Is this transformer normal?

[Chris Carlen]

Greetings:

I bought a 500VA 115V 50/60Hz Hammond Isolation transformer.

When I first plugged it in, it made a chattering sound, so I thought
maybe the circuit breaker was acting screwy.

Then I powered it from a variac and pushed towards 140V, which made the
chattering happen again. Disassembling the cover and then powering it
revealed that the chatter was caused by the magnetic field making the
cover buzz against the core.

Next I began to wonder why the field appeared to grow so strongly toward
voltages above 120V, so I made some plots. The plots show that the
magnetizing current grows faster than a linear relationship with
voltage, and that the inductance vs. current is downright bizarre.

You can see the plots here:

http://home.earthlink.net/~crobc/misc/images/hammond-500va.png

Is it normal for an iron line frequency power transformer to be
operating at what appears to be very close to saturation?

How on earth could the inductance have a maximum?

Thanks for comments.

Good day!

 

[Michael A. Terrell]

Greetings:

I bought a 500VA 115V 50/60Hz Hammond Isolation transformer.

When I first plugged it in, it made a chattering sound, so I thought
maybe the circuit breaker was acting screwy.

Then I powered it from a variac and pushed towards 140V, which made the
chattering happen again. Disassembling the cover and then powering it
revealed that the chatter was caused by the magnetic field making the
cover buzz against the core.

Next I began to wonder why the field appeared to grow so strongly toward
voltages above 120V, so I made some plots. The plots show that the
magnetizing current grows faster than a linear relationship with
voltage, and that the inductance vs. current is downright bizarre.

You can see the plots here:

http://home.earthlink.net/~crobc/misc/images/hammond-500va.png

Is it normal for an iron line frequency power transformer to be
operating at what appears to be very close to saturation?

How on earth could the inductance have a maximum?

Thanks for comments.

Good day!

How much headroom do you expect on a 115 volt transformer before it
saturates? You are running it at 140/115 volts, or at 123.478% of the
manufacture's ratings. You can buy transformers with more headroom, but
they are more expensive, and a lot heavier.

 

[John Woodgate]

I read in sci.electronics.design that Chris Carlen <[email protected]

Is it normal for an iron line frequency power transformer to be
operating at what appears to be very close to saturation?

Yes, but the core assembly of your unit may be poor.

How on earth could the inductance have a maximum?

You measured the parallel inductance and resistance? The better
equivalent circuit has both series and parallel resistance. If you use
that model, you don't see an inductance maximum, IIRC.

 

[Tony Williams]

......... The plots show that the magnetizing current grows
faster than a linear relationship with voltage, and that the
inductance vs. current is downright bizarre.

Hysteresis loss is not linear with Bmax. Look up the
law of Steinmetz, which basically says;

1.6
Loss due to hysteresis, Wh = N * (Bmax)

This is why the input current rises faster than the voltage.

Is it normal for an iron line frequency power transformer to be
operating at what appears to be very close to saturation?

Voltage transformers have to be designed for least
variation in output voltage with load, and this means
minimum winding resistances and minimum leakage inductance.
The practical outcome of that is to design for operation at
the highest possible Bmax.

If you push Bmax to the limit you do get a transformer that
is just going into saturation on no-load. But when loaded,
the primary IR voltage drop reduces the effective primary
voltage, and Bmax drops to below Bsat.

There's an expt you can do..... Compare the temperature of
the laminations after a few hours of running at no-load
with the temperature after a few hours at 500W load. The
laminations will run cooler at full-load (and the noise
should be less).

 

[Harry Dellamano]

Hysteresis loss is not linear with Bmax. Look up the
law of Steinmetz, which basically says;

1.6
Loss due to hysteresis, Wh = N * (Bmax)

This is why the input current rises faster than the voltage.


Voltage transformers have to be designed for least
variation in output voltage with load, and this means
minimum winding resistances and minimum leakage inductance.
The practical outcome of that is to design for operation at
the highest possible Bmax.

If you push Bmax to the limit you do get a transformer that
is just going into saturation on no-load. But when loaded,
the primary IR voltage drop reduces the effective primary
voltage, and Bmax drops to below Bsat.

There's an expt you can do..... Compare the temperature of
the laminations after a few hours of running at no-load
with the temperature after a few hours at 500W load. The
laminations will run cooler at full-load (and the noise
should be less).

Hey Tony, great answer. Got to remember that.
thanks
harry

 

[Chris Carlen]

I read in sci.electronics.design that Chris Carlen <[email protected]
thlink.net> wrote (in <[email protected]>) about 'Is this
transformer normal?', on Wed, 12 Nov 2003:




Yes, but the core assembly of your unit may be poor.



You measured the parallel inductance and resistance? The better
equivalent circuit has both series and parallel resistance. If you use
that model, you don't see an inductance maximum, IIRC.

Since the resistance was so low, about 0.4ohms or about 1000 times less
than the inductive reactance, I ignored it. I can't imagine that the AC
resistance is any higher at 60Hz.


--
____________________________________
Christopher R. Carlen
Principal Laser/Optical Technologist
Sandia National Laboratories CA USA
[email protected]

 

[Chris Carlen]

Hysteresis loss is not linear with Bmax. Look up the
law of Steinmetz, which basically says;

1.6
Loss due to hysteresis, Wh = N * (Bmax)

This is why the input current rises faster than the voltage.

Yes, this would account for an upward curving I vs. V graph. I
understood the graph from the perspective of saturation causing loss of
inductance, and therefore decreasing inductive reactance at higher
current. Both mechanisms result in the observed shape of the curve.

But the maximum in the inductance is not explained.

Voltage transformers have to be designed for least
variation in output voltage with load, and this means
minimum winding resistances and minimum leakage inductance.
The practical outcome of that is to design for operation at
the highest possible Bmax.

If you push Bmax to the limit you do get a transformer that
is just going into saturation on no-load. But when loaded,
the primary IR voltage drop reduces the effective primary
voltage, and Bmax drops to below Bsat.

There's an expt you can do..... Compare the temperature of
the laminations after a few hours of running at no-load
with the temperature after a few hours at 500W load. The
laminations will run cooler at full-load (and the noise
should be less).

Actually, my experience is that transformers run very cool until heavily
loaded, when they get hot. If the winding I^2R loss dominates, then
this would be the case, and it would be impossible to differentiate the
laminations heating from the coil heating because they are in such close
contact and there is such a long thermal equilibrium time required.


Thanks for the reply, Tony!

Good day!


--
____________________________________
Christopher R. Carlen
Principal Laser/Optical Technologist
Sandia National Laboratories CA USA
[email protected]

 

[John Larkin]

Greetings:

I bought a 500VA 115V 50/60Hz Hammond Isolation transformer.

When I first plugged it in, it made a chattering sound, so I thought
maybe the circuit breaker was acting screwy.

Then I powered it from a variac and pushed towards 140V, which made the
chattering happen again. Disassembling the cover and then powering it
revealed that the chatter was caused by the magnetic field making the
cover buzz against the core.

Next I began to wonder why the field appeared to grow so strongly toward
voltages above 120V, so I made some plots. The plots show that the
magnetizing current grows faster than a linear relationship with
voltage, and that the inductance vs. current is downright bizarre.

You can see the plots here:

http://home.earthlink.net/~crobc/misc/images/hammond-500va.png

Is it normal for an iron line frequency power transformer to be
operating at what appears to be very close to saturation?

How on earth could the inductance have a maximum?

Thanks for comments.

Good day!

Sounds pretty normal. The magnetizing current at 140v is about twice
that at 120, which isn't unreasonable.

Transformer-type silicon-steel has its maximum permeability at some
non-zero excitation, which is why you can expect an inductance peak at
some moderate terminal voltage; later on, saturation pulls L back down
again. I see this in current transformers, where the phase shift is
highest at low currents. Really good stuff, like permalloy, is much
better.

The buzz is just poor mechanical design.

John

 

[Walter Harley]

I bought a 500VA 115V 50/60Hz Hammond Isolation transformer.

When I first plugged it in, it made a chattering sound, so I thought
maybe the circuit breaker was acting screwy.

Then I powered it from a variac and pushed towards 140V, which made the
chattering happen again. Disassembling the cover and then powering it
revealed that the chatter was caused by the magnetic field making the
cover buzz against the core.

FWIW, I encountered this recently with the power transformer in an audio amp
I was fixing. Asked around a bit and got a "don't worry about it." So I
didn't. The transformer appears to be working fine - not getting hot, in
normal operation, despite what appeared to be a high current draw and
noticeable buzzing. The amp is now back in regular service and the customer
has not mentioned any change in its behavior (other than the fact that it
now works), so I presume it's just the way that xfrm is.

-w

 

[John Woodgate]

I read in sci.electronics.design that Chris Carlen

Since the resistance was so low, about 0.4ohms or about 1000 times less
than the inductive reactance, I ignored it. I can't imagine that the AC
resistance is any higher at 60Hz.

So you measured the *series* resistance only. But part of the resistive
loss is hysteresis loss in the core, and that is best modelled as a
parallel resistance. It almost certainly isn't negligible at 140 V.

 

[John Woodgate]

I read in sci.electronics.design that Chris Carlen

Actually, my experience is that transformers run very cool until heavily
loaded, when they get hot. If the winding I^2R loss dominates, then
this would be the case, and it would be impossible to differentiate the
laminations heating from the coil heating because they are in such close
contact and there is such a long thermal equilibrium time required.

I'll get jumped on again by someone who wants to complicate the issue,
but if the transformer is well designed, the copper and iron
(hysteresis) losses will be approximately equal at full load. At no
load, there is no copper loss, but increased iron loss. Which effect
wins depends on the individual design, and the input voltage.

 

[John Larkin]

I read in sci.electronics.design that Chris Carlen


I'll get jumped on again by someone who wants to complicate the issue,
but if the transformer is well designed, the copper and iron
(hysteresis) losses will be approximately equal at full load. At no
load, there is no copper loss, but increased iron loss. Which effect
wins depends on the individual design, and the input voltage.

<Jump>

The iron losses are maximum at no load, and go down a bit as the load
increases, due to an effective reduction in primary voltage caused by
ohmic drops in the primary. Loading a transformer *decreases* the core
flux density. So unless the transformer is designed to run fairly hot
at no load (and they seldom are) full-load copper losses will be much
bigger than iron loss.

<\Jump>

John

 

[Spehro Pefhany]

I read in sci.electronics.design that Chris Carlen


I'll get jumped on again by someone who wants to complicate the issue,
but if the transformer is well designed, the copper and iron
(hysteresis) losses will be approximately equal at full load. At no
load, there is no copper loss, but increased iron loss. Which effect
wins depends on the individual design, and the input voltage.

My experience with small E-I power transformers made with CRGO Si
steel is that the core losses should be much less than the copper
losses.

Best regards,
Spehro Pefhany

 

[John Woodgate]

I read in sci.electronics.design that Spehro Pefhany <speffSNIP@interlog

My experience with small E-I power transformers made with CRGO Si steel
is that the core losses should be much less than the copper losses.

How small? There are numerous reasons why a particular design deviates
from maximum efficiency for over-riding reasons. For transformers using
scrapless EI laminations, the winding area is too small unless you run
at an induction at normal supply voltage that causes excessive losses at
10% high supply voltage. Semi-scrapless laminations (where you get 2 Es
and 4 Is from each stamp) have double the window area and don't have
that problem, but the resulting transformer is larger.

 

[Tony Williams]

But the maximum in the inductance is not explained.

Magnetic materials manufacturers publish graphs for their
various sheet materials. A typical graph is of the gross
permeability versus +/- Bmax, at various frequencies.

This graph is useful in predicting the magnetising current
that would be drawn at various primary voltages and, to a
certain extent, could predict the primary inductance at
those voltages...... although I am always wary about using
the word 'inductance' in the context of a ferromagnetic
core swinging around a large B-H loop.

I have a few such graphs here, and they do show that the
gross permeability does indeed have quite a large peak.

For example, for 0.014in 4% silicon steel, the graph shows
that gross permeability peaks (at about 4000) when operating
around a B-H loop of about +/- 5000 gauss Bmax, and by +/-
10000 gauss it has dropped to about 2100.

The URL you gave crashes my browser, so I haven't seen your
figures for 'inductance' (or how you measured/calculated it),
but if your 115V transformer used a similar material then you
would see an apparent peak in the inductance somewhere in the
region of about 50V.

Actually, my experience is that transformers run very cool until
heavily loaded, when they get hot. If the winding I^2R loss
dominates, then this would be the case, and it would be
impossible to differentiate the laminations heating from the
coil heating because they are in such close contact and there is
such a long thermal equilibrium time required.

I have measured a 1KVA 230:115 isolating transformer (many
years ago), and the surface temperature of the laminations
did go down with loading.

 

[Spehro Pefhany]

I read in sci.electronics.design that Spehro Pefhany <speffSNIP@interlog


How small?

In the 3-6VA range (rated). Used in instrumentation. Keeping heat loss
down was important to me because it affected instrument accuracy for a
given tooling cost. The actual power consumption of the whole
instrument measured with a wattmeter at the supply terminals was more
like 1-1.5W.

There are numerous reasons why a particular design deviates
from maximum efficiency for over-riding reasons. For transformers using
scrapless EI laminations, the winding area is too small unless you run
at an induction at normal supply voltage that causes excessive losses at
10% high supply voltage. Semi-scrapless laminations (where you get 2 Es
and 4 Is from each stamp) have double the window area and don't have
that problem, but the resulting transformer is larger.

I think with the premium core materials you just can't get that level
of core losses and still have a transformer that works over supply
range and so on, but I didn't design these transformers myself, just
specified the core material. It's another data point. I don't doubt
that your statement is valid given some constraints..

Best regards,
Spehro Pefhany

 

[John Woodgate]

I read in sci.electronics.design that Spehro Pefhany <speffSNIP@interlog

I think with the premium core materials you just can't get that level of
core losses and still have a transformer that works over supply range
and so on,

It depends what you mean by 'premium'. If the material has too 'square'
a hysteresis loop (i.e. it goes 'bang' into saturation with no 'top
bend'), then indeed overvoltage spells real trouble. But I wouldn't use
such a material for a mains power transformer, for precisely that
reason. If I really *had* to, I'd put in a minimal air-gap to introduce
some top bend.

but I didn't design these transformers myself, just specified
the core material.

I'm surprised that a transformer designer would accept such a pre-
emptive specification.

These very small (3 to 6 VA) transformers normally have semi-scrapless
laminations, because insulation thickness doesn't scale with core size,
so it takes up more room in proportion in the window areas of small
cores.

It's another data point. I don't doubt that your
statement is valid given some constraints..

I'm not sure what you mean by that. I'm concerned with what theory and
practical lore works for the sort of laminated and toroidal power
transformers for electronic equipment, mostly below 1 kVA, that are used
by the millions. I'm not delving into hydrogen-cooled or superconducting
designs!

 

[Spehro Pefhany]

I read in sci.electronics.design that Spehro Pefhany <speffSNIP@interlog


It depends what you mean by 'premium'. If the material has too 'square'
a hysteresis loop (i.e. it goes 'bang' into saturation with no 'top
bend'), then indeed overvoltage spells real trouble. But I wouldn't use
such a material for a mains power transformer, for precisely that
reason. If I really *had* to, I'd put in a minimal air-gap to introduce
some top bend.

It's got much less core loss per lb at a given flux, and it costs a
fair bit more, so I call CRGO "premium". How do you put an air gap in
an EI transformer? They just buy the lams from some specialty steel
supplier or another, I don't think the xfmr assembly houses have much
control over the detailed design of the cores.

I'm surprised that a transformer designer would accept such a pre-
emptive specification.

If I told them to wind it with cotton-insulated wire and insulate it
with hockey tape, they'd have done that too. Can you guess the area of
the world we're talking about? (not that they'd have much use for
hockey tape). "Designer" may be too strong a word. We got several
thousand transformers from another company that were supposed to have
a wound screen (cheaper and more ungood than copper tape) between
primary and secondary and found that they had not understood the
instructions and simply taken the wire and terminated it under the
tape. Not only that, but being a (theoretically) disconnected wire
they didn't bother hipotting it or even testing it for isolation. It's
no fun when the 230V gets connected to earth (the metal housing),
which happened on 0.005% of the units Didn't deal with that factory
again. The part they did right was better (more reliable) than the
Japanese were getting, though.

These very small (3 to 6 VA) transformers normally have semi-scrapless
laminations, because insulation thickness doesn't scale with core size,
so it takes up more room in proportion in the window areas of small
cores.

I don't know or don't recall the details; they had standard EIxxx core
size designations. We made custom metal stampings or injection
moldings to cup the core so the size (all 3 dimensions) was important
to us. They could choose the lamination size and how many to stack up
(and those only within fixed limits if they wanted to use standard
stampings to house the transformer.

If the cost and weight of the transformer was paramount over
performance (low heating and decent regulation), we perhaps could have
reduced the size further, but I don't see how the copper losses could
have been made as low as the CRGO, those lams ran dead cold after
hours of operation, compared to the cheap ones that ran warm to the
touch .. or rather worse if one removed some lams, but that's another
story.

Best regards,
Spehro Pefhany

 

[John Larkin]

If I told them to wind it with cotton-insulated wire and insulate it
with hockey tape, they'd have done that too. Can you guess the area of
the world we're talking about? (not that they'd have much use for
hockey tape). "Designer" may be too strong a word. We got several
thousand transformers from another company that were supposed to have
a wound screen (cheaper and more ungood than copper tape) between
primary and secondary and found that they had not understood the
instructions and simply taken the wire and terminated it under the
tape. Not only that, but being a (theoretically) disconnected wire
they didn't bother hipotting it or even testing it for isolation. It's
no fun when the 230V gets connected to earth (the metal housing),
which happened on 0.005% of the units Didn't deal with that factory
again. The part they did right was better (more reliable) than the
Japanese were getting, though.

We're evaluating a Chinese company (they work through a US rep) for a
custom power torroid. Their construction looks fine, and their design
technique seems to be successive approximation via lots of samples.
Cheap though.

John

 

[John Woodgate]

I read in sci.electronics.design that Spehro Pefhany <speffSNIP@interlog

How do you put an air gap in an EI
transformer?

If you want a real air-gap you put all the Es through the winding one
way and cover their open ends with your gap material, then add the block
of Is. For a vestigial gap, don't put any gap material there. For an
even more vestigial gap, divide the E's into three piles. Put two
through one way and one through the other way. Add Is to taste.

You could have reduced the copper loss in your transformer with the cool
core by reducing the numbers of turns and using thicker wire. That would
have increased the core induction and thus the iron loss. You lose out
if you increase the iron loss until it exceeds the copper loss.