[Cncstarter]

Hello:

I just ordered a kflop and had some questions regarding timing. I thought that I had figured it out before ordering but maybe not.

My intention is to use the Kstep initially and then move to the Leadshine closed-loop stepper setup when more funds were available. In my research I had seen the issues with the minimum step pulse for the kstep and other drivers but determined that the leadshines at 2.5 us minimum were within the spec for the kflop. This was the only timing number I had seen.

Further reading though has me wondering if the kflop will work with the leadshine drives due to a 5 us setup time for the direction line. Another thread states that 4 us is the maximum direction setup time though I can't find this figure in any documentation. Unfortunately I did not notice this sooner. Is this an issue? There is some mention of settings to comply with a gecko drive direction setup time but I can't find any specific timing numbers. Can the Kflop work with the leadshine drivers? Having a closed loop system isn't going to be much of a leap forward if it stumbles with every change in direction.

Hopefully this is a non-issue but if not could someone point me to a solution?
Thanks!

[wentbackward]

Unfortunately I think you are correct:

http://www.cnczone.com/forums/dynomo...ml#post1505472

If you want to continue using the KFlop for the interim period until you upgrade, you might be able to use a resistor/capacitor and an inverse schmidt trigger or similar to prolong the signal, but the top speed is going to be impared.

- Paul

[Cncstarter]

Hi Paul:
Who needs 256 microsteps at 3000 rpm with a stepper?

For all the customization options with kflop this seems a pretty basic over-sight. I suppose it's hardcoded in the FPGA rather than programmable. I did some looking around and the gecko drives were the only ones I found that specified a setup time less than 5 us for the direction line. The leadshines, leadshines rebranded, and leadshine clones make up nearly all the rest of the drives (really!) and every single one specifies 5us (or longer for a couple analog one), if they specify this parameter at all.

At any rate I thought about it some more and because an extreme stepping rate is definitely not required there's another similar solution to a schmidt trigger that might be a little more precise.. At 100khz, for example, there is plenty of time to simply delay the step pulse with a delay line (LTC 6994-2) since all a guy needs to do is move it back 2.5us relative to the direction line change and doesn't need to do anything with the direction line itself. 100khz would still be 1500 rpm at 4000 steps per revolution and would give me 5us either side of the direction change as per the spec. Might have to change the pulse direction but it works for me anyway.

Thanks for the response and I guess it was only a momentary panic attack...

[wentbackward]

It's definitely in the FPGA setup that the step time is configured. Link is here.

The interesting part to save you clicking through and reading it are...

FPGA(STEP_PULSE_LENGTH_ADD)=32 + 0x80;   // set the pulse time to ~ 2us and pulse the Step High

The pulse length may be adjusted from 1 to 63 of 16.67 MHz clocks. Which corresponds to 60ns to 3.78us.

So the max length of the pulse is 3.78us. Running at 2us is ok for the gecko's I'm using. God only knows what I'll run into when I try to convert my unbranded chinese made CNC router.

[Cncstarter]

Thanks wenbackward, I did see that parameter (and checked the specs) before I bought the kflop.

There is another related constraint though that is not documented on the website and that is the interval between the direction line changing and the step pulse. All of the leadshine related drives that specified this figure use 5us (or more). The gecko's were the exception at 200ns but there is a note with the documentation you quoted that even for these the step pulse might need to be set inverted to work.

I'm just speculating but it looks to me like the FPGA is set to change the direction line at exactly (or nearly) the same time as the leading edge of the step pulse which by itself would be incompatible with everything that has a setup time. By setting the drive to step on the trailing edge of the step pulse the setup time for the direction change is the same as the pulse width and would inherit the limit of 3.78 us from the step pulse setting. By delaying the entire step pulse with a delay line more time would elapse between the step triggering and the direction line change giving the driver time to cope. So long as the step rate at the time the direction line changes is less than 100khz there would also be lot's of time for everything to settle given a 5 us parameter (x2 gives 100khz). The FPGA is much more difficult to change then a c program in the dsp but there really should be oodles of time for the kflop to wait since the stepper pulse rate should be quite slow when it is changing direction -- slamming into reverse at 5000rpm is going to leave some skid marks.

If you look up the chip I referenced Linear has an easy spread sheet that calculates the required resistor value for a given delay. A couple bucks more than a schmidt trigger type circuit but much more elegant. I should note though that I do not yet have the components to test this but it's a very simple solution. By leaving the width of the pulse alone and only putting it of a couple uSecs relative to the direction change the overall step timing isn't really changing. Again though this would not work if the pulse rate at the time of the direction change is more than 100khz or a 10 usec period, and the circuit will ignore pulses occurring faster than the delay time (a plus if it is noise).

Thanks!

[TomKerekes]

Hi cncstarter,

KFLOP was targeted and designed for systems with high step rates (up to 2.5MHz) so it was an oversight on our part not to provide longer step pulses or more direction setup time. The Step pulse is delayed 0.25us from a direction change and the step pulse size is programmable from 0.06 - 3.78us. This provides a maximum direction setup time to the trailing edge of the step pulse of 4us. I believe the Leadshine specify a minimum of 5us. Actually they specify a requirement of more than 5us direction setup time to both the leading and trailing edges, but that doesn't really seem to be the case. I suspect it is only a requirement to the active edge. I'm not aware of any other mainstream drives where this is an issue. We may add more setup time in a future code release but for now that is what it is. If you would like to return your KFLOP for a full refund contact our Support by email.

Regards

[Cncstarter]

Hi Tom:
I was curious and did quite a bit of poking around and couldn't find any drive except the gecko that specified anything less than 5us. Many didn't have any timing information at all of course. From what I could see any digital drive that runs with a processor (rather than an FPGA, or CPLD or whatever type of programmable logic) may have trouble with the timing.

At first glance this did look like a show stopper and I was worried it wasn't going to work without a lot of messing around.

It dawned on me though that the step rates are going to be low at a direction change and that the pulses can simply be delayed without any sort of logic attached to the direction line. If all the axis are delayed equally no one would be the wiser. I think the chip from linear will work fine and by the time I save up enough for the drives and motors you may have updated the system and made the problem go away....

I can certainly understand the need for a high aggregate speed rate for all axis but don't really understand the point of 2.5 mhz to a stepper motor. At any rate taking in to account the 5us before and after a direction change that still leaves around 100khz, worst case, which is way plenty enough for me. I think longer step pulse durations and the ability to program the delay between the control signals would be very useful. With all the other configuration options it would make sense. I do understand though that there is more to it than changing a couple lines of code.

Thanks I think this is solved!

[bhurts]

I have been reading along on this thread for a few days. It seems all this discussion is focused on the idea that sometime in the future you may want to switch to leadshine closed loop steppers. You are going to use kstep until that time. If you read through the dynomotion forum both here and at the Yahoo forum which has much more info you will see what drives have been used with k-flop successfully. You will also see where a few people have run into drives that are incompatible.

As for 2.5mhz being to fast for stepper? Remember k-flop is not a smoothstepper. It is a general/multi purpose motion controller. It (sometimes with the use off expansion boards) will control just about anything. It is used for servos brushed and brushless by quite a bit of the people using it in addition to steppers. It can run analog servo amplifiers as well allowing re-use of existing servos and amps on older industrial machinery. It also can take direct control of the servos and close the loop inside k-flop directly.

I suspect you will probably be fine with kstep and not need to upgrade at all. If you do upgrade I'm sure the closed loop steppers are great. You would probably be better served at around the same price skipping the steppers and getting servos.

Ben

Sent from my HTC6435LVW using Tapatalk

[Cncstarter]

Hi Ben:

The same parameters apply to Servo drives as well as steppers. All of the processor-controlled digital drives are going to have timing parameters and from a brief look around the servo drives varied a lot more than than stepper drives in terms of requirements. I am sure there are high end servo drives that take pulses in the microwave spectrum but the common (and cheaper) ones were all about 500khz max -- roughly twice as fast as a stepper drive. Some of these (leadshine again) still had setup requirements of 5 us even at 500khz though curiously they did not have the post-pulse timing requirement so Tom may be right that this can be ignored.

I don't think digging through entire forums to see if someone might or might not have got something to work is a good way to select tools. Maybe I am old fashioned. I don't have time to read every post. Even if something did not work it could be more about the operator's abilities than the product's. Half the time the forum questions have nothing to do with the (eventual) forum answers so thread titles are often meaningless. I had a hard time finding any detailed comparisons between the smoothstepper and the kflop, for example, so I don't think there is enough quality information on forums to make any informed choices. For the record I could find no mention of the drive system (let alone specific part numbers) I am interested in linked with the kflop before I ordered it; I looked. Off the soapbox...

I am sure a large number of applications with the kflop are going to interface with an external drive (servo or stepper) using the step/direction interface. The timing requirements should be stated clearly along with current, voltage etc. I was able to find the maximum pulse width parameter in the setup information on the website but could not find any other restrictions.. All of the better quality drives will have a datasheet that state these things unambiguously and it would be helpful to have matching data on the dynomotion site. The electronics industry is built on this idea. With the flexibility of the kflop I didn't expect this to be fixed value.

I may never order the closed loop drivers and motors.. An over-due client may pay up this morning and I could order them this afternoon. I posted this as I was worried that this was going to limit my choices before the thing had even arrived. I did consult briefly with Tom concerning the built in closed-loop with kflop before I ordered it. He was very helpful but the leadshine setup still has some appealing points that the kflop will not do. I can use the kstep for additional functions so it would not go unused if I use the closed-loop drives and motors for the main axis.

Sorry for the long winded response...thanks

[bhurts]

I was not trying to compare the k-flop and smoothstepper where step dir is concerned. I was just saying that it offers more choices then just step dir output. With kanalog you can run analog drives with k-flop controlling the loop and the drive is open loop. With snapamp you can control 2 stepper motors, two brushless servos, or 4 brushed servos. I am just saying that k-flop is not just for step dir. Many of the people using it are controlling servos of some form or another and k-flop is closing the loop. It is also capable of double closed loop with encoders and scales. I have read up on the leadshine closed loop steppers and in my mind it is more of a fad. I would think that servos would be a better option if you want closed loop. I am not making any assumptions about what will work best for your application. In truth if it works for you it's the right choice.

Ben

Sent from my HTC6435LVW using Tapatalk

[Cncstarter]

That is also why I ordered the kflop over the smoothstepper and I think it will work fine. With the increased sophistication I wasn't expecting a constraint with the basic step/dir output though (other than the pulse length which I had already checked) but in any event I've figured out an easy enough solution. Putting together an order for digikey tomorrow and can confirm it works soon.

I have found some user input regarding the closed-loop steppers though not a lot. There is a lot of editorial/manufacture info as I'm sure you've seen. If you consider my goals more as adding startup-torque to a servo system rather than adding error tracking to a stepper system you'll have an idea of what I would be hoping to accomplish. I also think the machine I am putting together was designed for steppers more than servos and needs some torque or the added expense of a gearhead motor. Compared to the price of name-brand, higher-speed nema 34 steppers and name-brand drivers the leadshine system cost is not much more really (< $50 with Lin and Gecko for example or about the cost of the encoder). It's considerably less than a quality servo system of equal power. If the steppers I currently have aren't up to the task (which I suspect might be the case) I would rather invest a little more in something that is going to make a better machine. Note too that I am building a rack and pinion type woodworking machine rather than a metal mill.

Thanks...