phase angle convention

Suppose we have a 3-phase, Y-connected power system. Call the phases
A, B, and C. Assume A peaks at T=0, B peaks next (120 degrees, 5.6
msec later) and C peaks last. Call the phase angle of A zero.

So, what's the convention for the sign of the phase of B? Is is +120
degrees, or -120?

Similarly, if the current in an inductor lags the applied voltage, is
the current +90 or -90 degrees relative to the voltage?

I used to know this stuff.

John

 

Does it matter what you call it ? Or are you having to write a
specification ?

...Jim Thompson

 

I'll most likely be flamed for this.
it's W,X,Y,Z W = Ground X,Y and Z = phase 1,2 and 3
Phase 2 (B) would be +120 since it don't take place until
the time displacement of 120 degree's later.

As far as the inductor. It's know to be -90 only because the
effects that take place were due to something that happen 90 degrees
before..

Ok, Guys. flame my ass [email protected]

 

I'm (still) doing the firmware for a dds-based waveform generator. If
the user specifies a phase angle for a waveform, I'm adding a value to
the memory address pointer that aims into the 2048-point sinewave
lookup table. So if he says to set channel 3 to +45 degrees, I'm
adding 256 to the ch 3 address pointer. That makes the peak of the
resulting sine wave come out 1/8 of a cycle earlier in time.

I like that, but is it what electrical guys expect? By my convention,
to generate a textbook 3-phase set on channels 0, 1, and 2, I'd have
to send the serial commands

0Phase 0
1Phase -120
2Phase -240

which might be weird to power folk.

I did take two semisters of Electrical Machinery, but it's been a
while. Class average on exams was, as I recall, about 15. I ran in the
mid-40's, which was an A.


John

 

Seems to me that the user should legitimately be able to enter either
+45 or -315 and get the same results:

PhaseAngle = GetUserInput();
offset = (PhaseAngle/360) * 2048;
if(offset < 0) offset = offset + 2048;

Or something like that...

 

Certainly; everything is mod 360, and -270 is the same as +90. But I
still want to get the sign right.

Actually, I get the user input (say, 22.5) as a signed 32-bit integer
that's scaled in millidegrees (22500), multiply by 1024, divide by
5625, and the low 16 bits becomes the phase rotator integer which is
poked into the fpga, the high 11 bits of which are the actual memory
address adder. All in assembly. The scaling takes two instructions on
the 68332. The "rational" math avoids rounding errors which you'd
probably get with floats.

John

 

I would say phase B lags phase A, so the sign of the phase B should
be negative, WRT A.

Again '-' for the same reason.

Maybe I did too. ;-)

 

Check out a HP/Agilent box as the electronics industry tends to follow
them for their conventions. A HP8904 Multifunction Synthesizer would be
a good place to start.

Andrew

 

For Va to peak at t = 0 ( w = 21.6E3 degrees per sec):

va = V * cos( wt )

If Vb peaks 5.6 ms later:

vb = V * cos( wt - 120 )

wt - 120 = ( 21.6E3 * 5.6E-3 ) - 120 = 0

-90 degrees. Using the same logic as above, if the voltage peaks at t =
0, then
v = Vp * cos( wt )

If the current peaks 4.167 msec later (90 degrees at 60 Hertz), then

i = ( Vp / Zl ) * cos( wt - 90 )

 

Shouldn't be. Power folks also consider 360 degrees full circle. Now
when they introduced a 400 degree convention I bailed ;-)

 

Congratulations, you're a physicist.

The sign of the phase depends on your Fourier transform sign convention,
and physicists and engineers use opposite ones. We'd all like to have a
positive frequency be exp(i omega t) and exp(i k dot x). We also all
like to consider a general plane wave as going in the positive x
direction.

Unfortunately you can't have all those things at once--a positive-going
wave is either (1) exp(i(k dot x - omega t)) or (2) exp(j(omega t - k
dot x)). Historically, EEs have cared more about temporal behaviour, so
they choose convention (2) and physicists more about spatial behaviour,
so they choose convention (1).

Applying a time delay means that the measured phase corresponds to an
*earlier* time, i.e. t' = t-tau. The phase is just the imaginary part
of the exponential, so if (as an EE would say) a positive frequency is
exp(j omega t), the phase is increasing with time. Thus that phase lag
gives a *negative* phase shift, corresponding to an earlier time. So
the AC leg that comes to a peak 1/3 cycle later is -120 degrees.

(And yes, this also applies to phase lags that don't introduce delay,
e.g. RC integrators.)

Cheers,

Phil Hobbs

 

Don't worry. That convention is now gon.

 

I still remember many EE exams where my brain went like this: I know
it's wrong but I've got to do it anyhow so that I pass. Like impedance
matching of a large transmitter final amp which, if it was ever done,
would result in flying glass and a 5-alarm blaze.

 

John Larkin posted to
sci.electronics.design:

I am not really a power type. But as i understand the conventions
phases a, b, and c in your definition are called 0, 120, and 240
degrees. As stated, these will represent lag values. For normal
power engineers just use 3-phase, with phase rotation a, b, c which
will indicate the order of positive peaks. For other persons you may
want a different interface (read modal interface). Some would call
this a suboptimal solution. My response to them is show me a better
one.

 

John Larkin posted to
sci.electronics.design:

I presume that you are familiar with with binary angle measurement
system (BAMS) where the (fractional) msb represents 180 degrees. It
makes a whole lot of angle and transcendental function computation
much easier. Establishing the equivalence of -90 and +270 is
trivial in this mode.

 

Joerg posted to
sci.electronics.design:

It should not have had that effect in a correctly designed unit.
Quite the opposite in fact.

 

That's exactly what my final 16-bit binary is.... 0 to 65535 == 0 to
359.99 degrees. The msb is 180.

My users program the thing in serial ascii, in degrees, so I convert.

John

 

I'm getting the general impression that a positive angle represents
lag to the power people. But they rarely use signed angles... they say
"45 degrees lagging (or leading)." That makes math funny, but maybe
they don't do much math.

John

 

John Larkin posted to
sci.electronics.design:

Thanks, i thought i understood the subcontext in the thread.

 

John Larkin posted to
sci.electronics.design:

Unfortunately true, they would rather use tables.