Forum Update - Announcement about May 10th, 2018 update and your password.

Electrical Gurus Needed :) ---- Measuring The Millivolt With High Accuracy

1235

Comments

  • Tracy
    Ahh, that chart is sort of what I was looking for. I looked up my 4 amp ATC fuse, and the chart says 5mV across it would mean 219 mA flowing through it. That translates to 5/0.219 = 22.8 milliohms. Checking other mV readings for that fuse in the chart--what a surprise!--They obey Ohm's Law.

    I had measured 30 milliohms using the ohmmeter (Agilent 34401 6.5 digits) earlier on my specimen. To check, I hooked it up to a constant current power supply (Agilent 6612) set at 0.219A. The voltage across the fuse was 6.6mV, which does translates to 30 milliohms. So there is a start on an error bar, 22.8 mΩ vs 30mΩ.

    In the ballpark anyway.

    I would assume that fuses are much like a resistor in having precision values, and it would probably all depend up the accuracy of some sort stamping die.
    About the Adafruit breakout board. Assuming you do remove the shunt to try it out, do add two resistors ~1kΩ between the (+) and (-) inputs and your fuse under test, and on the breakout side a pair of back to back parallel silicon diodes to clamp the differential voltage. Run the ADC on its own battery power, not from the car's power.

    I believe the last breakout board that you referred to was the ADS1015, however several breakout boards have been mentioned in this discussion. For clarification, specifically which breakout board are you referring to in your last comment?


    Novel Solutions - http://www.novelsolutionsonline.com/ - Machinery Design • - • Product Development
    "Necessity is the mother of invention." - Author unknown.

  • Yea, but the INA219 part# is speaking to me, because I live in a 219 area code

    And now repeating part of Tracy's last post, which also includes the numbers 219...
    Ahh, that chart is sort of what I was looking for. I looked up my 4 amp ATC fuse, and the chart says 5mV across it would mean 219 mA flowing through it. That translates to 5/0.219 = 22.8 milliohms. Checking other mV readings for that fuse in the chart--what a surprise!--They obey Ohm's Law.

    I had measured 30 milliohms using the ohmmeter (Agilent 34401 6.5 digits) earlier on my specimen. To check, I hooked it up to a constant current power supply (Agilent 6612) set at 0.219A. The voltage across the fuse was 6.6mV, which does translates to 30 milliohms. So there is a start on an error bar, 22.8 mΩ vs 30mΩ.

    I am just saying... :)


    Novel Solutions - http://www.novelsolutionsonline.com/ - Machinery Design • - • Product Development
    "Necessity is the mother of invention." - Author unknown.

  • 219 = 3 times 73 as prime factors. And 73 means "best regards" in ham radio lingo--Psychic!

    About the precision of the resistors in a fuse, I wonder. Different manufacturers could use different metal alloys that melt at different temperatures and exhibit a different DC resistance, but still meet all the specs as a fuse. Above the rated value, the resistance becomes time-dependent, 20% above rating, a slow burn, resistance gradually increasing as metal wafts away.

    About the breakout boards. Looking at either the INA219 or the ADS1115/ADS1015 Adafruit breakouts, the inputs to the ADC are connected directly to the breakout i/o pins. I'd like to see more isolation to protect from accidents, but without disturbing the mV measurement. I'll try to diagram it out later today when I have time.



  • Tracy
    About the precision of the resistors in a fuse, I wonder. Different manufacturers could use different metal alloys that melt at different temperatures and exhibit a different DC resistance, but still meet all the specs as a fuse.

    I am certain that you are quite correct. My fuses appear to be made from some type of aluminum alloy.
    About the breakout boards. Looking at either the INA219 or the ADS1115/ADS1015 Adafruit breakouts, the inputs to the ADC are connected directly to the breakout i/o pins. I'd like to see more isolation to protect from accidents, but without disturbing the mV measurement. I'll try to diagram it out later today when I have time.

    Thanks Tracy, I surely would appreciate a diagram or schematic of the intended preventative fixes.


    Novel Solutions - http://www.novelsolutionsonline.com/ - Machinery Design • - • Product Development
    "Necessity is the mother of invention." - Author unknown.

  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 21,966
    edited May 8 Vote Up0Vote Down
    I suspect that the metal-to-metal interfaces where the fuse is plugged in will have a higher resistance than the fuse itself -- especially in a car, where there can be corrosion. If that's the case, it begs the question of where to do the probing. (What was I thinking? :) )

    -Phil
    “Perfection is achieved not when there is nothing more to add, but when there is nothing left to take away. -Antoine de Saint-Exupery
  • jmgjmg Posts: 11,657
    About the breakout boards. Looking at either the INA219 or the ADS1115/ADS1015 Adafruit breakouts, the inputs to the ADC are connected directly to the breakout i/o pins. I'd like to see more isolation to protect from accidents, but without disturbing the mV measurement. I'll try to diagram it out later today when I have time.

    Certainly a good idea, and quite tricky to provide protection and not disturb too much the mV reading.

    Many of those IV chips have a Sense pin that is also a bias pin, and needs some uA of current, with poor tolerance and tracking. ie 20uA into even 100 ohms, is 2mV, 20uA into 50 ohms, is 1mV
    That can be zeroed, with a shorted probe type step.

    The better INA233 lacks a curve that does show in the INA219, (?)
    Figure 8. Input Currents With Large Differential Voltages(VIN+ at 12 V, Sweep Of VIN–)

    That suggests a +12 or -12 swing does have some inbuilt current limit, to something under 1mA ? - ie maybe clamp Zeners to device GND pin are enough, with modest RS (10 ohms?) ?


    For more rugged protection, Diode/Zener clamps is a start, but the series R ?

    There are resettable fuses from Murata called positors, that spec 100 ohms Trip 45mA, or 47 ohms trip 61mA, and the PTC nature ramps to 100x ~ 1000x the normal R
    Those are ~27c /100

    Or, another choice is the more esoteric of a JFET or depletion mode mosfet (Maybe BSS139 + Rgs) - these parts are inherently on, and current limit at higher voltages.

    A MMBF4393CT-ND is 8c/100, and Spice says typical Rds of 40 ohms, Typical Id(0) of 18mA. Adding Rgs improves Id - eg 22R is 12mA, 47R is 9mA, but those R's add to the Rds, so it's a compromise...

    or, another possible limiter, is the NSI50010/NSI45015 etc Current regulator diode - around the origin, they have the same action as a depletion mosfet, but a much more tightly controlled final limit current.
    - found a NSI50010 strip here, and they measures 108~126 ohms, so unbiased, they do look like simple resistors, as the graph suggests.
    That's a 10mA current regulator diode, so higher current ones, will have lower Rs. 20~30~50mA ones, could be 60~40 ohms ?
    Unlike the SOT23 fets, these parts are designed to cope with running heat, and come in a selection of currents and thus resistance.

    All of these semi devices are one-way protection, so using 2, one per probe, means one is diode-drop biased and the other is in limiting mode.
  • Use a HX711 which was meant to be a cheap front end for scales and thermocouples. You can get breakouts from eBay for a few bucks and there is example code in the OBEX.
  • jmgjmg Posts: 11,657
    localroger wrote: »
    Use a HX711 which was meant to be a cheap front end for scales and thermocouples. You can get breakouts from eBay for a few bucks and there is example code in the OBEX.

    Interesting part, broadly similar to the NAU7802, (also both in SO16N), with a quite simple serial interface.
    NAU7802 has much more comprehensive choice of gains, and calibrate/offset registers, via the i2c bus
    The HX711 looks intended for ratiometric reading, but it has easy to get breakouts, and it might be 'good enough' - does look easy to try, and the NAU7802 is quite similar on the analog front end, so it could swap-in later if higher performance is needed.
  • The HX711 also has a wicked thermal drift so be sure to zero often! Placing my finger on it I could observe the output rise/fall.
  • Tracy AllenTracy Allen Posts: 6,262
    edited May 9 Vote Up0Vote Down
    So Bruce, here are the diagrams I promised.
    A generic input protection looks like the first diagram attached. The probe is touching the two sides of the fuse. If the fuse happens to be blown with 12V across the contacts, the fault current (red arrows) will be diverted through 200Ω and the diodes. The diodes are back to back so that they function no matter which way the probes are applied. Either way, the ADC in your gizmo sees only +/-0.7 volts across the differential inputs. Both the diodes and the resistors have to be suitably sized to handle the possible fault current. With the values shown the fault current would be 60mA, and the resistors would have to be rated at 1/2 watt. You could use rectifier diodes, maybe in a DO219 package. (Had to sneak in the magic number 219!). As already pointed out, there can be error sources that would enter into this. These are indicated as "?µA". The resistors might change the gain of a followup circuit or they might incur voltage drops due to bias current that need to flow in the circuit. For the ADS1115 or ADS1015, the sources of those currents are pretty well laid out in the data sheet. At best, some of the errors would be predictable and could be subtracted out in firmware, whereas others such as temperature drift and noise would be more problematic. The resistor values at the input could be lower, but the fault currents become unwieldy. About the INA219, the exact topology of their input circuit is hidden and seems to be kind of weird and unbalanced at the two inputs, so it would take experimentation. For example, 20µA of bias current amounts to 4mV of offset across 200Ω, which is substantial in relation to the normally expected fuse voltages.

    The second diagram starts with a classic differential amplifier made with an op-amp at the front end. The two 1kΩ resistors determine the gain and also act as the protection. The fault current here would be only 6mA, so the resistors could be 1/4 watt, and the diodes could be cheap 1N4148s. The differential amplifier as shown would have a gain of x100, and the output is offset by 1.2V (or, could be ADC range/2), so that it can accommodate currents in either direction (probe reversal). The reference and op-amp output feed a differential input of the ADC. This is basically the same shunt reference circuit as is shown on page 32 of the ADS1015 data sheet. They add some additional filters. The resistors have to be matched, like 0.1%, and the op-amp should be a precision type. My favorite is the LTC2054 (auto-zeroing, <5µV offet, input bias current 1pA, near rail to rail i/o).
    388 x 182 - 10K
    639 x 260 - 15K
  • Tracy

    Thanks for getting back with me and posting the diagrams. I ordered the parts necessary to construct the second diagram.
    About the Adafruit breakout board. Assuming you do remove the shunt to try it out

    Just for clarification... When you used the word "shunt", you were referring to the INA219 breakout board... Is this correct? Because I would assume that a shunt would be across the inputs, and the ADS1015/ADS1115 breakout boards do not have a resistor associated with any of the inputs.

    Reference schematic for Adafruit ADS1015/ADS1115 breakout boards: https://cdn-learn.adafruit.com/assets/assets/000/036/145/original/sensors_schem.png?1475541078


    Novel Solutions - http://www.novelsolutionsonline.com/ - Machinery Design • - • Product Development
    "Necessity is the mother of invention." - Author unknown.

  • kwinnkwinn Posts: 7,873
    idbruce wrote: »
    Tracy

    Thanks for getting back with me and posting the diagrams. I ordered the parts necessary to construct the second diagram.
    About the Adafruit breakout board. Assuming you do remove the shunt to try it out

    Just for clarification... When you used the word "shunt", you were referring to the INA219 breakout board... Is this correct? Because I would assume that a shunt would be across the inputs, and the ADS1015/ADS1115 breakout boards do not have a resistor associated with any of the inputs.

    Reference schematic for Adafruit ADS1015/ADS1115 breakout boards: https://cdn-learn.adafruit.com/assets/assets/000/036/145/original/sensors_schem.png?1475541078

    I'm sure he meant the shunt resistor mounted on the INA219 breakout board. Worth trying that across a fuse or two to see how well it works.
    In science there is no authority. There is only experiment.
    Life is unpredictable. Eat dessert first.
  • kwinn

    Yeah, I was pretty sure he was referring to the INA219 board, but then I think he changed gears and drew up diagrams for the ADS1015/ADS1115.

    Whatever the case, I already have the ADS1015, and soon, the ADS1115 and the INA219 will be delivered, as well as all the parts, per his second diagram.

    Perhaps his circuitry would be good for all three of these breakout boards.


    Novel Solutions - http://www.novelsolutionsonline.com/ - Machinery Design • - • Product Development
    "Necessity is the mother of invention." - Author unknown.

  • jmgjmg Posts: 11,657
    idbruce wrote: »
    Whatever the case, I already have the ADS1015, and soon, the ADS1115 and the INA219 will be delivered, as well as all the parts, per his second diagram.

    Whichever board you select, I'd suggest doing your own plot the same as INA219 Figure 8. Input Currents With Large Differential Voltages(VIN+ at 12 V, Sweep Of VIN–)
    ie place a 1k in series with the pins, and sweep vDiff while you measure the mV across 1k.
    That shows the operating current, and clamp current pathways.
  • idbruceidbruce Posts: 5,655
    edited May 16 Vote Up0Vote Down
    Whichever board you select, I'd suggest doing your own plot the same as INA219 Figure 8. Input Currents With Large Differential Voltages(VIN+ at 12 V, Sweep Of VIN–)
    ie place a 1k in series with the pins, and sweep vDiff while you measure the mV across 1k.
    That shows the operating current, and clamp current pathways.

    I may need some help with that, once I get everything set up :)

    Anyhow, I thought I would take a moment and update this thread.

    At this point, all the ordered parts have arrived, except an order from SparkFun, which includes a SOT23 to DIP Adapter, for mounting the LTC2054 that Tracy suggested.

    While waiting for the ordered parts, I accidentally stumbled across an item, which should improve the versatility of my tool, but this item has presented a different set of design challenges and necessitates the need for ordering additional parts.

    And of course, a variety of user interfaces keep running through my mind, but I am doing my very best to keep it cheap, simple, durable, and an overall nice tool to own for troubleshooting current draws. Whatever the end result, I will eventually need to order additional parts to make it all usable for the intended application.

    EDIT: As a side note... The replacement of the battery, still seems to have fixed my problem, but I am still carrying around the booster pack, just in case :)


    Novel Solutions - http://www.novelsolutionsonline.com/ - Machinery Design • - • Product Development
    "Necessity is the mother of invention." - Author unknown.

  • idbruceidbruce Posts: 5,655
    edited May 19 Vote Up0Vote Down
    OOOPPPPSSS!!!

    I overlooked the 1.2 VREF required for Tracy's #2 diagram.

    Any recommendations for my best overall through hole options?

    EDIT: I would imagine that I want this VREF to be accurate as possible. Should I invest in a high accuracy VREF IC?


    Novel Solutions - http://www.novelsolutionsonline.com/ - Machinery Design • - • Product Development
    "Necessity is the mother of invention." - Author unknown.

  • jmgjmg Posts: 11,657
    idbruce wrote: »
    OOOPPPPSSS!!!

    I overlooked the 1.2 VREF required for Tracy's #2 diagram.

    Any recommendations for my best overall through hole options?

    eg AZ432BZTR-E1 is a TO-92 1.25V 1% shunt ref, but you could 'get it functional' using 2 diodes in series.
    AZ431LAZTR-E1 claims 0.5%, 1.24V AZ432AZTR-E1 is 0.5%, 1.25V
  • jmg

    I was just looking at what I would assume is a voltage divider using a zenier diode: https://www.nutsvolts.com/uploads/wygwam/NV_0907_Malone_Figure01.jpg

    I am assuming that in your suggestion, these two diodes would be attached to the + side of 3.3V regulator?


    Novel Solutions - http://www.novelsolutionsonline.com/ - Machinery Design • - • Product Development
    "Necessity is the mother of invention." - Author unknown.

  • jmgjmg Posts: 11,657
    idbruce wrote: »
    jmg

    I was just looking at what I would assume is a voltage divider using a zenier diode: https://www.nutsvolts.com/uploads/wygwam/NV_0907_Malone_Figure01.jpg

    I am assuming that in your suggestion, these two diodes would be attached to the + side of 3.3V regulator?

    Depends on where the 1.2v is referenced to - as above looks to be just lifted above ground, to allow +/- signals, so the diodes are GND connected and series connected, to a resistor to 3v3. Try 10k.

  • Depends on where the 1.2v is referenced to - as above looks to be just lifted above ground, to allow +/- signals, so the diodes are GND connected and series connected, to a resistor to 3v3. Try 10k.

    Thanks jmg


    Novel Solutions - http://www.novelsolutionsonline.com/ - Machinery Design • - • Product Development
    "Necessity is the mother of invention." - Author unknown.

  • Tracy AllenTracy Allen Posts: 6,262
    edited May 19 Vote Up0Vote Down
    Hi Bruce,
    The reference does not need to be accurate, and it doesn't even need to be 1.2V. I chose 1.2V because it is a common and because it is about 1/2 the 2.56V range on the ADS1115/1015. To get started, use a red led forward biased at about 5mA. (Like that zener circuit but diode forward biased. ) Leds make passable voltage references.

    The exact reference voltage does not matter, because the ADC is measuring differential between the reference and the amplified input. The reference should be steady though, otherwise it affects the gain of the amplifier away from the target x100.
  • I've tested some of these fuses and I'm getting good results with my LF356/Ina117 combo. I'm passing 1Ma through a 30 amp plastic auto fuse and it yields about 0.02 mv on my 6 digit fluke. My output of my INA117 is 125mv for 1 ma, 250 mv for 2 ma, 500mv for 3 ma etc. 1v for 4 ma etc.
    A common sleep mode for a cars ECU would 25-50 ma. The 15 amp fuse yield the same at 1ma current flow.
  • Tracy

    Thanks for the additional input.

    I will be soldering up some stuff here in a little while and figuring out my test setup.


    Novel Solutions - http://www.novelsolutionsonline.com/ - Machinery Design • - • Product Development
    "Necessity is the mother of invention." - Author unknown.

  • I typically use this circuit for shunts works excellent. I have it scaled by a factor of one hundred. 0.5 mv input 50mv. 10mv input 1 VDC output. No drift and dead on linear.
  • My typo 0.5 mv input and 50mv output, 10mv input and 1vdc output etc.
  • jmgjmg Posts: 11,657
    edited May 23 Vote Up0Vote Down
    Mickster wrote: »

    Nice instrument, but gets confusing in the finer specs, as the table talks about B35+ models not B33 ones ?
    some show Hz and some show Hz/% ?

    eBay also finds this...
    https://www.ebay.com/itm/RD-UM25C-UM25-Bluetooth-Tester-Meter-Ammeter-Voltmeter-Battery-Charge-Measure/302696604759
  • Mickster
    Probably off on a tangent but FYI:

    :) Considering that I am almost always secretive, tangents should always be expected :)

    In my opinion, all input is good, except for negative naysayers that add their two cents :)


    Novel Solutions - http://www.novelsolutionsonline.com/ - Machinery Design • - • Product Development
    "Necessity is the mother of invention." - Author unknown.

  • Don't forget contact points of your probes on the fuse. With the small amount of mv your measuring any bad contact will produce an error. Like a cold junction on a thermocouple. Might want to use a scope probe setup.
  • ErNaErNa Posts: 1,090
    Hello idbruce, followed the discussion for a while but not closely, I remembered a solution to measure starter battery load, because at start the current can be 1kA, but bleeding would happen with 1 mA, so it is a problem for automotive industry. Eventually I found this link: https://www.hella.com/emergency/en/Intelligent-Battery-Sensor-543.html so you can see, there is a solution, and it must be cheap, otherwise there is no change in the automotive industry
Sign In or Register to comment.