Pico Systems boards

Hey John,

I have a stepper based system running on a Dell that I got for free. The main problem at present is that the latency on it is rather bad so I have a very limited base period for software stepping. I was wondering if the Universal Stepper board might help me out there? I also have servo thread latency of, I think, about 0.3ms. There seems to be no way to work around nor disable much in the BIOS. I have found no way to reduce the latency thus far.

I was looking at your board and thinking that would make a nice way for me to offload the step generation and so forth and simplify the hookup of the various limits, E-Stops, and Charge Pumps.

Does the servo thread latency I have represent a problem still for using this board? Right now I am limited on my machine to 75IPM (1/8 microstep) because any faster and the steps can't seem to keep up. The PC is great overall so if I can get the time critical stuff offloaded then maybe I don't need to try to build a compatible system.

photomankc wrote:

Hey John,

I have a stepper based system running on a Dell that I got for free. The main problem at present is that the latency on it is rather bad so I have a very limited base period for software stepping. I was wondering if the Universal Stepper board might help me out there? I also have servo thread latency of, I think, about 0.3ms. There seems to be no way to work around nor disable much in the BIOS. I have found no way to reduce the latency thus far.

300 us is QUITE bad. If the absolute worst is 300 us, then a USC board should work, but it is running a bit close to the edge for a 1 ms servo cycle. I wouldn't worry about 100 us, which would make a computer practically worthless for a software step system. But, 300 us is really up there. What I worry about is that there might be a worse latency in the several ms range every once in a while.

I was looking at your board and thinking that would make a nice way for me to offload the step generation and so forth and simplify the hookup of the various limits, E-Stops, and Charge Pumps.

Does the servo thread latency I have represent a problem still for using this board? Right now I am limited on my machine to 75IPM (1/8 microstep) because any faster and the steps can't seem to keep up. The PC is great overall so if I can get the time critical stuff offloaded then maybe I don't need to try to build a compatible system.

First, I might try an old PCI video card and see if you can get the latency down, onboard video chips and some particular combinations of video cards and drivers are usually the cause of bad latency.

But, a USC will definitely ease the CPU load. As long as it can be serviced every millisecond, and your trajectories are not hurt by only updating velocity every millisecond, then you will see a big improvement in smoothness and the ability of the motor to track the commanded velocity. If you have 300 us jitter in the timing of step pulses, then when you go over 1000 (micro)steps/second you are getting an occasional 33% jump in step timing when the latency hits maximum. That is catastrophic for a software step generation system. The USC will give fine-grained step timing to 300,000 steps/second, where the granularity reaches 3%.

By the way, what Dell model is that?

Jon

Bob La Londe wrote:

dewey525 wrote:

I currently have installed the size VFD you are needing. A Hitachi sj200-075lfu 200v class. Which was recommended to me by its manufacturer to run my 3ph spindle. It cost right at $775.00 usd and will give you 3.75kw 3 phase from a single phase supply. This is currently being implemented in an upgrade to our mill. These VFD drives are de-rated for single phase as you know to no more than 50% its rated 3ph capacity . You could dl a manual for it from drives warehouse to find more information.

I will definitely download the manual and take a look. I had about given up on getting full power without setting up a phase converter. Most of the VFDs I have looked at don't list single phase input specs for their models over 3.7Kw.

I should mention that before all this derating stuff became well-known, I put a 1 HP 3-phase ONLY VFD on my 1 Hp Bridgeport about 12 years ago. I don't run it 8 hours/day, but it is a "production" shop, meaning I make parts for sale, some of them can be seen on my web pages. Now, I even do rigid tapping on it, which is pretty much a torture test with constant reversing. So, I have been running a 1 Hp motor on a 1 HP VFD not rated for single phase input for 12 years with NO problems! It definitely will wear out the filter capacitors faster than if it was fed from 3-phase, but I can easily imagine I could get several more decades out of it.

Remember, these things are rated to run without ventilation in a closed cabinet at 50C for 80,000 hours under full load. If you actually run under those conditions, then you'd better consider the derating seriously. In a home shop, maybe with the VFD mounted out in the open air, and used tens of hours a week, component life will be much less of a worry.

Oh, just to clear up a possible misconception, a 3-phase rated VFD will generally provide full power to the motor when running off single phase power. (The exception is a drive which has phase loss sensing, which won't run at all.) What the problem is is that the input rectifier and capacitor bank will run hotter than normal, and this will shorten their life.

Jon

When choosing a VFD we found that the standard for most manufacturers is single and 3 phase input for drives lower than 3 hp without de-rating them. But the higher HP drives mostly do not even list that they will accept a 1 phase input. So we started calling manufacturers to find ours rated for 5 hp which has a capacity for 3 ph input / output 10 hp at 240 volts.

Now Im not arguing the fact that they would provide 3 ph from a 1ph input without de-rating,( manufacturers as we know try to limit liability issues concerning their products ) but Im not one to argue with manufacturers on the limits of their products. Nor did I want to risk a 5 hp 3 phase motor. The price was acceptable for assurance from the manufacturer that it would work as intended for our use.

I like the VFD for over the rotary phase converters which were really about the same price. The rotary required more energy to run a single piece of 3 ph equipment than does the VFD. Now if I had more than 1 3ph machine I may have went with the rotary to convert single to 3 ph.

jmelson wrote:

photomankc wrote:

Hey John,

I have a stepper based system running on a Dell that I got for free. The main problem at present is that the latency on it is rather bad so I have a very limited base period for software stepping. I was wondering if the Universal Stepper board might help me out there? I also have servo thread latency of, I think, about 0.3ms. There seems to be no way to work around nor disable much in the BIOS. I have found no way to reduce the latency thus far.

300 us is QUITE bad. If the absolute worst is 300 us, then a USC board should work, but it is running a bit close to the edge for a 1 ms servo cycle. I wouldn't worry about 100 us, which would make a computer practically worthless for a software step system. But, 300 us is really up there. What I worry about is that there might be a worse latency in the several ms range every once in a while.

I was looking at your board and thinking that would make a nice way for me to offload the step generation and so forth and simplify the hookup of the various limits, E-Stops, and Charge Pumps.

Does the servo thread latency I have represent a problem still for using this board? Right now I am limited on my machine to 75IPM (1/8 microstep) because any faster and the steps can't seem to keep up. The PC is great overall so if I can get the time critical stuff offloaded then maybe I don't need to try to build a compatible system.

First, I might try an old PCI video card and see if you can get the latency down, onboard video chips and some particular combinations of video cards and drivers are usually the cause of bad latency.

But, a USC will definitely ease the CPU load. As long as it can be serviced every millisecond, and your trajectories are not hurt by only updating velocity every millisecond, then you will see a big improvement in smoothness and the ability of the motor to track the commanded velocity. If you have 300 us jitter in the timing of step pulses, then when you go over 1000 (micro)steps/second you are getting an occasional 33% jump in step timing when the latency hits maximum. That is catastrophic for a software step generation system. The USC will give fine-grained step timing to 300,000 steps/second, where the granularity reaches 3%.

By the way, what Dell model is that?

Jon

The base thread shows about 25us latency, it's not enough to be catastrophic but enough to mean that my speeds on 1/8 micro-step are limited. I have runt the test many times and that is the number that always comes back. The motors run decent at low step rates like 1/2 stepping but not so good at higher micro-step rates. The servo thread hits 300us but letting it run never sees it go beyond that. I think I just have to admit I have to go for a new PC. Its an Optiplex GX620 and there is no expansion ability. Do you know of a similar small footprint PC that is known to work? I don't want a tower PC as the cabinet I built was based on something small like this one.

Thank you sir. I will check it out.

Question though..... how is the para-port connected on it? I see a pin header, do they make a standard 25pin connector that terminates in a pin header socket?

Ah yes, I've seen them before just forgot about that. Thanks.