[boydage]

Hi,
incidentally can you confirm that the reduction ratio is exactly 60:1?

Craig

(smile) - you read my mind. Commercially made TB170 4th Axis with that in the specs. I am going to double check that today.....

[joeavaerage]

Hi,
one suggestion is to disconnect the servo from the reducer and then call a 360 degree rotation of the A axis. That should in turn rotate the servo by 60 revolutions, but it should stop exactly
in the same position. That would determine whether the servo is playing up or maybe the gear reducer is not a true 60:1. This last possibility would not seem very likely, but you have to ask the question.

That breakout board I think has 5V outputs? 5V is not enough for signaling the drive with open collector inputs. Do you have some sort of amplifier (BJT or MOSFET) at each output of the
breakout board? Open collector signalling is 24V with Delta drives. The same inputs are used but sans a current limit resistor which means that a 5V differential signal is enough.

Differential signaling requires two wires for each signal. They are usually twisted together. When one wire has a high voltage (5V) the other is low (0V) and at signal change they swap over.
I uses an AEIC7272 ic. It has four differential outputs, also called line drivers. It needs a simple signal input which can be 3.3V or 5V and it outputs that signal on two output pins.
The swing of the output pins is 2V less than the output supply voltage. In my case I used a 7V output supply and therefore the output swing of the differential (line) is 5V.

Craig

[boydage]

Hi,
one suggestion is to disconnect the servo from the reducer and then call a 360 degree rotation of the A axis. That should in turn rotate the servo by 60 revolutions, but it should stop exactly
in the same position. That would determine whether the servo is playing up or maybe the gear reducer is not a true 60:1. This last possibility would not seem very likely, but you have to ask the question.

That breakout board I think has 5V outputs? 5V is not enough for signaling the drive with open collector inputs. Do you have some sort of amplifier (BJT or MOSFET) at each output of the
breakout board? Open collector signalling is 24V with Delta drives. The same inputs are used but sans a current limit resistor which means that a 5V differential signal is enough.

Differential signaling requires two wires for each signal. They are usually twisted together. When one wire has a high voltage (5V) the other is low (0V) and at signal change they swap over.
I uses an AEIC7272 ic. It has four differential outputs, also called line drivers. It needs a simple signal input which can be 3.3V or 5V and it outputs that signal on two output pins.
The swing of the output pins is 2V less than the output supply voltage. In my case I used a 7V output supply and therefore the output swing of the differential (line) is 5V.

Craig

Thanks. Ok. Its def a 60/1 ratio. I have set my steps at 690 velocity 6000, acceleration a modest 600, with my gear ratio 16000/414 and as previously it is appearing nice and accurate to visual standards whereas it was losing about 2deg last night. I slowed the velocity down to 200, very slow but the same result. This is where it goes bad for me it will not retain this accuracy and will start losing position at some stage, because I am going around in circles (gee literally haha). I just dont want to destroy another part which is an issue here. Nothing worse than not trusting equipment.


Anyways. I am going to continue and I am going to upgrade to differential signaling because that appears to be a way forward. I do have the grounds connected on my transistor board, using the 5v signal on the base to get the 24v signaling. It felt weird to do that but its how it was drawn up and is working. Seems open to noise this way? I have just noticed I have not earthed the chassis of the Delta Driver so will do that too as the motor is grounded through that.

Did you wire up as per C4-1?

[boydage]

Yep 5 hours on it today. Unuseable even at 0.5rpm it lost a deg on 3 rotations. Im pulling it off and installing a Nema 34 which I know will be accurate. A shame really I thought I was stepping up not down. Sorry Craig I know you are happy with your setup but I cant get a response out of Delta and its either a faulty unit, or its not suitable for an ESS setup, or I have done something wrong. Expensive piece of rubbish is how I feel about Delta now especially with their lack of support. We learn and sometimes the expensive way.

[joeavaerage]

Hi,
you have a signaling fault. By all means swap out the servo, but you're going backwards. You spent big money to get this servo and it will eat any stepper you put it up against, you
should persevere and get it working.

Delta and its either a faulty unit, or its not suitable for an ESS setup

F****ing Rubbish!

Post the circuit you have used to drive the Step Dir inputs of the servo. What pins in the CN1 connector have you used?

Did you wire up as per C4-1?

Yes, page 3-30. Note that it is for 5V differential signaling. Note also that the differential signal drive the input photodiode and is in no way referenced to either the 0V COM of the
servo drive nor the 0V COM of the breakout board and therefore at least for the SIGN and PULSE signals the two 0V COMS need not be connected. However for the ENABLE, and RESET inputs
they are required to be connected.

Pins 37 and 39 form the SIGN/-SIGN pair and pins 41 and 43 form the PULSE/-PULSE pair.

Your arrangement should be as in C3-1 on page 3-28. Note that is this arrangement the 0V COM of the breakout board must be hardwired to the 0V COM of the servo drive.
Do you have pin 14 of the CN1 connector wired to the 0V COM of your breakout board?.

I would presume you have connected pin 17 to pin 35 (in CN1) to pull one side of the SIGN and PULSE photodiodes high? Likewise I have to assume you have hooked the collector of the amplifying
transistor to pins 37 and 41 respectively?

Craig

[boydage]

Gee. I have 16 parts to manufacture and almost, almost shelved the Delta and put a Nema on to get them done. Thanks for the words Craig.

Ok. I put some proper troubleshooting into the system last night with the mindset its almost impossible there is a fault in the new components. The Delta display can be set to show pulses which is the indicator and a feature I didn't know about. 1deg shows exactly 691pulses. Back to zero 1 pulse (should be zero but more to come)

I ran a series of positioning moves, some with a 30sec pause. My first issue, during the pause I saw a couple of non commanded steps received by the drive. Then, when it got sent back to zero, it was displaying 318 pulses. I have noise in my circuit and a lot of it. That cycle, 318 pulses infact which is the exact amount of inaccuracy. I did try some filter settings on P100 but no fix. Not quite sure where that noise is coming from as its well shielded and I have iron ferrule rings around my conductors (if thats the name for them) I did order a AEIC7272 last night but it will be a few weeks to arrive. I wonder what else I can do in the meantime.

I will draw up my diagram later today I need to work today. It is to the letter though. I do have some PC817 Optos here I was wondering if they might suffice as they would completely isolate the BOB in the place of sharing the ground in a 2222 circuit? But, regardless, having something to troubleshoot in proper focus is a way better feeling huh?

[joeavaerage]

Hi,
yes that is all consistent with a signaling error. While that may be troublesome to you, I presume also its much better than if there were a genuine fault in the servo.

I will draw up my diagram later today I need to work today. It is to the letter though. I do have some PC817 Optos here I was wondering if they might suffice as they would completely isolate the BOB in the place of sharing the ground in a 2222 circuit? But, regardless, having something to troubleshoot in proper focus is a way better feeling huh?

No, it is not necessary. The input to the servo drive are photodiodes, but it does require that the 0V Com of the drive be connected to the 0V Com of the breakout board, and that is perfectly OK.
I've been running Delta servos like that for over ten years.

I presume you have a current limit resistor in the base circuit of the 2N2222 BJT? But do you have a pulldown resistor? It is common, even normal, for a breakout board to have a low state output
voltage of 0.5V to 0.8V. That is uncomfortably close to having the BJT turn on despite your wish that it turn off. I have used BJT's but you do need to pay attention to the interface to the BoB.
I now prefer a MOSFET, a 2N7000 is as good as anything. Requires no current limit resistor, nor a pull down resistor. The Gate Threshold voltage of a 2N7000 is typical 2.1 V, well and truly clear
of any residual low state voltage of the BoB.

Single ended signaling (also called open collector) means that you have a single wire to each drive input (unless you can twist it with the 0V Com wire), and that single wire is both
noise prone and inclined to not reliably switch with excess stray capacitance. The wire should be short. I had single ended signaling operating reliably with Delta drives at 200kHz with a wire length
of about 600mm....so you don't have to be anal about it, but having the drive on the other side of the room from the BOB is not a good idea.

Craig

[joeavaerage]

Hi boydage,
I think the short term answer is to use a pull down resistor with you existing 2N2222.

As a longer term solution I would use this circuit. Its based on the AEIC7272, but has two non-inverting Schmitt trigger buffers. The buffers ensure that no 'flakiness' of the Bob
can do any harm. Note I have placed some 2k2 resistors as pull downs at the input of the buffers. Leaving them out leaves them at high impedance and thereby subject to noise.
The best way to control noise is to pay attention to impedance throughout your circuit. Very high impedances are subject to dV/dt noise whereas very low impedances are subject to dI/dt noise.
Terminating a circuit in a known reasonable resistance gives it the best possible immunity to noise without all the ferrite-ring/shielding/anal grounding palava.

The AEIC7272's you ordered....are they SIOC packages? I rather think they are, I'm not even sure that you can buy them in DIP packages, in fact the data sheet does not mention DIP packages.
I presume therefore you have given some consideration how you are going to use them? SOIC is just small enough that soldering direct to the leads is near impossible.

Second issue is that the AEIC7272 requires two power supplies. One is the Vcc Bias supply, ie the regular 5V from your BoB, but also another supply (sharing a common ground)
known as Vcc Drivers for the outputs. The output voltage swing is within 1.9V of the Vcc Drvier supply. Thus if you want 5V output swing then you need a 6.9V (7V amongst friends)
supply. Its a bit of a drag having yet another supply, but the sureity of signaling the servo drive makes it worth it.

Craig

[joeavaerage]

Hi boydage,
just as a heads up, if you do use differential signaling along the lines of the circuit I've posted be aware that you can hook up the two wires to either the SIGN or PULSE (Direction and Step
respectively) backwards. It will do no harm, but it wont work. You can try reversing the ActiveHigh/ActiveLow in Mach4 but it will not change the situation.

STEP+ in the above circuit must go to SIGN (pin 39) and STEP- must go to /SIGN (pin 37)
DIR+ in the circuit above must go to PLUSE (pin 43) and DIR- must go to /PULSE (pin 41)

It took me a couple of very frustrating hours to work out this little detail!!

Craig

PS: Whoops...just noticed an error in my circuit diagram, namely I have the Step+ and Step- labels switched. Only really makes any difference when you try the get the drive to work,
if it does not work one way then swap polarity. Just a detail that can catch you out. As I have drawn it:

STEP+ in the above circuit must go to SIGN (pin 39) and STEP- must go to /SIGN (pin 37) should in fact read:
STEP+ in the above circuit must go to /SIGN (pin 37) and STEP- must go to SIGN (pin 39)

Sorry for the confusion.

[boydage]

I appreciate your help I know it takes time to do that. Its been a ride and yes correct I am stoked to have something I can focus on now as it was invisible to me preciously. Out of interest I adjusted my 2N2222 circuit and even after installing a pulldown resistor I still got noise or un-commanded pulses. And, infact when I enable the servo it pulses 2.

Noted on the above messages I will be building it as such. Gee also noted on the 7V I have a LM317 regulator here which I will use. Thats now 6 different voltages in my machine.

Here is the thing I had some PC817 optos which I wired into a small board. Now with the grounds totally isolated it was perfect, all except that I am unable to get any kind of speed I would expect. But for now, that's absolutely fine because it is as accurate as I need it to be. I will be smart in my programming until the parts arrive for the differential signaling which I will sort in 3 weeks or so. Also am pretty happy with the rigidity of the 4th axis I do need to be careful with some of the more agressive operations in certain directions - quite gnarly just how much forces are pushed into the workpiece?

Agree the SOIC is a bit of a challange I purchased two just for that reason. I should be ok though with a nice new tip for my soldering station. Craig did you print a PCB for your own setup? A young colleague of mine has his made offshore for some of the electronic inhouse components we manufacture. Quite cool really. I have often thought about milling my own out but in reality I just dont have the time with a very diverse home business and a 4 day week employment. It removes some of the fun when upgrading my machines as I need them done for an actual job. I went for a 4th axis over an auto tool changer and reckon its a much better investment.

Curious if you have a brake on your 4th? I have built a hydraulic station, with SSRs to run it, operate the brake, with sensors going to Mach 4 to be controlled by a M10 and M11. It is working ok but I am considering using the Delta drive to take control of the brake in the place of Mach 4. I will keep the sensors though and write a monitoring code if in the case the brake fails because I found that a 12mm endmill on a reasonable cut is actually able to move an unbraked 4th axis even with the 60/1 gearing on the servo.

[RaderSidetrack]

That a fact? Would you care to share your findings?

He won't be responding to your query. That was a Spammer, just posting nonsense to disguise his spam link (in very small type -- hard to see). He has been banned :nuke and his original post removed. And I edited your post above to remove his spam link from the quote in your post.

[boydage]

He won't be responding to your query. That was a Spammer, just posting nonsense to disguise his spam link (in very small type -- hard to see). He has been banned :nuke and his original post removed. And I edited your post above to remove his spam link from the quote in your post.

Wow a new level of desperation huh? Penis enlargment pills? I guess that takes that particular loser to a new low.

[joeavaerage]

Hi boydage,

Craig did you print a PCB for your own setup? A young colleague of mine has his made offshore for some of the electronic inhouse components we manufacture. Quite cool really.

Yes, the circuit board is my own. I make it myself using my machine, that is and always was the major work stream for it. It has taken quite some time and practice to get it to the state of usefulness
as it is now. Unless you have 100hrs or so to devote to refining your procedures I would not bother. Having said that I will make a PCB for you. I am very practiced at them
now, and a new design to a finished PCB, of this size and simplicity, is under an hour.

Curious if you have a brake on your 4th?

No I do not. I did not in fact believe that a 19.5:1 reduction worm drive would backdrive....but it does. Should the servo depower or be un-enabled, the weight of the trunnion table can cause it to backdrive.
Thus when I require the trunnion to be stationary it is the servo that is required to hold position, and this it does rather well, but is still somewhat 'springy' by comparison to a hydraulic brake.
The bigger issue is the C axis platter. It has a 6.75:1 worm reducer and it does backdrive. With say 20Nm applied to the platter it will cause the platter to rotate half a degree or so before the servo kicks in and
tries to restore its commanded position. Having said that under all normal operating conditions I have not seen the platter deflect in that manner.

There is no opportunity to build and fit a brake and yet retain the current size. I could certainly do so on the trunnion as either end of the trunnion is available to me to fit a brake disc, but not so the platter.
It would require a major re-design and almost inevitably a 'higher stack height' which I would not be prepared to tolerate. That would be a case of tying to solve one source of compliance only to invite a worse
source of compliance. Compliance is this context is equivalent to 'springiness'.

Craig

[boydage]

Hi sorry for the delay I work around NZ with the end of day sorting out paperwork. The neonatal intensive care unit in Hamilton is getting an expensive birthday so I am running that.

That would be a fantastic result if you would be able to supply a board? Let me know any costs and I can send you a courier bag too. Details will be easier via email [email protected] The good thing being you already have the design :-)

I see your point re brake? I also see a good design challenge right there for sure. Working in medical I managed to score a hydraulic pump out of an operating table for mine although needed to derate it all the way down to 500psi for the spec of the 4th axis. I have been working with a piece of equipment lately that has swinging arms with the latest mod being a brake on those. Its just a tiny stepper on a acme screw that pulls the brake on with a rudimentary strain gage to give feedback. Very low profile and electronically controlled without oil is nice.

As for me I have managed to get through a 16 part order about halfway. Where it was a 5 x flip operation that took over an hour per part, is now a 2 operation part on the 4th axis that takes 20min. I am now just trying out just how aggressive I can be with that 4th axis before it moves on the brake. Also noting the challanges of programming so I dont smash into the 4th axis. I went for the 170mm model for its weight and rigidity. If I want my gantry to pass over the top I need to raise it by 40mm which is not a huge job but one all the same. I guess I will leave it on the Y axis for now.

[joeavaerage]

Hi,
I anticipate that the majority of the five axis toolpaths I will generate are simultaneous five axis, in which case brakes are not used. The servo and gear reducer must
withstand the cutting forces. This is what I designed for and while the capacity is low, say 16Nm on the C axis, it is adequate and showing good rigidity relative to its size.

Brakes on both A and C axes would be nice for when the rotary axes are fixed, say a 3 axis path, or a 3+1, or a 3+2 path, they are all possible with the setup as is. Could I make it better....
yes....probably I could. Am I going to bother?.....no....not in the immediate term. I'm hoping to add ATC before spending more time and money on the trunnion/platter.

Craig

[joeavaerage]

Hi, boydage,
I did not have any AEIC7272 IC's left. As it turns out I need some for a breakout board I'm making for a friend (actually employee) whom is building
a machine......so I ordered some on Monday 5/2/2024, I've just taken delivery of them this morning (Monday, 12/2/2024) which is pretty damned good in that they came from my 'local' Element14 warehouse
which is in Sydney.

The board I've already designed and will mill it out today (probably).

Craig

[joeavaerage]

Hi,
made the board, populated it and went to test it. They have supplied AEIC7273, not AEIC7272. They are open collector line drivers. Don't know whether I ordered wrongly
or they supplied wrongly, but either way the IC is wrong. Be another week before the right ones turn up.

Craig

[joeavaerage]

Hi,

the IC has arrived and it works fine.

Craig