Bias current sensor for floating earth

1. Bias current sensor for floating earth

Hello guys,

I need to take some idea about how to design a power electronic PCB to measure the isolation between high voltage DC link and a floating EARTH or Chassis. Have a look to the next pic:

There are two possibilities or cases:
- CASE A: The positive side of the DC link is connected accidentally to EARTH.
- CASE B The negative side of the DC link is connected accidentally to EARTH.

I need to design an simple analogue solution to identify every CASE and to measure the current drive to EARTH. Any idea?

2. Re: Bias current sensor for floating earth

So you're looking to detect a ground fault where either your DC link positive or negative is shorted to earth?

You can directly measure the voltage of either node to Chassis and see if one is equal to chassis. You can meet safety if you choose sufficiently high impedance resistors to do this.

Alternatively you could inject an AC voltage onto your isolated supply through a capacitor (safety rated) and measure how much current it draws. If its truly floating no current should be drawn (except the parasitic C across the isolation barrier). If its shorted your cap will draw more current. Not sure if your application can allow such an excitement.

3. Re: Bias current sensor for floating earth

Exactly! I need to detect a fault ground condition and it is not possible to inject any AC signal because the DC bus is powered by a group of photovoltaic panels capable to produce 1500Vdc.
Therefore I need a solution based for example on some power resistors, transistors and opto couplers...do you any similar schematic to take as a starting point?
Greetings

•

4. Re: Bias current sensor for floating earth

Hi,

You could use a high ohmic voltage divider connected to earth.
Then measure whether the voltage remains symmetric ... with respect to the DC bus voltage.

Klaus

5. Re: Bias current sensor for floating earth

Hi!
That's could be a solution but on normal working mode, the Earth should not be connected to the DC bus. It should remain floating. Just in case if something is wrong on the DC bus, it could be that one of the nodes will be short-circuited to the Earth. If I use a resistor divider connected to Earth, I would connect the nodes of the DC bus to Earth during the normal working mode...
Finally how could I simulate on Ltspice circuits with floating grounds?

6. Re: Bias current sensor for floating earth

Hi,

the Earth should not be connected to the DC bus. It should remain floating.
Voltage measurement as well as current measurements need two nodes.
"Floating" means "undefined voltage" Thus you cant find out what voltage is "good" and what voltage means "earth leakage".

Thus I doubt you can work with true floating nodes. At least I don't know any reliably working method.

Btw: Floating or not floating ... there is no ideal "high imoedance" ... so, where do you set the limit?

Klaus

7. Re: Bias current sensor for floating earth

Think of it this way: There is no such thing as 'floating'. Any two points on earth have some amount of resistance and capacitance between them.

And all isolated power supplies have R and C across the isolation barrier (maybe on the order of 250pF and 10MOhm) and an insulation breakdown limit usually between 2 and 10kV.

So Klauss's suggestion is valid. You can already assume that your 'floating' output is within a few kV of earth and already has maybe 10MOhm of resistance to earth.

Now you can see that adding an additional high ohmic resistive path probably doesn't matter.

And that's how you need to model it in simulation: one ground with your 'floating' source tied to it through maybe 10Meg and 250pF.

1 members found this post helpful.

8. Re: Bias current sensor for floating earth

IMD (insulation monitoring devices) are specified by IEC 61557 and available on the market. For unearthed (IT) power supplies, these devices are basically implementing the operation principle sketched by asdf44 in post #2, inject an AC voltage respectively current and measure the response. Bender is the inventor and a leading manufacturer of IMD. https://www.bender.de/en/products/in...oring-overview

9. Re: Bias current sensor for floating earth

Hello guys!
I would like to share with you a possible solution to the EARTH fault detection problem. In the next pic it is possible to see how the schematic looks like.

In normal working mode, FAULT_VPAN+ and FAULT_VPAN- signals are at high level.
When VPAN+ is short-circuited to EARTH the FAULT_VPAN+ signal is at low level.
When VPAN- is short-circuited to EARTH the FAULT_VPAN- signal is at low level.

However this solution needs some improvements:
1) Which commercial parts of transistors and diodes do you recommend me?
2) R_ISO is used to configure the isolation break down threshold. I mean VPAN+ or VPAN- could be short-circuited through a series resistor. Like this:

However I need more accuracy, because I set the R_ISO to 12K and it is still given FAULT. Any idea ?
Greetings

•

10. Re: Bias current sensor for floating earth

By the way is there anyway to attach Ltspice files to this forum?

11. Re: Bias current sensor for floating earth

Originally Posted by flote21
By the way is there anyway to attach Ltspice files to this forum?
1) File type .asc needs to be changed to .txt

3) Advise readers your file is LTSpice simulation netlist, and they must change file type back to .asc

12. Re: Bias current sensor for floating earth

Hi again!

SCH.txt

Furthermore you cand find also the LTSpice file on WeTransfer. If you want to have look, you can find it here:

Thanks to everybody

13. Re: Bias current sensor for floating earth

Your monitor circuit is generating huge leakage currents and power dissipation and surely no acceptable IMD solution.

What is your of insulation resistance threshold specification?

- - - Updated - - -

I expect that an IMD circuit should comply with IEC 61557-8 Insulation monitoring devices for IT systems

Besides a reasonable limit for leakage currents generated by the device, it should be able to detect both asymmetrical and symmetrical earth leakage. The presented circuit is however blind against symmetrical leakage by design. An IMD can't detect all kinds of ground faults without utilizing an AC test voltage or current.

14. Re: Bias current sensor for floating earth

Hi

Thanks for your quick replies to everyone.

The insulation resistance threshold should be in the range of 20k and 80k with steps of 10k. I was thinking to use a Potentiometer....

Right I need to be in compliance with the IEC rule and UL...

Any idea to make this design better?

•

15. Re: Bias current sensor for floating earth

Originally Posted by FvM
Your monitor circuit is generating huge leakage currents and power dissipation and surely no acceptable IMD solution.

What is your of insulation resistance threshold specification?

- - - Updated - - -

I expect that an IMD circuit should comply with IEC 61557-8 Insulation monitoring devices for IT systems

Besides a reasonable limit for leakage currents generated by the device, it should be able to detect both asymmetrical and symmetrical earth leakage. The presented circuit is however blind against symmetrical leakage by design. An IMD can't detect all kinds of ground faults without utilizing an AC test voltage or current.
For the symmetrical earth leakage I will use a current sensor like a LEM or something like that. Therefore I will cover all the cases. However I am bit worry about the huge leakage currents you are talking about. Where do you see that? And what type of design or technologies you will use to reduce them?

16. Re: Bias current sensor for floating earth

The idea of an ungrounded network (IT network) is to avoid short circuit currents in case of a single ground fault while being able to detect the fault.

The 10 k resistors in your circuit cause a short circuit current as high as 150 mA in case of an asymmetrical ground fault. The measurement circuit impedance to ground is about 5k (10k || 10k). IEC 61557-8 requires at least 45k (30 ohm/V).

For the symmetrical earth leakage I will use a current sensor like a LEM or something like that. Therefore I will cover all the cases.
How? the 10 k resistors are grounding the network midpoint. If you add another pair of resistors to ground downstream to the load or in the solar panel, the circuit stays in balance and no additional ground current is generated.

A state of the art IMD is described e.g. here https://patents.google.com/patent/US9069025B2/en

17. Re: Bias current sensor for floating earth

Hi

When you are talking about symmetrical earth leakage, are you talking to have a short-circuit in VPAN+ and VPAN- at the same time? I mean something like this:

In the previous case, you are right, it is not possible to detect that issue. However the specs of my design does not describes such a problem. At the beginning, I need to detect earth leakage in VPAN+ and VPAN- individually.

I have been researching in other solution and maybe this one is better:

The previous desig works fine however it is not possible to configure a R_Leakage threshold...any idea?

Please find attached the ltspice simulation model here: (remember to change .txt to .asc)

Greetings

SCH2.txt

18. Re: Bias current sensor for floating earth

A symmetrical fault example, referring to your previous design. How would you detect it?

- - - Updated - - -

Please find attached the ltspice simulation model here
The file is different from posted schematic (incomplete). The new scheme also seems to involve new problems like needing kV rated transistors for the current limiter circuit.

However the specs of my design does not describes such a problem. At the beginning, I need to detect earth leakage in VPAN+ and VPAN- individually.
If your design is somehow related to IEC 61557-8 is needs to detect symmetrical ground faults as well. As already mentioned, it's not possible without using an AC or pulsed DC test voltage.

19. Re: Bias current sensor for floating earth

Hi!

I don't think so, I have to be in compliance with the IEC 61557-8. But I will ask to the standards department to be sure.

Please find attached again the LTspice file. I was wrong before.

https://we.tl/t-7w831vNWIP

Notice the attached files are also wrong. I donīt know how to delete them. So please download the good one from we-transfer

Greetings.

--[[ ]]--