disks no longer read. The graphics cards, including my wire wrapped video board are all worse than a Raspberry Pi or Beaglebone so the reality is I should really just put the stuff up on EBAY. Hard to let go and yet I know I won't do anything with it.

What I have on hand, like the dsPIC30F5011 based modules that I designed and built are way more powerful than an entire S100 mainframe. Funny eh? One of those to control my power draw bar. Here are 3 of them. The gold box xIM are the PIC based units. BIM has analogue, PWM, CAN bus, RS485, Relay drivers, and is set up for 0-36V input measurement along with thermister measurement. The B stands for Battery Interface Module and it started life as a BMS for large batteries. The other one is a GIM for Generator interface and controlled an ONAN diesel Genset with a small plug in board and that's what I'm using for the Power Drawbar. There is one more version not in the photo that has 6 more relay drivers and some I/O plus RS232 all on the RH connector instead of the A/D inputs.

Finally the white box (xMU) has another 16 bit processors (9S12) and has 5 CAN bus channels, 6 relay drivers, 6 dry contact to ground inputs, USB and an internal add on board with Real time clock and micro-SD for logging. The xMU one was designed in 2009 and initially a Rings Management Unit for the 2010 Winter Games in Vancouver. Now i use it for other projects to talk to other CAN devices like Motors, BMS, Solar Chargers.

 

[Susquatch]

I still have all the old S100 boards and a couple of boxes along with my PERSCI 8" drive. An original 8" Turbo Pascal Disk etc. Probably about 10 years ago I tried to boot once of the machines and the disks no longer read. The graphics cards, including my wire wrapped video board are all worse than a Raspberry Pi or Beaglebone so the reality is I should really just put the stuff up on EBAY. Hard to let go and yet I know I won't do anything with it.

What I have on hand, like the dsPIC30F5011 based modules that I designed and built are way more powerful than an entire S100 mainframe. Funny eh? One of those to control my power draw bar. Here are 3 of them. The gold box xIM are the PIC based units. BIM has analogue, PWM, CAN bus, RS485, Relay drivers, and is set up for 0-36V input measurement along with thermister measurement. The B stands for Battery Interface Module and it started life as a BMS for large batteries. The other one is a GIM for Generator interface and controlled an ONAN diesel Genset with a small plug in board and that's what I'm using for the Power Drawbar. There is one more version not in the photo that has 6 more relay drivers and some I/O plus RS232 all on the RH connector instead of the A/D inputs.

Finally the white box (xMU) has another 16 bit processors (9S12) and has 5 CAN bus channels, 6 relay drivers, 6 dry contact to ground inputs, USB and an internal add on board with Real time clock and micro-SD for logging. The xMU one was designed in 2009 and initially a Rings Management Unit for the 2010 Winter Games in Vancouver. Now i use it for other projects to talk to other CAN devices like Motors, BMS, Solar Chargers.

Pretty sure I still have a few old 8" drives too. But I knew even at the time that they would only be parts for other stuff. When you watch tape drives turn into way bigger 5.25 drives with 8" coming in the side door but never making it into the regular stuff because way bigger 3.5 stuff was arriving and then the first hard drives doubling every year and look where we are now! Omg! Bigger drives just empowered bloat ware.

And here we are - two old dinosaurs lookin at the kids and sayin - if you only knew where I have all been.......

Too bad the tree tops don't taste so good anymore......

I have no regrets.

Do you really think you can sell that stuff on ebay? I may have to open some of those drawers and start selling some of it. Better that then the kids and wife just trashing it when I leave this playground.

 

[jcdammeyer]

Do you really think you can sell that stuff on ebay? I may have to open some of those drawers and start selling some of it. Better that then the kids and wife just trashing it when I leave this playground.

It's become a bit of a nostalgic hobby for some. S100 Parts or from Bulgaria S100 in Bulgaria

 

[YYCHM]

Bigger drives just empowered bloat ware.

Boy... Ain't that the truth.

 

[jcdammeyer]

I heated up the 3/8" drive part of the deep socket to yellow and then let it cool slowly. No problem drilling and tapping now even if I did accidentally make the hole a bit too far down. The socket did not discolour down to the 12 point drive part so I'm going to guess it's still reasonably hard.

Next step is a sensor for tool down so the wrench isn't started until the socket is all the way onto the drawbar bolt. Then look at a new drawbar bolt head that can be hardened so it can't wedge the socket by deforming.

More with pictures when I have them.