Power Stability for electrochemical sensors ?

Go To Last Post
50 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


Hi All

 

I am working on a test project that will require an electrochemical N02 sensor.  This is my design as shown below. It works very well. But I have noticed that the power supply must be very stable, otherwise you need to let the sensor stabilizes before having any accurate results. 


The sensor is connected to a PCB, that has a GSM module. I  am suspecting, that there could be some circumstances where the current consumption for the module could have been high(e.g. 2A),causing the voltage drop for the sensor. This would lead to false sensor reading, till that sensor restablizes. 


Has anyone else experienced such preformance or worked on such types of sensors?



 

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Provide another power regulator for the GSM module separate from the sensor regulator, then the GSM module will not load your sensor regulator.

 

Jim

 

 

FF = PI > S.E.T

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The GSM module is powered using Li-Poly Batteries 3.7-4.2V,

 

The GSM module PCB has also got Step-up DC-DC to achieve 5.5V.

 

I am using that 5.5V to power the sensors PCB. The sensor PCB uses that 5.5V as input to it LDO , to achieve 2.5V for the sensor circuit.

 

 

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Why have R14?  It certainly won't do much except maybe add some noise.

 

Why use 5 resistors to make a voltage for the + pins?    It's a little roundabout, but ok; why not make it once & send to both opamps, for better matching (they will then use the exact same voltage).

You might add a small cap to this voltage (+pins)... its generated through some high ohms, so you can quiet it down.

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

With that LDO, the GSM module would have to load the 5V5 down a LONG ways to change the sensor voltage. If the GSM module can do that, then your DC-DC converter is badly designed.

 

I would be more concerned about bypass caps for the sensor. The op-amps seem under-bypassed but might not be since there is probably only one pair of power pins for each package.

 

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

avrcandies wrote:

Why have R14?  It certainly won't do much except maybe add some noise.

 

Why use 5 resistors to make a voltage for the + pins?    It's a little roundabout, but ok; why not make it once & send to both opamps, for better matching (they will then use the exact same voltage).

You might add a small cap to this voltage (+pins)... its generated through some high ohms, so you can quiet it down.

 

I will look into this, my design is based on the ref design from the manufactures of the sensors. 

 

 

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

djoshi wrote:
... that has a GSM module.

... the current consumption for the module could have been high(e.g. 2A)

An amp or two is approximately correct.

Has to be GSM?

Reasons : NB-IoT, LTE-M1; LoRa network? (in-place, or, DIY with relatively inexpensive gateways and servers)

Bulk capacitance, or a capacitance multiplier, "may" be the solution.

djoshi wrote:
... till that sensor restablizes. 
Dwell? (akin to digital oscilloscope sample&hold then digitize and compute)

Ideally, control latency << WAN latency

 

P.S.

Otherwise, voltage regulator LDO PSRR can be more than enough though the high-PSRR LDO are usually CMOS with a relatively few that are also low noise.

There are unconditionally stable SMPS PMIC though with significant ripple (filterable?  capacitance multiplier before LDO?)

Conditionally stable SMPS PMIC "may" have low enough noise though must analyze, measure, and evaluate stability given the cellular module's high step load.

RC LPF on the op amp power "may" be enough.

Evaluate op amp PSRR and noise; compare with other op amps (zero drift and auto-zero op amps may have low enough noise, a LNA before an op amp may be a fit)

 


The Art of Electronics 3rd Edition | by Horowitz and Hill

Download a sample chapter

[page 49, bottom of right column]

...

We discuss this in some detail in §8.15.1 (“Capacitance multiplier”), where we show the measured effects of prefiltering versus the brute-force approach of piling on lots of output capacitance (Figure 8.122).

 

LDO Regulators with Ripple Blocker Products | parametric-search | Microchip Technology though Ripple BlockerTM is restricted input voltage and IIRC has some noise (LDO, load switches)

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:

With that LDO, the GSM module would have to load the 5V5 down a LONG ways to change the sensor voltage. If the GSM module can do that, then your DC-DC converter is badly designed.

 

I would be more concerned about bypass caps for the sensor. The op-amps seem under-bypassed but might not be since there is probably only one pair of power pins for each package.

 

Jim

 

Makes sense as if the GSM did cause that, then I would also see some reconnecting issue with the module, I think maybe I need to add some filtering between the GSM PCB and sensor PCB. 

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

gchapman wrote:

djoshi wrote:

... that has a GSM module.

 

... the current consumption for the module could have been high(e.g. 2A)

An amp or two is approximately correct.

Has to be GSM?

Reasons : NB-IoT, LTE-M1; LoRa network? (in-place, or, DIY with relatively inexpensive gateways and servers)

Bulk capacitance, or a capacitance multiplier, "may" be the solution.

 

djoshi wrote:

... till that sensor restablizes. 

 

Dwell? (akin to digital oscilloscope sample&hold then digitize and compute)

Ideally, control latency << WAN latency

 

P.S.

Otherwise, voltage regulator LDO PSRR can be more than enough though the high-PSRR LDO are usually CMOS with a relatively few that are also low noise.

There are unconditionally stable SMPS PMIC though with significant ripple (filterable?  capacitance multiplier before LDO?)

Conditionally stable SMPS PMIC "may" have low enough noise though must analyze, measure, and evaluate stability given the cellular module's high step load.

RC LPF on the op amp power "may" be enough.

Evaluate op amp PSRR and noise; compare with other op amps (zero drift and auto-zero op amps may have low enough noise, a LNA before an op amp may be a fit)

 


The Art of Electronics 3rd Edition | by Horowitz and Hill

Download a sample chapter

[page 49, bottom of right column]

...

We discuss this in some detail in §8.15.1 (“Capacitance multiplier”), where we show the measured effects of prefiltering versus the brute-force approach of piling on lots of output capacitance (Figure 8.122).

 

LDO Regulators with Ripple Blocker Products | parametric-search | Microchip Technology though Ripple BlockerTM is restricted input voltage and IIRC has some noise (LDO, load switches)

 

 

Yes this issue, does seem to be from the GSM PCB board.  Is there anything that can be done using some capacitors or inductors that i could mod on the PCB?

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You are missing the most important detail: the ground connection of the yellow box on the left.
It must be the same point as op-amps, otherwise any high-current may affect all.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

djoshi wrote:
I am using that 5.5V to power the sensors PCB.
AoE3 capacitance multiplier reduced an Apple ten watt USB wall wart noise density by two orders-of-magnitude; possibly more than enough at 1 Hz (hopefully the GSM messages are "short" with a period greater than 1 ms [sic])

The Art of Electronics 3rd Edition | by Horowitz and Hill

Download a sample chapter

[page 14, middle of right column]

8.15.1 Capacitance multiplier 578

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

grohote wrote:

You are missing the most important detail: the ground connection of the yellow box on the left.
It must be the same point as op-amps, otherwise any high-current may affect all.

 

There is no ground connection to the sensor

 

This is what i am using

NO2-B43F (alphasense.com)

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

gchapman wrote:
8.15.1 Capacitance multiplier 578

 

Where do i find this?  What page?

 

As page 578, is not in this pdf

Thanks

Regards

DJ

Last Edited: Mon. Apr 4, 2022 - 08:34 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

djoshi wrote:
Is there anything that can be done using some capacitors or inductors that i could mod on the PCB?
Yes though the design is dependent on GSM message duration and frequency.

MCU commonly transform high-rate data into low-rate data (messages; compression, periodic signals can under-sampled without information loss given reasonable jitter and a short enough sample duration)

Sampling and messaging might be interleaved.

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

At the moment, I am taking a 20 samples once a minute. Maybe i can change my design to continuously take samples and average them between the TX sample rate.

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


djoshi wrote:
Where do i find this?
AoE3 is a hardcover book.

The Capacitance Multiplier | audioXpress

[top]

Figure 1: The “capacitance multiplier” circuit.

For stability, AoE3 adds to the NPN's base a series 100R or a ferrite bead.

To reduce bipolar avalanche, AoE3 adds a one amp rectifier from NPN emitter to RC junction.

Adjust R to have dVCE of 600 milli-volts to 2 volts.

A guess for your use case is a jellybean NPN as some 100 mA and 150 mA LDO and linear regulators are near jellybean.

 


The Art of Electronics | IndieBound.org

A 1+ hour drive to the next city with a larger university.

 

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

gchapman wrote:

 

 

 

djoshi wrote:

Where do i find this?

 

AoE3 is a hardcover book.

The Capacitance Multiplier | audioXpress

[top]

Figure 1: The “capacitance multiplier” circuit.

For stability, AoE3 adds to the NPN's base a series 100R or a ferrite bead.

To reduce bipolar avalanche, AoE3 adds a one amp rectifier from NPN emitter to RC junction.

Adjust R to have dVCE of 600 milli-volts to 2 volts.

A guess for your use case is a jellybean NPN as some 100 mA and 150 mA LDO and linear regulators are near jellybean.

 


The Art of Electronics | IndieBound.org

A 1+ hour drive to the next city with a larger university.

 

 

 

 

Thanks, I guess I could use 2n3905  NPN.

 

So I can place this on Vin of each Sensor PCB.

 

In my can, Vin will be 5.5V, so I need to adjust that for R.

 

What I am thinking about doing is making a simple proto PCB, with the following capacitor multiplier then it goes to my Sensor PCB.

 

So it would look like GSM PCB--> Proto PCB Filter--> Sensor PCB.

 

It seems that these sensors only consume few mA.

 

 

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

djoshi wrote:
average
Averaging can be an issue due to ESD/EMI/EFT/lightning though that NO2 sensor is low impedance; there may be a functional safety requirement due to a safety standard or specification (built-in-test, sensor shorts or opens)

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Safety standards as in?

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

djoshi wrote:
Proto PCB Filter
Protoboard for proof-of-concept, maybe first prototype, and patches.

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes something to neatly join to PCB with a filter.

 

Or I could make a i2c ADC board , with the filters on a proper PCB, as I am in process of an i2c ADC PCB for another design and I could just try add this into it that PCB.

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

White goods, industry (mechatronics, etc), automotive

examples : SO2 sensors are near some oil and natural gas wells, CO sensors in households, O2 sensors in confined spaces and with firefighters.

 

Functional Safety | Microchip Technology

Functional Safety? | AVR Freaks

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes,

 

I will check that webinar out.

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

otherwise you need to let the sensor stabilizes before having any accurate results. 

 

???

 

It is hard to fix a problem when you haven't yet identified the source of the problem.

 

Is it the problem the GSM modem's current draw, or not?

The Thread above seems to go back and forth on this.

 

Do you have a PCB with just the sensor, its power supply, and the op-amps?

Is that stable?

 

Do you have test data showing stable data, and erroneous data whenever the GSM modem is active?

 

If the sensor runs 24/7, then it should "always" be "stable".

 

You need to identify if you have a power supply problem, some noise in the sensor, RF interference, other errors creeping in here and there, ...

 

You also have to take into account the EMI of the environment in which the circuit is eventually expected to operate.

 

JC 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

DocJC wrote:
If the sensor runs 24/7, then it should "always" be "stable".
Thank you for stating complete stability.

LDO are very typically conditionally stable and are usually simple to make stable though some LDO stability is dependent on the input current loop's impedance.

LDO control current loops (immediate, load) aren't effective beyond one to ten megahertz; EMI will fold into LDO control.

Some op amps have enhanced EMIRR.

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Your schematic is too low resolution for me to examine properly.

 

  • I cannot see clearly what's going on with the final stage. That's not something I' immediately recognise. Can you post a better image.
  • What does the AUX electrode respond to ? The datasheet you linked makes no mention of it. It's usual for CO sensors to have AUX outputs but sensors for other gases generally have N/C for that pin.

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The problem does not seem to be the GSM module or PCB, as the current consumption is as specified. 

When we leave the sensor running over a long duration such as a week, this issue is noticed once every few days or the odd sample. It does not happen consistently, but when the issue occurs it can be easily be noticed on the plot.

 

Yes EMI is also considered, I am also thinking this could be the issue, maybe I need to use the Metal enclosure.  As this sensor issue, can be easily noticed, I can always use software to correct it, but I would prefer to solve it on the hardware first.

 

At first, I thought it could be GSM mode in the module, so yesterday I forced the module in NB-IoT so the power consumption is far less, to confirm if it's a high current issue or something else. So let's wait a few days.

 

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Is this better

Attachment(s): 

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


Why do you have this 1M

It's already bias at Vmid, so this just add an noisy error term.

If needing the highest precision & stability1M isn't so good to set the gain (such as R4)...moisture and dirt will have a much larger effect.  Use lower, unless you really need micropower. 

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

djoshi wrote:
... GSM ...

When we leave the sensor running over a long duration such as a week, this issue is noticed once every few days or the odd sample.

Here, LTE transmit power seems to occasionally greatly increase even though the distance between the LTE device and the cellular tower is constant; IDK why unless like conversation (be patient, don't shout, get a word in edgewisesmiley

djoshi wrote:
Yes EMI is also considered, I am also thinking this could be the issue, maybe I need to use the Metal enclosure.
Usually proximity to the PCB ground plane is enough except for inter-PCBA resonance.

 

PCB-Chassis Ground Connections - Avoiding Pitfalls Due to Unintended System Resonances | Technical Tidbit - September 2011 by Douglas C. Smith

Technical Tidbit - June 2006 - Measuring Structural Resonances

 


Mobile Phone Induced EMI | Technical Tidbit - August 2007

Mobile Phone Induced Circuit Failure | Technical Tidbit - July 2007

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes, I have also noticed the same with LTE and NB-IOT in terms of power consumption. 

 

I will read the links sent.

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

avrcandies wrote:

Why do you have this 1M

It's already bias at Vmid, so this just add an noisy error term.

If needing the highest precision & stability1M isn't so good to set the gain (such as R4)...moisture and dirt will have a much larger effect.  Use lower, unless you really need micropower. 

 

 

I simply used a reference design from the sensor manufacturer. I guess i need to look into a better design. 

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


gchapman wrote:

Here, LTE transmit power seems to occasionally greatly increase even though the distance between the LTE device and the cellular tower is constant

Hmmm, how do you know your using the same cellular tower, if all the channels are busy on that tower, it will find another that may be further away and need more power to reach it!

 

 

why is this circuit duplicated, would not one ref. be enough for both opamps?

 

Jim

 

 

 

FF = PI > S.E.T

 

Last Edited: Tue. Apr 5, 2022 - 04:30 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

why is this circuit duplicated, would not one ref. be enough for both opamps?

We are circuit brethren (#4) 

 

Then both sets will have the exact same voltage, so both outputs will be more relatable.

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

That's possible (AT&T and UScellular towers are both visible though not equi-distant)

 

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

My design is based on the following circuit reference

 

ISB (alphasense.com)

 

I am looking for another application notes, that had more information

 

 

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

OK The better resolution schematic helps greatly. I can now see you are trying to coerce a sensor driver that inherently demands a split supply onto a single supply framework.

This isn't a particularly easy task.

 

  • How are you generating VMID ?
  • I'd also expect to see U4A/Pin-3 supplied with VMID .
  • All those biasing voltages (Voff etc) must be really stable and referenced to VMID .

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Extermely sorry, did not release the power supply circuit was actually cut out. This shows how Vmid is generated.

Attachment(s): 

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What about Vcc, it must be well filtered (not from a switcher). Probably a separate linear regulator.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

At the moment VCC comes from my GSM module board, which is simply 5.5V from a Step-up DC-DC.  I do have the option of using 4.2V, directly from the battery. 

 

I could add the capacitor multiplier filter to connector before the Sensor PCB is powered or does have to be on the actually sensor PCB?

 

What is the best type of power supply for these types of sensors?

 

When creating this sensor board, based on the reference design, as it includes its own LDO, I felt that would be more than adequate but seems that the DCDC could be creating the noise.

 

 

Thanks

Regards

DJ

Last Edited: Tue. Apr 5, 2022 - 09:36 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

djoshi wrote:
capacitor multiplier

 

Try it, but be prepared for a loss in the voltage. LDO should be a better solution.

Every DCDC is creating a noise, it is intrinsic for them.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The only issue that could arrise is when the battery starts to get low, will have enough power for ldo on sensor board. I need to re check the minimum vin to get 2.5V.

 

 

 

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

djoshi wrote:
This shows how Vmid is generated.

That looks fine.

 

  • The AlphaSense link did explain (but only to a degree) what the AUX electrode measured. They claim it's for baseline current compensation. Are you performing a differential measurement (ADC1 - ADC2) ?
  • I guess you copied the reference design; but why are different amplifiers used in what are essentially identical final stages. It's true that ADC1 will exhibit much more swing than ADC2, but it still looks odd..

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

djoshi wrote:
I could add the capacitor multiplier filter to connector before the Sensor PCB is powered or does have to be on the actually sensor PCB?
For proof-of-concept, easiest else PCB.

A capacitance multiplier is effective on noise and less effective for low frequency PSRR whereas LDO add noise yet typically have enough low frequency PSRR (when LDO are stable)

Does the GSM's boost converter have an intermittent low frequency oscillation?

djoshi wrote:
What is the best type of power supply for these types of sensors?
Power supply criteria by priority : stability, noise, efficiency

  • Battery or cell (unconditionally stable, extremely low noise [discharge])
  • Zener diode [current-limited] (unconditionally stable, very low noise, extremely reliable)
  • Shunt regulator (some are unconditionally stable else simple to stabilize, low noise)
  • Series-pass linear regulator (a relative few are unconditionally stable and/or ultra-low noise)
  • SMPS ("might" be unconditionally stable though with ripple, noisy though a few have reasonable noise)

djoshi wrote:
... but seems that the DCDC could be creating the noise.
boost -> LDO (potential stability issue, disagree on noise per post #27)

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

gchapman wrote:

djoshi wrote:
average
Averaging can be an issue due to ESD/EMI/EFT/lightning though that NO2 sensor is low impedance; there may be a functional safety requirement due to a safety standard or specification (built-in-test, sensor shorts or opens)

 

Seconded on this.  It depends what your measurements look like - if they're mostly stable, and a few wacky outliers, those outliers will badly screw up your average.  Two mathematical suggestions:

 

1) Take a small odd number of samples (say 5, or 7) and use the median value.

2) Throw out the outliers (you get to define them!), then take the average of the rest.

 

Both should work well for the example given - lightning, or other sudden upset, rather than a slow-motion drift.  For slow drifts, these tricks won't help much...  S.

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


grohote wrote:
LDO should be a better solution.
Each has a place.

Some LDO are high-PSRR and ultralow-noise; similar for capacitance multiplier though not high-PSRR at low frequencies (off-line power)

A relative few LDO are unconditionally stable; a complete capacitance multiplier is unconditionally stable.

 

P.S.

A jellybean capacitance multiplier fed by a not jellybean USB wall wart nearly matches a somewhat expensive bench power supply on noise (from one who's leery of wall wart quality)

Conversely, inexpensive bench power supplies by a near jellybean 723 can be common (easily made stable, very low noise)

 


LT3042 20V, 200mA, Ultralow Noise, Ultrahigh PSRR RF Linear Regulator| Analog Devices

NCP110: LDO Regulator, 200 mA, Low Vin, Ultra-High PSRR (onsemi, PMOS, very low noise)

LT3088 800mA Single Resistor Rugged Linear Regulator | Analog Devices

REG1118 800-mA, linear voltage regulator | TI.com

 

The Art of Electronics 3rd Edition | by Horowitz and Hill

Download a sample chapter

[page 14]

8.15.1 Capacitance multiplier 578

[book page 579]

Figure 8.122

About Apple USB power adapters - Apple Support

[1/4 page]

10W USB power adapter

Apple 10W USB wall wart appears to be EOL; was 34USD at a huge US retailer.

Keysight / Agilent / HP E3610A DC Power Supply, Dual Range, 0 to 8 Vdc @ 0 to 3 A, or 0 to 15 Vdc @ 0 to 2 A, 30 W

The Art of Electronics 3rd Edition | by Horowitz and Hill

Download a sample chapter

[page 30, left column]

Figure 9.5. 723 regulator: configuration for Vout < Vref , with 2 A current limit.

UA723 150-mA, 40-V, adjustable linear voltage regulator | TI.com

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

At the moment, I am taking 100 samples after every minute.

 

Would it be better to take 1 sample every 1 seconds, and then average over 60 samples?

Thanks

Regards

DJ

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Would it be better to take 1 sample every 1 seconds, and then average over 60 samples?

 

Depends on the noise.  Say it comes in little blasts.  A blast happens just when you take 100 samples all together...so they are all way off.

You may also consider 50/60Hz line noise and  sample against that repetitive interference.   How much noise are you seeing?  Maybe none/little in the first place?

Probably do a fast acquisition of, say 32 samples as fast as possible with the adc* and avg them (just total & div by 32), then exponential weighted average (the easiest type to do)** over a longer interval, say every second.

 

*since the ADC can only do maybe 15K samples a sec, this itself will take over 2ms, so a 100us noise spike will be filtered.

 

* there is no "N", you just have, say, exp_avg= (15*exp_avg + avg32_result)/16    ...as this repeats forever with new readings, the exp_avg is formed for your final result.

 

You can get fancier...for example, if a sample suddenly deviates too far from a previous value or known avg (careful!), you might toss it in the trash as an outlier.   You could take 50 samples and sort/bin them and only keep the 30 most central values and avg those.  Most of this is for situations where you really have noise troubles, or other statistical needs (for example, not all noise is zero-mean).

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Tue. Apr 19, 2022 - 04:40 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You might need to update us as to where you are with the project.

Did you overcome the GSM  module induced nose ?

Or are you asking about general background noisy measurements ?

 

Don't lose sight of what you're trying to measure here. A few 100nA per ppm of NO2. Quiet supplies and careful PCB layout regarding analogue grounds are essential.

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

djoshi wrote:
N02

SVS: Nitrogen Dioxide Over the United States, 2005-2021 (USA NASA)

[after 14s video]

Nitrogen dioxide can impact the respiratory system, and it also contributes to the formation of other pollutants including ground-level ozone and particulates. ...

at 0:51 for about 10s in X Class Solar Flare, Uranus, Big Interview | S0 News Apr.20.2022 - YouTube (3m1s)

 

"Dare to be naïve." - Buckminster Fuller