[mike_Kilroy]

Your not out of your league, your doing fine! I ran to office today and found my old datametrics catalog did not survive our last move so dead end there. sorry. You do know that if you can hook it up and move the axis with assumed 1000ppr, you can imperically figure right number by saying move 1" and measure it' then just scale the 1000 up or down to get it to move right distance.

Glad Electrocraft sent you the right data! My old book was here but did not have the E727 in it. Their spec sheet tells everything you need to know.

You can put upto 9.2amps into the motor forever without overheating for 26#-in cont torque. Most of the time folks use a 2xpeak drive on this kind of motor so if u have 10 amp drive, at 20 peak you get 52#-in torque max. so ur 30a8 at 30amp peak/15cont is good choice current wise. If you want full torque rating of the motor you should have 9.2a available continuously for it.

She also stated that the max current for this motor is 9.2A, but I never got from her what the max motor voltage (i.e. Which winding) it is.

I bet she said/meant the max continous current for this motor is 9.2amps - because the spec sheet shows this as the continuous rating and the peak torque is 2000oz-in or 2000/45= 44 amps peak. Typically on that era motor this is the demag current - the point that if you exceed you break the motor (this motor is designed with ceramic motors - tell tale is the derate of Ke/Kt per degree c heatup).

for top speed, I would assume their 80v no load test line and their 2400rpm max no load line mean the motor max speed rating is 2400v since both these comments jive.

if you use worse case Ke of 36 then you need 36*2.4=86 volts to get to 2400rpm no load. If you want some torque up here too you need to add some more voltage for the IR drop - the current to make torque. In this size motor the 9.2a rating is usually for upto about 1/2 speed so you might say around 1200rpm then it would drop down to maybe 1/2 at max 2400rpm. lots of assumptions saying this but they did not include a speed torque curve so we gotta guess this one. so say 4amp is available continuous at 2400rpm; add another 4a*.8ohm=3.2v for IR drop to make 13#-in of torque. If you want to assume more worse case, assume 10amp at max speed so 10*.8 or 8v more. so if you need all 2400rpm and some torque up there, you should have around 8+86=94v here.

as al says, if you don't have an amc db unit then the dc bus will likely pump up another 10v during decels so drive should be able to do 94+10=104v before faulting. if you use the 30a8 at say 75vdc to allow for decel regen of 10v, and you want say 13#-ft at speed, then you will get a speed of:

v=ir+Krpm*Ke or 75=4a*.8ohm+K*36 or K=1,990 rpm max -- is that enough for you?

I believe amc still sells their db kits for use on power supplies.

Lastly, you asked about the max voltage rating of these motors and why. the other fellows comment of pitting the commutator if you exceed rating is right; the max voltage is a point where the amount of sparking is acceptable to the design. the key is: there is a max bar to bar voltage that can be switched on a commutator for a given amount of sparking. this bar to bar rating determines the max voltage of the motor design. so it is usually the same voltage for a given motor size and design; to get a higher voltage rated motor you gotta change the commutator design - by adding more bars so the bar-bar voltage is less than one with less bars. once the bar-bar max voltage is deterimined, then the motor designer decides how many turns of wire to put in the motor slots. the more turns the higher Ke/Kt. now it is just a matter of picking the number of turns to match the required top speed. of course less turns means lower Ke so faster speed, but also lower Kt so more current, so as the turns goes down, the wire size used in the slots gets bigger to handle the higher current.

[Al_The_Man]

There is also some good information in the videos on tuning on the Galil site.
On-Line Tutorials - Galil: We Move The World
Galil amps are relabeled AMC, they show what happens if current limit is reached during high accel rates etc and you essentially lose control when this happens.
Although you may not use Galil, the videos pretty much apply to whatever servo system is used.
Al.

[torinwalker]

Fantastic. I see now the relationship between all the values and I can resolve this now. Thank you (again.)

I am looking at using two Keiling KL-8285 transformers (Unregulated 1230W, 82VDC/15A toroidal with dual caps and bridge, est 1Vrms ripple @ 9A) in parallel. They're $239 each.

It's the closest power supply I can find with the right voltage and current (for the right price, too) that meets all of my demands, including the limitations of the drivers.

Using your calculations and solving for K, I get 2188RPM at the estimated 4A. In fact, graphing from 0.5A to full 9.2A continuous, I get a linear RPM range of 2011 to 2211 rpm. If I had spent another $400 on Mesa electronics H-bridge drivers instead of the 30A8's, I may have been able to get another 200 rpm. That's 10% for $400. I think the top end is good enough. This machine will be fast enough, Mike. Thanks.

As you guys pointed out very nicely, if the motor pumps energy back into the caps and the drive shuts down, I'll be in big trouble. My choices are to choose another supply with a lower voltage and lose the top end speed (forget it), decel very slowly, or best of all, to put clamps on the motors.

Is that what you meant when you said AMC db's? I couldn't find "db"s easily.

AMC makes an SRST80 (shunt regulator) that clamps to 80V and dissapates 95W. That should keep the decel spikes at bay. They don't look terribly expensive, but considering that I bought a drive from eBay for $35 that would sell normally from AMC at $500, perhaps I'll wire together my own clamp. I was hoping I wouldn't have to create a PCB - it's such a pain in the ass.

[Al_The_Man]

DB, Dynamic braking and Shunt regulator are similar, the DB usually acts across the motor whereas the AMC Shunt regulators operate across the P.S. DC bus.
The circuit is fairly simple, a voltage comparator can be used to switch a Hexfet into a low ohm, high wattage resistor.
The beauty of the Toroidal transformer is that it can be easily tailored or modified by removing (or adding) winding for the correct voltage.
Small overwinds can also be wound on for low voltage auxilliary supply.
I get mine from Antek also on he is on ebay under the name Jonango, he makes them to order also.
Transformer alone or complete supply.
Al.

[mike_Kilroy]

DB, Dynamic braking and Shunt regulator are similar, the DB usually acts across the motor whereas the AMC Shunt regulators operate across the P.S. DC bus.
Al.

sorry for the confusion; we use DB to mean the resistor for shorting across the PS DC bus in our servo world, so same thing. Yes that STS thing is what I meant.

Sounds like you have it under control. Only thing you might consider is HOW much power you will put back on the dc bus. Sounds like you might be able to estimate it by the mechanics. The energy (power x time) you put back into PS is E=1/2mv^2 or 1/2Jw^2 where m is mass, v is velocity. a cheat to figure out how much is if you can figure how much torque you will suck out of the rotating load at max speed to make it go down to 0 speed; just use HP=TN/5252 where T is torque in #-ft, N is max speed stopping from in rpm. then HP*746watts/HP gives you the watts of decel. from this you use power/volts=amps and you see how big a Hexfet you need and how much current you will drw into the db resistor. or just guess your 9amps from your motor rating

also, remember that energy absorbed into the ps by the caps is E=1/2cv^2; so you can see you can simply add more caps to the bus to absorb more energy; might be you can stick enough capacitors in there to not need db at all.... double the caps and get double the stopping energy! You can get into these equations and really make some good guesses! for instance, lets say you use 40,000mfd of caps on your 82vdc bus. If you let it rise 4volts to 86v max, they will absorb 1/2(.04)(4*4)=.32joules or .32watt-sec. If your decel time is 0.1 sec, your dc bus will rise from 82 to 86v if you send in 3.2watts of power.... that will be 86v*.04amps from your motor. nothing! but let the bus go up 25v and watch: 1/2(.04)(25^2)=12.5watt-sec or 125/107=1.1amp on your motor. Make the caps 400,000mfd instead of 40,000 and now they will absorb 11 amp during that 100msec stop; no db circuit or resistor needed...but this amps and this short time may not be enough to stop your motor and load so this is probably not very useful info by itself.
(hope I did my quick math right)....

Im next going to look at the kenetic energy of stopping just the motor from 2000rpm...

motor J is .12oz-in-s2 or .000086Kg-m-sec^2. 2000rpm=209rad/sec
so energy to stop your motor from 2000rpm =.5*.000086*209^2= 1.9 Kg-m energy. converting this to watt*sec I get 18 joules or watt-sec....so letting your dc bus go up 25v above, with 40,000mfd worth of caps that absorb 25watt-seconds of energy, the kenetic energy in your motor alone spinning at 2000rpm is 18 watt-seconds, so it will stop and your dc bus will not go up the full allowed 25vdc you designed to.... so add about 1/2 the motor inertia worth of load inertia and you now pump your dc bus up 25v over nominal to stop from 2000rpm.... knowing this, if I didnt mess up the math, you may not require your Hexfet circuit at all with around 40kmfd caps - if your load inertia is about 1/2 the motor inertia; increase it to 80,000mfd and you can see you can stop almost 3x the motor inertia too in this 25v pump up.

[torinwalker]

I just measured my wall outlet this morning - it reads 125.6VAC. So I'm thinking about the power supply I chose and that I might consider something lower. The 82V version I selected earlier was supposedly tested at 120VAC, so who knows - if my wall shows 126V, then my 82V supply might end up providing nearly 86V and trip the DC bus voltage limit.

Keling offers a 72V version that is probably rated at 115/230V. If the voltage goes up by 126/115V = 1.095 (as I've just demonstrated), the 72V supply will still be at 72 * 1.095 = 78.88V. I should have considered this earlier - it's a good thing I didn't just dive in without looking.

I like the capacitor idea. The problem with the keling supplies is that they come with two puny caps. I'm looking at a company here called plitron.com that manufacturers toroidal transformers in 1.5kva flavours and in particular have one that takes 115/230V in and puts out 50VAC. At 1.414 minus the diode drops, I'm looking at 230VAC in, 69VDC unregulated out. If the power is 125.6/251.2V as it was this morning, the DC output would be 75.8. Then I can add my own monster caps and forego the shunt with ample room up above to absorb the regen spikes.

Ehm... Let's say, for example that I had a supply at 75V and my motor put out 12V during regeneration but I had large caps to absorb the spike. Wouldn't my drive see that 86V momentarily while the power rings back and forth on the bus until the caps charge? Caps don't charge instantaneously which means that power remains on the bus until it is absorbed by everything in its path - the motor and transformer windings, the cables conveying the power, everthing. But the caps are the least path of resistance (well, so are the cables, but wait for my point), so if I worked out the actual impedances in this complex waveform, I'd probably discover that the majority of power charges the caps rather than presenting to the drives, and the rate the drives produce the current is less than the charging rate of the caps.

So I guess I answered my next question which was going to be whether the drives would see that 86V spike momentarily. The answer, if I choose my caps carefully, is no.

Perhaps I'm overthinking this because the shunt follows the same rules - the 5ohm resistor can only dissipate power at a specific rate and that voltage would necessarily have to hang on the bus until it is fully absorbed. Yes, obviously I'm overthinking this.


Torin...

[mike_Kilroy]

Ehm... Let's say, for example that I had a supply at 75V and my motor put out 12V during regeneration but I had large caps to absorb the spike. Wouldn't my drive see that 86V momentarily while the power rings back and forth on the bus until the caps charge?

Perhaps I'm overthinking this because the shunt follows the same rules - the 5ohm resistor can only dissipate power at a specific rate and that voltage would necessarily have to hang on the bus until it is fully absorbed. Yes, obviously I'm overthinking this.
Torin...

Not the way it works; the E=.5cv^2 is continuous; as you pump energy into the caps the nominal voltage across them rises linearly - no jumps. they obsorb this instantly for all intents and purposes. If you scope it you will see your nominal 75,76,77,78,79, etc as the dc bus rises. no spikes. linear. if you calculate a 12v rise, then it will continue up like this to the top 87v then when that energy stops coming it will linearly start going down again. no spikes.

If you use DB circuit instead, it will rise 75,76,77,78,79 to the turn on setpoint of the comparitor - say 82v - which turns on the hexfet shorting the 5 ohm R across the bus, the voltage drops linearly to the turn off voltage - say 79v - then if decel is not done and more power is coming in still the process repeats. So the dc bus now looks like a sawtooth as this cycle continues to the end of the need.