All About Circuits Forum Op~Amp Current Regulator?
 Register Blogs FAQ Members List Today's Posts Search Today's Posts Mark Forums Read

 General Electronics Chat Discussion forum for general chat about anything electronics related, including asking questions about material in the All About Circuits E-book, Worksheets, and Videos.

#21
07-12-2009, 11:04 AM
 millwood Guest Posts: n/a

now, the 2nd chart. Rather than one transistor of hFE=200-ish, we are going to use a "super transistor" in the form of a darlington. its effective beta would be 40k + (200x+ bigger than in the first chart), and will drop two Vbe, right?

again, the green trace is the output current and the red trace is the effective beta of the output device.

question:
1) do they look identical to the chart where no transistor or just one transistor is used?
2) why?

once you work out the math - which is extremely simple, it will be clear to you. and it will also help you understand negative feedback and how it works here.
Attached Images
 VCCS - Darlington.PNG (31.8 KB, 9 views)
#22
07-12-2009, 04:10 PM
 ELECTRONERD Senior Member Join Date: May 2009 Location: Outer Space Posts: 1,145

Thanks everyone! I notice that these circuits have low CMRR. I thought of using an in~amp instead, but I intend to learn about op amps first and how to cope with these constraints.
#23
07-12-2009, 04:23 PM
 ELECTRONERD Senior Member Join Date: May 2009 Location: Outer Space Posts: 1,145

Op~Amps are completely new to me and so I don't know how to apply the math here to get want I would like. Also, I noticed a couple schematics that had high value resistors at the output of the op~amp; why do they do this? I think It might have been for an oscillator circuit.

I went ahead and picked up a book from the Library, "Op amps for Everyone," by Ron Mancini. It seems to be a good book and could help me get the fundamentals down and also the math.

Thanks!
#24
07-12-2009, 04:32 PM
 millwood Guest Posts: n/a

there are quite a few such books on the net. Analog / Walter Jung has an excellent book on opamp, and full of historical perspective as well.

TI also has a great book on opamps.

the book you have is excellent as well.
#25
07-12-2009, 04:34 PM
 ELECTRONERD Senior Member Join Date: May 2009 Location: Outer Space Posts: 1,145

Quote:
 Originally Posted by millwood there are quite a few such books on the net. Analog / Walter Jung has an excellent book on opamp, and full of historical perspective as well. TI also has a great book on opamps. the book you have is excellent as well.
Thanks Millwood, you've been a big help. Could you explain how you came up with the resistor values of your last circuit? I'd appreciate it!
#26
07-12-2009, 05:08 PM
 millwood Guest Posts: n/a

Quote:
 Originally Posted by ELECTRONERD Thanks Millwood, you've been a big help. Could you explain how you came up with the resistor values of your last circuit? I'd appreciate it!
opamps are some of the simplest electronic devices to analyze. at least ideal opamps, .

two principles when it comes to (ideal) opamps's DC performance:

1) they have infinite open loop gain. That means that once feedback is applied, the voltage differential between an ideal opamp's inverting input pin and non-inverting input pin is always zero.

2) they have infinite input impedance. so no current is going into or output the input pins.

once you establish those two principles, the rest is a breeze.

1) by principle #1, you can that the voltage on the inverting pin is also Vin.

2) take a look at node A. you know that all current going into that pin has to sum up to all current going out of that pin.

so I3=I2+I4, thanks to principle #2.

3) but, since the voltage at node A is Vin, per Ohm's law, the following must be true:

I3=Vin/R3
I4=(Vref-Vin)/R4, and
I2=(Vo-Vin)/R2

so we have

Vin/R3=(Vo-Vin)/R2+(Vref-Vin)/R4

solve for Vo:

Vo/R2=Vin/R3+Vin/R2-(Vref-Vin)/R4, or
Vo=Vin(1+R2/R3) - (Vref - Vin) *R2/R4

the 1st term is the non-inverting gain for Vin, and the 2nd term is the inverting gain for Vref.

thus the output current on R1 would be Iout=Vo/R1.

4) take a look at node B. and you have the current going through the load R5

I5=Iout+I2.

if R2 >> R1, you have I2<<Iout, so

I5=Iout+I2 = ~Iout = Vo/R1.

as you can see, I5 has nothing to do with R5's value.

so when you play with the numbers, you want to make sure that the current going through the feedback network is sufficiently small.

R1 is selected so that it doesn't overload the opamp: it needs to be sufficiently large so the opamp works below its maximum output current specification.

R2/R3/R4/Vref are selected to give you the right Vin - Iout relationship.

well, that's just in a nutshell.
Attached Images
 opamp vccs.PNG (26.1 KB, 15 views)
#27
07-12-2009, 05:28 PM
 The Electrician Senior Member Join Date: Oct 2007 Posts: 1,722

Quote:
 Originally Posted by millwood whether you need a transistor depends on if the load can be floated, ie. both ends off the ground.
Fair enough, but the Howland circuit doesn't need a transistor to drive a grounded load, so what is the advantage of your circuit in this case?

I said "It seems to me that the output current will be strongly influenced by the hfe of the transistor, rather than being determined by resistor ratios."

You said:
Quote:
 Originally Posted by millwood no.
I wasn't talking about high beta transistors like you mention in another post. I was supposing that ordinary transistors with a wide production spread of betas would be installed in your circuit, and with fixed Vin and Vref, determine the load current. A 10 to 1 spread is not atypical; let the spread be 30 to 300, for example. Then the output current will have variations in the several percent range, enough to disqualify the circuit from being called a precision current source. If we similarly change the active device (the opamp) in the Howland circuit over the production spread of devices of the same type, the variation of output current will be orders of magnitude less.

A good way around this problem is to use a mosfet, such as the BS170. Since there's no base current, the output current won't show the variations of output current as we plug in various BS170s from the production spread.

I said "The original circuit can have an output impedance approaching infinity, and the current is determined by resistor ratios, and is therefore not much affected by temperature, or by parameter variations of a single transistor.
"

You said:
Quote:
 Originally Posted by millwood it will be affected by temperature, unless you go out of the way to take care of it.
I was referring to the Howland circuit, of course. I suppose one could say that "it will be affected by temperature" in the sense that every electronic circuit ever built "will be affected by temperature." to some minuscule degree. But if the opamp is a modern high-gain, low offset-tempco device, the output current change with temperature will be primarily determined by the temperature coefficient of the resistors. Decent resistors will have a tempco in the 10's of ppm range, and the change in output current due to that tempco will be entirely negligible compared to the effect of changes with temperature of a single bjt's parameters.

Good opamps are readily available, so you don't have to go much out of your way to get very good temperature invariance from the Howland circuit. You could get a multiple of matched, extremely low tempco resistors if you wanted ultimate performance.

Doing that in your circuit would be a waste as long as that single bjt is there; it will dominate the temperature variations, and you won't be able to get low ppm performance. Using a mosfet would be much better.

I said "Do you have an expression for the output impedance for your circuit? Won't it be essentially just the ro of the transistor? That may not be very high."

Quote:
 Originally Posted by millwood I am not sure if i can give you one but I am imagine that it is pretty high, given that it is a ccs.
The Howland circuit gets its high output impedance by balancing positive and negative feedback. In theory, there could be an exact balance, and Zout could be ∞. In practice, it can be trimmed to be very high.

Can your circuit be trimmed to improve the output impedance?

In another post you attempt to determine Zout of your circuit, but you've done it incorrectly. I'll address that in a response to that post.
#28
07-12-2009, 05:42 PM
 millwood Guest Posts: n/a

Quote:
 Originally Posted by The Electrician I wasn't talking about high beta transistors like you mention in another post. I was supposing that ordinary transistors with a wide production spread of betas would be installed in your circuit, and with fixed Vin and Vref, determine the load current. A 10 to 1 spread is not atypical; let the spread be 30 to 300, for example. Then the output current will have variations in the several percent range, enough to disqualify the circuit from being called a precision current source. If we similarly change the active device (the opamp) in the Howland circuit over the production spread of devices of the same type, the variation of output current will be orders of magnitude less.
that's only a problem if you haven't figured out how the circuit works. work out the math - feel free to use my examples above, and you will know that you are simply killing yourself over a problem that doesn't exist.

Quote:
 A good way around this problem is to use a mosfet
a good way around this problem is to understand it first and then see if the problem is for real or not.

Quote:
 The Howland circuit gets its high output impedance by balancing positive and negative feedback. In theory, there could be an exact balance, and Zout could be ∞. In practice, it can be trimmed to be very high. Can your circuit be trimmed to improve the output impedance?
I don't know that. since any disturbance in the howard's output is equally distributed to the inverting and non-inverting end, and it is just applied to the non-inverting end in this circuit, I think the answer is now.

But THAT has a few papers on that so do check them out.

Quote:
 In another post you attempt to determine Zout of your circuit, but you've done it incorrectly. I'll address that in a response to that post.
would love to see it.
#29
07-12-2009, 05:44 PM
 The Electrician Senior Member Join Date: Oct 2007 Posts: 1,722

Quote:
 Originally Posted by millwood here are two dc sweep charts, one for R5 (the load) at 22ohm, and another for R5 at 220ohm, or 10x bigger. you can see for yourself. or you can work out the math for yourself as well. BTW, I hope you know why the curve is a little bit flat on the right for the R5=220ohm chart.
This is not the circuit to which my comment referred.

You've changed the circuit. In the first circuit you posted (in post #9), there was no R5. I had to guess which resistor was intended to be the load and I guessed R1.

I will grant you that in this changed circuit, the drive R5 receives will be high impedance, an approximate current source.

But, you've measured the output impedance incorrectly.

In the first place, you should not connect the + input of the opamp to the - input. You should ground the + input because under normal operation of the circuit, it's driven from a zero impedance voltage source.

Furthermore, the output impedance is not the impedance at the node between R5 and R1 with respect to ground. The output impedance is the impedance between the opamp output and the node between R5 and R1; that's the impedance seen by R5.

What you should do is put an AC voltage source in series with R5 and determine the AC current out of that source. Then the ratio of that voltage to the current out of that source will be the output impedance.
#30
07-12-2009, 05:47 PM
 millwood Guest Posts: n/a

Quote:
 Originally Posted by The Electrician This is not the circuit to which my comment referred. You've changed the circuit. In the first circuit you posted (in post #9), there was no R5. I had to guess which resistor was intended to be the load and I guessed R1.
you guessed wrong then.

Quote:
 But, you've measured the output impedance incorrectly.
let me just say that you should take that issue up with those nice folks at Labcenter or Orcad/CADENCE because they apparently needed to revise their training manuals.

and those textbooks too, .

 Tags current, opamp, regulator

 Related Site Pages Section Title Worksheet Algebraic equation manipulation for electric circuits Worksheet Nonlinear opamp circuits Worksheet Active filters Worksheet Opamp oscillator circuits Worksheet Linear computational circuitry Worksheet Summer and subtractor opamp circuits Worksheet Inverting and noninverting opamp voltage amplifier circuits Worksheet Negative feedback opamp circuits Video Lecture Basic Amplifier Configuration (Part 1) - Op Amps and Op Amp Circuits Textbook Current mirrors : Bipolar Junction Transistors

 Similar Threads Thread Thread Starter Forum Replies Last Post TechnoMan The Projects Forum 9 11-15-2008 08:13 PM relicmarks General Electronics Chat 9 11-15-2008 03:18 AM kk57 General Electronics Chat 4 10-25-2007 08:51 PM childs72 The Projects Forum 14 03-26-2006 10:51 AM alphaone Physics 4 12-23-2005 06:03 PM

 Thread Tools Display Modes Linear Mode

 Posting Rules You may not post new threads You may not post replies You may not post attachments You may not edit your posts BB code is On Smilies are On [IMG] code is On HTML code is Off Forum Rules
 Forum Jump User Control Panel Private Messages Subscriptions Who's Online Search Forums Forums Home Electronics Forums     General Electronics Chat     The Projects Forum     Homework Help     Electronics Resources Software, Microcomputing, and Communications Forums     Programmer's Corner     Embedded Systems and Microcontrollers     Computing and Networks     Radio and Communications Circuits and Projects     The Completed Projects Collection Abstract Forums     Math     Physics     General Science All About Circuits Commmunity Forums     Off-Topic     The Flea Market     Feedback and Suggestions

All times are GMT. The time now is 02:25 AM.