Voltage-current amp

[Doug Williams]

Hi all,

I need some help with an analog circuit, thought maybe this would be the place to ask.  

I work in the EE dept. here at UC Berkeley, supporting the undergrad labs.  The controls systems class does a lab every year where they levitate a ball bearing with an electromagnet: http://www.youtube.com/watch?v=AV-0HNNkviA&feature=related

The "Current Amplifier" in the video was made many years ago by a guy who's retired now, so I don't have schematics or anything.  They're starting to wear out.  I'm pretty sure what we want is something that will push 1 amp through that coil for every volt that's input to it.  I have a decent supply of the now-obsolete LM12's, which the previous circuit is based around, that I could use to make something new.  We'd really prefer to just buy something.  Either way, I really don't have the analog chops to get it done in a timely fashion.  I don't really know where to start, or what to look for, to find something off-the-shelf.

Anyway, any suggestions for how to get this done with the LM12s I have, or, preferably, with an off-the-shelf solution, are much appreciated.

Thanks,

-Doug

[Windell H. Oskay]

> The "Current Amplifier" in the video was made many years ago by a guy
> who's retired now, so I don't have schematics or anything.
> They're starting to wear out.

What is it that's actually wearing out in the circuit?

> I'm pretty sure what we want is something that will push 1 amp
> through that coil for every volt that's input to it. I have a decent
> supply of the now-obsolete

> LM12<http://www.national.com/opf/LM/LM12CL.html#Overview>'s,

> which the previous circuit is based around, that I could use to make
> something new. We'd really prefer to just buy something. Either way, I
> really don't have the analog chops to get it done in a timely fashion. I
> don't really know where to start, or what to look for, to find something
> off-the-shelf.

Redoing this, yourself, from off-the-shelf components is potentially very
easy. One way is to use a high-current op-amp (e.g., LM12, or preferably
something in production like the OPA548T, but avoid the LM675), or you can
use a "regular" op-amp in combination with an output transistor. Do you
think that you'd be able to build a circuit like this if you had a circuit
diagram to work from?

My impression is that there are a number of OTS devices as well, but I'm
much less familiar with them.

[Doug Williams]

What's actually wearing out is a good question.  Based on the tests the course TA performed, it sounds like it's different on each board.  I replaced the LM12s on most of the boards before these tests were performed, since I'd been told by the guy who made them that that was the usual failure point.  

I'd be very comfortable doing the PCB layout and assembly if I had a good schematic, and I feel like I know enough about op amps to gin up a basic circuit, but I'm not confident in my ability to get all the details right and end up with a precise and robust circuit.

---------- Forwarded message ----------
From: Andrew Tinka <ti...@berkeley.edu>
Date: Wed, Dec 7, 2011 at 6:19 PM
Subject: Re: testing current amps w/ inductive load now
To: do...@eecs.berkeley.edu


Hi Doug,

Here's my test setup.

1. Function generator: BNC->two alligator clips connected to "input"
jack of current amp.  Oscilloscope channel 1 also connected to input
jack.

2. Oscilloscope channel 2 connected to the current sense resistor of the
current amp. (2x 2ohm power resistors in parallel = 1ohm).

3. Current amp output connected to maglev coil.  Oscilloscope channel 3
also connected.

4. +/-24V supply connected to current amp.


Photos attached: one with only the probe to the current sense resistor
attached, one with all connections.



Test:
1. Apply 10V P-P, 500 Hz to input.  Should see 10V P-P, 500Hz on sense
resistor.  Should see 500Hz sinusoid of some sort on the current amp
output.

2. Apply 10V P-P, 10kHz to input.  Should see 10V P-P, 10kHz on sense
resistor.  Should see 10kHz sinusoid of some sort on the current amp
output.


Notes:

I tried to figure out the impedance of the coil in order to set an
expected voltage output, however I was unable to get consistent
readings.  Either the coil is significantly non-linear or my lab
technique sucks.  Nevertheless, the amplifier spec (1V in = 1 A out) can
be verified by looking at the current sense resistor, as I mentioned
earlier.

Only one out of the seven current amplifiers passes the tests above.

Full results of the test:
https://docs.google.com/document/d/1roI9FTQ7J6bkPGGKJ_CBRlkWAUeJ6c7kmtU_7IDgy1E/edit?authkey=CJG2uLQL



- Andrew

-Doug

--
You received this message because you are subscribed to the Google Groups "SFanalogcircuits" group.
To post to this group, send email to sfanalog...@googlegroups.com.
To unsubscribe from this group, send email to sfanalogcircui...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/sfanalogcircuits?hl=en.

[Michael Prados]

Hey Doug,

Is there any way you could show us a schematic of the existing circuit, so we could try to do failure analysis on it?  The failure modes that immediately come to my mind are due to shorting the output or an inductive spike, and I think we could design around those.

Alternatively, maybe if you've got a stub version of your new schematic, we could critique and suggest additions to that.

Of course, a schematic could be something from your favorite schematic capture software, or it could be something scrawled on a napkin and photographed with a phone...

-Mike

[Jon Wagner]

From your language: "wearing out" I immediately think of capacitors
going bad. Not sure if that is the case, but certain types of
capacitors can get old and work poorly as a result.

What you're trying to do can generally be accomplished with this type
of circuit:
http://www.ecircuitcenter.com/circuits/curr_src1/curr_src1.htm

For your power range, you could use an OPA549 (big brother to OPA548
recommended below) instead of XOP1 and Q1. If you use for example a
100mOhm power resistor for the Rsense, you'd have to divide down your
input signal with a 10:1 resistor divider in order to get the desired
1 amp per volt.

Jon

On Mar 26, 11:02 am, Doug Williams <therm...@gmail.com> wrote:
> Hi all,
>
> I need some help with an analog circuit, thought maybe this would be the
> place to ask.
>
> I work in the EE dept. here at UC Berkeley, supporting the undergrad labs.
>  The controls systems class does a lab every year where they levitate a
> ball bearing with an electromagnet:http://www.youtube.com/watch?v=AV-0HNNkviA&feature=related
>
> The "Current Amplifier" in the video was made many years ago by a guy who's
> retired now, so I don't have schematics or anything.  They're starting to
> wear out.  I'm pretty sure what we want is something that will push 1 amp
> through that coil for every volt that's input to it.  I have a decent
> supply of the now-obsolete

> LM12<http://www.national.com/opf/LM/LM12CL.html#Overview>'s,

[Doug Williams]

Mike, unfortunately the closest thing I have to a schematic is the board you see in the video.  It's only two layers, so it wouldn't be impossible to reverse engineer the schematic, but it would be a bit of a pain.  Perhaps you can suggest some additions to the circuit proposed by Mr. Wagner:  http://www.ecircuitcenter.com/circuits/curr_src1/curr_src1.htm since I'll be driving a coil instead of a purely resistive load?

Speaking of that circuit, since the max input is meant to be +/- 10V, the max output should be +/- 10A, so with a 100Ohm sense resistor, I'd need a +/- 1000V supply, which doesn't quite seem reasonable.  Is there any reason I shouldn't use a 1Ohm sense resistor to keep my supply more reasonable?  Currently we have +/- 24V supplies (which are also starting to go bad :( )

-Doug

[Michael Prados]

Doug, I hope to look at this a little closer soon, but for now let me point out that Jon suggested 100mOhms (milli-Ohms,) aka 0.1 Ohms.

-Mike

[Jon Wagner]

I found a pretty cool online circuit editor and simulator and used it
to draw the circuit. Way easier than describing in words. Be aware
that this circuit won't simulate because it of course doesn't have the
OPA549 in its library.

https://www.circuitlab.com/circuit/py6tvu/current-source-for-coil-driver/

If you want to make this circuit better, you would create a separate
gain stage for the sense resistor instead of dividing down the input
signal.
Jon

On Mar 27, 10:30 am, Doug Williams <therm...@gmail.com> wrote:
> Mike, unfortunately the closest thing I have to a schematic is the board
> you see in the video.  It's only two layers, so it wouldn't be impossible
> to reverse engineer the schematic, but it would be a bit of a pain.
>  Perhaps you can suggest some additions to the circuit proposed by Mr.
> Wagner:  http://www.ecircuitcenter.com/circuits/curr_src1/curr_src1.htmsince
> I'll be driving a coil instead of a purely resistive load?
>
> Speaking of that circuit, since the max input is meant to be +/- 10V, the
> max output should be +/- 10A, so with a 100Ohm sense resistor, I'd need a
> +/- 1000V supply, which doesn't quite seem reasonable.  Is there any reason
> I shouldn't use a 1Ohm sense resistor to keep my supply more reasonable?
>  Currently we have +/- 24V supplies (which are also starting to go bad :( )
>
> -Doug
>

[Doug Williams]

Thanks Jon, I'll try prototyping this and see how it goes.

-Doug

[Jon Wagner]

Doug,
You might want to add a snubber circuit on the output of the OPA549.
It is described in the datasheet - basically to keep the signal
stable. You should also read the section on the current limit setting
using an external resistor. Add dc-link capacitors on the power
supply rails near the OPA chip.
Jon

On Mar 28, 12:31 pm, Doug Williams <therm...@gmail.com> wrote:
> Thanks Jon, I'll try prototyping this and see how it goes.
>
> -Doug
>

> On Wed, Mar 28, 2012 at 12:23 PM, Jon Wagner <wagnerdesigns...@gmail.com>wrote:
>
>
>
>
>
>
>
> > I found a pretty cool online circuit editor and simulator and used it
> > to draw the circuit.  Way easier than describing in words.  Be aware
> > that this circuit won't simulate because it of course doesn't have the
> > OPA549 in its library.
>

> >https://www.circuitlab.com/circuit/py6tvu/current-source-for-coil-dri...

[Doug Williams]

Okay, I've made some changes to the circuit you sent.  It now looks like this: https://www.circuitlab.com/circuit/h3ycgs/current-amp/

With a 1.87kOhm resistor to limit the current to 8A, not shown because it doesn't seem like I can add pins to the OPA in this program.  

I'm not exactly sure how the snubber circuit works.  The diodes are meant to keep the output from going above the positive rail or below the negative rail, yes?  Why do they need to be schottky diodes?  How much current should I spec for them?  The RC is just a filter to blunt any voltage spikes, yes?  Should it be connected to ground, or just in parallel with the load?

Thanks much for your help!

-Doug

[Jon Wagner]

Looks right to me - the snubber, diodes, and caps are all in the right place and look like reasonable values.  The snubber just limits the dv/dt on the output, which you need for inductive loads such as motors/relays/coils.

http://en.wikipedia.org/wiki/Snubber 

I forgot to ask what bandwidth you need for the controls solution.  You should be able to do 10KHz without a problem.  You could get to 100KHz bandwidth if you put gain on the sense resistor and ditch the input resistor divider.

Jon

 

> > > > To post to this group, send email to sfanalogcircuits@googlegroups.com

> > .
> > > > To unsubscribe from this group, send email to

> > > > sfanalogcircuits+unsubscribe@googlegroups.com.

> > > > For more options, visit this group at
> > > >http://groups.google.com/group/sfanalogcircuits?hl=en.
>
> > --
> > You received this message because you are subscribed to the Google Groups
> > "SFanalogcircuits" group.

> > To post to this group, send email to sfanalogcircuits@googlegroups.com.

> > To unsubscribe from this group, send email to

> > sfanalogcircuits+unsubscribe@googlegroups.com.

> > For more options, visit this group at
> >http://groups.google.com/group/sfanalogcircuits?hl=en.

--
You received this message because you are subscribed to the Google Groups "SFanalogcircuits" group.

To post to this group, send email to sfanalogcircuits@googlegroups.com.
To unsubscribe from this group, send email to sfanalogcircuits+unsubscribe@googlegroups.com.