Through-hole, diy friendly µcurrent?

[Chipset]

Hi there! First time posting here, been a long time youtube subscriber but never got around to visiting the forums all that much for some reason. I'm somewhat new to electronics but I like to think I know what I'm talking about, at least most of the time.
You read the title of the thread but before I get into that let me start by saying I'm not doing this because I'm a cheapskate. As an underage student I live on about $1400 a year. Sure, I live at home and don't pay the bills but that's still not a lot of money to fuel the financial black hole that is the electronics hobby. Right now I'm desperately in need of a new multimeter and I've got about half a dozen projects on the backburner because I'm saving up money for parts. Dropping $50 on a µcurrent just isn't realistic at the moment, sorry Dave!
I'll probably get one at some point in the future, probably after school when I have an employment (or wellfare  ) but the reason I made this thread is just to explore my curiosity.

Now as I said, you read the title, you know what this is about. I had a peek at the circuit diagram and a lot of the design choices looked to me like they where made for two main reasons: to keep it all on one board - no loose components, and to make it run off of a single 2032. Omitting those two design guidelines would it not be possible to build a µcurrent using through-hole jellybean components?
I hopped onto a circuit simulator and shaved as much off of the design as I could to get down to the basics of how it worked and got it to work perfectly. I then breadboarded it with an LM324 (what I had available) and as you'd expect it failed miserably. Any number of things could have gone wrong, the chip could be busted, my multimeter could be malfunctioning (not unsurprising as the amp ranges have already died entirely, why I'm desperate for a new one), I could have breadboarded it wrong or it could just not work with the 324. Frankly I didn't bother troubleshooting as I realized I'm in a little bit over my head, trying to reverse engineer a circuit like that.
I decided to ask for help before proceeding, not only for the reason above but because it's very nearly 6am and I'm dying to get to bed.

If this triggers anyone's interest please feel free to toss your ideas and thoughts into the ring!

[IanB]

Dave has a video on the design choices behind the uCurrent I'm sure. Watching that will definitely help you in picking the right components. I think the choice of op amp is going to be the central focus and picking the right one will make or break the circuit.

[free_electron]

324 is not usable. You need a way better opamp than that.

[EEVblog]

My design choices were for several reasons:
- MAX4238/39 because I was using it in another design at the time, and it's a cool chip, so simple used it again.

- The split rail system with the opamp was done because the design needed to go both positive and negative of course, and it's easier to do low battery detection with just one battery than two. So you could easily use two batteries and do away with the battery detect and split rail opamp if you really wanted.

- The decision to go all SMD was simply design elegance. I liked the idea of the PCB as a nice professional front panel, and with such a simple circuit, it makes sense to mount it on the back. A second PCB and wiring would have been so 1980's old-school.

You can use any auto-zero/chopper amplifier with similar specs. I think some are available in DIP.

Dave.

[Chipset]

Thanks dave, I'll be sure to pick up some better op-amps next time I shop for components and continue this project then.
While waiting for better op-amps though, I am curious as to why the 324 didn't work. Right now I just wanna know if, precision be damned, it can be made to work at all with a 324?
I'll try breadboarding it again and messing with it later today. As I said in the opening post though, if anyone has any ideas, I'd love to hear them.

[SeanB]

Offset voltage, drift, input current, bandwidth and just too small a signal relative to the noise are probably the reasons why, just off the top of my head.

[ejeffrey]

The LM324 has a worst case bias current of 250 nanoamp at 25 C (although typical is only 20).  That makes the nanoamp scale unusable.  The offset voltage is 7 millivolt.  The full scale reading only corresponds to a 10 millivolt input, so that makes every range useless.

There are better conventional opamps than the 324, but only an auto-zero opamp is going to perform suitably on all three ranges.

[Chipset]

The LM324 has a worst case bias current of 250 nanoamp at 25 C (although typical is only 20).  That makes the nanoamp scale unusable.  The offset voltage is 7 millivolt.  The full scale reading only corresponds to a 10 millivolt input, so that makes every range useless.

There are better conventional opamps than the 324, but only an auto-zero opamp is going to perform suitably on all three ranges.

Thank you for explaining that, as I said I'm gonna pick up some better op-amps next time I shop for components but at least now I know it's fruitless trying to make it work with the parts I have at hand. I'll bump this thread in a couple of weeks or so when I get more money and I can order the op amps.

[T4P]

Just remember to get a chopper amp

[Zad]

Also, don't forget Maxim do free samples.

[free_electron]

Lmc66xx are sub femtoampere opamps. Those would work...

[ejeffrey]

What is the offset voltage on them?  A couple of millivolts?  The ucurrent needs sub nanoamp bias current _and_  <10 microvolt offset voltage.  Only an auto-zero (chopper stabilized) amplifier can meet both of those requirements.

You could redesign it as a transimpedance amplifier instead of a current shunt with a voltage amplifier.  That would let you use basically any good FET input amplifier, but it would make the milliamps range untenable as the opamp must supply the full source current.

[timelessbeing]

Dave, would you consider selling me a naked µcurrent PCB?

[ElectroIrradiator]

The TLC2652ACP from TI looks like it might be a candidate for a through hole µCurrent then. It is available in a DIP8 case, has <5uV offset voltage and 150pA input bias current. Would require a few additional passive components though, and the chip costs twice as much as the MAX4238.

I haven't actually seen the schematic for Dave's contraption yet, so somebody else should probably try and check my suggestion against the original design.