noise in current while probing

[metebalci]

I have an oscillator (a few kHz) like circuit (on a protoboard) which I want to probe the current on one wire. Around 10kHz bandwidth is fine. The current is in 1-5 mA range. When using a sense resistor (1 ohm) at low side, I see lots of noise particularly >10 kHz. If I lowpass filter, it gets better, but it is hardly usable. If I use a current sense amplifier (INA214 at the moment, 100x gain) with the same sense resistor, the result is much better.

Is this because of the increased visibility/effect  of the noise at low voltage levels on the scope (picoscope 4424a) ? and/or is it because current sense amplifier also improves it in other means (which is?) ?

[tggzzz]

I have an oscillator (a few kHz) like circuit (on a protoboard) which I want to probe the current on one wire. Around 10kHz bandwidth is fine. The current is in 1-5 mA range. When using a sense resistor (1 ohm) at low side, I see lots of noise particularly >10 kHz. If I lowpass filter, it gets better, but it is hardly usable. If I use a current sense amplifier (INA214 at the moment, 100x gain) with the same sense resistor, the result is much better.

Is this because of the increased visibility/effect  of the noise at low voltage levels on the scope (picoscope 4424a) ? and/or is it because current sense amplifier also improves it in other means (which is?) ?

No schematic nor photo of your circuit/waveforms => I suspect you are a beginner. (We all were once )

Cured by INA214 making a differential measurement => I suspect common mode noise.

Beginner => I suspect you think all points on a wire connection are at the same voltage. In practice you may need to look at databooks and textbooks to see a conductor's resistance and inductance. ("Protoboard" => if using a solderless prototype board, then those will be much larger.[1][2])

Try using your scope to make a differential measurement, and see if that improves the waveform.

[1] for reasons why and the alternative construction methods, see https://entertaininghacks.wordpress.com/2020/07/22/prototyping-circuits-easy-cheap-fast-reliable-techniques/
[2] Cue people claiming they have used solderless breadboards without problems. They may well have done, but they were sufficiently knowledgeable to recognise and circumvent their non-ideal characteristics

[metebalci]

I have an oscillator (a few kHz) like circuit (on a protoboard) which I want to probe the current on one wire. Around 10kHz bandwidth is fine. The current is in 1-5 mA range. When using a sense resistor (1 ohm) at low side, I see lots of noise particularly >10 kHz. If I lowpass filter, it gets better, but it is hardly usable. If I use a current sense amplifier (INA214 at the moment, 100x gain) with the same sense resistor, the result is much better.

Is this because of the increased visibility/effect  of the noise at low voltage levels on the scope (picoscope 4424a) ? and/or is it because current sense amplifier also improves it in other means (which is?) ?

No schematic nor photo of your circuit/waveforms => I suspect you are a beginner. (We all were once )

Cured by INA214 making a differential measurement => I suspect common mode noise.

Beginner => I suspect you think all points on a wire connection are at the same voltage. In practice you may need to look at databooks and textbooks to see a conductor's resistance and inductance. ("Protoboard" => if using a solderless prototype board, then those will be much larger.[1][2])

Try using your scope to make a differential measurement, and see if that improves the waveform.

[1] for reasons why and the alternative construction methods, see https://entertaininghacks.wordpress.com/2020/07/22/prototyping-circuits-easy-cheap-fast-reliable-techniques/
[2] Cue people claiming they have used solderless breadboards without problems. They may well have done, but they were sufficiently knowledgeable to recognise and circumvent their non-ideal characteristics

It is chua’s oscillator, current at L. It is low side, so I was thinking if there is any other reason than common mode.

Not concerned about actual value, so not concerned about the resistance of the wire, but waveform is noisy, and there is less I guess amount of parasitics with a single resistor.

It is on the low side, and I dont have a proper differential probe and subtraction in the scope is always worse, but being in low side I dont need a differential measurement or I am already doing against GND.

Both are on protoboard, and pretty low freq. and slow rise times, only change is opamp vs no opamp, so I dont think it is related to protoboard mainly. I cant do deadbug over copper at the moment so that is what it is right now.

I will attach two images soon.

[tggzzz]

I have an oscillator (a few kHz) like circuit (on a protoboard) which I want to probe the current on one wire. Around 10kHz bandwidth is fine. The current is in 1-5 mA range. When using a sense resistor (1 ohm) at low side, I see lots of noise particularly >10 kHz. If I lowpass filter, it gets better, but it is hardly usable. If I use a current sense amplifier (INA214 at the moment, 100x gain) with the same sense resistor, the result is much better.

Is this because of the increased visibility/effect  of the noise at low voltage levels on the scope (picoscope 4424a) ? and/or is it because current sense amplifier also improves it in other means (which is?) ?

No schematic nor photo of your circuit/waveforms => I suspect you are a beginner. (We all were once )

Cured by INA214 making a differential measurement => I suspect common mode noise.

Beginner => I suspect you think all points on a wire connection are at the same voltage. In practice you may need to look at databooks and textbooks to see a conductor's resistance and inductance. ("Protoboard" => if using a solderless prototype board, then those will be much larger.[1][2])

Try using your scope to make a differential measurement, and see if that improves the waveform.

[1] for reasons why and the alternative construction methods, see https://entertaininghacks.wordpress.com/2020/07/22/prototyping-circuits-easy-cheap-fast-reliable-techniques/
[2] Cue people claiming they have used solderless breadboards without problems. They may well have done, but they were sufficiently knowledgeable to recognise and circumvent their non-ideal characteristics

It is chua’s oscillator, current at L. It is low side, so I was thinking if there is any other reason than common mode.

Not concerned about actual value, so not concerned about the resistance of the wire, but waveform is noisy, and there is less I guess amount of parasitics with a single resistor.

If a solderless breadboard, the contact resistance will be variable and may be significant.

If long wires, their inductance can cause "noise" when currents change.
If long wires, "noise" can be coupled by stray capacitance and mutual inductance[1]

It is on the low side, and I dont have a proper differential probe and subtraction in the scope is always worse, but being in low side I dont need a differential measurement or I am already doing against GND.

At low frequencies subtraction in the scope should be sufficient. Try it before saying it won't work.

When you have wires plus joints with inductance and resistance, where exactly is your reference point. The resistance and inductance are a principal cause of common mode "noise".

Both are on protoboard, and pretty low freq. and slow rise times, only change is opamp vs no opamp, so I dont think it is related to protoboard mainly. I cant do deadbug over copper at the moment so that is what it is right now.

I will attach two images soon.

[1] Here's an example of "stray" mutual inductance used as a circuit element. At least Tektronix had the decency to include the inductors in the schematic; the flat bent horseshoe is L70, the wire hoop partially hidden behind it is L69, and their mutual inductance forms a transformer

[metebalci]

all same circuit, sense-only images has only 1 ohm sense resistor, low-side. one image is with 10khz filter with dsp.

ina images have the same sense resistor at the same location, only difference is INA 214 is powered and measurement is taken from its output rather than the sense resistor.

all other settings (other than the channel range because of INA's 100x gain) is same. I3 channel is AC coupled.

both improves naturally with 10khz filter but sense-only is very noisy without filter. the channel bandwidth is 20MHz, I use direct cables no probes, I guess 20MHz captures many things around and it is not probing correct but anyway there is a big difference between the no lpf images of both, that is what I am trying to understand.

[metebalci]

and the main circuit is straight-forward chua's circuit or oscillator from Kennedy's 1993 paper as in the image. Only difference is I have the 1 ohm sense resistor on the low side of L.

[metebalci]

same idea, but with math channel, subtracting the voltages around the sense resistor. I did not measure anything so I cannot quantify but maybe a little improvement over single channel images before, but still I think the ones with current sense amp is better.

[tggzzz]

Need the actual schematic including all components and their values. You will need at least two more capacitors than shown.

"The difference between theory and practice is that in theory there is no difference, but in practice there is a difference".

[metebalci]

If a solderless breadboard, the contact resistance will be variable and may be significant.

If long wires, their inductance can cause "noise" when currents change.
If long wires, "noise" can be coupled by stray capacitance and mutual inductance[1]

I have an idea about the limitations of breadboard and I guess there are variety of noise sources around. What I dont understand is there is a very little change here (and the change is actually a negative in terms of parasitics for the addition of current sense amp I believe), but there is a considerable difference and the major difference is the signal level, but I am not 100% sure if it can be attributed only to the signal level, hence my question.

Something I take from your messages and I should try is to minimize the size of sense resistor (it is a standard size THT part at the moment). I understand it can be picking up noise, and this noise is maybe rejected by the opamp, on the other hand without opamp and with low signal levels, the effect of picked up noise can be (visibly) amplified.

At low frequencies subtraction in the scope should be sufficient. Try it before saying it won't work.

images of math channel of subtraction is attached in the previous post. but I am not using a probe, so it is still not a very proper differential signal just around the sense resistor (it includes leads and some breadboard effects).

[metebalci]

Need the actual schematic including all components and their values. You will need at least two more capacitors than shown.

"The difference between theory and practice is that in theory there is no difference, but in practice there is a difference".

This is the actual schematic (without the current sense part and power supplies and bypass caps etc.). I dont have the full exact schematic of the circuit I have, I am modifying it on the breadboard. Parts are as same as in the same paper (below image). Opamp is TL082. I am using a trimmer instead of potentiometer but with the same value. Opamp power supply is +-9V, with standard decoupling caps, current sense amp power supply is same 9V, reference input is 5V coming from an LDO. Sense resistor is 1ohm, at the low side of L.

[jonpaul]

scope images seem like a lissajous figure.

Please post EXACT LINKS TO ORIGINAL PAPERS AND DESIGNS

a photo of the unit and EXACT SCHEMATIC with test set-up is helpful

j

[metebalci]

scope images seem like a lissajous figure.

Please post EXACT LINKS TO ORIGINAL PAPERS AND DESIGNS

a photo of the unit and EXACT SCHEMATIC with test set-up is helpful

j

Yes XY views.

Paper, I dont know if it is publicly available (but I saw it can be found on some sites):
https://ieeexplore.ieee.org/document/246141

Only photo I can share is below and it is changing. Pico is PicoScope 4424A. Power supply is a 5V USB battery, DC-DC converter makes +-9V (top middle on breadboard), negative resistance/chua’s diode implementation with dual opamps is on the right side, L and 2x C network in the middle, trimmer is on the right for convenience, current sense amp on the left, powered by same 9V, LDO on the left gives 5V ref to current sense amp.

(The circuit on right bottom of breadboard is another experiment, not powered here)

(The LEDs are only to see if pos/neg rails are powered)

[tggzzz]

scope images seem like a lissajous figure.

Please post EXACT LINKS TO ORIGINAL PAPERS AND DESIGNS

a photo of the unit and EXACT SCHEMATIC with test set-up is helpful

j

Only photo I can share is below and it is changing. Pico is PicoScope 4424A. Power supply is a 5V USB battery, DC-DC converter makes +-9V (top middle on breadboard), negative resistance/chua’s diode implementation with dual opamps is on the right side, L and 2x C network in the middle, trimmer is on the right for convenience, current sense amp on the left, powered by same 9V, LDO on the left gives 5V ref to current sense amp.

At last!

You shouldn't place things for "convenience", but for electrical behaviour, including "parasitic" resistance and inductance.

There are lots of wires and "joints" with variable resistance all over that board.

You have different probe's shields connected to different points. That can introduce noise due to currents flowing through resistance (and inductance) not shown in your schematic (nor your thinking).

You should look at the time domain oscilloscope waveforms, to assess any correlation between "noise" on each channel. Only after that is acceptable should you look at the XY plots.

Try connecting two probes to the exactly the same strip on the solderless breadboard. What do you see?
Try connecting two probes's shields to the exactly the same strip on the solderless breadboard. What do you see?

[metebalci]

At last!

You shouldn't place things for "convenience", but for electrical behaviour, including "parasitic" resistance and inductance.

There are lots of wires and "joints" with variable resistance all over that board.

You have different probe's shields connected to different points. That can introduce noise due to currents flowing through resistance (and inductance) not shown in your schematic (nor your thinking).

You should look at the time domain oscilloscope waveforms, to assess any correlation between "noise" on each channel. Only after that is acceptable should you look at the XY plots.

Try connecting two probes to the exactly the same strip on the solderless breadboard. What do you see?
Try connecting two probes's shields to the exactly the same strip on the solderless breadboard. What do you see?

I understand your point and I at least to some extent understand the issue with the breadboards. I think I mis-asked and/or mis-explained the reason of my question or this post, also I should have put another title I think. Maybe because of English as my second language.

I am probably going to repeat myself, so excuse me for that.

I am not looking a way to improve the situation on the breadboard using only a sense resistor (I already have a current sense amp solution which gives me more than I need). I am already trying different things (using a current sense amp was one), just to make the trial a bit better. However, this is not going to live on the breadboard. If I want to carry on a proper experiment, I am going to have it on a proper PCB. Naturally I cannot share every step I am taking neither every waveform I am seeing, this would be too much time consuming.

The scope channel and the probe (actually a coax + adapter) is the same for current sense amp output measurement and the sense resistor voltage measurement. Hence I dont think that can be the reason. It may not be 100% correct, but I think the effect of breadboard's imperfectness should be more pronounced with current sense amp because it involves more wiring.

There are old research, and the ones I saw is not the same element values as they can be calculated differently but the circuit topology is the same, that is using only the sense resistor for measuring the current passing through L, and it is a bit difficult to judge but it seems like it was OK (at the moment, I see it as not OK on my prototype). Naturally it was not a circuit on the breadboard, possibly a PCB.

Coming back to my actual question that it seems I couldnt manage to ask, in other words, taking the breadboard out of the picture; if I have this circuit (with only the sense resistor) on a PCB, can I have the same current measurement that I do with the current sense amp in terms of less noise ? would they be of equal or not very different quality ? in other words, is the noise I see without the current sense amp only due to breadboard ? I understand if the answer can be we cannot know without trying.

I understand you want to see actually what is going on, the circuit, the wiring etc., but my question was more broad -I should have been more clear about this-, that is why I didnt want to mention details of the circuit at first. This is not the only circuit I want to measure the current with similar properties.

Basically my question is this. I have a project, I have to bidirectionally measure current, a few tens of kHz bandwidth is more than enough. I can use 1 ohm (or less) sense resistor (but not more), either low side or high side not important, and high side is 9V max, and the current is around 1mA. When I say measure, I mean measuring voltage with the scope I have (20MHz, input sensitivity minimum 2mV/div, 45uV noise on 10mV range, 12-bit ADC). These are the things I cannot change, but I can change anything regarding to power supply (for convenience I am using a 5V to +-9V DC-DC converter here, I can also use a split rail linear regulator if it matters), and I can use other ways to 'sense' the current (e.g. using a current sense amp, and I will actually need a differential amp as well to shift the level) as long as the effect on the circuit is the same as merely having the sense resistor. So the question, given these above, is it OK to measure the current just as a single ended voltage measurement on a 1 ohm current sense resistor at low side ? or is it not and I need another way (which probably involves using a current sense amp) ?

[tggzzz]

To summarise your response.

You have a problem with a circuit built on a solderless breadboard.
You want us to remotely diagnose the problem.
You haven't shown us the complete schematic, despite several requests.
You haven't shown us waveforms we have requested.
You won't try quick and easy suggestions we make because you don't think they are relevant.
And on top of that you really want answers about a completely different construction!

Good luck!

[David Hess]

To see if the problem is common mode noise, connect the probe signal to the same point as the probe ground connection to the circuit.  Then short the probe signal to the probe ground without any connection to the circuit to measure the inherent noise of the oscilloscope and probe.

10 kHz might be below the point where the flicker noise rises on the PicoScope 4424A, so the noise could be from the PicoScope.

[metebalci]

To see if the problem is common mode noise, connect the probe signal to the same point as the probe ground connection to the circuit.  Then short the probe signal to the probe ground without any connection to the circuit to measure the inherent noise of the oscilloscope and probe.

10 kHz might be below the point where the flicker noise rises on the PicoScope 4424A, so the noise could be from the PicoScope.

You mean the noise floor including the probe or something else ? I measured the noise floor, it is around 90uVrms with the probe and not very different without the probe. This is comparable to 1mV, so that was my initial thought that the primary benefit of current sense amp here is the gain.

[metebalci]

I saw in a paper they used a transimpedance amplifier, and I also tried one now instead of the current sense resistor and the amp. The result is definitely comparable, and probably better. Even without any LPF, I dont see any major issue. Also there is no need for level shifting.