5FP7 project starting with the focusing coil.

[AdrianPH]

Jeffrey

I have been thinking about this in my head most of the day as well as trying to do the house work, so while waiting for the oven to get to temperature, I have tried to put it down in this post what I have been trying to figure out, so please let me know if I am following the correct idea, this is not for you but to try and get it in my mind.

An imaginary transformer is made of Resistance in the winding which will not decrease with frequncy and the Reactance of the inductor part of the winding determined by the value of its inductance in H and the frequency of use.

So just pulling figures from my last post. A transformer for 50 Hz has 2000 turns in the primary on a nice iron core and has a DC resistance of 300 Ohms, the Inductance is 10 H.

At 50 Hz the reactance of the core will be 3,141 Ohms ( 2 x PI x 50 x 10 ) so the overall Impeadance of the transformer would be SQRT(300 x 300 + 3142 x 3142) = 3,156 Ohms. So at 230 Volts the current through the transformer is 230/3156 = 73 mA.

This 73mA being just below the knee of saturation as tested in the previous post and with 2000 turns we have an ampere turns value of 146 AT.

Down at 12.5 Hz we still have 300 Ohms for the resistive value but the reactance of the coil is now 785 Ohms ( 2 x PI x 10 x 10 ) and the impedance of the transformer is now 840 Ohms so at 230 Volts the current would try to be 230/840 = 274 mA which at 2000 turns would be 548 AT. taking it well into saturation. So to get back to the a safe 73mA primary current would require the voltage dropping to 840 Ohms x 73 mA or 61 Volts.

Am I on the correct lines or not?

Adrian

[Ed_Dinning]

Hi Adrian, a quick and dirty method to check saturation is to plot an I/V curve for the transformer at its frequency of operation, or use 50Hz and extrapolate.
The "knee point" of the curve should be quite obvious, this is where saturation is starting. It is then possible to get a turns/ volt figure to stay below saturation for that specific core.

Or, using the transformer equation N= V/ 4.44, F, B, Ae
N=turns
4.44 constant for a sine wave (4.0 for Sq wave)
F = freq
B =working flux density
Ae, cross sectional area of centre limb in M^2

For Stalloy B = 1.0, for Unisil (GOSS material) it can go as high as 1.5 and the curve into saturation is slower

Plenty of things to play about with there.
Note that on a laminated transformer the gaps between E & I lams should be tightly closed or saturation will start earlier

Ed

[AdrianPH]

Hello Ed,

As I tend to play a lot and use old transformers I have no idea what the EI cores are made of, most of the time it is difficult to determine what EI designation they are as many I have seem to be mid way between what info I can find. If memory serves me correctly that is why I was given the information in post 60.

I guess that if I used that information I could feed it back into the formula and work out a value for B and what would it be for a sawtooth waveform?

Jeffrey has been trying to get me to understand the de-rating for different frequencies specifically 12.5 Hz which would be the frame frequency of a sawtooth for NBTV. I need to make a transformer up somehow. So as I am a bit thick with these things I have been trying to understand where the values for de rating comes from rather than just accept it because I read it. Hence my post 61 in tying to get my brain ramblings down to get checked if on the correct line of thought.

Adrian

[Mark Hennessy]

It might help to think about the transformers that you find in switched-mode power supplies.

How can they be so very small compared to a conventional 50Hz type that passes similar powers?

It's because they are dealing with much higher frequencies. Simply put, a switched-mode power supply turns 50Hz into DC, then turns it back into AC, but at 50kHz or 100kHz or more.

Power is energy over time; joules per second. When you run 50Hz mains through a transformer you have 50 cycles per second; whereas for a 50kHz switched-mode power supply you have 50,000 cycles per second.

So you can see that a 50Hz transformer has to transfer 1000 times as much energy per cycle to achieve the same power throughput.

And that means that the higher the frequency, the smaller the core of the transformer can be, because it has to store less energy during each cycle.

Conversely, if you slow down to 25Hz, then you've doubled the energy per cycle needed to maintain the same power throughput. That means the core will need to store double the flux than it did at 50Hz. Or 4 times for 12.5Hz. If it's not big enough to do this, it will saturate. And bear in mind that standard mains transformers are designed to be just starting to saturate at 50Hz. To bring the flux down to a level that the core can cope with, you must reduce the voltage by basically the same factor you reduced the frequency by.

When thinking about all this and trying to convey a "hand-waving" understanding of the basic idea, I don't find it helpful to consider the DC resistance or inductance of a transformer. Just think of it as an energy storage device. Though the DC resistance is basically what you're left with when you hit saturation, as it is no longer an inductor at that point.

Another way to approach this might be to think about audio. The signal level at which a transformer will saturate depends on frequency, for all the reasons discussed to far. Analogue tape is another possible "angle" that might help to get a handle on this, perhaps. Depends on what you've done before - analogies are only useful if you know the thing being used.

Sorry if all that is a bit simplistic, but hopefully it'll help you get to the nub of the question.

Mark

[AdrianPH]

Mark i will get there eventually, I have not needed to make my own transformers before getting into valve tech. I have built a small valve audio amp with a push pull output transformers for a small less than 1 watt amp, but was basically told the formula to use put the numbers in and off I went. I changed a secondary winding on a mains transformer simply by counting the turns as I took them off, I knew what the secondary voltage was before hand so I could wind a different secondary knowing what the TPV was. Other than that I just bought them in for any typical application.

The Scroggie Foundations of Wireless book I read talks of ampere turns to generate a back EMF, so it may be the generation of this back emf I am not understanding. Much of this project is falling apart, I calculated that I need 13.6 Volts peak to peak to swing the line trace from top to bottom, I am actually needing double that and I do not know why. The transformer I made up for that at 400Hz working is an air cored one as it is basically a Single ended driver amp, and looking at my turns per volt I could have gone way over the top. Much of this is try it and see how it works try to understand why it does or why is does not. I will stick with it and get there eventually. I need sleep, I was up into the early hours yesterday so I better not push it to long this morning.

Cheers

Adrian

[Mike Watterson]

I think on transformer size vs frequency a square-law might be involved?

An LOPT for EHT is a special kind of inverter, so needs to be single ended as it's the switching off of the valve on flyback that generates the voltage pulse. But a straight drive of scan coils, especially for frame, either direct or via a transformer can advantageously be push pull. It only reduces the core problems to being AC, rather than the DC bias on single ended. A given pair of valves in class B rather than a single valve can give nearly four times the power.

However AC only, nor push pull solve the core issues. Really it's the magnetic field, thus current in the deflection coil that counts. In that case a transformer is a current ratio transformer, but the voltage ratio is the inverse of the current ratio. So 10:1 voltage step-down is 1:10 current step up.
The core saturation is related to turns per volt on the primary and the inverse square of the frequency as the core (even with AC only) can saturate with NO load at all with insufficient turns or too low a frequency. Hence a 220V marginal transformer on 240V can heat up with no load, that can be solved easily. A marginal 110V transformer for 60Hz may have a problem with 110V at 50 Hz, but that has no simple solution, except maybe to try 90 to 100V.

Even in 1980s some very lower power isolated industrial cards were using as high as 1 MHz so as to use a high profile quad DIL 2KV isolation transformer. One core was power and CPU clock, the other 3 were shift clock, data in and data out. 4 x 1N4148 as the "power rectifiers" as the CMOS CPU at 1MHz system clock was very low power.

Subwoofer or woofer filters may use very large air core coils as laminate cores would be very large anyway and add some distortion.

[Ed_Dinning]

Hi Adrian, be careful studying Scroggie, it is good theory but he uses the imperial system of measurements that confuse the hell out of most people. (note that I'm perfectly happy with mechanical measurements but electrical units wind up with all sorts of strange constants.

The old transformers you are dealing with are most likely to be Stalloy, so operate at 1 Tesla or less.
T/V are inversely proportional to frequency, so use 2 times the turns that you would for 50Hz for the 25Hz frame rate.

The 4.44 constant is related to the Sine wave form factor (1.11) (1.0 for a square wave.)

As a first approximation the flyback is a small portion of the total cycle and is effectively an HF component, so ignore.

The above is an Engineering approach to get you close, not a detailed mathematical analysis, I leave that to others

If you can, use an autotransformer as that will reduce the size of wire and winding. This works well for ratios of about 1:1 to 3:1

Ed

[ppppenguin]

What Ed says, about using twice the number of turns for half the frequency (or 4 times for a quarter) is near enough the same as what I was saying about using a transformer at half (or quarter) of its nominal rated voltage. But Ed has put in a more fundamental way.

[AdrianPH]

Hello all, today has been spent dealing with general family matters so not had much time to think on this lot today.  I must get to sleep before midnight this evening. 

Long post very sorry.

I was reading some downloads this morning and as Ed has mentioned turns out it is using Imperial measurements for all core sizes and talks about flux density in lines per square inch, stacking factors of 0.92 for 1 x 1 interleave and 0.95 for a butt stack.  I would have thought it should be the other way round with interleave at 0.95 but no matter  The downloads were Wolpert Power transformers and audio transformers.  I did not get to far before I gave up for sleep.

I bought a job lot of transformers from Ebay some time ago, and where I can, I strip them down for the laminations it does unfortunately mean I do not know the core material details, or what VA the transformers were as they are not marked as 12 Volt 500ma etc.So any home wound transformers have been a finger in the air approach.

The transformer I wound for the Line scan, which has a frequency of 400Hz, I calculated last night had a core section of 6 x 10^-4 m².
when I made it up it was all just guessing but used some figure I had written down for an audio transformer.  I used 14 turns per volt, a butt stack with a piece of high temp tape as an air core as there was a DC bias current flowing as well as the scan. 
This could well have been over the top, I know I am lacking how to work this out better.

I have also bean using peak to peak measurements when trying to work things out dealing with the sawtooth waveform and as jeffrey pointed out the rms values would be a lot less.
For example the line scan coils when driven with DC just needed 12.5 volts to get full deflection, the tester said the coils inductance was only 61.6mH so a little more maths and calculated the scan voltage should only be 13.7 Volts call it 14 Volts peak to peak.  So I picked 14 Volts for the secondary.
I needed a peak current of 204 mA to drive the spot from one side to the other and using a EL84 I wanted to drop this to a more manageable figure for the anode current so used an 8.8:1 ratio, so the primary current should be in the region of 24 mA plus DC.
so with these figure the primary voltage peak to peak would be around 123 Volts so with the 14 TPV I used  1724 turns for primary and 196 turns for secondary, I just  guessed I would be OK, it does work, in a fashion.  The average current flowing through the anode is 24.2mA as I monitored the voltage drop on a 100 Ohm resistor last night, so I believe that the line scan coils and transformer are supplying some current back into the system just after flyback, fly back is massive so I have very heavily damped the primary side of the transformer using caps and resistors.

I am not sure how some of the issues would be reduced if rather than 123 and 14 Volts, I used 43.5 and 5 volts being rms for the windings in the transformer.  A previous comment from Jeffrey, leads me to believe not a lot as the inductance of the transformer has little effect it all being from the scan coils and the square of the turns ratio, which I was able to understand.

During the measurements i also found that the actual peak to peal scan voltage on the coils was 28 Volts and not 14, the primary scan voltage peak to peak voltage is 230 Volts rather than 123 Volts, so I know I could have some errors with adding a few extra turns on the secondary, but for the voltages to be nearly double what I worked out get me wondering what else i have missed.

I would like to do the frame scan using a push pull ECC82 as again suggested by Jeffrey, but I need to understand what size core is needed at 12.5 Hz.  Testing with DC the resistance of the scan coils are 511 Ohms, the frequency and inductance of the coils means at 12.5 Hz, 50 Volts peak to peak would do a full scan,  So 18 Volts as Jeffrey posted at 511 Ohms is only 0.634 watts.  But before I go any further I could do with understanding why the line scan is double the voltages of the DC tests?

It will be Monday or Tuesday before I can lift a soldering iron as I have a few other items to sort out.

War and peace over with.
Stay safe.

Adrian

[Diabolical Artificer]

But before I go any further I could do with understanding why the line scan is double the voltages of the DC tests? Is this because of back EMF? If you shove current through a coil of wire, it creates a magnetic force, stop the current,the magnetic field collapses creating a voltage. That's why we have to put diodes on relays and OPT's have to be rated for higher than HT. Hope I got that right and it's what you meant.

Re, building tfmr's, a lot of the tomes on the subject are heavy going but there's a few tricks to sidestep the maths. To find out if you have enough turns on a primary, use a lightbulb in series with your winding. I use my variac, lamp limiter combo to test tfmr's. For toroids to find number of turns divide 42 by area of core, don't know if this works for EI and for other frequencies. It might be that you could use that "formula" and double NT for half the frequency?

This video explains some of the mysteries of tfmr's - https://www.youtube.com/watch?v=4UFKl9fULkA This page is useful - https://engineerexperiences.com/design-c...tions.html and this one - http://dicks-website.eu/coilcalculator/index.html

Good luck, Andy.