INA193 High side current sensing, ground problem?

[Atom]

hello! i'm working on a "power meter" that should be able to measure current and voltage of a dc power source, but i have a couple of problems

1- The ina193 and the load are connected to 2 isolated power supply, the DO NOT share a ground, if i try to measure the current its spot on. almost perfect, if i connect the grounds togheter the output voltage of the ina 193 is shifted by 20 to 30mV upwards, note that i'm using a switchmode power supply to supply the load so i thought it could be noise from the switchmode---it isn't, i've checked whit the oscilloscope and there wasn't any switching noise... that's STRANGE

2. i started measuring the current with an ads1115 and an arduino to serial monitor...everything good, so now i wanted to get the voltage on the load, so i added the resistive divider 39k and 3k for a maximum input voltage of 70v .... 70v in 5 out so the adc could read it.

3 the resistive divider shares a commoun ground with the load thats obvius, but if i want to measure voltage (from the divider) and current from the INA193 they HAVE TO share ground since i have only one adc

4now heres when the magic smoke escaped i connected the 2 ground togheter (1.5Kohm load ) and i started to increase the voltage from about to 5v to 40v and when i was almost at 40V the ina193 burned ...waited a few seconds and the 5V regulator started burning too (i'think it was burning because of the INA193 failed shorth circuited).

5 soldered new INA 193 and AMS1117 5.0v now whitout the 2 grounds connected i can measure current and voltage perfectly (with the multimeter not the adc).
if i connect the 2 f*****g ground togheter the amps measurment goes up of about 500mv ....

EXAMPLE SMPS SET TO CC 1A LOAD 4.7OHM (it is in CC mode so 4.7Vout from the SMPS)

i attach the load and i can read 2V from the INA193 that's good 1A=2V because 0.1ohm(shunt)*1A*20(amp of INA193)= 2V
i go to read the voltage from the divider ..everything good , THE GROUND ARE STILL NOT CONNECTED

if i try to connect the 2 grounds the Voltage of the SMPS jumps to 5.5V ....now something is backfeeding voltage into the SMPS  but i can see what it is, has anybody some solution? here's the schematics

[Marco]

The polarity of Vcc is wrong? Hard to see how you're burning up the INA otherwise.

[Atom]

nop, the polarity is good ... i've done further testing, When Vcc is lower than 5v so for example VCC = 3V if i connect the ground togheter i get an increased output voltage from the ina 193, and if a put an ammeter in series with the groun lead i see 500ma or even 1 amp flowing from ground of psu1 to ground of psu 2...WTF.. i'll do a video so its easier to understand.

[Jeroen3]

Do you short the grounds at the supply or trough your board?
Are the supplies properly floating?

In any case, if you connect a floating supply on the shunt of the INA, it will fry itself. That would have been ~115 Vac common mode, not 80Vdc.

[Atom]

Do you short the grounds at the supply or trough your board?
Are the supplies properly floating?

In any case, if you connect a floating supply on the shunt of the INA, it will fry itself. That would have been ~115 Vac common mode, not 80Vdc.

i short them by connecting them with a cable at the supply , i'll do a video.

[Atom]

i've done the video i'm uploadign it now, it'll take a while because i have a s**t upload speed,

[SiliconWizard]

Not connecting the grounds together, as it is, is a mess. The return path for the INA193 would be the resistive divider (without a return path, an instrumentation amp won't work properly), which itself breaks isolation (but doesn't make the two grounds share a common potential...)

What is VCCmin? Could VCC be used to power your whole circuit (via a step-down converter), so you'd avoid all the trouble with an additional PS?
Do the two sides have to be isolated (in which case, your approach won't work anyway as it breaks isolation)?

It's possible that your two power supplies are actually NOT completely isolated from each other, which could explain your troubles.

Show us more about the two PS.

[namster]

First , the Absolute Supply Voltage is 18V the common mode rating of -16V and 80V is the actual VCM rating for the inputs , i purpose a modification in the schematics , you have to replace the regulator U2 with another wich can have input of 70V .
For ADC you can connecte the two ground together and supply the arduino with the out of regulator U2 , Or use a external ADC with seprate Ground (Digital and Analog ground ) .

[Atom]

First , the Absolute Supply Voltage is 18V the common mode rating of -16V and 80V is the actual VCM rating for the inputs , i purpose a modification in the schematics , you have to replace the regulator U2 with another wich can have input of 70V .
For ADC you can connecte the two ground together and supply the arduino with the out of regulator U2 , Or use a external ADC with seprate Ground (Digital and Analog ground ) .

1 i dont think it will work , first of all finding a linear reg with a vin of 70v is going to be difficult ..and probably pricey.
2 by doing like in your schematics the current measured would be ILoad +ILin Reg, i only want i load to be measured.

[Atom]

Not connecting the grounds together, as it is, is a mess. The return path for the INA193 would be the resistive divider (without a return path, an instrumentation amp won't work properly), which itself breaks isolation (but doesn't make the two grounds share a common potential...)

What is VCCmin? Could VCC be used to power your whole circuit (via a step-down converter), so you'd avoid all the trouble with an additional PS?
Do the two sides have to be isolated (in which case, your approach won't work anyway as it breaks isolation)?

It's possible that your two power supplies are actually NOT completely isolated from each other, which could explain your troubles.

Show us more about the two PS.

1 i know not connecting ground is a mess, infact even in the datasheet they connect ground togheter.

2 i'm pretty sure the psus are isolated ..they are bench psus and their ground isnt "earthed"

3 i dont care about using an additional power source , i dont care for the isolation either (i'll isolate the i2c bus of the adc).

4 where did you rad Vccmin ? i dont want to use a step down converter because i want the INA 193 to always work...for example the load is powered by 1 V ..and the INA193 has its own 5V(VDD) power supply ..by doing so i can measure the current even if VCC is 1V

pics of the 2 psus

[SiliconWizard]

4 where did you rad Vccmin ?

I didn't, that's why I asked.
If Vccmin was high enough, then you could use a step-down converter. That was the reason for asking.

i dont want to use a step down converter because i want the INA 193 to always work...for example the load is powered by 1 V ..and the INA193 has its own 5V(VDD) power supply ..by doing so i can measure the current even if VCC is 1V

OK that was the answer. VCCmin would not be high enough for the above solution.

Have you tried connecting the - terminal of both PSUs together, not connecting the + terminals to your circuit. And measure voltage between each + terminal and the common - ?
Just to check that they behave properly in this setup.

[Atom]

Have you tried connecting the - terminal of both PSUs together, not connecting the + terminals to your circuit. And measure voltage between each + terminal and the common - ?
Just to check that they behave properly in this setup.

i'll try to do that.

here's the video https://drive.google.com/file/d/1pPiyfZVf9aA8VIY2rUKaO65FmYq3KZjR/view?usp=sharing

[namster]

for low  cost regulation you can use a zener diode , by choosing the right value of  R3 , in previous post i connected the régulator before the shunt resistor that was error on my part , the INA193 brunout beaucoup of high supply voltage ! not because you connect the ground together .
for more viable version you can replace your adc with another wich have separate ground ( Analog and Digital ) .

[Jeroen3]

Good video!

Mistake 1: at 2:58 you connect the second power supply, before it's ground was made common. This creates a surge of current through the INA. However, it doesn't in your case, read below.
Mistake 2: at 7:26 you short the ina output.

Either: the problem seems to be your Delta Electronic power supply is sinking current. Somewhere supplies are not as isolated at you think they are.

Or you board accidentally parallels the power supplies, trough the INA. Does it get hot when you run an amp trough it?
You might have fried internal R1 with mistake 1 causing the INA to lose it's common mode capabilities.

_{By default the Delta one does not come with banana plug on the front, are the connections on the rear touching something?}

[Atom]

Good video!

Mistake 1: at 2:58 you connect the second power supply, before it's ground was made common. This creates a surge of current through the INA. However, it doesn't in your case, read below.
Mistake 2: at 7:26 you short the ina output.

Either: the problem seems to be your Delta Electronic power supply is sinking current. Somewhere supplies are not as isolated at you think they are.

Or you board accidentally parallels the power supplies, trough the INA. Does it get hot when you run an amp trough it?
You might have fried internal R1 with mistake 1 causing the INA to lose it's common mode capabilities.

_{By default the Delta one does not come with banana plug on the front, are the connections on the rear touching something?}

thanks, yes at 7.26 i shorted the output of the INA193 in fact i kinda pull off the cables really quickly ...the ina didnt get damaged so its all good, i have another 7 so i shoulfìd be able to solve the problem without frying them all.

yep the delta psu is sinking current.

now ive done further testing and there are 2 situations but befor lets clarify some things:

VCC is a voltage variable from 0-70V , this voltage is the voltage applied to the load ...the ina193 has a voltage rating of 80V so it should be fine.
VDD is a voltage bigger than 5V so the 5V lin reg can regulate the voltage(this is the "logic" supply voltage)

SITUATION 1. VCC<5v (why 5v ? because its the supply voltage of the ina)

when connecting the 2 grounds togheter i get huge currents passing trough the "common ground lead"... from 500ma to even 1A, the readings from the ina gets all wrong and the voltage VCC gets pulled up to 5 V

its seems like the voltage regulator (the 5V one is bringging up the VCC rail)

SITUATION 2. VCC>5V

connecting the 2 grounds togheter i can see a current of 5ma flowing at about VCC =12V... bringing up the voltage of VCC brings up also the current in fact at VCC=40V the current measured is 40ma, at this point the INA193 relases the magic smoke and burns.

any help is appreciated

[Atom]

for low  cost regulation you can use a zener diode , by choosing the right value of  R3 , in previous post i connected the régulator before the shunt resistor that was error on my part , the INA193 brunout beaucoup of high supply voltage ! not because you connect the ground together .
for more viable version you can replace your adc with another wich have separate ground ( Analog and Digital ) .

i pushed the ina at max 40V and it burned WITH the ground in common.. now ill try to push it to 70V without the common ground and see if something happens.
i dont mind purchasing an external and isolated psu fot the ina. i could also wound additional spires on a toroidal transformer and have a lot of isolated outputs(thats what i'v planned to do)
i could replace the adc but this one is really cheap at about 1 euro and its a 15bit one ..thats really good for the price
the ina does not burn out from the high voltage, it has a separated power supply @5V

[namster]

this kind of IC have to be connected with one ground for correct operation (supply and Vin ) , i think that you have to review your design.
the tooridal transformer isn't a good idea , in a DC voltage the transformer will be saturated .

[Marco]

any help is appreciated

It would probably help if you admitted to yourself that you're almost certainly doing something wrong and one of the myriad assumptions you are making is wildly flawed. If you blew up the INA193 you either have far more voltage on the supply leads, input or even output than you think you have.

[Jeroen3]

when connecting the 2 grounds togheter i get huge currents passing trough the "common ground lead"... from 500ma to even 1A, the readings from the ina gets all wrong and the voltage VCC gets pulled up to 5 V

Didn't you notice the amount of current flowing is equal to the current limit set on your delta power supply?

Power the INA supply from a 9V battery, see what happens.

[Atom]

problem solved.

while soldering a ground cable on the pcb the insulation melted ..melted the isulation of another cable..one of the 2 from the sense lines of the shunt... so the 2 cables where touching..

here's testing at 50V 2.4A

[Atom]

this kind of IC have to be connected with one ground for correct operation (supply and Vin ) , i think that you have to review your design.
the tooridal transformer isn't a good idea , in a DC voltage the transformer will be saturated .

why would be a toroidal transformer be bad ?..i added 4 isolated secondaries and i plan to use this one as VDD for the ina193 op amp and adc

i wanted to add more input frotection for the adc and for the VCC input so i added a fuse and a MOV

for the adc input i added 2 schottky diodes for clamping after the ina 193

buffer for the resistor divider ...RC filter and schottky clamping, is this enough? should i add someting else?

should the ina193 output be buffered by another opamp? i dont think so, let me know what you think

[namster]

RC filter have to be calculated according to your sampling Frequancy , for schottky clamping it can be replaced by a zener diode of 5V to protect the input of Arduino
I thought I understood that you will connecte the primary of transformateur in output of Opamp for isolating the output , that's why i said  that will not work because it will be saturated in DC mode , for now I have not yet understood how you will use the transformer ? , maybe  for the linear power Supply ? to continue the discution please clarify with a schematics . 

[Atom]

maybe  for the linear power Supply ? to continue the discution please clarify with a schematics . 

yes the transformer will be used for the linear psu before the bridge rectifier, there are other "safety" measures that i can adopt?

[namster]

Ah ! , the toroidal transformer is a good choice , it have a better efficiency than a classic transformer ! , you can also add a EMI filter fuse and varistor , a Series Regulator Diode between Vin and Vout of your regulator , to protect it against reverse voltage
https://www.digikey.com/Web%20Export/Supplier%20Content/Lambda_285/PDF/TDKLambda_all_about_emi_epmag.pdf
http://literature.cdn.keysight.com/litweb/pdf/5952-4020.pdf (39)