HV AF transformer

Discussion in 'General Electronics Chat' started by ariemeir, Apr 4, 2013.

  1. ariemeir

    Thread Starter Member

    Jun 24, 2011
    68
    1
    Hi guys,

    I wanted to thank the senior members of the forum who are amazingly helpful !

    I am looking for a audio frequency (100hz - 20khz) transformer capable of
    taking a roughly 12v peak to peak signal to 2000v peak to peak. budget is somewhat of an issue, i am trying to keep costs for this part < $50.

    The application is a scientific instrument that requires high voltage, low curren t AC pulses (more or less sine wave shaped) , where the key parameter is the voltage, so i am worried less about noise and frequency shifting issues.

    I plan to drive this transformer from an audio amplifier which i in turn drive from a signal generator.

    The current output requirements are roughly 100ma, this is flexible, i could go as low as 10ma. Thanks to #12, a senior member here, I am looking at a 70w transformer: (P=Vrms*I=1000*0.707*0.1)

    I am trying to absorb a lot of information about transformers in a short amount of time, hence i appreciate your tolerance to my probably fundamental questions.

    It seems from a search in mouser/digikey that when choosing an AF transformer, it is the impedance ratio that matters more than the output voltages. I am assuming that this is because the purpose of a transformer in audio circuits is imp.matching.

    On the other hand, flyback transformers seem to work at these high voltages and more, but they typically operate on a higher frequency band (>100khz) and if i understand correctly, the frequency of the output is not exactly related to the input signal, so i cannot easily control it.

    Am i missing something fundamental which dooms my search for 2kv p2p, audio frequency transformer ?

    At the end of the day, i am almost ready to roll my own like this gentleman shows http://www.sm0vpo.com/power/diy_transformers.htm
    But i am wondering if there is a better way.

    Thanks for all your advice.

    Kind regards,
    Lenny
     
  2. timescope

    Member

    Dec 14, 2011
    298
    44
    Samsung LCD TV model LE40R88BD power supply unit has a large inverter transformer that supplies high voltage to the ccfl backlights. Something like this may be suitable.

    Timescope
     
    Last edited: Apr 4, 2013
  3. #12

    Expert

    Nov 30, 2010
    16,277
    6,788
    $126 plus shipping.

    First, the math, then the research.
    You need about 7070 ohms to 8 ohms @ 70W
    The 1650PA seems close.
    Do the math on that one.
    If it isn't good enough, consider the 1650RA.
    Now you adjust with things like:
    E2/E1 = N2/N1
    I2/I1 = N1/N2
    Z1/Z2 = N1^2/n2^2
    to find out how well your audio amp driver will fit the needs.
    Play around with 4 ohm output transformers, calculate what the other impedance would be, look for something that fits that impedance had has a 4 ohm output that you could drive.

    http://www.mouser.com/Passive-Components/Transformers-Audio-Signal/_/N-5gbg/

    (With AG gone, some of this is going to fall on me, I guess.)
     
  4. ariemeir

    Thread Starter Member

    Jun 24, 2011
    68
    1
    Thanks again #12,
    I will look into the details of it.

    I think i am going to take a small simple transformer (50hz) , take it apart and mess with it by driving it from my signal generator around 50hz and get a better understanding of the relationships between the parameters involved.

    Regards,
    Lenny
     
  5. #12

    Expert

    Nov 30, 2010
    16,277
    6,788
    You might as well. You aren't even going to "roll your own" unless you learn the math and you aren't going to see a 60 watt transformer for $50 retail.
     
  6. timescope

    Member

    Dec 14, 2011
    298
    44
    The transformer I referred to is the black one on the left of the circuit board on the ebay page.
    Here is a link to the LE40R88BD power supply schematic.
    The transformer part number on the schematic is UU32.5 which I Googled and ended up somewhere in China where the only info I got was that it is rated at 200watts, is made by Favotel and the information contained in the attached pdf.

    Timescope
     
  7. ariemeir

    Thread Starter Member

    Jun 24, 2011
    68
    1
    timescope, thank you for the resource.
    Will look into it.

    Regards,
    lenny
     
  8. ariemeir

    Thread Starter Member

    Jun 24, 2011
    68
    1
    Looking into alternative designs to reduce cost, I have seen that high voltage swings can be achieved by bootstrapping opamps with variable supply voltages.

    Besides not isolating the high voltages from the micro-electronics, and the added complexity of circuitry, what are the advantages of using a transformers as opposed to bootstrapped opamp circuit ?

    Thanks for the advice,
    Lenny
     
  9. The Electrician

    AAC Fanatic!

    Oct 9, 2007
    2,281
    326
    In your first post, you said you could allow the output current to be as low as 10 mA. At that limit, the needed power would be 7 watts rather than 70 watts. Even if you don't go that low, you could have a transformer rated substantially less than 70 watts with the concomitant reduction in cost.

    Could you provide more detail on the nature of the pulses you will be generating. Do you really need a low frequency limit of 100 Hz? You say the pulses will be more or less sine shaped. Is it a single pulse of variable width, occasionally long enough to need the 100 Hz low end?
     
  10. #12

    Expert

    Nov 30, 2010
    16,277
    6,788
    (1) There is no such thing as a 2000 volt op amp. (2) You would be using a dual output power supply of +/- 1000 volts and controlling thousand volt (rated) transistors. First, you have to build the power supply! Compare that to the simplicity of connecting an audio amplifier to a transformer. Most electronics nerds have an audio amp laying around for just such needs.
     
  11. ariemeir

    Thread Starter Member

    Jun 24, 2011
    68
    1
    Thanks #12,

    I was referring to a technique i've seen here:
    http://www.edn.com/contents/images/45890.pdf

    According to my undestanding, by varying the supply rails, one could have a regular (say +- 30v rated) opamp to provide a wide swing of voltage since the opamp
    operates in a virtual supply it is comfortable with.
    I've seen that digikey does carry low-cost 600v bjts, and high voltage diodes are easy to come by so this provoked the idea of using bootstrapping.
    As for a power supply, i have a step up converter that can generate 450v ~10ma.

    http://www.ebay.com/itm/151010612850?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649.

    This gives 200v , but a simple voltage doubler takes it higher and it also gives more power:
    http://www.ebay.com/itm/320831453118?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649

    It is definitely not 2000, but what i am thinking is breaking my experiment into two parts ( first test at 450 and later go higher).

    What do you think ?

    Regards,
    Lenny
     
  12. ariemeir

    Thread Starter Member

    Jun 24, 2011
    68
    1
    Dear The Electrician, hi,

    Thanks for your reply.

    My chain of flow looks like this:

    function generator (sine or square pulses,1v p2p) =>|
    audio amplifier (~24v p2p) =>
    further amplification/transformation
    (my goal is to take this 24v p2p to roughly 2000v p2p)

    The low frequency limit of 100 is somewhat flexible, it could be pushed up to say 250 if that changes matters.

    The pulse typically come in a train of 10-100 pulses. For example : for a 1khz sine/square wave, i will run the sine for 100ms yielding 100 cycles. For the square wave i often use relatively low duty cycles (<10%).

    I do not combine several frequencies in one pulse.

    I made additional simplifications to my system and 10ma is something i could
    live with. I've noticed that the reduction in the cost of the transformer is not proportional to the power rating, e.g.
    this item rated at 10w, costs $90,
    http://www.mouser.com/Search/ProductDetail.aspx?qs=gF4h2Znc3LAkjUnxHoRKrg==

    whereas this, 60w rated one costs $140 (50% more, but 6 times more power):
    http://www.mouser.com/ProductDetail/Hammond/1650PA/?qs=/ha2pyFaduiMxiiDLsQadr8Sa8s3DeJyeCH/tXHG4p0=

    It makes me wonder what is the root of this discrepancy.

    Hope this provides more information.
    Thanks again.
    Lenny
     
  13. ariemeir

    Thread Starter Member

    Jun 24, 2011
    68
    1
    Dear #12,
    I've spend some time thinking of the numbers here and seeing where can i afford reducing the specs to cut the project costs.

    I am a bit confused and was wondering if you could clarify a few things:

    First, the sides: my goal is to have input @ low (say 110vac) voltage so it should be running high current which is translated to high voltage low current on the other side. So based on this reasoning i thought that i need
    low impedance at the primary coil and high impedance at the secondary.
    But you have mentioned 7070 to 8ohms, so i followed that train of thought,
    but http://www.stereophile.com/content/listening-57-page-2 explains this topic consistent with your specifications : high to low impedance yields step-
    up-transforming.

    Could you provide some intuition here : i am thinking about magnetic induction in the core: the primary coil runs some current, induces magnetic
    field "felt" by the secondary. Ampere's law states BL = mu N I.
    If the field B,L,mu are the same, we get N1 I1 = N2 I2 as you mentioned.
    But now, N1/N2 = I2/I1 , so if i want a reduction of say 10 in current, i need N1 to be 10 times smaller, hence the small impedance on the primary. This
    contradicts reality.

    Where am i wrong ?

    Also, i've noticed that the Hammond and other transformers specify their impedances with a suffix ct. the nearest thing i could find explaining it was
    ct=current transformer but it didn't really make sense since i am not really
    trying to measure currents. Could you clarify this please ?


    Kind regards,
    Enjoy the weekend,
    Lenny


     
  14. tubeguy

    Well-Known Member

    Nov 3, 2012
    1,157
    197
    In this case ct means center tap which is a tap at the mid point of a winding.
    Sometimes used for a Ground reference connection.
     
  15. ariemeir

    Thread Starter Member

    Jun 24, 2011
    68
    1
    #12, thanks for your guidance.
    I've messed with some transformers i've picked up in urban ore (berkeley) this weekend, and i think that connecting several trasnformers in reverse might give me at least partially the voltage swing (and very partially the power) that i am aiming for eventually.

    I'm trying to estimate the power losses vs. frequency (because my 50hz transformer responds all over the audio spectrum with the voltage going higher as i go > 10khz),
    but that's a separate story.

    Just wanted to say thanks again.
    L.
     
Loading...