Roll your own Output Transformers

[MatthisD]

Just thought I would share since we're on the topic of making bobbins. Just had this 3d printed for me, waiting on wire and a winding machine to be built.

Woe!  I don't know anything about 3D printing other than it exists and that making custom bobbins would be fantastic. I struggle to make decent bobbins, its not too bad for large cores but really tough for small ones. Did you need to draw this with software yourself?

 

[gemini86]

It will eventually be a v72 plate choke. I used google sketchup to make the model and then export to .dae file. Then head over to shapeways.com and upload your model for an instant price. They do large batches all at once, so it takes about 2 weeks to get it in the mail. I actually screwed this one up. It should have been .814 inches tall, when instead i made it. .841 inches. Also the wall thickness could go up some. I'm thinking about just making it work.

 

[abbey road d enfer]

How much did they charge for that one? (if it's ok to ask)

 

[gemini86]

It was like 5.80 something plus shipping. Worth it, imo.

 

[abbey road d enfer]

It was like 5.80 something plus shipping. Worth it, imo.

That's a very good price. We have a number of prototypes done in stereolithography; each cost about $400; admittedly,  they are bigger than your bobbin, but I think I'll ask them for a quote.

 

[gemini86]

wow... That's some serious bones. You can also get it in a bunch of colors as well. If I decide to start making more, I'll probably go with a cool red or something snazzy.

 

[CJ]

if you have any issues with flanges warping you can fab some inserts.

also, a piece of white or yellow paper background will help you spot the wire>

 

[All Ears]

I am now planning on winding a bifilar version of the 2503  (k.i.s.s principle at work here)

I have been told that multifilar winding can increase the capacitance.

Are there any practical methods I use during winding to reduce the capacitance while still retaining all the benefits of the bifilar/multifilar winding ?

Cheers

 

[joaquins]

for output multifilar tx capacitance may not be a problem, in a 1:1 capacitance will carry a lot of HF that you want, with  different ratios capacitance may help.

Also for such low Z at output stage ussaly seen with a 2503 5nF or 6nF you'll get won't be so much.

My advice, don't worry about it for this task, worry for it in an input tx.

JS

 

[gemini86]

In this case the capacitance is a feature, not a bug...

 

[abbey road d enfer]

I am now planning on winding a bifilar version of the 2503  (k.i.s.s principle at work here)

I have been told that multifilar winding can increase the capacitance.

Are there any practical methods I use during winding to reduce the capacitance while still retaining all the benefits of the bifilar/multifilar winding ?

Cheers

Unfortunately there is no magic method for significantly reducing capacitance in multifilar winding. You have to think about what you intend to do with the resulting xfmr.
Parasitic capacitance in multifilar xfmr's can be a nuisance, affecting the final performance. You have to consider the xfmr itself AND the attached circuit.
In most cases, the a large portion of the inter-winding capacitance will be loading the output stage, which puts a strain on its stability.
I was first attracted to this phenomenon when I looked at Jensen's application note
http://www.jensen-transformers.com/as/as025.pdf
Note the ferrite bead at the buffer's output.
In the case of a 1:2 application, the advantage of better coupling is somewhat countered by the increase of capacitance, resulting in a sharper roll-off, albeit at a higher frequency due to lower leakage inductance.

 

[All Ears]

Ok - thanks guys - so it looks like it is something I can't do much about interwinding capacitance nor worry too much about in a 1:1. Certainly mic input would be another case entirely (I am not sure if I dare to attempt to wind my own mic transformers at this point)

In that link from Abbey Road I see something Jensen terms an "output load isolator"  - it seems its function is to reduce effects of cable capacitance on amp stability

http://www.jensen-transformers.com/datashts/oli3.pdf

According to the notes, it does not affect audio band.
Does anyone have any expreience with these ? What are they actually doing ? Why don't we see this more often given that it is only a resistor and a very small inductor in parallel and it protects against RF interference ?  It actually looks like the inductor is actually wound over the top of the resistor as a former.

I see them being used in the SCA API clone going into the output trafo:
http://www.seventhcircleaudio.com/A12/A12R17/docs/a12_sch.pdf

As usual, every answer raises another question.

 

[abbey road d enfer]

According to the notes, it does not affect audio band.
Does anyone have any expreience with these ? What are they actually doing ? Why don't we see this more often given that it is only a resistor and a very small inductor in parallel and it protects against RF interference ?  It actually looks like the inductor is actually wound over the top of the resistor as a former.

In other app notes from the Jensen site, they are refferred to as "ferrite beads". Anyway, they are are inductors with damping provided either by an actual resistor (inductor wound on resistor body) or by the equivalent loss resistor (ferrite bead). If the inductor was not damped enough, the combination with the load capacitance would create a severe impedance dip that would be problematic for the output stage.
It's a "one-size-fits-all" compromise; thorough optimization would involve tuning the inductance and the resistance for B'worth or Bessel response.
The "winding-over-resistor" construction is quite common in power amps.

 

[bruce0]

Just thought I would share since we're on the topic of making bobbins. Just had this 3d printed for me, waiting on wire and a winding machine to be built.

Rodney:

Very cool! Is the part at all flexible? I did a 3d part (not with that company) and it was brittle.  You might want to wind with "walls and inserts" (see CJ's winding picture) because the winding tensions compound to make some pretty impressive forces.

b

 

[CJ]

you want the capacitance of the quad windings to pass the high end,

this is how the 2503 works, they have the C take over right when L craps out,

it is a tuned coil, 2.5 turns per inch times 100 feet = flat from 10 hz to 1,000,000 hz,

think of this xfmr as a piece of audio coax good to 1 meg,

there is always a price to pay when improving one aspect of a transformer, in this case, it is phase angle, but do we care about phase at 1 megahertz we think not,

 

[gemini86]

Just thought I would share since we're on the topic of making bobbins. Just had this 3d printed for me, waiting on wire and a winding machine to be built.

Rodney:

Very cool! Is the part at all flexible? I did a 3d part (not with that company) and it was brittle.  You might want to wind with "walls and inserts" (see CJ's winding picture) because the winding tensions compound to make some pretty impressive forces.

b

It's actually very flexible... That's kindof a problem here, because as CJ has pointed out, the copper winding will tend to push the sides out. I'll need to use spacers and stuff to hold everything together during winding and then wax or varnish the sections to glue them in place.

I'm not sure what the material is, but at shapeways.com they call it "white, strong and flexible". It can even be used to make wristbands and stuff.

 

[All Ears]

I see - got it - thanks CJ

I understand Gemini's comment now: "In this case the capacitance is a feature, not a bug..."

Perhaps we should refer to these audio transformers as "iron filled caps"

At what point does "L crap out" ?  Presumably somewhere between 10k and 20kHz ?

 

[abbey road d enfer]

At what point does "L crap out" ?  Presumably somewhere between 10k and 20kHz ?

No rule of thumb here. It depends on several factors. Let's assume 1:1 ratio, 600r load and nominal inductance 0.5H. A cheap xfmr with coupling factor K=0.992 (Ll=4mH), HF corner will be just about 15kHz. Typical of two-layer construction (power transformer style).
Medium quality with k=0.995 (Ll=2.5mH) will be 20kHz. Typical of 3 or 4-layer construction.
A more expensive xfmr with k=0.998 Ll=1mH) will go up to 50kHz. Requires some serious sandwiching.
Bifilar winding yields typical k=0.9997

 

[CJ]

after about 1 k hertz the inductance starts to really drop sharply due to core loss,

this is for a core made from .014" EI  lams (Ni or steel, does not matter much),

but even if you only have 10 henries left out of 300 at say, 5000 hz,, you still have

XL=6.28*5,000*10= 314,000 ohms reactance

so as L drops, w , (2 pi f),  at 5 k Hz will generate enough ohms for a tube load,

here are some curves from a 300 Henry plate choke - L vs f,

the top curve (red) is achieved by stacking the core in a different way.

 

[All Ears]

Hmmm...the plot thickens

I think I "knew" more about transformers before I started asking questions. Now I know I know nothing !

In that graph that is an amazing difference in response achieved simply by different stacking.

But speaking of freq response....I have also become intrigued by many of the "less desirable" audio transformers I see around for resale. I am not sure if these transformers come from old radios or TVs or military or whatnot but I have seen oodles of interstage or line matching transformers (some gapped, some not) in widely varying impedance ratios with quite limited frequency responses.

Eg - you might see  these 60's or early 70's era interstage transformers from companies like Bendix, Triad, Freed, HP et al in their UTC style looking shielding cans with freq response marked on the can as [200Hz - 4kHz]... or perhaps [100hz - 10kHz]...these seem to be two common ranges marked on some audio trafos I have seen (many also seem to be marked with a maximum operating aeronautical altitude (!) in feet).

Now this intrigues me - have they been able to band-limit those transformers for a particular purpose ? How much are those specs printed on the side accurate ? Do those freq response specs only mean something when the trafo is in the circuit they were designed for ?

For example - I have here a Triad "Line Matching" transformer in a black metal can. It seems to be a type that is common to many manufacturers (radio?). It has four coils as follows:
[a)150z b)150z c)150z DC capable 50mA d) 600z centre tapped] and the freq response is marked as 200 - 4000k and power is listed as 6mW. Not sure why they say only one coil is DC capable - as all four coils share the same core which must be gapped. Anyone know what this type is for ?

Now band-limited trafos are not going to excite most studio gear builders round here ...but it interests me. I am thinking some of these cheapies might give a pretty nice  band-limited, saturated sound for some purposes  - they might also sound crappy...but I am interested in exploring different tonal options with these kind of things and even possibly switching between trafos in circuit for tonal colour changes.

Any thoughts ?