RK05 servo adjustment fun

Hi Tony, On 12.04.2010 22:21, Tony Duell wrote: I'm not 100% sure but what I found out is the following two things. 1. The RK05F I've been dealing with shows strange behavior on the sin and cos signals: They behave as if there's a low pass somewhere in the path. But there's no capacitor. And I don't think that there's a broken trace on the servo board (G938-YA, -YA is for RK05F with some resistor values different). So when you adjust the amplitude to 10V peak to peak, you get very small amplitudes at higher speed. The oscilloscope images in the manual show the same behavior - but only very weak. I can imagine that it has something to do with the photo diodes in the transducer aging. This problem lead to my drive overshooting on long seeks. No real overshooting: When I monitored the velocity feedback signal, everything looked good. The carriage didn't really overshoot and also reached slow motion as planned before stopping. And still it went to far. The problem seemed to be that with the low sin and cos amplitudes it missed some pulses and counted erroneously while traveling at high velocity. What I have done is deliberately turn up the sin and cos gain potentiometers. Adjusted them to approx 16V peak to peak instead of the recommended 10V. That allowed me to turn up speed without getting seek problems. But it's still not completely good: I could not go faster than about 90ms full stroke time. But the official timing is 70ms for the RK05F. But it seems to work reliable at the moment. 2. At some drives you encounter the problem that the heads won't be properly retracted. That's a bad problem. When you look at it, you get the impression that something is too weak to pull the heads up the unloading ramp. First idea would be to turn up the power amplifier gain to solve the problem. Indeed the problem shows up when you turn down the amplifier gain. But just turning up the amplifier gain seems to be useless in most cases. I now thing that the problem is NO lack of force. Put your drive's heads somewhere onto the disk. Then hold the carriage while setting the drive to unload. You will realize that the slow motion retract is done stepwise and with good force. If you turn off the power amp and unload the heads by hand you will see that the force needed to manually unload is smaller than the slow motion drive delivers. So that should not be the problem. It is something with the servo control. There seems to be a critical point when the outer limit signal goes high. If you push the head just a little little bit further - it will go on retracting to the end! So there seems to be one critical position. The servo electronics have to be convinced to step over it when unloading. What helped me: Velocity amplitude. While dealing with my main problem I realized that there's a much stronger relationship between the velocity amplitude adjustment and the head unloading problem. There seems to be some kind of threshold somewhere in the circuit. If you're below that, it won't unload. If you're over, it unloads. and it looks like it either works or not. So this is how I got that particular drive working. I just fixed the next drive, a RK05 without J or F. But that had only a broken Zener diode on the servo board... Here's my recommendation for drive adjustment (read the manual first to know how it should be done): 1. Adjust everything like explained in the manual. 2. If you encounter seek overshoots at correctly adjusted velocity: Increase sin and cos amplitude to 16V p-p if they break in too much while moving. 3. If you still have overshoots, probably less than before, decrease the velocity amplitude. This is the point where you start running your drive below specified speed. Try to lower the speed as little as possible. 4. If you now have a problem with head unloading, dig into the sin and cos amplitude problem if it exists. 5. Adjust the power amplifier gain -- if you like. That's the least important adjustment in the whole machine. Conclusion: * Too low velocity amplitude -> No head unloading * Too high velocity amplitude -> Overshooting due to count errors (NOT because of going beyond mechanical specs!) * Sin and Cos amplitude too low -> probably impossible to find an adjustment where heads unload properly and drive works fine. I have good reason to believe that the transducer speed is the limiting factor for seek speed: The velocity gain adjustment adjusts the velocity *command* to the closed loop servo system. So it also determines the maximum head travel speed (35ips says the manual). The servo system doesn't seem to even get near to its mechanical abilities. So the only explanation for the limitation I find is the slow transducer system. Ugly! The fixed regulated travel speed also implies that the amplifier gain has only very weak effects on the seek time: If you watch the full stroke waveform you'll see that most of the time the heads are on full velocity or in controlled slowdown. What you can alter by amplifier power is only the raise time in the beginning. The normal value is 14ms for the heads to go to full speed. 14ms of 90ms which you can shorten or prolong a little. That's all. The drive I'm just bringing up doesn't show sin and cos amplitude problems. It just doesn't show any problems yet. I'll see if I can adjust the speed up to the mechanical limits of the servo system. Up to the point where it it is near real overshooting. I'll post the max speed I could go at while still passing the drive control test :-) I would like to get some feedback of other people dealing with RK05 drives. I'm especially interested in other opinions concerning my sin and cos amplitude problem. Sorry for the chaotic explanations but as things are circular dependent I didn't know where to begin :-) Best wishes, Philipp -- http://www.hachti.de