EEVblog #1318 - What's State-of-the-Art in µCurrent Opamps?

[David Hess]

Would it be difficult to implement this 'on the cheap' outside the confines of an integrated circuit?

In the original context, it was always implemented with discrete parts so that an NPN log amplifier on an integrated circuit could accept positive inputs.  It helps to make up for the lack of good PNP matched pairs.

But the requirements are the same as any log amplifier except that temperature compensation is not required; it requires a precision operational amplifier and a matched diodes or transistor.  I assume that a translinear version using the transistors as transistors instead of diodes would perform better but maybe not since in this case, the extra errors from diode operation cancel out.

Now whether that is the best way to implement a bipolar input log amplifier, I do not know.  Maybe it is better to just implement two complete log amplifiers with opposite polarity although this requires a good PNP matched pair for the positive input.

[David Hess]

I found another example where the circuit idea I mentioned was used.  Check out figure 9 on PDF page 5 of PMI (Precision Monolithic Incorporated) application note 106 which is now available as an Analog Devices application note with PMI references scrubbed:

https://www.analog.com/media/en/technical-documentation/application-notes/28080533AN106.pdf

The circuit itself may be useful to some people for current measurement but note that it is *not* logarithm responding.  On the other hand, no temperature compensation is needed as with a square and square root circuits based on the same idea.

[SilverSolder]

I found another example where the circuit idea I mentioned was used.  Check out figure 9 on PDF page 5 of PMI (Precision Monolithic Incorporated) application note 106 which is now available as an Analog Devices application note with PMI references scrubbed:

https://www.analog.com/media/en/technical-documentation/application-notes/28080533AN106.pdf

The circuit itself may be useful to some people for current measurement but note that it is *not* logarithm responding.  On the other hand, no temperature compensation is needed as with a square and square root circuits based on the same idea.

Nice app note - I love that era of analog circuitry...

[edit] I have to think a little about the Figure 9 circuit, it is kind of cool...


The circuit in Figure 25 is more like what I was trying to do earlier, and I was also trying to partially use the logarithmic capabilities of the P/N junctions...

I think an approximation like that with 2 or 3 lines could work very well, and also be easy to understand on a linear (non-logarithmic) scope display...  like instant automatic range switching.

[David Hess]

I have to think a little about the Figure 9 circuit, it is kind of cool...

I always have to study it again when I encounter it but it gets easier each time.  I do not work with translinear circuits often enough to be comfortable with them.

[schmitt trigger]

Join the chorus in praising this old app note.

And one knows it is old when it shows both a NAB and RIAA amplifiers!

I ignore whether this level of analog engineering proficiency is still actively growing, or will it die when the old beards start retiring. I suspect the latter.

[SilverSolder]

Join the chorus in praising this old app note.

And one knows it is old when it shows both a NAB and RIAA amplifiers!

I ignore whether this level of analog engineering proficiency is still actively growing, or will it die when the old beards start retiring. I suspect the latter.

The essential thoughts (or "transfer functions" of the analog circuits, if you like) are still with us, whether implemented as software algorithms (e.g. in a DSP) or implemented by taking advantage of the behaviour of fundamental components.

What I admire about the old style analog designs is how much complexity can be covered by the natural behaviour of a handful of basic components...

Even today: if you want to make something with a microprocessor, it sometimes helps to process the signal in the analog domain before converting to digital (e.g. to get away with a slower or less feature rich processor, for example)

[SilverSolder]

I found a temperature compensated log op amp circuit in "Operational Amplifiers 5:th Edition, G. Clayton, S. Winder, Newnes, 2003"




Looks like the temperature compensating ability comes at a price:   simulating it in LTSpice shows that it is only capable of about 3.5 decades of operation, which is significantly less than the unstabilized single transistor circuits.

[EEVblog]

A further complication is that the uCurrent has to be able to deal with both positive and negative currents, so it isn't strictly speaking enough with a log transfer function...

Most people don't use the uCurrent in bipolar mode, so I'm thinking any new design might just drop that.

[EEVblog]

±75ppm/C is not ±15ppm/C but 0.28 vs $4 for Vishay Y14870R01000B9R

Vishay is 0.1% part. I think the whole point of using precise parts was to avoid calibration.

Correct.

[SilverSolder]

A further complication is that the uCurrent has to be able to deal with both positive and negative currents, so it isn't strictly speaking enough with a log transfer function...

Most people don't use the uCurrent in bipolar mode, so I'm thinking any new design might just drop that.

Funny, the true AC capability is one of the things I like the most about it - I frequently use the uCurrent to measure small AC currents (and voltages, in low impedance circuits) where it works as an "oscilloscope booster" or "DMM booster" for frequencies below 250KHz.  (Perhaps create a topic "Creative uses of the uCurrent", could be interesting and create some buzz?)

Even for the pulsating DC use case, sometimes the current might have unintentional excursions into the negative side due to the way the DUT (mis)behaves - which you would then not be able to see.

If the user reverses polarity on a unipolar device, there would now be no signal displayed - user can no longer tell the difference between 'no signal' and 'reverse polarity', which for sure would lead to some wasted time here and there.

To my mind, the bipolar capability of the original uCurrent was a stroke of genius that it would be sad to lose!

[EEVblog]

To my mind, the bipolar capability of the original uCurrent was a stroke of genius that it would be sad to lose!

Noted!
Looking at using a 9V battery supply for the new design, so enough range to give decent bipolar supply.
Could do some sort of user switched split rail system to get the best of both worlds though.

[MLXXXp]

Even for the pulsating DC use case, sometimes the current might have unintentional excursions into the negative side due to the way the DUT (mis)behaves - which you would then not be able to see.

Allowing the virtual ground offset to be changed, as I've done, can give a larger positive range while still being able to see small negative swings.
Modification number 4:
https://www.eevblog.com/forum/crowd-funded-projects/current-gold-on-kickstarter/msg448208/#msg448208

[SilverSolder]

To my mind, the bipolar capability of the original uCurrent was a stroke of genius that it would be sad to lose!

Noted!
Looking at using a 9V battery supply for the new design, so enough range to give decent bipolar supply.
Could do some sort of user switched split rail system to get the best of both worlds though.

9V battery would be awesome, I go through a lot of CR2032 batteries -  50 hours sounds like a lot, but sometimes an experiment can run a long time and it becomes an issue.   The extended range would be super useful too, of course.  You probably wouldn't need a switchable split rail system with 9V because the overlap between the ranges would now be "big enough" to be able to find a home...   But a switchable split rail would definitely be a cool thing for edge cases!


Another idea for the suggestion box:   Quite frequently, I am trying to monitor small currents between large pulses.  The uCurrent as it stands doesn't work so well for that, because the 10K shunt is in use to measure the small currents - sadly, the 10K resistance stops the DUT from being able to draw its occasional "breath" of high current, so it malfunctions.

A possible solution to that is to add a couple of (switchable) shunt by-pass diodes for the 10K shunt -  perhaps a pair of low Vf Shottkys, or even transistors wired as diodes to minimize drop.  That way, you can monitor ultra low currents over a long time period on your scope, and not have the DUT crash if it occasionally needs to inhale a large pulse of current.   

I usually end up putting a diode externally to work around this problem.  - Maybe something better than a diode could be used, to give a more clearly defined characteristic, but the ability to actually measure the high current pulse is not necessary.  All we are looking to do is allow the DUT to keep working when drawing too much current for the nA range.  After all, you can just switch to a higher range and measure the pulse there, if that's what you want to do.

[Edit: the 10 ohm shunt should probably also be included in this scheme, as it too is capable of dropping enough voltage during a pulse to worry the DUT]

[David Hess]

9V battery would be awesome, I go through a lot of CR2032 batteries -  50 hours sounds like a lot, but sometimes an experiment can run a long time and it becomes an issue.   The extended range would be super useful too, of course.  You probably wouldn't need a switchable split rail system with 9V because the overlap between the ranges would now be "big enough" to be able to find a home...   But a switchable split rail would definitely be a cool thing for edge cases!

I was thinking about that earlier; instead of lower noise or higher precision, micropower operation might be more useful for extended operating time.  Micropower chopper stabilized operational amplifiers are available but you have to specifically select them; just because something is CMOS does not make it low power.  Bipolar precision parts may have an advantage here with lower noise and higher bandwidth.

[EEVblog]

To my mind, the bipolar capability of the original uCurrent was a stroke of genius that it would be sad to lose!

Noted!
Looking at using a 9V battery supply for the new design, so enough range to give decent bipolar supply.
Could do some sort of user switched split rail system to get the best of both worlds though.

9V battery would be awesome, I go through a lot of CR2032 batteries -  50 hours sounds like a lot, but sometimes an experiment can run a long time and it becomes an issue.   The extended range would be super useful too, of course.  You probably wouldn't need a switchable split rail system with 9V because the overlap between the ranges would now be "big enough" to be able to find a home...   But a switchable split rail would definitely be a cool thing for edge cases!


Another idea for the suggestion box:   Quite frequently, I am trying to monitor small currents between large pulses.  The uCurrent as it stands doesn't work so well for that, because the 10K shunt is in use to measure the small currents - sadly, the 10K resistance stops the DUT from being able to draw its occasional "breath" of high current, so it malfunctions.

A possible solution to that is to add a couple of (switchable) shunt by-pass diodes for the 10K shunt -  perhaps a pair of low Vf Shottkys, or even transistors wired as diodes to minimize drop.  That way, you can monitor ultra low currents over a long time period on your scope, and not have the DUT crash if it occasionally needs to inhale a large pulse of current.   

I usually end up putting a diode externally to work around this problem.  - Maybe something better than a diode could be used, to give a more clearly defined characteristic, but the ability to actually measure the high current pulse is not necessary.  All we are looking to do is allow the DUT to keep working when drawing too much current for the nA range.  After all, you can just switch to a higher range and measure the pulse there, if that's what you want to do.

My new design will have automatic range switching.
I don't know about having a long term battery just for the edge cases where you need it for long term data logging. Perhaps just an external DC jack for that? Wire in your own external battery if needed.

[SilverSolder]

[...]

My new design will have automatic range switching.

That sounds very cool!  It would have to switch very fast to avoid interrupting the current.  I imagine if the user selects a range manually, it could automatically "uprange" and immediately fall back to the user selected range again when the pulse is gone.

I'm struggling to see, though, how you would tell which range was in use, when you look at the data later?   (i.e. the pulse would now look like it was part of the low current?)   This might be hard to crack without a separate "range selected" terminal of some kind.

I don't know about having a long term battery just for the edge cases where you need it for long term data logging. Perhaps just an external DC jack for that? Wire in your own external battery if needed.

An external DC jack is a much better idea.  ...Also, you can get rechargeable 9V Li-Ion batteries these days...   perhaps the DC jack could optionally be a way to recharge the internal battery in the uCurrent?  It is a little inconvenient to open and close with 4 screws every time you need to replace the battery when it is in heavy use...

[EEVblog]

That sounds very cool!  It would have to switch very fast to avoid interrupting the current.  I imagine if the user selects a range manually, it could automatically "uprange" and immediately fall back to the user selected range again when the pulse is gone.

Yep, it'll do that.

I'm struggling to see, though, how you would tell which range was in use, when you look at the data later?   (i.e. the pulse would now look like it was part of the low current?)   This might be hard to crack without a separate "range selected" terminal of some kind.

I've got that covered 

An external DC jack is a much better idea.  ...Also, you can get rechargeable 9V Li-Ion batteries these days...   perhaps the DC jack could optionally be a way to recharge the internal battery in the uCurrent?  It is a little inconvenient to open and close with 4 screws every time you need to replace the battery when it is in heavy use...

I'm weighing up the pro's and cons of a rechargeable solution. I like the simplicity of a 9V battery (or two for an ultra dynamic range mode). But yes, it needs to be changeable without screws into a plastic case.

[SilverSolder]

I'm weighing up the pro's and cons of a rechargeable solution. I like the simplicity of a 9V battery (or two for an ultra dynamic range mode). But yes, it needs to be changeable without screws into a plastic case.

I also like the 9v battery (or 2x 9v battery) idea.  If the battery is accessible without opening the case, the uCurrent probably shouldn't get involved in battery charging at all.  We can even live without a DC jack, as external power could be applied directly to the 9V battery connector(s) for long term data logging. 

The user can then choose the battery chemistry that works for them, and whatever charging technology works with it.

Some rechargeable 9V batteries are a little larger than the standard ones...  hopefully there is room for a "generous" battery compartment!

For rechargeable use, I was thinking of the li-ion versions of the venerable 9v battery - e.g. similar to these ones: 

[David Hess]

The problem with rechargeable versions of the standardized 9 volt battery is that there are so many different kinds.

Energy density is better with cylindrical cells if lower voltage is acceptable.  I remember ads from either Maxim or Linear Technology about "how to suck a battery dry" which compared dual AA cells to 9 volts with various power management solutions.

Supply current for a pair of MAX4239s is 1.2 milliamps?  That can be reduced to 1/10th pretty easily and an extreme design could be 1/50th.  How much bandwidth is required when the measuring instrument is a multimeter?

[SilverSolder]

The problem with rechargeable versions of the standardized 9 volt battery is that there are so many different kinds.

Energy density is better with cylindrical cells if lower voltage is acceptable.  I remember ads from either Maxim or Linear Technology about "how to suck a battery dry" which compared dual AA cells to 9 volts with various power management solutions.

Supply current for a pair of MAX4239s is 1.2 milliamps?  That can be reduced to 1/10th pretty easily and an extreme design could be 1/50th.  How much bandwidth is required when the measuring instrument is a multimeter?

The instrument is most often a 'scope, in my case at least.  So the current uCurrent current  [ ] bandwidth of about 250KHz is about right.  Could live with a little less, but it is nice to see the "shape" of the current consumption.

[Kleinstein]

There are two uses:
1) measuring current with the scope

2) measuring current with a voltmeter with reduced burden voltage. With old DMMs this may add another current range. It may also improve on the choice of actual shunts in a newer meter - modern auto-scale meters may have a limited choice of shunts. The cases where the reduced burden is really needed are relatively rare. It may still be handy to simplify things.

For the DMM ind DC one would usually only need low BW, like some 100 Hz.

For the scope use one would like to have a relatively high BW, but also low noise. Especially with the scope the low offset is not really needed, low noise would normally be preferred. The same is true with AC measurements.

With the low burden voltage the dynamic range is naturally reduced. So the µC current is not the right starting point to measure the highly variable current of a µC. This would be another piece of equipment, e.g. using voltage regulation with a FET and measure the current at the other side of the FET.

[SilverSolder]

[...]
With the low burden voltage the dynamic range is naturally reduced. So the µC current is not the right starting point to measure the highly variable current of a µC. This would be another piece of equipment, e.g. using voltage regulation with a FET and measure the current at the other side of the FET.

I am struggling to understand.  Isn't the current (and therefore also the dynamic range) the same on both sides of the FET, so the problem of measuring it is the same?  (I do see that you could use a larger shunt resistor in front of the FET,  but I don't see a change in dynamic range of the current)

[EEVblog]

I'm weighing up the pro's and cons of a rechargeable solution. I like the simplicity of a 9V battery (or two for an ultra dynamic range mode). But yes, it needs to be changeable without screws into a plastic case.

I also like the 9v battery (or 2x 9v battery) idea.  If the battery is accessible without opening the case, the uCurrent probably shouldn't get involved in battery charging at all.

I've got a neat idea for an external 9V battery.

We can even live without a DC jack, as external power could be applied directly to the 9V battery connector(s) for long term data logging. 

For my idea the contacts would be external, so would still need a DC jack or screw terminals. I presume DC jack is preferable in that case?

[EEVblog]

Supply current for a pair of MAX4239s is 1.2 milliamps?  That can be reduced to 1/10th pretty easily and an extreme design could be 1/50th.  How much bandwidth is required when the measuring instrument is a multimeter?

The instrument is most often a 'scope, in my case at least.  So the current uCurrent current  [ ] bandwidth of about 250KHz is about right.  Could live with a little less, but it is nice to see the "shape" of the current consumption.
[/quote]

For the new design I'm targeting around 4-5 times that bandwidth, and lower noise. Comes at the expense of battery life though, hence the 9V vs coin cells.

[SilverSolder]

We can even live without a DC jack, as external power could be applied directly to the 9V battery connector(s) for long term data logging. 

For my idea the contacts would be external, so would still need a DC jack or screw terminals. I presume DC jack is preferable in that case?

A DC jack is a nicer solution than screw terminals, especially if user has more than one external power source.  For convenience, the unit could be supplied with a couple of spare DC plugs or even pigtails, so user can easily make custom cables for external power sources?