High voltage voltage divider minimum current?

Hello

I need measure high voltate , for example 1kV - 8kV ,
if use 50M compensated high voltage resistor divider will be on

8kV current over divider 160uA and Power disipation 1,3W
1kV current over divider 20uA and Power disipation 20mW

The question is in which current the resistive divider works OK and at what already leaked currents cause an error, which also prevents the indicative measurement (error is over 5 or 10%)

Hi,

I don't think in uA when I calculate such dividers.
I think in Ohms.

Example:
Your 50M divider for 8kV (10kV including headroom) down to 5V.
--> then the lower resistance is about: 50M x 5V / 10kV = 25k.

This is about the source impedance of the now divided signal.
For my taste it is too high impedance to drive an (SAR) ADC directely.
I'd use an amplifier between divider and ADC.

About current.
For sure a current of 10uA may cause a lot of error...on the other side: to generate a current of 10uA into a 50Mohms resistor needs 500 Volts..?.where do they come from?
You may say this current may be introduced in the 25kOhms....but then you are on the low_voltage side of the divider and you need a rather low ohmic error resistor to generate this current.

But if you want to improve it, then use an inverting Opamp circuit, 25k as feedback resistor .... and place a GND_guard around your inverting input node. --> Zero voltage prevents current from flowing...

Klaus

where do they come from?

Click to expand...

I need about a high voltage source for testing, actually, an arc generator in the gas.
It would probably be enough to have two or tree levels of High voltage, , but why not make your life uncomplicated
I think about flyback converter with measuring output voltage and feedback, with setting of output voltage.

My idea
Primary side 24V and MCU controled MOSFET.
Secondary side - voltage divider for measure output voltage, secondary HV side full isolated i.e. voltage to primary by linear opticali isolated ampliier

Regulation loop full digital ADC measure output voltage ove oticaly isolated amplifier and set duty on PWM output driving MOSFET. Safety regulation measure over current transformer current over MOSFET .
Max output power 10W i.e 1,25mA on 8kV.

I have calculated the transformer and it seems that it can be realized, but it would want an ETD core with 6 to 8 sections. As they have HV transformers for CCFL.

My first resistor divider was 5 x 4,7M || 4x 10nF C. I was a little bit surprised at that power dissipation 2.7W on 8kV :???:

Hi,

What you are doing is highly dangerous!

I don´t like the idea of generating high voltage without safely limiting current to a non dangerous value.
You say digital regulation. You need to ensure that a software or hardware fail will lead to switch OFF output voltage.

My first resistor divider was 5 x 4,7M || 4x 10nF C

Click to expand...

What does this mean?
Are the resistors for voltage measurement?
Is each capacitor in parallel with one resistor?
If both YES, than I don´t think this is a good idea, because the capacitors influence your measurement.

Klaus

What you are doing is highly dangerous!

Click to expand...

It depends on the angle of view.
1. Pn the one hand I measure the current at the primary and this is a function of the secondary current. The truth behind flyback inverters is indirect.
2. Voltage is not dangerous, Dangerous is a current over the heart greater than 30mA. It is forbidden that the 10W max power supply gives a 30mA current on high voltage level.
Such a car ignition has 20kV no current limitation and can not kill you. Like an electric fence. there is a limit only to the maximum energy in the puls, for example 10J.

Look at the madmen with tesla transformers, higher voltages and higher powers. :smile:

Hi,

1. Pn the one hand I measure the current at the primary and this is a function of the secondary current.

Click to expand...

I don't think this is true for flyback. I rather think the priary average current is about proportional to output power.
This means if you have 1/10 of the output voltage you get ten times the current.... like 800V, 12.5mA..

I know about voltage and currents.

Klaus

Referring to the initial question, I would go for 100 MOhm up to GOhm voltage divider.

Safety level of high voltage power source.

Well, let's take a closer look at the issue of safety

On picture is an internationally recognized graph of current danger through the human body
Zone AC-1 nad AC-2 (all under green line) is is a safe area without painful physiological effects
Zone AC-3 is safe areea with with unpleasant, but non-health-enhancing effects, you may get convulsions, but it will not kill you
Zone AC-4 (all over red curve c1 ) a dangerous current which can cause the fibroilation of the heart and the victim to kill

In other words
Current 30mA is in all time under red cuve.
Current 100mA is safe in time shorter than 500ms
Current 500mA is safe in time shorter than 10ms
Current 200mA is safe in time 100ms with reserve.

OK if have power supply with max output power 10W
everything fails, yet we are sure that max output current 200mA is possible max on voltage 50V, becouse P=U*I
50V is safe DC and AC voltage in all time and all environment (becouse human body have any resistence)!
I suspect that a 10W voltage source can not kill humans, under no circumstances, it prevents the resistance of the human body.

Let's take a look at real HW implementation of power supply.
For example when use as control IC MCU STM32F303.

Timer in MCU as PWM control MOSFET on primar 24V side timing for example 60% charging flyback transformem 20% dischrge and 20% as dead time.
Timer in MCU can be stoped by using internal hw comparators, this works on hardware level. If on internal comparator bring the voltage from the current measurement over the mosfet.
We have, in my view, a functional limitation of maximum power relized on hardware level.
Information about output voltage level is on MCU avalible over opticaly isolated amplifer. STM32F303 have two watchdog Windows and independen, I thing WWDG is good choice, becouse use same clocs as PWM times. If we set this timer on example 10ms we have hardware sure, MCU and PWM will be stoped when every 10ms, not refresh WWDG , ADC not measure output voltage and output voltage in time 10ms after start PWN not over 500V.
If all work we will measure every 10ms output voltage and if not over 500V (any error or big load on output make savety STOP. If sw not works make safety stop WWDG.

10W power supply on 500V can delivery max 20mA and 20mA is safety current level on every time and every voltage level.
In other safety level is possible relize anu over current protection on charging capacitor in primari side.
What do you think about it?

Referring to the initial question, I would go for 100 MOhm up to GOhm voltage divider.

Click to expand...

1G divider on 500V have 500nA current. they are so small currents that they also affect the change of air humidity or dust on PCBs.

The essence of my question was whether these high ohmic dividers are not more of a weather detector, device age and operator mood than an exact voltage meter

Yes, and my answer is that the high ohmic resistors can be used with proper instrument design. Obviously we need to talk about design details.

In good old days, voltmeters were rated by the current loads, for example we had voltmeters rated at 10k/V- that simply means that the meter takes 100uA to produce a deflection of 1V.

The nomenclature is no more because modern digital meters offer input impedance about 1000 times higher. But the problem comes with high voltages.

High voltage power supplies cannot supply high currents- particularly electrostatic voltage generators, like Wimshurst machine or van de Graaf machine (they can happily produce a few microamps but that is too little for accurate measurements).

Tesla coils are another class of high voltage and high frequency machines that produce voltages that are rather tricky to measure accurately.

I strongly suggest you use a 1uA resistance divider (high value resistors are tough to get but that is a different story- PCB layout must be guarded) and use an electrometer input op-amp for the final measurement.

It used to be challenging.

In these applications, construction techniques mean everything.

Back in the late 1980s, I once saw a disassembled high voltage meter. The resistor divider board was encapsulated in potting, to prevent both corona discharge and board leakage effects. I would assume that prior to encapsulating the board, it had been thoroughly cleaned to remove any ionic contamination.

And the divider output had a Teflon-insulated wire which was fed to an opamp whose input had a guard ring.
The opamp used was one of the first ones that had nanoampere input currents: the CA3140.

The resistive value of the top branch was too high for me to measure. But I could measure the bottom branch and with the known divider ratio I could calculate its value, if I remember properly it was 1 Gohm.

carpenter said:

Voltage is not dangerous, Dangerous is a current over the heart greater than 30mA...

Click to expand...

Your point is fundamentally valid but the numbers are wrong I believe. Arteries and veins are full of blood that is rather highly conducting. Even 1mA is sufficient for a normal human adult to get a fatal shock if the current passes through the heart.

If the current does not go through the heart, you have a great escape! If the skin is wet, the chances are very slim. Dry skin resistance is very high and offers considerable safety protection.

Car ignition voltage and current can kill if applied with a correct period (2-3 pulses with 0.5 to 1.5s gaps)- that can match the heart beat and will stop the poor heart from beating ever again!

carpenter said:

On picture is an internationally recognized graph of current danger through the human body

Click to expand...

Can you please provide a reference to the picture you gave?

I do not think people have done actual experiments on that and the results are based on models. There are some interesting aspects in the graph and I would like to learn more about it.

The picture empirically and statistically expresses the influence of electric current on humans body.
It is part of the International Standard IEC 60479 Effects of current on human beings and livestock
there are several versions, and the images are slightly different in them.

Here is one version, in better picture quality

Thank you for the details; one relevant paper is in Chinese and I cannot read it- Weineng Wang, Zhiqiang Wang, Xiao Peng, Effects of the Earth Current Frequency and Distortion on Residual Current Devices, Scientific Journal of Control Engineering, Dec 2013, Vol 3 Issue 6 pp 417-422- the abstract is in English and clear:

"Abstract: Residual current devices characterized by simplicity and cost effectiveness holds a wide range of applications in the protection against electric shock that is one of the important detriments to the safety of human and the performances of electrical apparatus. This article puts the focus on the description of the basic structure and principles of residual current devices, particularly the effect of the earth current frequency and distortion on the mal- and miss operation of electrical apparatus is investigated, which is of reference value for the proper selection of electric apparatus.
Keywords: Residual Current Devices; Electric Shock Protection; Earth Current; Miss Operation; Mal Operation"

IEC 60479 costs CHF 450 (what is there that common man should not know?)

And the abstract is really bland:

"IEC 60479-1:2005+A1:2016(E) For a given current path through the human body, the danger to persons depends mainly on the magnitude and duration of the current flow. However, the time/current zones specified in this publication are, in many cases, not directly applicable in practice for designing measures of protection against electrical shock. The necessary criterion is the admissible limit of touch voltage (i.e. the product of the current through the body called touch current and the body impedance) as a function of time. The relationship between current and voltage is not linear because the impedance of the human body varies with the touch voltage, and data on this relationship is therefore required. The different parts of the humanbody (such as the skin, blood, muscles, other tissues and joints) present to the electric current a certain impedance composed of resistive and capacitive components. The values of body impedance depend on a number of factors and, in particular, on current path, on touch voltage, duration of current flow, frequency, degree of moisture of the skin, surface area of contact, pressure exerted and temperature. The impedance values indicated in this technical specification result from a close examination of the experimental results available from measurements carried out principally on corpses and on some living persons. This technical specification has the status of a basic safety publication in accordance with IEC Guide 104. The contents of the corrigenda of October 2006 and June 2013 have been included in this copy. This consolidated version consists of the fourth edition (2005) and its amendment 1 (2016). Therefore, no need to order amendment in addition to this publication."

I was simply curious how and from where they get these data for these graphs?

Hi,

the given graph in post#8 is for 50Hz/60Hz AC.

Usually DC is more relaxed. The only worse (I know) with DC is, that it increases decompensation of water into oxygen and hydrogen. Gas production.

I did some kV measurements. I always kept current below 0.5mA - even if one device fails.
* for 2500V I used 3 pieces of 2.5MOhms resistors in a string (each rated with 2.5kV). So if one of these 3 resistors fail and create a short circuit, then there are 2 others giving a resistance of 5MOhms to keep the current below 0.5mA.
For sure one can say this is too much safety.

Klaus