Apocalyptic harmonic 3 phase rectification issues

Thread Starter

KevinEamon

Joined Apr 9, 2017
281
Hi guys;
Hope everyone's staying safe out there. So I'm in a bit of a pickle here.

I have this ltspice circuit. Now usually I'd have this to play with and I'd edit it, and see what comes out. The lecturer has said however as everyone cannot have access to it, with the plague, then it can't be given....Anyways... 1585934553888.png
1585934730590.png

It follows immediately with a table needed to be completed with Ton, Tdelay and Tperiod for each switch 1-6. There's other questions. But i'd like to understand this first.

Now I probably want to make this Spice circuit myself so I can have a look at it. I'm wondering why the pulse values are missing?

Also see this: 2 switches are always on. 61 | 12 | 23 ...

Now I think I'm safe saying these are the 61st harmonic frequencies and such.
But I was thinking they were related to Hz values...

I'm just a little confused tbh. I want to create this circuit anyways so I'm going to start the body of it. Any thing you can spot that might assist me in the general right direction would be mucho appreciated.
 

Thread Starter

KevinEamon

Joined Apr 9, 2017
281
1585942053709.png

Ok so I have the circuit completish. Apart from the common at the 3 phase source. At least I think it's 'common' there on the original schematic = The lil 'C'.
I've put grounds in instead maybe that won't be accurate, though I think in principal it should be.

All the pulse switching sources are set like this atm. Though I'm guessing there's something in the delays of these that are going to change things.
1585942301154.png

I'm currently getting what seems to be a full wave rectified output at the moment with a Pk V of 230

I've also found these equations. Am I on the right path here ?

1585942468942.png
 
Last edited:

Thread Starter

KevinEamon

Joined Apr 9, 2017
281
Right so I've been messing about. Now I know this is completely wrong. But I suppose it shows some intuition of the subject.
I'll be honest I'm swimming about in the dark here.

So I had to fill in this table.

From what I understand, I have to alternate the switches to achieve better harmonics.

For example SW1 would have 50% of the first part of the duty cycle. And S4 would pick up on the second half.
SW2 2 should be 60 degrees out of phase with SW1. But there doesn't seem to be a phase angle category in these pulse parameters, so I changed the delays instead.

When I ran the simulation the pulse wave came out like the Fig. 2 there. So I marked down 5ms for the period...

SwitchesTdelay (s)Ton (s)Tperiod (s)
SW1
0​
1n5ms
SW2
0.333333333​
1n5ms
SW3
0.666666667​
1n5ms
SW4
1​
1n5ms
SW5
1.333333333​
1n5ms
SW6
1.666666667​
1n5ms
1585954146713.png



I'm going to be honest here folks - I have no freaking idea what I'm doing :D

Any help would be great, even to tell me I'm an idiot and point to the bits that wrong.

Any idea how he's getting 210V Peak at the output? It's 230V in.
 
Last edited:

tsan

Joined Sep 6, 2014
116
Also see this: 2 switches are always on. 61 | 12 | 23 ...

Now I think I'm safe saying these are the 61st harmonic frequencies and such.
But I was thinking they were related to Hz values...
I think that 61, 12 and so on refers to NMOS switch numbers. First switches 6 and 1 are conducting, then switches 1 and 2 etc. Not related to harmonics.

Ok so I have the circuit completish. Apart from the common at the 3 phase source. At least I think it's 'common' there on the original schematic = The lil 'C'.
I've put grounds in instead maybe that won't be accurate, though I think in principal it should be.
Don't put ground instead of common. Now the voltage supplies are short circuited at least during the half cycle, when the NMOS body diode is conducting.

Figure 3, the output waveform? What is looks like?
Upper NMOS gate drive voltage supplies are referenced to ground, which is strange. Normally for correct operation, gate drive is supplied between gate and source, but perhaps example output has been simulated with the given circuit.
 

Thread Starter

KevinEamon

Joined Apr 9, 2017
281
Thx guys. Ok I think I understand this a little better now. My main issue is how these delays should be given to the Spice file.
Fig1. below, Is the output waveform we were given. I was thinking the Y axis represents Pk voltage - but then the source voltage is 230V and the max for the 3 phase here is 210V. The 0.7 drop in the 2 mosfets wouldn't account for that?

The rectified voltage looks like a pk, it seems to be higher on the axis than the average and RMS values in the little display box
1586030551719.png

Now I haven't reconfigured the silly circuit I posted yesterday. I think I might have been attempting to build some type of inverter, rather than a simple rectifier. Confusing myself is possibly the best explanation for what I was doing...

Anyways forget that so - 120 degrees is 2pi/3
Period is
1/50
The period I'm thinking should always be 20ms
ton should be 0.0001s

Hang on... is ton the 120°
120 degrees is 6.667ms then, that seems too big?
20ms/3 right?

Sigh...
But why make a table that has the same value for ton and tperiod for every switch?

So confused
 
Last edited:

tsan

Joined Sep 6, 2014
116
Fig1. below, Is the output waveform we were given. I was thinking the Y axis represents Pk voltage - but then the source voltage is 230V and the max for the 3 phase here is 210V. The 0.7 drop in the 2 mosfets wouldn't account for that?
The simulation result is from different circuit than shown on post #1. V(a,n), V(b,n) and V(b,n) doesn't exist on the given circuit, but I assume for example V(a,n) is the same than V(Van,c), but the voltage is set to 210V.

I have this ltspice circuit. Now usually I'd have this to play with and I'd edit it, and see what comes out.
Perhaps just change the voltage to 210 V to match the example simulation result? Or ask correct circuit and simulation result from the lecturer.

Output voltage shows normal 6-pulse diode rectifier output and mosfets are not needed. I guess the idea is, that mosfets should not conduct at all. Von is given as 1 volt, which most likely doesn't turn on the mosfet. Depending on the NMOS model, it is possible that mosfet never turns on and the mosfets behave just like diodes. Ton setting etc. doesn't matter in this case.
 

Thread Starter

KevinEamon

Joined Apr 9, 2017
281
Thx guys - those calcs will come in handy @JoeJester and I've emailed the lecturer @tsan.
I'm not sure if this will upload but I've attached what I've got in the spice file atm.
What I'm really trying to work out - is how to introduce that delay of 120 degrees into the Spice file.

I tied off that common point with a label. I'm not sure if that's the correct way to do it. But I seem to be getting a peak across the load of 400V on the 3 phase - though my average output voltage is surprisingly similar to the graph the lecturer provided.
Hmmm... I think you might be right tsan there may be something wrong with that graph we were given, anyways I've emailed him so awaiting an update on that.

Amazing guys - but could someone tell me - which of the values in post #1 is one I need to edit to introduce the 120 degrees and will the other table values remain constant? Is that right?

1586105538683.png
 

Attachments

JoeJester

Joined Apr 26, 2005
4,259
You know there are three phases .... spaced 120 degrees apart
You know how to calculate the start time of each of those phases, and their negative transistions.
You know the frequency is 50 Hz.
You know the gate has to be on for 120 degrees.

You should be able to draw the time line for each phase.
 

tsan

Joined Sep 6, 2014
116
But I seem to be getting a peak across the load of 400V on the 3 phase - though my average output voltage is surprisingly similar to the graph the lecturer provided.
Supply voltages on the example output are 230V. Green line shows 210V, but it is not on the peak. Output voltage is 400V peak also on the example. If you have a printer, you can print the output and check with pencil and ruler. Your result is correct and this is without mosfets switching. Required 120 degrees phase shift comes from the line commutation of mosfet body diodes.

Amazing guys - but could someone tell me - which of the values in post #1 is one I need to edit to introduce the 120 degrees and will the other table values remain constant? Is that right?
It follows immediately with a table needed to be completed with Ton, Tdelay and Tperiod for each switch 1-6.
About other table values remaining constant. Only Ton, Tdelay and Tperiod is asked on the table and on the screenshot on post #1 only Ton, Tdelay and Tperiod are emphasized with yellow squares.

With Von given as 1volt, (at least lower) mosfets will not turn on so IMO pulse times has no effect. Short test releaved, that current on standard LTspice NMOS model mosfet is very low also with 10 V gate drive. I'm interested to see, if pulse time settings have any effect and especially if settings can improve already correct output.
 

Thread Starter

KevinEamon

Joined Apr 9, 2017
281
Guys I have to explain something. I usually get high marks because I work hard, but tbh I'm dumb as a brush especially when it comes to very simple things. I have to ask very many stupid questions before the ahaa moment drops. Then I'm usually fine understanding the complex. I am uncomfortably aware that this really annoys people...
You both are trying to be coy and pointing with your elbow at what must seem to you to be amazingly obvious. But not to me.
What I see here is that this thing seems to be working basically on it's own.
So what the heck is the story with this table?
The only thing I've managed to work out is that I was confusing this circuit with an inverter. I think I was trying to make one, albeit very n00bishly.
I mean as far as I can see the fundamentals of KCL will control this circuits on it's own, based on potential as the phases cycle through, their transitions. So like wth is with this table?
 

tsan

Joined Sep 6, 2014
116
What I see here is that this thing seems to be working basically on it's own.
So what the heck is the story with this table?
I think this is a trick question (I hope the term is correct) and Ton, Tdelay and Tperiod does not affect to the output voltage. At least I didn't see any change on the output, when I tried constant 0V, constant 1V and so on as Von. I also replaced mosfets with diodes and output voltage was the same. There is an inherent body diode in a mosfet so mosfet works like a diode when turned off. It works on these diodes. I have worked with variable frequency drives so I recognize/know 6-pulse rectifier, which helps in this exercise.

1V Von made me think, that mosfets do nothing significant, because 1V is very low gate voltage. It is like turned off all the time (upper mosfet gate drive connection is strange, but does not affect to output voltage). I tested NMOS model with 10V gate voltage and current was still about one mA only. It seems to be that LTspice NMOS model is like that. More info here,
https://electronics.stackexchange.com/questions/424924/ltspice-simple-nmos-circuit

The only thing I've managed to work out is that I was confusing this circuit with an inverter.
It is an inverter too, but flow of energy is from DC side to AC side when working in inverter mode. Then you have to have mosfets switching. When flow of energy is from AC to DC, it can work without mosfet switching and it is like a diode rectifier. This seems to be case on this exercise. Mosfets can be switching in rectifier mode too, but the output voltage is not like with diodes and it requires inductance on the AC side. Or at least all practical circuits I have seen has an added inductance on the AC side in "switched rectifier" mode.
 

tsan

Joined Sep 6, 2014
116
If this is not a trick question, perhaps the idea is to turn on the mosfet at the same time than its body diode conducts. You have calculated Tperiod and Ton already, but the delay has to be found. One method is to start saving from zero (instead of from 10 ms) and find the delay for the first 120 degree current pulse. Switch 1 current starts from 1.6667ms on below screenshot so that is the delay for V1. I added resistor because mosfet M1 current was strange (Giga amperes).

start_of_simulation.png

Here is also V1 gate drive voltage plotted. V1 gate drive voltage is 1V at the same than the body diode conducts.

mosfet_gate_drive.png

It is possible to find out the settings for other mosfets the same way. Slightly modified simulation file is attached.
 

Attachments

schmitt trigger

Joined Jul 12, 2010
201
To gain a basic understanding of what a 3 phase full bridge rectifier operation looks like, replace all Mosfets with plain diodes.
As simple as that.

Having said this, why are you using N-chan Mosfets? In real life, its intrinsic diode will conduct naturally, even without any gate drive.

After you have completed the full diode bridge, replace again with N-chan Mosfets, but short every gate to its respective source. The operation will be identical.
 

Thread Starter

KevinEamon

Joined Apr 9, 2017
281
Thx guys @tsan my lecturer sent me back this

1586206617330.png

I'm going to get a chat with him so we'll see what happens. I' getting very slightly different measurements. From each phase to the common... I think LTspice does that though. For some weird reason...
Or it might be the difference in that Trans setting.
Anyways thanks for all the help guys. I need more info from the lecturer. So we'll see what happens.
 
Top