# High Voltage (0-600V) DC Voltage Sensing, Shared Ground with Isolated Converter and HV-

I'm designing a circuit to detect when the voltage across two DC input lines (HV+ and HV-) is greater than 60V. The voltage will vary from 0-600V. A seperate supply of 12v and 5v is available for use. The output side of the sensing circuit (and if used the 12v and 5v supply) must be galvanically isolated from the high voltage lines.

Initial Idea

My initial idea was to use a 12v zener diode voltage divider with an opto-isolator as shown in Fig 1. R2 and D2 are modelling the opto-isolator.

The value of R1 is set so that once the input voltage V1 reaches 60V, enough current (5mA) gets through to meet the forward current requirement of the opto-isolator diode. The main issue with this circuit is the large amount of power (~43W) burnt across R1 (be it a single large resistor or several smaller ones) once the input voltage reaches 600V.

Possible Solution?

I was hoping that an isolated DC/DC converter (such as https://uk.rs-online.com/web/p/isolated-dc-dc-converters/1896931/) could be used to supply an IC (such as a comparator) to reduce the current through any voltage divider used between the HV lines, as shown in Fig 2.

My question is whether the converter can share a ground with the HV line (ie HV- connected to converter ground)? And should you have time and/or are willing, suggest alternative solutions to this issue?

Thank you

• How much precision do you require, regarding this 600 V specification? (Obviously, from the schematic you show, you are permitted to have a galvanic connection between these two HV lines.)
– jonk
Mar 14, 2020 at 19:13
• The maximum voltage between the HV lines will be 600V. The only requirement for the circuit is that it must sense when the HV lines have surpassed 60V. This is simply an on/off signal, ie when V1<60V, sig_out = 0, and when V1>60V sig_out = 1 (or visa-versa). It is not an issue if this triggers too early, ie when V1=30-60V, so if the solution has low precision, the trigger voltage can be set lower so the circuit will definitely output high whilst V1>60V. You are correct in saying the HV lines can be connected through a load. Cheers Mar 14, 2020 at 20:44

if you need precision your "possible solution" is on the right track, but because isolated DC-DC converters typically have poor regulation you probably want to use a voltage reference instead of a divider for the inverting input, or use a LM431 which has an internal voltage reference.

simulate this circuit – Schematic created using CircuitLab

• Cheers for the help :) Mar 14, 2020 at 21:15
• Nice optimalisation of my answer! Mar 14, 2020 at 21:36

The isolated DC-DC you picked has an isolation rating of only 1kV. That's not enough to meet safety standards nor is it in my opinion enough to use on a 600V bus. The idea can work, and you can tie the DC-DC isolated ground to the HV-, but you probably want a converter with "reinforced" isolation of at least 5kV.

You can read a good explanation of the various types of isolation HERE.

• Thank you very much, I was unaware of the details of isolation ratings, and had naively assumed they were a "for safe constant operation" kind of thing. I will definitely read into this further. Mar 14, 2020 at 20:57

This is just an idea but may be useful.

Figure 1. NE-2 type neon lamp powered by alternating current (AC). Source: Wikipedia [https://en.wikipedia.org/wiki/Neon_lamp].

A small electric current (for a 5 mm bulb diameter NE-2 lamp, the quiescent current is about 400 µA), which may be AC or DC, is allowed through the tube, causing it to glow orange-red. ... The lamp glow discharge lights at its striking voltage.

A small electric current (for a 5 mm bulb diameter NE-2 lamp, the quiescent current is about 400 µA), which may be AC or DC, is allowed through the tube, causing it to glow orange-red. The gas is typically a Penning mixture, 99.5% neon and 0.5% argon, which has lower striking voltage than pure neon, at a pressure of 1–20 torrs (0.13–2.67 kPa). The lamp glow discharge lights at its striking voltage.

From memory these breakdown at about 60 V which would suit you.

CircuitLab won't load for me right now but the idea is to make a series RC circuit with R connected to V+ and C connected to ground. Connect the neon indicator across the capacitor and make your own opto-isolator by attaching the light sensitive receiver of your choice to the neon. It's going to flicker so you would need to address that in your software.

Will it work? I don't know.

Figure 2. Only the cathode lights up. +DC (left), -DC (center), AC (right) supplied to NE-2 type neon lamps.

CircuitLab is working again.

simulate this circuit – Schematic created using CircuitLab

Figure 3. The idea. Component values not calculated.

• Thank you very much, I will look into this. A few complications would be the relatively abusive envirnoment this circuit may find itself in (high vibration and inertial forces), and also that the whole system must be hard wired electronics without software control. However these can probably be addressed with the right mounting / damping etc. Cheers again for the help Mar 14, 2020 at 21:04

I'd use a LM431 as comparator in the HV lines.
Because my phone doesn't support the editor, below a modified schematic:

The LED represents the LED from the optocoupler. Because the HV can become 10x as big (60V to 600V) I'd choose an optocoupler with min CTR > 100% and a resistor (Rlim) that limits forward current from 1 mA @ 60V (that's what the LM431 needs as well!) and so 10 mA @ 600V.
Drawback is this resistor still dissipates 6W at 600V.

UPDATE
As Jasen points out, the cathode voltage should be limited. This can be solved by adding a resistor parellel resistor (Rp) to the LM431 and optocoupler's LED, so it makes a voltage divider with the current limiting resistor. Choosing the value of this parallel resistor equal to the current limiting resistor, limits the voltage to max 30V (because then the LM431 starts working).

The LM431 has a threshold of 2.5V, so, keeping in mind Iref, you can calculate R1B and R2B to get 2.5V when the HV becomes 60V.

If I recall correctly, LTspice also has this (or an equivalent) shunt reference in its library.

If 6W dissipation is too much, you'll need a DC/DC converter:
disconnect the optocoupler's current limiting resistor from the HV lines and feed it with the DC/DC converter.
Because this DC/DC converter is locally (between the HV lines) the converter itself does not need to be isolated.
But I think this converter will be too costly to be interesting.

• lm431 might not be happy with that setup when the input voltage is between 40V and 60V Mar 14, 2020 at 20:25
• @Jasen Good point. The LM431 from TI only allows for 37V max. Mar 14, 2020 at 21:04
• Cheers for reply, didn't know such a thing as the LM431 existed Mar 14, 2020 at 21:25