Free Energy | searching for free energy and discussing free energy

TK:

Easy fix, just put two ammeters on the high side, one for each voltage.

MileHigh

P.S.:  PW, welcome back!  Please forgive my somewhat amateur "brute force" designs.  I know that you are _The Man_ when it comes to this stuff and I am just a Joe Blow.

TK:

Easy fix, just put two ammeters on the high side, one for each voltage.

MileHigh

P.S.:  PW, welcome back!  Please forgive my somewhat amateur "brute force" designs.  I know that you are _The Man_ when it comes to this stuff and I am just a Joe Blow.

MH,

I can't agree with most of your post, particularly the Joe Blow comment, I always enjoy reading your posts.

I thought I'd present some alternate circuit possibilities for hi-side sensing.   

I still do design work so I have to keep up with new IC options, etc (mainly linear and data acquisition).  Some of the new function specific IC's are hard to beat using discrete components, particularly with regard to resistor matching (laser trimmed), thermal stability (low drift) and hi DC accuracy/low offset. 

Also, keep in mind that most manufacturers will send out free samples...

BTW, I still have a full set of National Blue data and linear application books from eons ago, and all the "cookbooks".   

(I did, however, get rid of all my old issues of "Photonics" ...) 

PW

 

TK:

About your new clip.  Perhaps there is less power out because your drive coil is coreless and therefore stores much less energy.  That's not necessarily a bad thing if you are comparing input power vs. output power vs. RPM efficiency.  It depends on the measurements of course and the timing timing timing.  It's possible that a drive coil with a core stores "too much energy" and you put in disproportionally more energy per drive pulse than you get compared to the push that you deliver to the rotor.  If your coreless drive coil can make the rotor spin at about the same speed as the same coil with a core, that might be true.  Then the core itself may have a fat hysteresis loop or a skinny hysteresis loop.  It's the area inside the loop that is equivalent to the lost energy.  A certain amount of power gets lost in the sloshing of the magnetic domains in the core.

Anyway, it spins and it outputs real joy buzzer pulses!

MileHigh

Now that I've got the drive coil mounted the way it is, I no longer can just slide a core into it like I did at first. But it's pretty clear from those early trials that the core, at least the bolt I used, didn't really help much. It did change the waveform, it did warm up quite a bit, it did put asymmetric loads on the rotor, but the RPM and acceleration were about the same. I don't know what would happen at higher power levels with a core, though. I did run the coreless version at 36 v input and there wasn't much gain from that either. Again, I seem to have accidentally (on purpose) hit a sweet spot right out of the gate, as it were.

P.S.:  Did you ever try putting a super-high voltage cap as the load on a pulse motor just to see what happens when the cap voltage reaches crisis levels?  I figure the diode starts to reverse-conduct after every coil pulse which would make life very harsh and difficult for the diode.

No, and I'm hesitant to do it right at the moment as I am running low on spares. After the major rebuild I made an error hooking it back up and blew a mosfet and one half of the original TL082. I'm down to one spare TL082 so I don't want to take any risks at the moment. But once I'm stocked with spares again I will try it and see what happens. If the diode breaks down though I'm pretty sure it will take out the mosfet and the op-amp. (My error actually even fried a trace on the little RS PCB, like blowing a fuse.)
I'm just using a 1n4007 rectifier in there at the moment but I think something like MUR1560 or other beefy ultrafast diode would be better.

I didn't want to put ammeters on the high side because I want eventually to be able to just replace them with current-sensing resistors and scope hookups, and I don't have any differential voltage probes, so I need to keep the probe references at the system ground or negative rail.

Meanwhile... the, er, research continues. Now don't go off half-cocked, you know that I'm not making any claims in the following video, but what I am illustrating has caused others to draw conclusions that might be a bit over the top and premature.

http://www.youtube.com/watch?v=xajlH6wkj_0

TK:

If you are tempted you could have another go with your op-amps for your current sensing.  I am assuming this would cost a fraction of what differential probes would cost.

You may have some "signal" type transformers or something that will work fine.  Preferably a 1:1 ratio but even that's not that critical.

You do the dual 9-volt battery setup with a single inverting op-amp and you build this:

[current sensing resistor] -> [inverting op-amp] -> [signal transformer] -> scope channel.

The op-amp is converting the input voltage to a low-impedance output voltage.  That means that when the op-amp drives the primary of the signal transformer, within limits it can supply increasing current when it outputs a DC voltage.  In other words it can energize the inductor in the transformer.  That generates a DC voltage on the transformer output.  It may sound funny to state "the basics" but the important thing to realize that this gives you the ability to pass an uncorrupted waveform from the primary to the secondary down to a fairly low frequency.  In other words, this will give you a "poor man's differential probe" that should work find above a nominal frequency.  The frequency might be quite low, perhaps 10-15 Hz.  Likewise, the larger the transformer the more "headroom" you have for coupling low frequencies.  You know those transformers that might be one-inch cubed?  Something like that, substantial.  I am guessing smaller might work too.  You don't want to get too large because we are going to make the assumption that the smaller the transformer, the better the high frequency response.

Simple test:  You breadboard your inverting amplifier and feed it with a square wave from your signal generator.  The output goes to your signal transformer, and the secondary of the signal transformer goes to your scope probe.  Suppose you start with a 200 Hz square wave.  Your scope channel should show you a very faithful, isolated replication of that wave form.  Then start lowering the frequency.  If you are lucky, let's say you get down to 10 Hz before you start to notice that the transformer is "choking" and failing to replicate the low and high "DC" levels.  If we assume that this is all true, then if the pulse motor is pulsing at say 30 Hz, then the the secondary waveform on the transformer should be an exact replica of the waveform on the primary.   So for five dollars or less, you should be able to enjoy all the benefits of a differential probe using an op-amp, with the understanding that as long as your pulse frequency is above a certain threshold (and you will know that threshold frequency) then you will see a nice clean waveform on your scope display.

When you think about it, it's a nice little potential project.  Two 9-volt batteries, current-in and current-out terminals, an inverting amplifier, perhaps a trimpot arrangement for your input resistance, a trimpot arrangement for your feedback resistance, a trimpot to zero the op-amp output when there is no signal, and the signal transformer, and isolated ground and signal terminals for your scope probe.  You can adjust the gain of the amp whenever you want, and away you go.

Again, there are two major assumptions here: 1) with the op-amp driving the transformer input you will get faithful signal propagation above a certain low pulse frequency, and 2) the op-amp and signal transformer bandwidth will be high enough to give you a quite faithful waveform reproduction.  Perhaps when you compare the original signal and the transformer-coupled signal on your scope for typical pulse motor current waveforms, you will barely notice the difference.

There is also a limitation.  We are assuming that most current waveforms in a pulse motor are of the form <nothing><pulse><nothing><pulse>.  In other words there is no DC bias in the waveform.  If there was a DC bias in the waveform, that would mean there was some constant current flow.  Naturally you can't scope a DC bias.  The op-amp output will induce a constantly rising current waveform in the primary of the transformer.  Eventually the op-amp will crap out or the transformer will get saturated and start to heat up and you will lose all of the signal on the secondary.  But for fun, you could buffer the main op-amp output into a second amplifier.  You could connect a pair of LEDs to the second op-amp output.  So those LEDs could show you "above ground" and "below ground" signal activity - in other words, the direction of the current flow and the approximate magnitude of the current flow.  Just a neato "LED scope" to let you know if something is alive and pulsing without even having to hook up your scope.  If you tuned it right, you might be able to decently judge current flow and direction just by the apparent brightness of the LEDs.

If you don't have a high-bandwidth signal passing transformer then you could probably find something excellent on DigiKey.

So, isolated signal probing is just an op-amp away!   If PW is watching, he could "polish" my comments and give you the real deal also.

It would be a fun and easy project to do.  As you might imagine the identical concept could be done for an isolated voltage probe also.  I think the key is having the right transformer.

MileHigh

P.S.:  There may be an issue that would have to be investigated.  I am wondering what would happen if you say feed a square wave into the input that varies between say zero and five volts (like a current waveform).  Every positive pulse in the square wave will induce current to flow in the transformer primary.  When the op-amp output goes to zero volts the primary of the transformer will still be conducting current.  The op-amp output is an active ground at this time so the transformer primary inductor sees that as a short circuit and current keeps flowing.  So that could be problem, every positive pulse will induce more more and more current to flow into the primary until it saturates.  So you may have to add an AC coupling stage before the main inverting op-amp.  Well, it could be more of a challenge than I thought.  There might even be something in the app notes that would be much better....

TK:

You just blew me away with that amazing link.  80 KHz bandwidth and it's fully isolated!  Just requires 5 volts!  It's quite amazing and just gives you all the crap I discussed in one tiny board!

It's great but at the same time, for some and I for sure for you, there is the satisfaction and fun of building something.  But it's still damn amazing.  I also suspect that you can get much higher bandwidth with the op-amp and signal transformer.

MileHigh

Yah, gotta get me one of them thangs.

But you are right, I could build what you suggest with parts on hand, for nothing. Probably will go for the store-bought solution though, until I discover it can't work right for some reason.

Meanwhile, some unpowered rotor rundown data is coming up, with and without propeller load, CW and CCW:
 
http://youtu.be/_zOwadZH6T0
(still uploading)

(boring video, but confirmation of 90 second rundown from 2000 RPM)

TK:

I think I fixed my problem.  You know how when a scope is AC coupled you can see how the AC coupling reacts to a DC step input, and it takes several seconds for the trace on the scope to "settle."  It's just a fairly long time-constant AC coupling action we are seeing there and we have to do the same thing with op-amps.

New pipeline:

[current sensing resistor] -> [inverting op-amp] -> [AC coupling with long time constant] -> [unity gain buffer with offset compensation] -> [signal transformer] -> scope channel.

Note this is still a one-chip solution.

We need to do an AC coupling with a long time constant so that it allows low pulse frequency signals to propagate without noticeably distorting them.  We only need one zero-offset point so the unity gain buffer is the best place to do it.  That way the signal transformer primary will have no current flowing through it when the current sensing resistor senses no current.

Let's pick a time constant of five seconds.  That's longer than we typically see for the AC coupling for a scope.

C = 1000 uF,  R = 5 Kohm gives you a time constant of five seconds.  As you can see you could easily make the time constant much larger if you want.

The output from the inverting amplifier will connect to one side of a 1000 uF non-polarized capacitor.  The other side of the capacitor connects to the non-inverting input of the unity gain buffer.   There is also a 5 Kohm resistor connected between ground and the non-inverting input of the unity gain buffer.  So the 5 Kohm resistor is the "DC bias bleed-off resistor."

With this configuration it will be just like an AC-coupled scope trace but with a much longer time constant.  This will ensure that low frequency signals are properly coupled and imperceptibly distorted and the signal coupling transformer primary does not get saturated with constantly increasing current gong in the same direction because all the DC bias gets removed from the signal.

If you notice both the AC coupling and the op-amp driving the signal transformer are designed to be "friendly" to low frequency signals so that you can look at an undistorted current pulse waveform that might be pulsing at 10 Hz.  At the same time any average DC bias will be removed from the signal after about 25 seconds due to the RC coupling circuit.

Again, this is a design without any finesse in designing op-amp circuits but it should give you want you want:  isolated current sensing so you can use an ordinary scope probe anywhere in the circuit.

MileHigh

TK:

About the apparent lack of RPM change in your clip.  Note that you are tapping into the coil discharge while the main agent causing the push on the rotor is the energizing of the coil.  So in theory tapping into the coil discharge is soaking up the "left over" energy and the pushing on the rotor has already been taken care of.

MileHigh

@Tinselkoala,

Lenz delay is caused by a phase shift that speeds the rotor up. Increasing the load beyond your "left over" energy can produce it. I suggested attaching a 120 volt LED, which has sophisticated circuitry in the base. The back spike power will not illuminate it to full brightness, but draw an additional amount that should cause a lag and result in the speed up effect. I got this kind of reaction with my Bedini SSG'S, and think it's worth a try!


The bonus feature of your MHOP circuit might include a possible continuing increase of acceleration with the simultainious readjustment of dwell.