250mV + and - from a single output bench supply?

[Chris Wilson]

I am building the circuit attached and want to test it on a breadboard before building the phase meter part of the .pdf file to drive it. It is to motorize a variometer in a low frequency antenna matching coil.. As such I need to feed a maximum of 250mV into the input of the motor drive circuit, and swing it from positive to negative and vice versa and check the motor drive goes clockwise and anticlockwise as it should. How can I get a low current source of plus and minus 250mV from a single output bench supply, should I be thinking in terms of a resistive potential divider?

Thanks.

[Ian.M]

See the chassis symbol after the first OPAMP?  That OPAMP is splitting the supply to provide a virtual earth and your test signal must be referenced to that.  The easiest fix would be a 100K pot wired across D1,D2 (which provide the threshold voltages for the comparator) feeding a non-inverting OPAMP stage with a gain of approx 1.5. (i.e resistor from -in to virt. earth needs to be about double the feedback resistor)  Add 0.1uF to the virtual earth from the pot wiper for decoupling. 

[Kalvin]

You may want to add a series resistor to the output of the op amp generating the virtual ground in order to make it more stable.

[Chris Wilson]

If I got my dual output supply from work could I mimic the meter output of the phase meter circuit and run the + and - in where it is intended in the motor drive circuit?

Thanks for the replies.

[Ian.M]

You want quick&dirty?  Connect a 100R pot across two AA batteries. For easier adjustment, add a 330R resistor in series with each end of the pot track. Take the output from the strap between the two batteries and the pot wiper.

[dom0]

That virtual earth TL084 is going to be rather unhappy.

Here's a circuit that's essentially unconditionally stable, low cost, and higher output current. Output resistance a couple Ohms. https://www.mikrocontroller.net/topic/258266#2675651

The 7667 seems like huge overkill to me here.

[Chris Wilson]

In what way will it be unhappy? I am not a social worker or psychologist, and so long as it works and doesn't set alight I am happy to ignore its mental well being, barsteward that I am I was hoping not to have to veer off at a tangent as I far too inexperienced to evaluate other options effectively.

Thanks.

[Chris Wilson]

You want quick&dirty?  Connect a 100R pot across two AA batteries. For easier adjustment, add a 330R resistor in series with each end of the pot track. Take the output from the strap between the two batteries and the pot wiper.

Quick and dirty should be fine! Just looking for a sanity check that it all could work before making time consuming hardware and then finding it has "issues". Are you saying I can, using the above quick and dirty method, inject + / - where the phase meter connection is shown? Thanks again Ian.

[dom0]

In what way will it be unhappy?

The TL0xx have a high output resistance and poor phase margin at unity gain, so these circuits are almost always unstable and oscillating. Sometimes it's not directly visible on the scope, but only evident by high current consumption of the OP.

It might work in this application, but good engineering it isn't.

[Ian.M]

OPAMPs don't like driving large capacitive loads directly.  That rail-splitter breaks all the rules and is likely to become unstable and maybe overheat the OPAMP.  There is negligible load on the virtual earth so I'd simply delete the first OPAMP and its 47K resistors, which would let the 4.7K resistors and diodes divider at the comparator provide the virtual earth.  If you need that virtual earth for other circuits, the first OPAMP stage could be turned around to bufer it, but you'd still need to avoid direct capacitive loading on the output of an OPAMP that isn't specifically rated to drive them..

[Chris Wilson]

Thanks again, I am well out of my comfort and knowldge zone, all I can say is I am told I need to build just the phase meter section of the circuit in the .pdf file, and instead of the meter, (or possibly as well as the meter, unsure of that...) I feed where the meter is / was into the control circuitry. Another poster on a UK forum has commented on the instability potential, when the circuit deviser posted his schematic, by saying:

Can I warn against one feature shown in the circuit diagram
The two 10uF electrolytics connected from the output of the first opamp to the supply rails should not be there.   They cause instability as opamps are not designed to drive capacitive loads and frequently go unstable.


If an opamp is used to deliver a mid voltage rail like this, the temptation to add extra decoupling should be avoided.   The opamp output gives a very low source impedance  that will not benefit from any extra decoupling.


If you do not get nay instability using that circuit as it stands, then you are very lucky  - or perhaps it is oscillating and not noticed


I recently fell into this trap quite accidentally.   It is so "natural" to want to  decouple supply rails with capacitors placed everywhere  that I placed several 100nF caps on a PCB without thinking.  All of them had to be removed to stabilise the resulting circuitry."


So I am certainly taking heed of all these warnings!


The UK poster went on to say:

"Actually, looking again at the circuit, it is not necessary to use an opamp in that position to define the mid rail for the comparator reference.    Just two equal; value resistors and one electrolytic decoupling the junction to ground (Definitely not two caps to top and bottom in the original configuration shown ... think about it...) 


You only need an opamp to define a stiff mid rail voltage where there is likely to be any significant current draw.  Or many stages need to be referenced to it - such as a multi stage filter.    For single stage configurations, just two resistors and a bit of decoupling is enough".

So I think I will be breadboarding it less the first op-amp bit and less one of the 10uF caps, if you are in agreement?

Thanks again everyone.

[Ian.M]

The reason for two caps is that it keeps it symmetrical during power-up.  If you only use one, the motor is likely to drive hard one way for half a second or so while the cap charges to its final mid-rail voltage.  As I said above, all you need to do is delete that first OPAMP and the 47K resistors from that schematic.  The virtual earth (mid-voltage rail) will then be provided by the existing chain of 4.7K resistors + BAT42 diodes, and decoupled by the 10uF caps.

[Chris Wilson]

OK Ian, I have nearly all the components and should be able to breadboard it next week, I'll post my results, many thanks for your advice.

[Ian.M]

Beware: the phasemeter is grounded to true earth ground.  To use it with your motor drive circuit, either all that's got to be fully floating (not good with motor drive currents to deal with) or you've got to float the phasemeter.  To do that, break the bottom ground connection and put a 0.1uF ceramic cap in parallel with a 470pF one between the bottom rail and earth ground.  The coax sleeve, the sockets and the case should still all be hard grounded.  Unless you want the RF Voltmeter/Ammeter, you could delete everything to the right of R2.  If you haven't got a suitable meter, I'd replace the meter with a 150R resistor and put 0.1uF across the wires you are bringing out as close as possible to where they come out of the case + twist them together and loop them several times through a ferrite ring to try to keep the RF where it belongs and away from the OPAMP.

[Chris Wilson]

I breadboarded this today, and although it works the 1 megohm trimmer pot needs to be set at circa 1k (near at the end of its travel) to give a suitably reduced sensitivity to get the drive motor to stop at a given point. I think this is the purpose of the 1 meg pot? I am using the quick and dirty split 1.5 volt dry cells to input a dummy phase signal, and this is the circuit I built shown below. I may have misunderstood how the original circuit was suggested to be modified? Thanks.

[Ian.M]

What's it doing if the pot's set too high?

[Chris Wilson]

Hi again Ian. I didn't have a 100 Ohm preset to dummy the phase meter, so I used a 330 ohm preset pot, wired across two used 1.5 volt cells, outer legs of the preset to 3V, inner leg to my circuit, and centre tap of the 2 cells also to the input of my circuit. With the 1 meg multi turn preset in the circuit set much more than 1K Ohm the test preset becomes very very touchy and the drive motor just keeps running in one or the other direction. By getting the circuits 1 meg pot down to around its minimum of just under 1K Ohm the test pot is much less sensitive and I can use it to drive the motor either way with movements of the pot that are not super critical. I am not sure if my test is valid, or if the phase meter part of the circuit, currently un-built, will be outputting less "severe" voltage swings than my test pot and batteries.
I hope the above is clear, it's difficult to describe the interaction. Thanks.

[Ian.M]

Part of the problem is that you have a proportional term and an integral term (set by the feedback cap in parallel to that 1M resistor.  If the integral term diminates and input is too large for too long, the cap charges up till the integrator 'rails', and the motor wont stop until you either reverse the input long enough to get it back to the dead-band of the comparator output stage or zero it and wait for it to decay back to the deadband.  As you have no warning its approaching the deadband.  you are almost bound to overshoot and rail it the other way.   Try connecting a DMM on volts (or better yet: a  >10K/volt center zero analog voltmeter)  between the output of the OPAMP with the preset and the virtual earth to give you some warning that you need to zero the simulated phase PDQ.

It will be somewhat better behaved on the actual variometer as that will close the feedback loop, with presumably a reasonably linear (or at least smooth without local minima or maxima) feedback signal via the phasemeter, but it will still get a bad case of the jitters if you turn up that feedback pot too much in an attempt to reduce the phase error.

If you want to do better, you'd need a PWM motor drive with a carefully tuned P.I.D. control system.  The important difference is that the motor slows down as it approaches the zero phase point.

Don't forget limit switches so it cant slam the varistor mechanism into its end stops.

[Chris Wilson]

Hi again Ian, I experimented with a volt meter on the output pin 7 of the first opamp and virtual ground and it gave me a much better feel for how it interacted. It's now very linear and controllable from the test cells and preset. I also found dropping the supply from 12V to 9V slowed the drive motor to a crawl, given even at 12V it's a 1 RPM motor. I will build the phase meter stuff next and see how it interacts with that now. I really appreciate your patient help.

I am slightly familiar with PID stuff as it's used for various control functions on automotive engine management systems, as I am a race car engineer by day, but for now I will stay with what I have and try and keep it simple, again, thanks very much!

[Chris Wilson]

Can anyone please explain why the circuits below appear to work functionally the same? I breadboarded it and failed to connect the 47k divider resistors initially, but it seemed to work nonetheless. When I noticed I had hit the wrong "socket" on the breadboard with the divider and rectified the mistake it seemed to work just the same! Thanks, sorry for the bandwidth, but not understanding will bug the hell out of me I am presuming the second schematic is how it SHOULD be...Or have I misunderstood?

[Ian.M]

They aren't needed.  The series chain of the two diodes and the 4.7K resistors split the supply to give the virtual earth anyway.  I'd keep the 10uF caps in circuit though.

[Chris Wilson]

Ahh, OK, sorry to have been a pain, much appreciated Ian, have a good weekend.