Step-up transformer design

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have been using these software based tools with a good amount of accuracy to measure transformers with lately.

Visual analyzer,

Visual Analyser details

Sometimes it is a PIA to setup and work, But when it is cooperating it gives great results.
A 192Khz capable sound card is desired as well.

Then there is this program,

WB6DHW

I use this same interface for all of the programs.
This program works very good as well but it doesn't have as fine of a results that VA produces when using a higher resolution sound card.

I have determined that both of these programs determine the same results many times over and RLCMeter seems to be the easiest to use with consistent results and those results are here,

http://www.diyaudio.com/forums/software-tools/212908-exploring-visual-analyser-va-3.html#post3383721

The accuracy of the reference resistor is the key to good results as does the sampling frequency as well as the sample rate.

I recently have discovered this program as well for plotting Impedance curves,

AcustikA | signal analysis software tools

It gives a great generalization of what kind of impedance curve the transformer has with a good amount of accuracy.

Although I have been having issues determining the accuracy of this program using different sound cards I have not figured out yet what the issues are yet.
But I can say that it is within or better than about 2% to 5% and a worst case of about 10% depending on the setup and reference resistor value.

I have been using a 1kohm reference resistor and seems to give the best and closest results so far than any of the lower values that I have used with this program.
I haven't tried any higher values with this program yet and is something that I am in the middle of at this very moment.

And again some of the results are different with different cards and other cards produce the very same results with another.
The difference is typically small depending on the situation as I have shown here,

http://www.diyaudio.com/forums/soft...peaker-impedance-tool-simpes.html#post3402200

and earlier in that same thread.

But the data is very workable!!

I have verified the accuracy of the other two programs VA and LMSBridge with various capacitors and inductors and checking them using the resonance technigue.

And I verified the SimpleS impedance plots vs frequency using a reactance calculator found online and found them to be very close knowing the values determined from the other two programs.

You will find out the once you set the measurement frequency above the impedance peak it will reflect a capacitive reactance on the primary winding as measured in VA and if you divided this value by the transformation ratio squared this will tell you the overall capacitance on the secondary side of the step up transformer.

I have verified this but adding some extra capacitance large and small to the secondary winding and it exactly worked out in the math with the two different values that VA produced

I have verified this using the very same painstakingly long handed method that Bolserst describes by measuring the voltage drop across the primary winding and the current in the winding by using a resister between the winding and ground.
I used VA to do this as my meter is not accurate for measuring AC voltages at frequency's above the normal line frequency's of 50Hz to 120Hz.
Using a transformer test jig that I describe here,

http://www.esldiy.com/esldiy/forum/index.php?topic=177.msg311#msg311

The measurement I did in VA exactly matched the capacitance I found using the long handed method.

The resonate methods using leakage inductance was similar and matched as well but are more difficult to do accurately at times.
Therefore I had lots of data to compare too.

The electrostatic shield can give you some strange results depending on the connection configuration of the winding's as I have found out while testing the Antek AS-1206 using the resonate mode technique for leakage inductance and stray capacitance.

Overall it does add to the total capacitance of the transformer, regardless of what it does to the resonate frequency.

In my situation I ended up shorting one of the 120V winding's during a core saturation stress test and the poor quality/amount of insulating material cause a breakdown failure to the electrostatic shield and burned the winding.

Luckily I was able to strip it down to the burned spot and repair the 7 or 8 damaged winding's and make the core work again.

This is the main reason I have not presented any data in this particular core yet.
And lately I have been hashing through these programs to determine their level of accuracy.

But so far everything has been looking pretty good.

I sure have learned a great deal more since I first started with the stuff and these few programs really helped a lot !!!!
And many thanks to you guys for helping as well!!!

Cheers!!

jer :)
 
Last edited:
What is the negative side effect of extra capacitance in this application (dynamic driver and non-inductive resistor as load)? 120 turns is still quite low, ESL output transformer with several thousand turns would be a different story...

The extra winding capacitance will unnecessarily restrict the upper limit of the transformer bandwidth which, when driving a predominately resistive load like yours, will be defined by resonance of the winding capacitance with leakage inductance. When ordering stock power transformer you have to live with what is available, but I'm still pretty confident most small toroidal power transformers with 1:8 turns ratio would meet your needs.

Now, if you can customer order, have them wind all 240 turns of the secondary as one continuous winding with 2mil mylar insulation between layers to minimize winding capacitance further. Also, divide the turns per layer such that each layer completely cover the primary winding which should be wound to cover 80% or more of the core circumference to minimize leakage inductance.
 
The extra winding capacitance will unnecessarily restrict the upper limit of the transformer bandwidth which, when driving a predominately resistive load like yours, will be defined by resonance of the winding capacitance with leakage inductance. When ordering stock power transformer you have to live with what is available, but I'm still pretty confident most small toroidal power transformers with 1:8 turns ratio would meet your needs.

Now, if you can customer order, have them wind all 240 turns of the secondary as one continuous winding with 2mil mylar insulation between layers to minimize winding capacitance further. Also, divide the turns per layer such that each layer completely cover the primary winding which should be wound to cover 80% or more of the core circumference to minimize leakage inductance.

The 0,44m2 thick secondaries will be approx 4 metres long in total (secondaries in series) and the series res of the secondaries (in series) will be under 0,5 ohms (propably under 0,3 ohms). Can this amount of wire cause much of a capacitance when bifilar wound?

Which is more important in my application, to reduce stray capacitance (= to use normal winding) or to reduce leakage inductance (= to use bifilar winding)? I read that bifilar wounding reduces the leakage inductance to 1/3 - 1/2 compared to normal winding.

It might likely that both kind of transformers perform well enough (freq response wise) in this application. How about the signal integrity, which winding style sounds better, has anyone compared? Bifilar wound transformers have cleaner square wave response (less ringing) and behave better with pulse signals. I think the common Antek trafos, used as ESL step up frequently here, had bifilarly wound HV-windings?

I asked the manufacturer to drop out the electrostatic shield.
 
Last edited:
I think in your application Bifilar winding to reduce the leakage inductance would be a good route to take as your total turns are much less than that is used for an ESL setup.

Therefore your transformers stray capacitance will be quite low anyhow and having only an 1:8 ratio won't make the capacitance factor near as bad as a 1:80 ratio will.

Taking measures to reduce the leakage inductance would be a big plus and should keep the resonance well above the audio band as well as reducing the ringing.

You can estimate very closely your capacitance by knowing the total surface area of the winding and the insulation thickness to the core or adjacent winding of concern.

The Antek AS-1206 core that I am working with shows to have about 450pf or so of capacitance when the 120V winding are in parallel and is typically double this when they are in series.

I don't have any exact measurement as I will have to go through my old data to sort them out.
My old data was before I stripped them down.

Right now the capacitance between the two 120V winding is reading 1100pf on my meter this is with each winding having about 520 turns each.

Before I stripped off the electrostatic shielding and 6V winding's this capacitance measured to be about 1300pf between the two 120V winding's.

The electrostatic shield was nothing but a 1 inch wide aluminium foil tape around the outside circumference on top of the 120v windings and then the 6V windings were wound on top of that.

It had its own separate connecting wire.
Connecting this wire seemed to add another 150pf to 200pf to the total capacitance of the transformer and this was relavent as it showed up in the impedance tests at 20Khz.

As I had mentioned I did burn this transformer and so it throws off my train of thought for testing and it is now not exactly the same as when I started.
Right now I have duplicated the 26 turn 6V winding with some speaker wire and it show to be about 100pf to the 120v winding's.
This is actually a bit less than what I first measured with the original winding's with the shields in place.
I think closer to half of what it was originally was.

100pf seems high to me for only 26turns but then again it is pvc jacketed speaker wire and is still in zipp form and not separated.
At least it is on the primary side of things.
It is basically 13 bifilar turns in series right now.

jer :)
 

Attachments

  • AS-1206 shield.jpg
    AS-1206 shield.jpg
    95.6 KB · Views: 453
  • AS-1206 fixed b.jpg
    AS-1206 fixed b.jpg
    94.8 KB · Views: 110
  • AS-1206 fixed.jpg
    AS-1206 fixed.jpg
    91.2 KB · Views: 115
  • AS-1206 bad spot b.jpg
    AS-1206 bad spot b.jpg
    205.3 KB · Views: 347
  • AS-1206 bad spot.jpg
    AS-1206 bad spot.jpg
    80.8 KB · Views: 437
  • AS-1206 original b.jpg
    AS-1206 original b.jpg
    120.8 KB · Views: 443
  • AS-1206 original.jpg
    AS-1206 original.jpg
    110 KB · Views: 447
Last edited:
For the record here are the impedance plots for the Antek AS-1206 as it sits right now from the 26 turn winding (6V) and the 120V winding's open and not connected to anything.

The lowest frequency core saturation points are about 300hz to 330hz maximum with a 40Vrms signal applied to the 6V windings.

Some of the graphs for those can be found here but with the windings connected for 12V and some in 6V configurations at different levels -6db apart starting at 20Vrms and 40Vrms,

A TEST JIG FOR FINDING ESL STEP-UP TRANSFORMER PARAMETERS

jer :)
 

Attachments

  • test4 1k a.jpg
    test4 1k a.jpg
    60.5 KB · Views: 160
  • test4 1k b.jpg
    test4 1k b.jpg
    51.3 KB · Views: 135
  • test4 1k c.jpg
    test4 1k c.jpg
    36.2 KB · Views: 117
Last edited:
Thanks geraldfryjr for your advice and pictures. My biggest concern is that how the transformer affects the transparency and whether or not it suppresses the natural dynamics and transient response of the compression drivers. We''ll see.
 
Last edited:
The 0,44m2 thick secondaries will be approx 4 metres long in total (secondaries in series) and the series res of the secondaries (in series) will be under 0,5 ohms (propably under 0,3 ohms). Can this amount of wire cause much of a capacitance when bifilar wound?
A rough estimate would be about 300pF - 400pF. Like I said before, in your application with so few secondary turns and 8:1 ratio it should be fine. In fact, if only two layers would be used when layer wound rather than bifilar there way not be much difference in winding capacitance. Layer winding becomes more important with increasing turn count.

Which is more important in my application, to reduce stray capacitance (= to use normal winding) or to reduce leakage inductance (= to use bifilar winding)? I read that bifilar wounding reduces the leakage inductance to 1/3 - 1/2 compared to normal winding.

Bifilar winding only reduces leakage inductance(LL) between the windings that are bifilar. So, you would be reducing LL between the two secondary windings.
But, what you care about is LL between primary and secondary.
Both LL & winding capacitance are equally important as the extent of the HF bandwidth is proportional to sqrt[C*LL(pri-sec)].

It might likely that both kind of transformers perform well enough (freq response wise) in this application. How about the signal integrity, which winding style sounds better, has anyone compared? Bifilar wound transformers have cleaner square wave response (less ringing) and behave better with pulse signals. I think the common Antek trafos, used as ESL step up frequently here, had bifilarly wound HV-windings?
Clean square wave response is directly correlated with frequency response. It comes from extended flat bandwidth without resonance peak at the top of the bandwidth. As you might recall, the bifilar secondary windings of the Antek can cause problems because of the high voltages generated between the wires which have only their coating to protect them from arcing rather than a nice layer of Mylar insulation tape. This is one of the things alexberg was warning about.
See geraldfryjr pics in post above, or HV test in the Antek thread here:
http://www.diyaudio.com/forums/plan...up-measurements-part-1-2-a-3.html#post2861741
 
Last edited:
Well I can tell you that in order to get a good square wave response you may need a fair amount of inductance on the primary winding to begin with.

Feeding my signal generator into one 120V winding and monitoring the other with I scope I get a very nice looking square wave From 10Hz clear up to 10Khz and not so bad at 20Khz.

But if I feed the signal generator into the 26turn winding then it isn't until when I get to about 4Khz to 5Khz that I get any thing that even remotely resembles a square wave.
Below that all I get is a pulse and the typical sign of ringing caused by the leakage inductance.

Although much of this can be due to my signal generator not being able to properly drive this winding with enough current as well.
I will hook it up to my power amp later and see what comes out and report back.
As I recall the square waves didn't look to bad under those conditions but I forget at what frequency's though.

This transformer did have a pretty low leakage inductance as stock and much of that may have been due to the Magnetic shield that was on it as well.
I still have that shield and I plan on checking out its effects.

I am just a little behind now due to burning the device but I did luck out and was able to repair it.
And since then I have been testing the software I mentioned that I found in order to make this study much easier to learn and understand.

jer :)
 
Shield has nothing to do with the "leakage" inductance which is due to coupling between the windings. Shield are to reduce stray (electro)magnetic field emanated by inductor/x_former/etc.

A shields affect on leakage inductance would be indirect.
Its presence does not directly affect leakage inductance but it may affect the distance between the primary and secondary which will change the leakage inductance.
 
The transformers' specifications that I ordered:

- Core's area ~11,5cm2, core material grain oriented silicon steel, Bmin = 1.7T
- 2 x 30 turn bifilar wound primaries (2,0mm2 wire)
- 2 x 120 turn bifilar wound secondaries (0,44mm2 wire)
- Primay is wound close to the core, secondary on top.
- Electromagnetic shield on the outer rim

I finally received the 1:8 step-up transformers. Primary leakage inductance (primaries in parallel, secondaries shorted) appears to be ~1-1.2µH @ 650khz and the secondary parasitic winding capacitance 570pF (secondaries in series) @ 550kHz. (I cannot change the measurement frequecies.) After I have build the interface boxes I will do some measurements w/wo the trafo's in the signal path. For me it's very intersting to see and hear how the step-up trafo changes the sound and measurements of a dynamic driver, haven't tried this before.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.