1. Home
  2. Electronics Design
  3. Is this circuit idea crazy?

Is this circuit idea crazy?

Problem: I'm cheap, I want to do a high-side current monitor, and low offset opamps aren't cheap.

Tools: it's all being driven from a microprocessor, so I can do oddball things if I want to.

Thought:

Just use a plain 'ol op-amp, and once every second or ten turn off the charging circuit & measure the current sense output. This should tell me the op-amp offset, which I can then subtract in software.

Problem, again:

The circuit I'm thinking of using (and many others, for that matter) are referenced to ground; negative offsets simply will not register.

Another thought:

Shift the output, in some controlled fashion, so that it never goes to zero, even when the opamp is at its _maximum_ (not _typical_, thank you Digikey) offset.

See the circuit below. It has a small error from the finite transistor beta -- but I ain't gonna worry about that, it'll be comparable to errors due to resistors being off. The total gain will be set by Rfb/Rin times the ratio of (Rshift|Rset)/Rtop, degraded a bit by the transistor beta. C1 will serve to keep things stable, and to act as a low-pass filter. When I turn things off to measure offset I will have to wait for the filter to settle -- but I can survive that.

An obsolete thought:

This is what I was originally going to ask, but then I thought of using the reference. So I'm going to ask it anyway, just to see what folks think.

Before it occurred to me that analog to digital converters imply references, I had thought of the below circuit, but with the Rshift resistor omitted (and this was the crazy part) Rbias set -- instead of equal to Rin|Rfb -- to zero, or at least to less than Rin|Rfb. Then the bias current of a cheap current-sourcing op-amp would lift the negative pin a little bit, and insure me some offset on my output.

So my question was going to be how stable folks would expect the op-amp offset to be, and whether or not the idea of intentionally inducing offset in this manner was a good idea, or just too weird.

Your thoughts:

So -- whadda ya think? Is it going to work? Chopper-stabilize your cheap op-amp by chopping the whole freaking supply!!

Rsense ___

from source >---o---|___|---o----------------------o-------> to load >

| | | | | | .-. .-. .-. | | | | | | | | Rin | | Rbias | | Rtop '-' '-' '-' | | Rfb | | | ___ | | | .-----------|___|----o | | | | | | | || | | | o-----||------. | | | | || | | | | | C1 | | | | | |\ | | '-----------)-o----|-\ | |< | | >-----o----| Q1 '------|+/ |\ MMBT3906 |/ | | U1 | o-----> current V >

___ | Vref o-------|___|----o Rshift | .-. | | | | Rset '-' | | === GND (created by AACircuit v1.28.6 beta 04/19/05

formatting link 

--
www.wescottdesign.com
Reply to
Tim Wescott
Loading thread data ...

Value of Rsense?

Current Range?

Accuracy (and offset) needed? ...Jim Thompson

[On the Road, in New York] 
--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
I love to cook with wine.     Sometimes I even put it in the food.
Reply to
Jim Thompson

Oh, yea -- details.

Current range is 2A. Accuracy wouldn't need to be much better than 50mA, but better is always -- better. I had been planning on 15m-ohm for Rsense, but that was limited by my desire to keep the op-amp cheap. Now I may lower it, and raise Rfb to compensate.

--
www.wescottdesign.com
Reply to
Tim Wescott

At 2 Amps, what output current from the PNP? ...Jim Thompson

[On the Road, in New York] 
--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
I love to cook with wine.     Sometimes I even put it in the food.
Reply to
Jim Thompson

You're wasting your time trying to gain it up... you also multiply the effects from the offset of the OpAmp by the gain.

Chopping could be worthwhile if you really need to cure the offset. Or spend $$$$ on a fancy OpAmp ;-) ...Jim Thompson

[On the Road, in New York] 
--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
I love to cook with wine.     Sometimes I even put it in the food.
Reply to
Jim Thompson

Which brings to mind... how fast is the current changing? Or is it relatively stationary? ...Jim Thompson

[On the Road, in New York] 
--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
I love to cook with wine.     Sometimes I even put it in the food.
Reply to
Jim Thompson

Not crazy, but several people make high-side current sense chips to do pretty much this function.

This might be good enough, for about 10 cents worth of parts...

ftp://jjlarkin.lmi.net/Isense.JPG

Actually, that may be too many parts....

John

Reply to
John Larkin

Man, you're a concrete thinker (:-).

I was thinking 2 - 3mA, if that looks good when I get into the detail.

--
www.wescottdesign.com
Reply to
Tim Wescott

Ahh, but that's the point -- I can chop it by switching off the charge command at which point (after settling) I'm ensured zero current. Then I measure, then I charge for a while.

It's a battery. It isn't going to change very fast unless something very bad is happening -- and that's what fuses are for.

I'll probably be sampling it at 10kHz, which is way more than enough.

--
www.wescottdesign.com
Reply to
Tim Wescott

My web browser doesn't like your FTP site.

And the high-side current sense chips cost way more than the parts I've slapped down. Unless you know of one that's less than $1.00 in quantity...

--
www.wescottdesign.com
Reply to
Tim Wescott

As usual, Larkin's sketch has no values, so it can't be challenged as crap. It _can_ work with matched transistors, it's a variation of...

formatting link

At the low sense R you want, I'd tend toward chopping it. Any reason you can't use a larger sense resistor? ...Jim Thompson

[On the Road, in New York] 
--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
I love to cook with wine.     Sometimes I even put it in the food.
Reply to
Jim Thompson

formatting link

John

Reply to
John Larkin

It certainly is not.

But why would it need matched transistors? He's going to auto-zero the offsets out.

That's crazy. It doesn't do anything like what he needs.

John

Reply to
John Larkin

Using a bipolar PNP transistor at Q1 is slightly crazy, a p-channel MOSFET doesn't have a beta-error, even if it is going to be more expensive.

The amplifier - U1 - is going to need to have an input common mode range that includes the positive rail, and input off-set specs have been known to be down-graded when the inputs are close to the rails.

You might think of trading off the price of an amplifier with an input common mode range that extends to the rails against the price of a more or less precise divider array pair to shift the voltage being measured away from the rail.

High-side current-sensing devices tend to put both the op amp and the divider into a single package, but I've never had occasion to use one and don't know what's available or the kinds of prices being asked.

As for genuinely low offset op amps, I always liked the Linear Technology LT1006, LT1013 and LT1014 single, dual and quad parts, but they all cost a couple of bucks, and their common mode range only includes the negative rail.

--
Bill Sloman, Nijmegen
Reply to
Bill Sloman

Neither does mine. I tried Firefox and Seamonkey. Then I tried an FTP program. That didn't work either. Weird.

mike

Reply to
m II

BCX71K has a min beta of 380 and a max of 630, so its beta error variation contribution will be a tenth of a percent or so. We pay 1.8 cents for them. The resistors will probably be the main gain error.

BCX70 and BCX71 are beta graded, so you can do things with them that give old farts apoplexy, like design things that count on beta.

John

Reply to
John Larkin

Grumble. Snarl. I suppose I should have a web site.

John

Reply to
John Larkin

Back when integrated circuit op amps still cost enough that we designed stuff with transistors - early 1970's - we had high beta npn's and pnp's.

I used quite a few BC184's and BC179's at that time. If we really wanted to minimise base current we'd stick two together in a Darlington or complementary Darlington pair - with something between 680k and 68k in parallel with the base-emitter junction of the second - driven - transistor to keep the collector current of the driving transistor in the high beta area, and to keep the bandwidth respectable.

A p-channel MOSFET did better, with fewer parts, and didn't take anything like as long to explain to the pointy-headed manager, which did save design time, which was the major cost for low-volume products.

--
Bill Sloman, Nijmegen
Reply to
Bill Sloman

w
l
e

or

o

the

r

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0|

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0|

=A0 =A0 =A0 =A0 =A0 =A0 =A0.-. | | =A0 =A0 =A0 =A0 | |

| | | Rin =A0 =A0 | | Rbias =A0 =A0 =A0 =A0 =A0 =A0

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0'-'

=A0 =A0 =A0 =A0Rfb =A0 =A0 |

=A0 =A0 =A0 =A0___ =A0 =A0 |

-|___|----o

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0|

| =A0 =A0 =A0 =A0 =A0 =A0 |

---. =A0 =A0 =A0|

| =A0 =A0 =A0| =A0 =A0 =A0|

=A0 =A0C1 =A0 | =A0 =A0 =A0|

=A0 =A0 =A0 | =A0 =A0 =A0|

=A0 =A0|<

=A0| =A0>-----o----| =A0 Q1

=A0 =A0 =A0 =A0 =A0 |\ =A0MMBT3906

=A0 |/ =A0 =A0 =A0 =A0 =A0 =A0 =A0|

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 |

=A0 =A0U1 =A0 =A0 =A0 =A0 =A0 =A0 |

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 o-----> current V >

=A0 =A0 =A0 =A0 =A0 ___ =A0 =A0 |

---|___|----o

=A0 =A0 =A0 =A0 =A0Rshift =A0 |

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0.-.

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| |

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| | Rset

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0'-'

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 |

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 |

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0=3D=3D=3D

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0GND

ADM4073 close enough?, analog say 0.99$, digikey 1.16$ @ 3000

-Lasse

Reply to
langwadt

Works fine for me. Firefox, DSL.

-- Cheers, James Arthur

Reply to
dagmargoodboat

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.