Your diagram doesn't show any part of the circuit connected to Earth.  So any connection would be stray capacitive or inductive coupling, and leakage through insulators that aren't perfect.

A typical multimeter has an input resistance of 10M ohms on volts ranges.  So every time you connect the meter, you're adding a new resistance to it, and changing the circuit you're measuring.  Each measurement is inserting that extra resistor in a different place, so the results are totally inconsistent with each other.

Hi Simon. Thank you for the reply.

Cay you please advise which part should I connect to the earth? The converter is supplied from a single phase TN-S system. I also read it is not recommended to ground the DC bus link.

Please DO NOT measure to earth with a meter - and earth nothing. !!  The usual arrangement for a 3 phase rectifier is a six diode bridge, and the DC bus negative is connected to each phase that is most negative in turn spending 1/3 of the time connecting to each. Similarly the DC positive is connecting by its diodes to whichever of the 3 phases is most positive, again connecting to each phase in turn for 1/3 of a cycle, but with the timing offset by 1/6 of a cycle from the negative side commutation.

So If you ground one side of the DC bus, you will blow 3 diodes and probably some hefty fuses as well.

If you really must measure the DC bus relative to ground, you need a scope. 

What you would see is that   the DC bus is a mixture of a DC offset, on average half the Dc bus voltage superimposed on a  wibbling up and down at 3 times the mains frequency of all the waveform crests by ~ sin 30 times the peak voltage of the cycle.

(actually the PFC >PAM confuses this somewhat but the 150Hz p-p wibble is more or less the same, the waveform is more insanitary being chopped by that circuit and the average DC is pulled over to one side - the side without the series inductor...)

If there was any sensible  'earth' reference to measure your 3 phases with respect to, it would be relative to  the bouncing centre of the DC bus. Now the voltages applied to the ends of the motor windings are not  sinewaves, rather each winding end is toggled between one side  of the DC bus and the other at a supersonic rate, with a rectangular switching pattern whose mark to space ratio ('duty cycle' ) is varied to give a similar motor  torque to a sine wave.

Maybe if you low pass filtered it you would see a component. at 50Hz or whatever the inverter rate is .

So why on earth do you expect a digital meter that has a peak measuring diode circuit, with a division factor that only gives the right RMS  when expecting sinewaves  to make any sense of this at all ?

If you must measure, you need a scope, and you need to know what to expect ;-)

On a meter you may measure the DC bus voltage between the two sides of the reservoir capacitor - that should be a DC with only a few tens of volts of ripple.

This explanation is as concise as I can make it, to fit in a lunch break, so if that means some parts are not clear, come back and myself and or others  can re-do those bits.

Mike

Your diagram doesn't show any part of the circuit connected to Earth.

Wouldn't the AC input N on the LHS normally be Earthed somewhere towards the source? So perhaps not really floating, but certainly the d.c. side will be somewhat detached from simple Earth potential by the rectification diodes.

   - Andy.

mapj1 said:

If there was any sensible  'earth' reference to measure your 3 phases with respect to, it would be relative to  the bouncing centre of the DC bus.

In the examples, I found it is usually a three-phase rectifier or inverter producing a DC voltage, so grounding the mid-point is done as shown in the Figure 2.b from Ref. 1. Fig. 9 in the same reference shows DC grid grounding systems, where DC bus midpoint point solid grounding gives the lowest common-mode voltage. 

In a three-phase supply system, none of the input phases are grounded, but in my application, I am using a single-phase TN-S system so isn't the neutral line already grounded at the substation? Is it safe to ground the DC mid-point in that case?

AJJewsbury said:

Wouldn't the AC input N on the LHS normally be Earthed somewhere towards the source?

That's my biggest concern for grounding the mid-point or anything else in the converter. I am supplying a converter from a single-phase TN-S source. In all cases I looked at, they use a three-phase system feeding a recertifier, and if I ground the mid-point there is no direct part between the input phase and the ground. But in a single-phase system, one of the input phases is grounded.

there is no direct part between the input phase and the ground.

errm, sort of but not quite. the lines are referenced to Earth (by the opposite end of the source windings) - so definitely not floating, but if you like Earthed but with an offset - i.e. Earth+240V (ac) - so Mike's comment to be careful what you Earth still stands I think.

    - Andy,

In a three-phase system between the phase and ground, there is a transformer winding impedance.

In a single-phase system the winding impedance is only between phase (L) and ground, while "phase" N is grounded. If I ground the mid-point it looks like I am shorting the input to the mid-point:

I would like to ground the DC link mid-point, and as I understand, for a three-phase input this is usually done. But should I do the same in a single-phase system?

I think that you are suffering from the delusion that Earth is something useful once you move into a circuit like this. It is not, simply a meaningless concept of no value. You ask yourself "why are you measuring this?". A better question would be "why am I measuring anything"? Because of the way that inverter drives work, using PWM and not AC voltages as such at all, you will measure a lot of very strange numbers, many of them not related to very much. As Mike says, if you want to investigate the circuit you will need to use an oscilloscope, and doing so safely really requires a pair of high voltage probes used differentialy. If you want to understand what happens if there is an Earth fault somewhere on the motor, the result will be a high current through quite a lot of parts of the circuit, and these will either be electronically limited by controlled self preservation of the electronics or trip the circuit protection.

On  a single phase system with a bridge, alternate sides of the smoothing capacitor are connected to mains L and N as the cycle reverses. Again, neither side is grounded, and the mid point is not safe to ground either.

Now, to make it clear how to visualise it, if  instead of L and N you had two lines in perfect anti-phase ( a split phase with a +L and a -L if you like) then the mid point of that would match the mid point of the DC rail. So shifting our reference frame by grounding the neutral then the mid point of the DC bus becomes centred about the centre of the mains, so about 120V RMS or so wrt neutral, and of  course the useful ends of the DC bus see the same AC plus a DC offset.

But neutral is not earthed, rather it is offset by a small load dependent voltage, that is the drop in the neutral wiring, much as the live is rather less at the load than at the origin so nothing is properly fixed relative to ground at all.

Note, re your comments on 3 phase, that if you take the mid point of two capacitors off a six diode bridge and force it to neutral you do not have the same circuit as I described. Instead of 300 Hz ripple and a bouncing centre you now have two completely independent single phase rectifiers, each with 3 diodes feeding one cap and 150Hz ripple. The ripples on the two caps are offset by 1/6 cycle so the effect is not dissimilar, but there is a 300Hz current flowing in the neutral and the DC bus voltage is not quite the same. Both flavours with and without neutral do exist in the wild but are not quite equivalent.

Mike