Ferrite Core Transformer testing

Thread Starter

SVS

Joined Aug 16, 2012
89
Hello Everyone,

I wanna test and optimize a Step up transformer (330V-6KV) transformer . It is a ferrite core transformer which works at 30-40 Khz .
But in my laboratory , due to some safety reasons , I am not allowed to work above 300V AC.
Is it fine if I perform the test routines using a Signal generator with the same operating frequency(30-40 KHz) but with a low voltage input?
1. I know i won't get a proper result if I wanna calibrate its efficiency.
But what about
Turns Ratio
Primary,Seconadary Inductance
Winding Resistance
coupling factor
Leakage Inductance

Self Capacitance.
Can I measure these parameters ?? Will it be accurate?
 

crutschow

Joined Mar 14, 2008
34,285
The inductance measurement accuracy at lower voltages would depend upon how close to saturation the transformer is being operated. Operating near saturation can reduce the apparent primary and secondary inductance. Other parameter measurements should be largely unaffected.
 

Thread Starter

SVS

Joined Aug 16, 2012
89
The inductance measurement accuracy at lower voltages would depend upon how close to saturation the transformer is being operated. Operating near saturation can reduce the apparent primary and secondary inductance. Other parameter measurements should be largely unaffected.
Thank you for the reply!!!

Usually during normal operation (i.e)
Primary side input Vin=330V (Square wave signal from a H-bridge) , F=33 KHz &&
We have a capacitve load of 200pF on the secondary side. So, I think that we are not close to saturation.

Now I wanna find the vakue of inductance with a low voltage input , and the same frequency !
Will I get a close value ?

If not can you please suggest me how to do it in a right or better way?
 
Last edited:

crutschow

Joined Mar 14, 2008
34,285
Usually during normal operation (i.e)
Primary side input Vin=330V (Square wave signal from a H-bridge) , F=33 KHz &&
We have a capacitve load of 200pF on the secondary side. So, I think that we are not close to saturation.
..........................
Why do you "think that we are not close to saturation"? Saturation can occur if the input voltage/frequency value is too high for the core and the number of primary turns causing the magnetizing current to be too high. It's independent of secondary load current. You have to know the design of the transformer to determine where it saturates.
 

szhighstar

Joined Jun 26, 2012
27
1.for inductance, you can test using LCR meter, like 3260B meter, note test frequency and voltage.
2.for Turns Ratio, you can test using 3260B meter (WK), it is very precision.
3.Winding Resistance, test using LCR meter.
4.Leakage Inductance, same as above, note short some pins.
5.Self Capacitance, same as above
6.SRF, you can test SRF.

 

Thread Starter

SVS

Joined Aug 16, 2012
89
1.for inductance, you can test using LCR meter, like 3260B meter, note test frequency and voltage.
2.for Turns Ratio, you can test using 3260B meter (WK), it is very precision.
3.Winding Resistance, test using LCR meter.
4.Leakage Inductance, same as above, note short some pins.
5.Self Capacitance, same as above
6.SRF, you can test SRF.

Thank you for the reply!! I have already done it with a LCR meter!!
I am using a signal generator to cross check my values!!
 

Thread Starter

SVS

Joined Aug 16, 2012
89
. It's independent of secondary load current. You have to know the design of the transformer to determine where it saturates.
I got your point.
How do I calculate the saturation current?
According to the earlier measurements I made, The primary current is 3,5 A (rms).

I have also attached the data sheet of Ferrite core which is in use.

Thank you in advance!!
 

Attachments

t_n_k

Joined Mar 6, 2009
5,455
I got your point.
How do I calculate the saturation current?
According to the earlier measurements I made, The primary current is 3,5 A (rms).

I have also attached the data sheet of Ferrite core which is in use.

Thank you in advance!!
How many turns on the primary winding?
 

t_n_k

Joined Mar 6, 2009
5,455
Thank You for the reply!!

24 turns in total on the primary side.
It is a 2 layer winding. 12 turns each layer.
Depending on your particular core the minimum Al factor could be 5400 nH per turn squared. This could be lower by 20% according to spec sheet.

With 24 turns this give a notional minimum magnetizing inductance of
L=24^2*0.8*5400*1E-9 H = 2.49 mH.

With a square wave voltage of ±330V at 33kHz & 50% duty cycle the peak magnetizing current would be ..

Ipk=(330V/2.49mH)*T/2=(330/2.49E-3)*15.15E-6=2A

This would imply a triangular magnetizing current of ±2A. In fact since in this case there is a symmetrical alternating square wave input voltage, the pk-to-pk current will probably be 2A (rather than 4A) and the triangular magnetizing current waveform would then be ±1A.

With the shorter length core this would give a maximum flux density of

Bm=μo*μe*N*Ipk/le=4*π*1E-7*1800*24*2/0.258=0.42 Tesla at 2A or 0.21 Tesla at 1A.

For N27 material this is not going to be near saturation - allowing for ~0.5 Tesla capability.

I'm not sure why your previously measured primary current was 3.5A rms. Did you mean you calculated this as the likely saturation current or a measured value? From my calculations it seems unlikely that the magnetizing current would be so high unless this value includes the referred secondary load current as well.

Perhaps you should also state which core configuration you are using as there are a few options in the data sheet you posted. Also posting a simple schematic of your circuit would make it easier to comment.

The presence of a 200pF load capacitance is probably going to radically effect the current as the capacitive load current draw will be significant. There may even be interaction with the secondary winding leakage inductance in the absence of any load resistance damping.
 
Last edited:

Thread Starter

SVS

Joined Aug 16, 2012
89
Depending on your particular core the minimum Al factor could be 5400 nH per turn squared. This could be lower by 20% according to spec sheet.

With 24 turns this give a notional minimum magnetizing inductance of
L=24^2*0.8*5400*1E-9 H = 2.49 mH.

With a square wave voltage of ±330V at 33kHz & 50% duty cycle the peak magnetizing current would be ..

Ipk=(330V/2.49mH)*T/2=(330/2.49E-3)*15.15E-6=2A

This would imply a triangular magnetizing current of ±2A. In fact since in this case there is a symmetrical alternating square wave input voltage, the pk-to-pk current will probably be 2A (rather than 4A) and the triangular magnetizing current waveform would then be ±1A.

With the shorter length core this would give a maximum flux density of

Bm=μo*μe*N*Ipk/le=4*π*1E-7*1800*24*2/0.258=0.42 Tesla at 2A or 0.21 Tesla at 1A.

For N27 material this is not going to be near saturation - allowing for ~0.5 Tesla capability.

I'm not sure why your previously measured primary current was 3.5A rms. Did you mean you calculated this as the likely saturation current or a measured value? From my calculations it seems unlikely that the magnetizing current would be so high unless this value includes the referred secondary load current as well.

Perhaps you should also state which core configuration you are using as there are a few options in the data sheet you posted. Also posting a simple schematic of your circuit would make it easier to comment.

The presence of a 200pF load capacitance is probably going to radically effect the current as the capacitive load current draw will be significant. There may even be interaction with the secondary winding leakage inductance in the absence of any load resistance damping.
Wow. Thank you for spending some useful time !!

I have attached a circuit which might be helpful !

1.We use a UI core combination. So , there will be a slight deviation from your calculation.

2.The resonant frequency of the circuit is approximately 29-30 KHz.
So, the current waveform at the primary side of the transformer will be sinusoidal.(Resonant converter topology)
This H bridge is operating with Zero voltage switching technique.

So, our actual aim is to keep the voltage constant(330V) and just by varying the frequency from 29 KHz - 33 KHz, we control the input power , thereby the input current only changes.

I used a Energy meter to calucate Vin, Iin, Apparaent power , Real power.
I varied the input power from 100 W to 600 W , i.e 29 Khz to 33 KHz,
My Input current was 0.76 A at 100 W and 3,6 A for 600 W.
But this need not neccesarily be my primary trafo current.

3. The main problem is, I used a LCR meter to find the inductance of the transformer (i.e) for the primary side , I gota value of 0.4037 mH at 29 KHz.

Why is there a big difference between calculation and measurement??
This freaks me out! Even i did the same calculation as you did!!
Maybe later I can also send you a picture of my transformer.

Thank You!!
 

Attachments

Last edited:

t_n_k

Joined Mar 6, 2009
5,455
So it boils down to whether you trust the LCR measurement or not.

You have a function generator at your disposal. If you have access to a reasonable quality two channel CRO and a couple of other items you can verify the primary inductance value.

Suppose the inductance is indeed ~0.4mH. At 10kHz this would have a reactance of around 25Ω.

So connect your function generator to the transformer primary wired in series with [say] a 100Ω ±1% resistor. Connect the generator ground to the resistor side and the generator active to the transformer primary side. Connect one channel of the CRO to the generator active and the other channel to the junction of the transformer primary and the resistor. Now one channel reads the total source voltage and the other reads the resistor voltage drop.

Set the generator output to sinusoidal mode with frequency 10kHz [exact as possible] at say 5 to 10V pk-pk output with zero offset.

Trigger the CRO off the channel connected to generator output active. Check the CRO shows a period of 100us for the generator signal.

Expand the CRO timebase [cal mode], and iteratively adjust channel vertical gains and trigger level to enable you to readily measure the time delay between the generator source AC voltage zero crossing and the 100Ω resistor voltage signal zero crossing. You would need to be able to resolve to say a tenth of a microsecond. If the CRO has two cursor display time delay measurement option or something similar then all the better. Call this time delay measurement ΔT.

Provided the frequency is 10kHz, the magnetizing inductance will then be given by

\(L_m=\frac{1}{200\pi}\tan{\( 2\pi \times {10}^4 \times \Delta T \ [rad])} \ H\)

For instance if ΔT=3.9 microseconds [usec]

\(L_m=\frac{1}{200\pi}\tan{\( 2\pi \times {10}^4 \times 3.9 \times{10}^{-6} \ [rad])} \approx 0.4mH\)

If say ΔT=16usec

\(L_m=\frac{1}{200\pi}\tan{\( 2\pi \times {10}^4 \times 16 \times{10}^{-6} \ [rad])} \approx 2.5mH\)
 

Thread Starter

SVS

Joined Aug 16, 2012
89
So it boils down to whether you trust the LCR measurement or not.

You have a function generator at your disposal. If you have access to a reasonable quality two channel CRO and a couple of other items you can verify the primary inductance value.

Suppose the inductance is indeed ~0.4mH. At 10kHz this would have a reactance of around 25Ω.

So connect your function generator to the transformer primary wired in series with [say] a 100Ω ±1% resistor. Connect the generator ground to the resistor side and the generator active to the transformer primary side. Connect one channel of the CRO to the generator active and the other channel to the junction of the transformer primary and the resistor. Now one channel reads the total source voltage and the other reads the resistor voltage drop.

Set the generator output to sinusoidal mode with frequency 10kHz [exact as possible] at say 5 to 10V pk-pk output with zero offset.

Trigger the CRO off the channel connected to generator output active. Check the CRO shows a period of 100us for the generator signal.

Expand the CRO timebase [cal mode], and iteratively adjust channel vertical gains and trigger level to enable you to readily measure the time delay between the generator source AC voltage zero crossing and the 100Ω resistor voltage signal zero crossing. You would need to be able to resolve to say a tenth of a microsecond. If the CRO has two cursor display time delay measurement option or something similar then all the better. Call this time delay measurement ΔT.

Provided the frequency is 10kHz, the magnetizing inductance will then be given by

\(L_m=\frac{1}{200\pi}\tan{\( 2\pi \times {10}^4 \times \Delta T \ [rad])} \ H\)

For instance if ΔT=3.9 microseconds [usec]

\(L_m=\frac{1}{200\pi}\tan{\( 2\pi \times {10}^4 \times 3.9 \times{10}^{-6} \ [rad])} \approx 0.4mH\)

If say ΔT=16usec

\(L_m=\frac{1}{200\pi}\tan{\( 2\pi \times {10}^4 \times 16 \times{10}^{-6} \ [rad])} \approx 2.5mH\)
#

Thank you once again!!

I have a signal generator , but it can produce only square wave or PWM.
And If i supply 5-10 V pk-pk, My secondary side output will be around 100V(eventhough 100 Ohm resistor acts as a voltage divider) , which i am not allowed in my lab without supervison.

So, is it fine if i use 0.5 V square wave as input to the transformer, and perform the test you mentioned above.?
 
Last edited:

t_n_k

Joined Mar 6, 2009
5,455
#

Thank you once again!!

I have a signal generator , but it can produce only square wave or PWM.
And If i supply 5-10 V pk-pk, My secondary side output will be around 100V(eventhough 100 Ohm resistor acts as a voltage divider) , which i am not allowed in my lab without supervison.

So, is it fine if i use 0.5 V square wave as input to the transformer, and perform the test you mentioned above.?
A square wave generator wont fit with my proposed test procedure and data interpretation above.

You could use the square wave generator with a different approach. You could set up the same circuit and attempt to measure the circuit time constant based on a step response approach with the square wave as stimulus. A 0.4mH inductance with a 100Ω series resistance would result in a 4us time constant which with a bit of care you could probably measure with the scope.

I'm surprised there isn't a suitable sine wave generator in an electronics lab.
 

Thread Starter

SVS

Joined Aug 16, 2012
89
A square wave generator wont fit with my proposed test procedure and data interpretation above.

You could use the square wave generator with a different approach. You could set up the same circuit and attempt to measure the circuit time constant based on a step response approach with the square wave as stimulus. A 0.4mH inductance with a 100Ω series resistance would result in a 4us time constant which with a bit of care you could probably measure with the scope.

I'm surprised there isn't a suitable sine wave generator in an electronics lab.
I found a sine wave signal generator. It can produce only 2V. But I will do the measurement and will post the test results!!
I think this value of magnetizing inductance could be the right one!

Thank you!!
 
Top