Autotransformer question

Okay, if the core losses are lower at full voltage, and I assume zero at no voltage in, then at what voltage is the core losses a maximum?

Dan

I forgot to mention -- there really ought to be handbook solutions to all this. I know that the ARRL Handbook used to have them for tube-amplifier power supplies, back when radio amateurs actually built their own equipment.

They may still, but I'd have to take a look.

Because the current it draws from the supply is not sinusoidal- it will be drawn in pulses at the peak of the sine wave, so the I^2*R heating of the windings is higher than for a sinusoid.

Here is a decent application note on the matter-- I use 1.7 as a rule of thumb, the document says 1.6~1.8 so we are in complete agreement- see the bottomline in the table on the last page.

Actually, now that I see your actual requirements of 30A peak per axis, a power supply of this type would definitely NOT be a good idea- it will current-limit at the rated current and your drive will stop working, whereas a xfmr + filter will only be limited by the circuit breaker you put in the primary.

BTW, toroidal transformers draw absolutely wicked primary surges on power up sometimes (depends on their magnetization state when they were shut off and where in the cycle it's turned on), so you may want to put a slow circuit breaker on the primary and a fuse on the secondary after the filter cap.

Because 20A out of the DC supply is way more than 20A RMS out of the transformer. You're pulling current out of the transformer for maybe

The transformer current rating is there to keep the coils from dissipating too much power, and virtually all of that is I^2*R losses, which is RMS current.

Decrease the cap values and it'll be better (at the cost of more ripple), but you'll never get it down to 1:1.

Cadillac supplies of this type use a choke-input filter; i.e. the rectifiers feed a honking big coil that then feeds a cap. They have far better voltage regulation as long as you're pulling some minimum current at the cost of buying that honking big coil and lugging it around, and at the cost of the output voltage being about 90% of the transformer's output RMS voltage, less the rectifier drop.

Use a choke-input supply, and the RMS transformer current will be almost exactly 100% of the DC output current, instead of way more.

choke ___ DC out o----. ,------o--->|-----o----UUU-----o-------o AC in )|( | | | )|( .--)--->|-----' | o----' '---o | --- | '---|

Tim, what is the part where I put ***** above?

Do I understand you right, that by adding a choke I would improve regulation?

If so, what size choke and cap would be appropriate here? I have some chokes that seem to be the right size.

I'm curious, I've always used 2:1 transformers (with bridge rec. and cap.) for AMC A8 series servo drives because they are a perfect voltage match and very cheap. They've never been a problem. But I have told a few paying customers to just do this. Bit of a different thing once you cross that line in the sand. Do torroidal have significant safety or other benefit?

Karl

Karl, what is your line voltage where you are?

My voltage is 124 VAC, usually.

That translates into 62 VAC on the secondary, and 87 volts after rectification, and I believe the drives shut down from that.

They are the four diodes inside your bridge rectifier.

Yes -- but keep in mind that it's probably not necessary, just neato.

Oy. It's handbook time.

For 3A ripple current, you want about 9mH (milli henry). Follow that with a 4000uF cap and you'll get 1V of ripple -- follow it with your

You have a pretty wide choice of choke values -- if you're just plucking them from a bin it's much more important that you get one that'll handle the current, then pair that up with your astonishingly large (for this application) 20000uF caps.

Note that at no-load, without a ballast resistor, your output voltage will still rise to 70V or so. I wouldn't worry about this, as it's still under the 80V limit imposed by your amps.

OK, I got it

Yep

I have 20a motor fuses on the incoming 220v. I was wiring that yesterday.

I think that it will work fine. In my mind, 30A "surge" amounts to fairly huge torques that probably are unnecessary. It would mean accelerations enough to make the mill dance on the floor or at least shake.

My back on the envelope calculations suggest that at 80v, I could get

They don't have as much external field so they're better for audio, instrumentation and CRT applications where that matters. They're supposed to be smaller, maybe more efficient (depending on materials used).

OTOH, they take more time to wind, so they tend to be more expensive (basically all the copper wire in each winding has to be wound onto a kind of bobbin that passes through the center of the core, then it's unwound onto the core. Ordinary E-I transformers can have the bobbins wound before the core is introduced, on a sort of special purpose CNC lathe. It's a lot faster as it's one process, not two, and a number of cores can be wound at once, and the machines also tend to be simpler and cheaper.

Here's an older toroidal winder in action:

You might want to have another look at some hysteresis curves!

Regards Mark Rand RTFM

He's right. The core losses will be way less, but of course the I^2*R losses stay the same for the same current. If you run at 1/4 voltage and the same current you'll have a very (electrically) efficient transformer that is four times heavier (and around that more expensive) than the "optimal" design (which likely was optimized for material cost reduction).

Lloyd, have you ever tested and measured transformers?

jsw

I'm putting my $5 that you will be scrapping the underpowered drives and power supply and replacing them with items that will power the servos at the same levels the Bridgeport engineers spec'd them for immediately after you attempt your first project requiring reasonably heavy cuts.

I will be the first to let you know that.

The motors are rated at 30 amps peak, 15 amps continuous, 145 VDC. At that current they develop stall torque 3.2 n-m. or 2.4 ft-lbs. Converted to force 2.4*2*12*10*0.8 = 460 lbs of force.

The speed will be about 2x less.

These drives do not lose out on the torque/force front much and the speed would be about 55% of the max speed derived from motor nameplate.

It is also not a given that the original Bosch drives were capable of delivering nameplate torque and speed to the motors.

I get 124 to 126 all the time. Don Foreman had told me what parts to get for regulators. But i tried without them and never had trouble. Its been eight years. And I know of many others doing the same.

Karl

Jim Wilkins fired this volley in news:848b287a-2434- snipped-for-privacy@s9g2000yqd.googlegroups.com:

Yeah, Jim. I was the senior technician of the Florida Transitron test and prototyping laboratory back in 1968-1970.

Done a few -- Every variety from nearly dc-saturated 400Hz aircraft types to fixed chokes to swinging chokes to special low-emf shielded IBM projects to industrial power distribution equipment. Every regimen, every type of application (except RF, which we did not build).

But whatever. Iggy made the right choice.

LLoyd

Lloyd is obviously not an electrical engineer. Core losses are almost perfectly proportional to excitation X frequency. So, there is NO WAY underexciting a transformer will overheat it. Try it if you have any doubts, but he's just wrong.

You WILL lose power rating proportional to the voltage reduction, so a

Jon

Can you define this "minimum required excitation"? I've never heard of such a thing. Induction MOTORS definitely do have such a thing, the excitation must be enough to develop the field in the rotor, or it llos like a transformer with a dead short on the output.

But, with transformers, the core flux is proportional to the applied voltage X time, and opposes the magnetizing current in the primary windings. This may not apply to substation-class transformers, but should be perfectly true of a 3 KVA autotransformer.

Jon