Voice Coil Rail

[ChrisR]

The first graph had a much smaller voltage range than the latter - I don't think I can have separate 'scope amplifier settings. Inever tried two scopes with this - haven't needed them with 13/14 year olds!

"If your simulator has a Current Controlled Voltage Source (CCVS) device -" nothing like that, not even a varactor diode.
That's why I made the differential amp - it's a form of CCVS. Voltage out is proportional to current through the 1 ohm resistor.

There IS an opto-coupler I could put in the FB - but I can't at the moment see how to avoid the crossover issue with that.

The 741 will act like a 741 in terms of some basic (I'm guessing which ones ) set of parameters, of input impedance, offset, slew rate, open loop gain, voltage range etc, I expect, having used the program . For the diff amp I chose "Ideal" as my op-amp.
It certainly doesn't always get things right, eg an OC output comparator will source a little current .
The program itself is speed-limited, you can slow it down x10, x1000 etc.

Worries about slew rate seems to be wasted to me - it's measured as at worst (iirc) 1 volt per microsecond. You must be stacking at a heck of a rate if you can't wait a few nanoseconds for it to settle a few mV.
Parasitic oscillations are more likely to be a problem surely?

Marking to do right now - students seem to think that comparator outputs are tri-state and turn OFF - oh dear, I failed somewhere .

[mawyatt]

Chris,

Must be fun working with young minds, I have fond memories of creating and teaching a series of graduate level RF IC design courses (I believe the 1st courses of this kind anywhere). The emphasis was on using non-traditional concepts at RF and Microwave frequencies for integrated circuit implementations. The courses were televised throughout the state and we had 35 students live for the 1st course (mostly foreign). I took advantage of these courses and hired my best students, since I had them for a semester I knew the really good ones!!

Agree with your diff amp use, you don't want any distortion here because it's not relative to the circuit, but relative to the measurement.

I have no idea how the op amp you are using is modeled, but regarding observable non-linearities, probably not an issue because of the high open loop gain any decent model should mimic.

Also agree with the slew rate concern, even with the slow general purpose op amps you would have to be stacking at a very high rep rate for this to become an issue. Not going to be a problem in reality!! Now trying to do these kind of things in well under a nanosecond, even less than 100 picoseconds, gets to be tough...this is where I work!!!

Oscillations shouldn't be a problem if you observe proper design and layout techniques with decoupled power supplies. However, if you use the Analog Devices high current op amp I previously mentioned, then be careful. This op is much faster and thus more prone to higher frequency instability caused by improper design, layout and decoupling.

Best,

Mike

[mjkzz]

I think the circuit can reach its desired current, even at "crossing point", it MIGHT just take a bit longer, so unless were are shooting 1000fps, it probably will no matter.

I think I will satisfy my curiosity with my scope if I can fix it.

[ChrisR]

Well, I don't know whether the op amp is modeled with group or phase delay, I can't reveal any evidence of it.
I don't really understand the circuit, yet, but I can't make it do anything Mike says it shouldn't.
(I don't quite know what to do with R1. )

Arranging the crossover point to be at 0v is tricky - an "offset" knob would be useful.

Here's a couple of screenshots with exactly the same circuit /settings as each other, save the frequency of the generator.

The line you can't see is the whacky one - green. That's the one which dips into notches and bounces or steps - depending much on the offset setting on the generator. But who cares, as long as red behaves itself?
(NB the gain of my floating ammeter is now 500)



Same - faster :
Well before this rate of change, momentum in the coil/mount and whatever else, would surely be interfering?

[mawyatt]

Well, I don't know whether the op amp is modeled with group or phase delay, I can't reveal any evidence of it.

Arranging the crossover point to be at 0v is tricky - an "offset" knob would be useful.

Here's a couple of screenshots with exactly the same circuit /settings as each other, save the frequency of the generator.

The line you can't see is the whacky one - green. That's the one which dips into notches and bounces or steps - depending much on the offset setting on the generator. But who cares, as long as red behaves itself?

(NB the gain of my floating ammeter is now 500)



Same - faster :
Well before this rate of change, momentum in the coil/mount and whatever else, would surely be interfering?

Chris,

Nice work sir!!

If you really want to show what this circuit is capable of, try this. Make a plot with VR1 at some value, maybe 1 ohm, 10 ohm, whatever, record your red trace. Now replace VR1 with back to back diodes , this will introduce a highly non-linear function in series with the load. Now rerun the simulation and compare the red trace with the previous trace. You should find they are identical!! Note how bad the op amp output looks, but the current thru the load is perfectly linear. How can this be??

Current mode negative feedback at work!!

Best,

Mike

[ChrisR]

There's a glitch with a 741, so your circuit's rubbish




Much better with eg a TL071 though.

Strangely intolerant of little random parasitics..

[mawyatt]

Chris,

That looks like SR limiting in the first plot with the glitches & dead zone around the center. You are moving the rail full swing in 500us, which is pretty quick. If you slow it down considerably these glitches should go away, or diminish considerably. Not sure why the amplitude is so high?? Putting the diodes in was to show the effects of a high non-linerarity in the feedback path.

The second trace with the faster op amp is responding to the delay caused by the shunt capacitor working against the 4 ohm resistor and the overall output impedance of the transistors, 1K resistor and op-amp. This delay coupled with the delay thru the large 0.21H inductor working into the op amp - input is creating an unstable system as you have shown. At some point the phase shift thru this network is approaching -180 degrees and the op amp provides another 180 degrees (at least 180, when you move out in frequency this delay can go beyond -270 degrees!). You are correct about sensitivity to parasitics as I mentioned earlier, best to stay away from fast analog electronics unless you absolutely need the speed, because the design complexity will go up exponentially!! The parasitics are the very reason for my prior comment about not using the high current (& speed) AD8397 op amp in a plug in breadboard!! Shown below.

"For you electronic types that are familiar and know how to work with fast electronics, fast analog op amps and other components can be tricky and don't tend to like plug in breadboards!"

Anyway, this is venturing into an area of higher speed linear and non-linear performance and one that could take weeks, months and years of time to address. Better left to graduate level courses tailored to this subject.

Best,

Mike

[ChrisR]

Yup, with an op amp which I used to consider normal, a TL071 or a CA3140, it's pretty hard to make it show glitches at all.

The L & C were chosen to resonate at 1kHz - I was trying to make it react; you wouldn't normally have unintended strays that size, I would hope.

Amplitude of red is just due to gain(500) in the differential amp so it shows up.

The highest frequencies I ever worked with while I was sill getting my hands dirty 30+ years ago, were baseband TV for cable networks, so ~6MHz, so not very high. It was rapidly getting easier to deal with, even at that time. I then went off at a tangent.

The circuit design software I have here handles breadboards, and is very fussy indeed. If your wires are too long, things "stop working". Overloaded components "explode", too.

If someone wants a student version, let me know. You can build/run "Genie" processor programs on it too.

[mjkzz]

The A's where the red (output) are flat is exactly what I expect, a little delay but I am surprised it is so little, ie so fast.

Not sure what the B's are, but it looks like circuit model is NOT based on schematics of an op-amp, because most op-amp output can not swing in full rail to rail range, in the circuit, it looks like the rail to rail is -15v to 15v, yet the red line can reach -15 (lower B) and +15 (upper B).

Even if not modeled with op-amp schematics, most simulation software should take this into account that op-amp output can not swing rail to rail, typically with 1V less on each side yet this is not the case, so this could be something to think about, though I understand that the red is output of "assistant" op amp for measurement.

Note sure what happens to the other "glitches", beside the spike in left A, the "bump" in both A's does not make sense to me. The spike does not make sense either.

[mawyatt]

Peter,

As Chris mentioned, he did this intensionally! The Red trace is from Chris's high gain (500) differential amplifier he's implemented to"measure" the load current. So the actual load current or output amplifier is not banging into the supply rails, the measurement amp is.

Regarded the A dead zones, these are probably realistic with the 741 op amp and ramp speed. However I'm not sure what's going on with the spike (damped ring) to the left, suspect this isn't going to show up in a real circuit. Also the deviation on the right A area until the result returns back to linear fashion, not sue this is real either.

One thing to note is Chris is using a high gain amplifier to measure the load current, this amplifier could be introducing things other than the rail clipping and is being asked to provide a high gain at a relatively high frequency.

Best,

Mike

[mawyatt]

Yup, with an op amp which I used to consider normal, a TL071 or a CA3140, it's pretty hard to make it show glitches at all.

The L & C were chosen to resonate at 1kHz - I was trying to make it react; you wouldn't normally have unintended strays that size, I would hope.

Amplitude of red is just due to gain(500) in the differential amp so it shows up.

The highest frequencies I ever worked with while I was sill getting my hands dirty 30+ years ago, were baseband TV for cable networks, so ~6MHz, so not very high. It was rapidly getting easier to deal with, even at that time. I then went off at a tangent.

The circuit design software I have here handles breadboards, and is very fussy indeed. If your wires are too long, things "stop working". Overloaded components "explode", too.

If someone wants a student version, let me know. You can build/run "Genie" processor programs on it too.

Chris,

I remember the old days of TV and cable, now TV and internet from space at Ka band!! I was flying back from San Diego recently and had broadband internet service on the flight, and it was as good as being in my office on the ground!! So my seat became my office and I could get some useful work done instead of watching an old movie or looking out the window!!

Thanks for taking the time to simulate all these areas, maybe your students could get involved and show some interest?

Best,

Mike

[mawyatt]

Chris,

I downloaded LTspice from Analog Devices (It's Shareware). They have an Op-07 op amp model which I used. This is a general purpose precision op amp, a step above a 741, but they have no 741 model.

Anyway, here's the schematic and plot for a slow 5 volt ramp producing a +- 100ma peak output load current. I was able to plot the actual load current - the current thur an identical 50 ohm sense resistor (R3) to show the error current (Blue trace). Note the scale, it's in 100s of nano amps.

Anyway, you, Peter, Ray and others can download the LTspice software and run all kinds of simulations, it even will plot device temperature under transient conditions!!

I've spent lots of time with simulators (used to write articles for Spice use), and even have to create my own simulators based upon Berkley Spice for RF and microwave use back in 1990 because there wasn't any good time domain RF simulators then, they all used Harmonic Balance which wasn't going to work for the things we wanted to do.

So I can say this is pretty good SPICE software considering it's free, not up the Cadence standards (which we use with ADS) but doesn't cost $50K either!!

Best,

Mike

[ray_parkhurst]

Is the op-amp model a behavioral type, or simulated subcircuit?

[mawyatt]

It's a subcircuit. But don't know to what level it's modeled.

[mawyatt]

Added a couple back to back diodes to show the effects of highly non-linear load characteristics. Not the lack of effect on the load current waveform as predicted.







Expanded ranges to show effects at crossover.




Added NPN and PNP collector currents to show when they supply load currents, and effects of 1K resistor.