uCurrent Gold and a lot of noise on oscilloscope

[jule553648]

Hello guys!

I was reading this forum for a long time, but now I have a question, so this is my first post.

I am dealing with AVR microcontroller and NRF24L01 RF transmitter.

And I would like to measure its consumption.

The peak transmitting  consumption of the whole circuit should be about 20mA, idling currents are sub mA.

I have a test circuit on breadboard and I am sending a packet every second.
 
But I have a lot of noise when measuring with uCurrent and scope - see the attachment.

Yellow line is output from uCurrent, blue is trigger signal from uC.

One is raw signal, other one is digitally filtered - the consumption footprint is corect.

I suspect the scope ground is giving me headaches.

I have uCurrent set on 1mV/mA.

If I measure uCurrent output with DMM (without scope attached) it shows like 0.2mV (of course it cannot detect transmitting pulses). This is correct measurement of idle current.

But when I connect probe positive end on Voltage output (+) on the uCurrent and scope ground on the (-), the DMM shows -11mV (negative!) If I connect just probe end or ground on  Voltage output (+) it says -6mV or -16mV. So the connection disturbs the uCurrent output
Whatever scope probe end I hookup to Voltage output (-) on uCurrent it is still OK. DMM shows 0.2mA.

So the problem is in Voltage output (+) pin of uCurrent.

Something on the scope disturbs the output. Not just the ground, because also just the probe end disturbs it.

The circuit is powered through AC adapter with transformer, so it is isolated from scope ground.
What could it be?

If I have to explain further, just say it.  I sometimes overcomplicate my descriptions.

thanks

[VK5RC]

What does a straight piece of wire do when connected to the uCurrent terminals,  you have rf floating around,  it is notorious for 'getting into places' . 

[bson]

Don't!  The uCurrent (-) output isn't grounded.  You can't probe it directly but need a differential probe, or use two scope channels as A-B.  (With either ground clip attached to a real ground point in your circuit.)  Connecting the scope ground to the uCurrent (-) terminal will short it and can damage either the uCurrent or the scope!  You're likely seeing some oscillating artifact of connecting one half of a differential amplifier to ground!

[PSR B1257]

Don't!  The uCurrent (-) output isn't grounded

The µCurrent is battery powered and therefore floating.

You can't probe it directly but need a differential probe

No, you don't.

[jule553648]

I tried differential, but it didnt work, (CH1-CH2) is still as noisy as before.

I am attaching my wiring.

uCurrent has both negative leads wired together  internally. I showed that on schematic.

Weird thing is, that the scope probe disturbs the (+) output of uCurrent. My DMM(Agilent) shows that there is an 10mV increase of the uCurrent output voltage(noise) after i hookup scope lead on the uCurrent. So the scope is somehow messing things up. I tried using 10X probe, but it is still the same problem.

I don't have scope on isolation transformer since I don't have it, but I doubt it would help, because the circuit power supply is isolated from ground.


@VK5RC
RF emission isn't the problem, because I transmit for only 1.75 milliseconds every second. Then the  RF module is at sleep.  But the noise is everpresent when scope is hooked up. I tried hooking 0.5 metre wire on (+) output of the uCurrent, but it doesn't disturb the DMM measurement. Only scope disturbs it.

[bson]

I took a closer look since I had simply observed the (+) output is not ground referenced.  What seems to happen is when you put your scope ground on the negative output you place your scope-to-power supply ground across a 0.1ohm resistor.  This seems to be a precision resistor since it measures to ~4 digits (just simple 2W mode, nulled), which means it's likely the main sense load.  Whatever it is, you're putting your scope-to-supply ground connection across it, which if it's the sense load will make the positive output amplify your scope ground noise.

[tautech]

The circuit is powered through AC adapter with transformer,

Two possibilities IMO. 1. ^^^^^^^ SMPS is it?

2. http://www.eevblog.com/2014/04/10/eevblog-601-why-digital-oscilloscopes-appear-noisy/
http://www.eevblog.com/2014/04/27/eevblog-610-why-digital-scopes-appear-noisy-part-2/

[mmusskopf]

Hello,

Just go my uCurrent recently I had some similar issues here. And yes, the noise captured by the scope on the mVs can be very annoying!

Ideally the uCurrent should have another configuration, something like 10mV/mA?! or even 100mV/mA?

Using 1mV/mA has too much noise and you'll need to filter/average the signal... the problem is that a short +/-10mA peak will be difficult to capture. This settings is only useful if you're measuring currents over 200mA and don't care much about +-10/20mA variations.

IMO the step from uA to mA is too big. As I normally need to measure currents between 1mA and 100mA a 10mV/mA (or over) would be great... I'll see if I can modify it, but I still haven't find time to open the schematic yet!

Btw, this is only an issue when connected to a scope, on a Voltmeter as everybody know works flawlessly!

Edit:
Just looking and should be the case of replacing the R1 from a 0.01 ohms to a 0.1 ohms... this should do the trick:
http://au.rs-online.com/web/p/surface-mount-fixed-resistors/6841649/

I'll double check once I get home before pay $15 for single SMD resistor 

Cheers,
Mike M.

[MLXXXp]

Just looking and should be the case of replacing the R1 from a 0.01 ohms to a 0.1 ohms

You can always just add an external shunt resistor across the input terminals when you need a different range. Set the range switch to nA and the internal shunt will be 10K. Putting an external 0.1R resistor in parallel with the internal 10K will result in a only a 0.001% error over having exactly 0.1R.

[mmusskopf]

That's a great idea... I was worried to have it in parallel with the internal resistor, but you're right, the 10K over a 0.1R will be insignificant! I'll get a few high precision TH resistors.

Cheers!

[borjam]

I have just ordered one from the Euro distributor.

If I understood the schematic well, it can be used *without* a differential probe, just being a bit careful.

I need to measure the consumption of a small microcontroller when sleeping, reading sensors, transmitting on Wi-Fi, etc. If I insert the µcurrent betweem the ground of the power supply and the microcontroller, J2 will be connected to ground and J4, the virtual ground, will actually be a ground terminal as well, perfectly safe to use with a passive probe. Right?

[Gyro]

Why not build your own?

With a rotary switch and proper choice of shunt resistors (ie. one per decade rather than one every 3 decades) you could get away with far less gain and better noise immunity for the same or better voltage burden. Less gain would also allow you to use a standard low Vos opamp rather than the chopper stabilised one to achieve better AC performance too.

It's an ideal 'build your own test equipment' beginner's project.

[MLXXXp]

If I insert the µcurrent betweem the ground of the power supply and the microcontroller, J2 will be connected to ground and J4, the virtual ground, will actually be a ground terminal as well, perfectly safe to use with a passive probe. Right?

Correct. And if the supply you're using to power the microcontroller is floating/isolated, you're extra safe.

[borjam]

Thanks

I need to get a differential probe, I've seen one from R&S no less, at a reasonable price and perfectly good for my purposes. But for now this will do, even requiring much more care.

[Simon]

I have just ordered one from the Euro distributor.

If I understood the schematic well, it can be used *without* a differential probe, just being a bit careful.

I need to measure the consumption of a small microcontroller when sleeping, reading sensors, transmitting on Wi-Fi, etc. If I insert the µcurrent betweem the ground of the power supply and the microcontroller, J2 will be connected to ground and J4, the virtual ground, will actually be a ground terminal as well, perfectly safe to use with a passive probe. Right?

Yes I have received your order but I am out of stock. I am about to have delivered a box of them from Dave but I don't know yet how many he sent I had told him I needed 10 but judging by the customs charge it may be 20 that he has sent.

The input and output ground on the microcurrent are connected together. You should not need a differential probe. The fact that the microcurrent is battery powered means that it will never be mains referenced therefore if you are connecting other isolated equipment to it you have the 1st step in your chain of isolated equipment.

[borjam]

Yes I have received your order but I am out of stock. I am about to have delivered a box of them from Dave but I don't know yet how many he sent I had told him I needed 10 but judging by the customs charge it may be 20 that he has sent.

I saw it was out of stock, don't worry

These thingies must be selling like hotcakes. How else can you measure the consumption of tiny microcontrollers without spending a fortune?

The input and output ground on the microcurrent are connected together. You should not need a differential probe. The fact that the microcurrent is battery powered means that it will never be mains referenced therefore if you are connecting other isolated equipment to it you have the 1st step in your chain of isolated equipment.

Yes, what I mean is, without a differential probe I can't connect it to an arbitrary point in a circuit with the oscilloscope earthed. But in my case, as this configuration is exactly what I need, it's not an issue.

[Simon]

Yes while your other option is to use an isolated oscilloscope. You could use a battery powered oscilloscope. My old Owon oscilloscope broke down because the power supply board broke. But as it has a battery input I have used a switch mode power supply that has an isolated output to power it onto the battery terminals so I now have an isolated oscilloscope. The other option is to use an isolation transformer to power your oscilloscope. I don't know how much differential probes are but if you are working on a low-voltage then using isolated equipment would not be a problem. Obviously I would not recommend connecting the microcurrent to the mains.

[ANTALIFE]

Dunno if this is relevant, but I had issues with using a microcontroller to read the current/voltage https://www.eevblog.com/forum/crowd-funded-projects/ucurrent-gold-problem/
In the end I had to run the output through an op-amp to read the current/voltage accurately.

[Gyro]

How else can you measure the consumption of tiny microcontrollers without spending a fortune?

Not difficult, you just need a series resistor and a bit of voltage gain. You can do it very cheaply.

[Simon]

How else can you measure the consumption of tiny microcontrollers without spending a fortune?

Not difficult, you just need a series resistor and a bit of voltage gain. You can do it very cheaply.

Good luck! what do you think the uCurrent is ? it is litterally that, a resistor and a 100x amp. The problem with dealing with such low voltages is offset voltage on the amp so the uCurrent was designed to resolve that with a good amplifier.

[Gyro]

Of course I know that. With the right choice of shunt resistors, sized for say 10mV full range (edit:burden) (no way excessive) then you don't need much gain to get up to usable level, x10 is plenty. The reason that the uCurrent needs so much gain is that it uses the same shunt resistor for 3 decades, fine for the top decade, but it leaves it scrabbling about in the noise for the bottom decade. One resistor per decade makes a lot more sense and eases the gain and noise problems. With 10mV full range burden per decade it's easy to do on a breadboard.

[Simon]

The reason that the uCurrent needs so much gain is that it uses the same shunt resistor for 3 decades, fine for the top decade, but it leaves it scrabbling about in the noise for the bottom decade. One resistor per decade makes a lot more sense and eases the gain and noise problems. With 10mV full range burden per decade it's easy to do on a breadboard.

wrong, there are 3 different shunt resistors for nA, uA and mA they are a 3 decades apart with a fixed amp gain of 100 so all your selector switch does is alter the shunt. The sunts are 10mR, 10R and 10K if memory serves.

[Gyro]

Yes I know there are 3 shunt resistors Simon, each serving 3 decades = 9 decades total!

What it means (for instance taking the 10k nA shunt) is that at 100nA it's dropping 1mV (ok), at 10nA it's dropping 100uV (hmmm) and at 1nA it's dropping 10uV (ridiculous!). With one shunt resistor per decade it would drop 1mV for each.

As I say, 10mV max burden wouldn't be excessive for pretty much any application, and still way better than off the shelf DMMs. It would be very easy to do.

Seriously Simon, let's not go there! I have no wish to trash Dave's product - I was merely pointing out that it would be easy and cheap for borjam to do what he wanted. Please don't force me into doing an eevblog style critique of the uCurrent (Quote: "Come on guys!" ).

[Simon]

Oh i see (and no need to get on your high and mighty) the problem then is that you have an ambiguos reading. The idea is that mV = nA, uA and mA, if you do each decade it will be so confusing it will make life harder. A multimeter reads in mV or V, to have 100mV per 1nA, 10nA, 100nA, 1uA, 10uA, 100uA, 1mA, 10mA, 100mA will be very confusing if your trying to use all ranges. Of course a more complex product could do as you want but it would be more expensive as it would need it's own display and a more complicated case.

[Simon]

There is of course nothing stopping anyone removing one of the shunts and putting in an exteral one at will.