DIY uCurrent Design

[Jay_Diddy_B]

Hi group,

I am the process of building my own ucurrent. I have deviated a little from Dave's original design. Some of the design decisions were made because of the parts I had to hand.

I certainly don't want to detract from Dave's uCurrent project.

Here is my version:

Schematic:



I have decided to use separate jacks for the three current ranges. This simplifies the switching. I have also used a SPDT switch for the power and a tact switch for the battery test. I used a Linear Tech LTC2050CS8 for the main amplifier

Component side of the board: (Not quite finished)



Top side:



I have made the board a little bigger, to fit the Hammond 1591 BBK box, I have not been able to find the UB5 Jiffy box outside of Australia.

Jay_Diddy_B

[dannyf]

Nulling would be very difficult in your design.

NI (? and now TI) has an old application note on some micro-current opamps (LM-something) in which they talked about quite a few ua/pa-level ampmeters.

[Jay_Diddy_B]

Hi Group,

Did Dave ever discuss the choice of Op-amp and what the effect of a different choice might have insofar as offset voltage, rail to rail, etc. ie, how much of the specs relate to the choice of a 3v lithium battery supply and which spec relate to measurement accuracy. Suppose you were to use a +/- 15V Op-amp powered from a bench supply. What could you use?

The key parameters in choosing and op-amp for this application are:

1) Vos

The offset voltage is multiplied by the gain of the amplifier. The idea is to get the burden voltage as small as possible, so that insertion of the uCurrent device has minimal impact on the circuit being tested. With a 10m Ohm shunt the voltage is 10uV / per mA of current. So a 10uV offset in the op-amp would be seen as 1mA of current flow. It is desirable to have as low a Vos as practical.



There is a group of op-amps that fall in the class of 'Zero-Drift' (Linear Technology), 'Ultra-Low Offset' (Maxim), 'Zero Drift' (Intersil). The key feature of all these op-amps is that they are chopper-stabilized.

A secondary feature is the temperature Coefficient of Vos.

2) Input Bias Current

The next consideration is input Bias current. This is current that flows in or out of the input pins of the opamp. Lower is better. It will add to or subtract from the current being measured.

Input Bias Current can vary depending on the common mode voltage of the inputs relative to the supply rails.

3) Gain Bandwidth Product, GBW.

The bandwidth of the amplifier is limited to the GBW of the opamp divided by the gain of the circuit.

4) Slew Rate

Slew rate can limit the large signal bandwidth of the opamp.

5) Rail-to-Rail IO

This parameter is not really important, because bad things happen if you operate the op-amp outputs or inputs near the rail. The gain is limited. The output voltage swing will determine the maximum current that can be measured on each range.

6) Minimum Supply Voltage

This is fairly obvious. 2.7V is desirable if you want to power the circuit from a single Lithium coin cell.

7) Supply Current

This will determine how long the battery will last. Very low power op-amps will tend to have low SR and low GBW

I hope that this answers the question.

Jay_Diddy_B

[Jay_Diddy_B]

Nulling would be very difficult in your design.

NI (? and now TI) has an old application note on some micro-current opamps (LM-something) in which they talked about quite a few ua/pa-level ampmeters.

The LTC2050 op-amp chosen has a typical Vos of 0.5uV and guaranteed offset of 3uV. When multiplied by the gain of 100. This gives offsets equivalent to:

     50uA typical, 300uA max on the 1000mA range
     50nA typical, 300nA max on the 1mA range
     50pA typical,l 300pA max on the 1uA range

http://www.linear.com/product/LTC2050

I don't see a need to adjust the offset.

Jay_Diddy_B

[george graves]

Subscribed.  I might learn something by building one myself!

[Skimask]

I think I know the answer to this already...but...

What happens if you're drawing say 10mA but plugged into the nA jack?
Conversely, maybe draw 1nA and plugged into the mA jack?

I'm guessing one will max out the reading (but obviously not be correct) and the other will likely show something close to zero...but you won't smoke anything, right?

[nctnico]

If you use 2 opamps from the same chip with the same amplification the offset between them is zero (give or take the tolerance of the resistors). You can use any cheap jelly-bean opamp. I use this trick for interfacing thermocouples to a differential ADC or 2 seperate ADC channels and then subtract.

edit: I have to take this back (and reclaim tuition fees). It seems that even though the transistors for both opamps are on the same die the differences between them can be quite large.

[dannyf]

the offset between them is zero

Fat chance.

[Jay_Diddy_B]

Hi Group,

I have been busy and built and tested my version of the uCurrent.

Blank PCB


Here is the component side of the PCB. The board contains an error. The pin connections for the TPS3809 are wrong.



Here is the other side of the board which is the front panel:



Assembled Board

Component side:



Front Panel Side:



Testing



This is the most challenging test. I took the difference between the 1.000V and the 1.018V outputs of a Fluke 732A DC voltage standard. Feed this through a 1M Ohm resistor to uCurrent nA Input. I got a reading of 17.8mV which corresponds to 17.8nA. The reading does bounce around a bit. This is because of the noise from the op-amp. This is the noise from the datasheet. It is amplified by the gain of 100 at the output of the uCurrent:



I had to turn off the fluorescent lights above my bench for this test. The noise from these lamps interferes with small signal measurements.




I tried to measure the 18nA directly with my Fluke 289. It read 0.1uA or 100nA.

Looks like it is working good.

Jay_Diddy_B

[SeanB]

Very nice job there.

[Kevin.D]

I think I know the answer to this already...but...

What happens if you're drawing say 10mA but plugged into the nA jack?
Conversely, maybe draw 1nA and plugged into the mA jack?

I'm guessing one will max out the reading (but obviously not be correct) and the other will likely show something close to zero...but you won't smoke anything, right?

Dont think so  , 10 mA through the 10k nA range resistor would be 100V on the  opamp input . So that opamp would  go POOF  . He  needs to consider input overvoltage protection for the opamp (upping r12 considerably but then more offsets from input bias/offset  currents to consider).

[KerryW]

You could try a circuit like this one.  It won't go as high in current, but how high of a current should a MICRO current adapter handle?

Down side - Current limited by power supply and op-amp+booster.  Needs low  input current op-amp

Up side - Reads + or - current.  No matched resistors, accuracy determined by range resistors.  Burden voltage - Vos of U1 +/- (output voltage / U1 open loop gain), or basically 0.

[nctnico]

the offset between them is zero

Fat chance.

I guess you never heard of matched pairs. Two opamps on the same die (chip) have exactly the same errors.

[dannyf]

Two opamps on the same die (chip) have exactly the same errors.

Fat chance.

[KerryW]

the offset between them is zero

Fat chance.

I guess you never heard of matched pairs. Two opamps on the same die (chip) have exactly the same errors.

A piece of string walks in to a bar and orders a beer.  The bartender looks him over and says "Are you a piece of string?".  The piece of string says "Why, yes, I am."  The bartender says "We don't serve pieces of string here" and throws him out.  The piece of string  twists himself up and lays down in the street and lets cars run over him for a while.  Then he gets up and goes back in the bar and orders a beer.  The bar tender looks at him suspiciously and says "Wait a minute, aren't you a piece of string?".  The piece of string says "No, I'm a frayed knot."

If that were true, all op-amps would have 0 Vos.  They would just put 2 on each die (or 4 for duals8 for quads) and have them cancel.

[NiHaoMike]

The active IV converter is a great idea for small currents. Very low input impedance (basically zero) and the offset of the opamp is not as critical.

[hans]

the offset between them is zero

Fat chance.

I guess you never heard of matched pairs. Two opamps on the same die (chip) have exactly the same errors.

A matched pair also has a tolerance, so not each one can be made perfect. For example, the MAT04 was specificied for a couple percent current gain matching. What makes you think it doesn't have that?

See datasheet AD708 "Matching characteristics"
Or OP227 from Linear
Or LT1024

The difference seems to be about half of the individual typical specification, so at best you're halving the offset voltage, but at 30uV it's not particularly exciting.

http://e2e.ti.com/blogs_/archives/b/thesignal/archive/2012/10/30/matchy-matchy-how-alike-are-dual-op-amps.aspx

[dannyf]

-The active IV converter is a great idea for small currents.-

agreed. Plus, it has the added benefit of being easily converted to a u-volt instrument.

[Jay_Diddy_B]

Hi,
 I have been following the great suggestions that have been made on the thread today . I have a couple of concepts to share.

Modification for IV converter nA range

This concept shows how to implement a current to voltage converter for the nA range while preserving the original circuit for uA and mA. The only downside to this scheme is that the nA range is inverted. Only one op-amp is used. I showed three circuits to illustrate the configurations.



Addition of protection components



This concept shows how protection can be added to the design withou interfering with the measurements. The diode connected transistors were used by Fluke in a lot of their DMMs. The circuit is protected to 1A and 40V. The levels found in low voltage circuits.



Jay_Diddy_B

[SeanB]

Good and usable, but is getting away from the original idea of simple, robust and cheap. Probably best would be a 2A picofuse on the common lead for the load, and just buy a strip of 20 or so and store in the uCurrent box bottom. Problem with the transistors is that they only zener above the opamp supply voltage, you might be better off there making a bridge out of 4 with another 2 to shunt the bridge, giving about 2v of safe input before the diodes conduct

[Jay_Diddy_B]

Hi,

I was drawing faster than I was thinking 

This is corrected:



Jay_Diddy_B

[nctnico]

the offset between them is zero

Fat chance.

I guess you never heard of matched pairs. Two opamps on the same die (chip) have exactly the same errors.

A matched pair also has a tolerance, so not each one can be made perfect. For example, the MAT04 was specificied for a couple percent current gain matching. What makes you think it doesn't have that?

See datasheet AD708 "Matching characteristics"
Or OP227 from Linear
Or LT1024

The difference seems to be about half of the individual typical specification, so at best you're halving the offset voltage, but at 30uV it's not particularly exciting.

http://e2e.ti.com/blogs_/archives/b/thesignal/archive/2012/10/30/matchy-matchy-how-alike-are-dual-op-amps.aspx

I definitely have to get my money back for 4 years of theory on analog IC design... Using dual opamps seems to work to cancel offset voltages to a certain extend but not as good as I expected.

[NiHaoMike]

I think active IV conversion would also be useful on the uA range. Also, it would be nice to maintain a low voltage drop during "overload" conditions, as might happen when measuring quiescent current of a microcontroller that draws a few mA pulses from time to time. A pair of Schottky diodes can do that, or even better, a pair of MOSFETs switched in with a comparator.

[Jay_Diddy_B]

Hi,

My final version of the uCurrent will use Tellurium Copper connectors. I have named the device after these connectors, it will be called:

CuTe Current 


Jay_Diddy_B

[dannyf]

Modification for IV converter nA range

Slight different implementation.