Another DIY Ribbon thread

[dhenryp]

I got my two types of ferrite toroids today CWS Coil Winding Specialist and tried winding some transformers. One is item F-150-W: $5.00 each (I think the code is Ferrite - 1.50 inches OD - Type W composition) and the other is F-114-H : $2.65 each (Ferrite - 1.14 " OD - Type H). Both have a .75 inch "hole". I first tried winding them both with 37 turns on the primamry and 4 (or maybe 5) on the secondary. I put 20 uf capacitor in series. The primary is a dead short at low frequencies so you need some sort of high pass filter. All I have on hand now is 16 gauge magnet wire so 37 turns the maximum I can put on one layer. I've got some smaller wire coming so I'll do some more experimenting then.

Both of these worked, but no well. I could drive both directly from my amp (i.e. no series resistor) and they sounded OK at low to moderate levels. Above this, it distorted badly. After trying various secondary winding turns I hit upon: 37 turns primary - 10 secondary using the F-150-W. This works really quite well. I can play it very loud with no really obvious distortion. This is not a very scientific analysis but it's the best I can do for the moment. Turning the amp almost to the max, the driver was too loud to comfortably listen to at three feet. I set up my Radio Shack Sound level meter about three feet from the ribbon. The Meter showed peaks of ~ 105db. I had it set at A weighting, fast response time.

I need to find a good book or website to help me understand better how to determine both the turns ratio and the number of turns. Below is the picture of my Version 1 transformer hooked up to the ribbon tweeter.

Denis

[tiroth]

Aren't audio transformers usually made from steel laminations or even permalloy/amorphous cores? I know you only need ~1kHz and up, but ferrite has extremely low permeability....W composition is u=6, permalloy is around 20,000 I think. Primary inductance must be very low as a result.

http://hyperphysics.phy-astr.gsu.edu...ic/indtor.html

I also think the turns ratio can't be optimized for a normal amp. Your ratio is 3.7:1 yielding an impedance transformation of 13.69:1; I wouldn't think the dynamic impedance of the system was above 1 ohm + the capacitive reactance.

If it works, it works though. I'm just armchair quarterbacking. You must have a forgiving amp.

You might consider looking into those small PC mount Talema toroids that Digikey sells. I think they are only about $9-11. I have used them for audio and >1 kHz the distortion is -100dB down. The 1:2 voltage ratio model I used was flat 20Hz-20kHz small signal. I believe they have ratios up to 48:1 if you do series primary and use a single (or parallel) secondaries.

Another idea for cores would be to get the Edcor line transformers and remake them.

[Variac]

Someone making stuff- how refreshing! I think in the other thread there was a picture of a matching transformer made just this way.

Lead on good man!!

[tiroth]

Repost of Talema data from my samples. I can send you a 35VA unit if you want to experiment. Maybe the leakage inductance is too high though. The smaller cores are way better on this measurement.

1.6VA as line input
==============
2:1
Primary inductance: 4.5H
Primary interwinding capacitance: 200pF

with 13k secondary load
-0.5dB at 20Hz
-0.3dB at 20kHz

The bad news is that distortion is on the order of 0.6% at 20Hz, -0.05% at 250Hz, 0.001% at 500Hz, and unmeasurable at higher frequencies. This would be excellant as line input for a tweeter amp, but the primary inductance just doesn't appear to be high enough for the low frequencies.

35VA as OPT
=========
27:1
Primary inductance: 146H
Primary leakage inductance: 10.5mH
Primary interwinding capacitance: 1.4nF

[LineSource]

I have built amps that can direct drive 0.1 ohm Apogee ribbons and was not happy with the sound. I then converted to a 5.8 micron ribbon that increase the restance to 0.35 ohms, tripled efficiency, and it sounded wonderful with the direct drive amp. Much better than with a transformer.

Sputtering very thin aluminum on a 2 micron thick film will create a more physically robust ribbon than pure aluminum, but also a low efficiency ribbon relative to its weight since the electrical resistance will be much higher than 100% AL. Most ribbons are also corrugated(pleated) to increase horizontal stiffness. A thin film ribbon may not hold this corrugation, and the ribbon is likely to "fold" from the magnetic field gradients.

Most commercial tweeters seem to use 8-10 micron aluminum, as this holds a corrugation well and is a good compromise between weight and physical robustness. 5.4 micron thick pure aluminum foil is thinnest I use on long, wide bandwidth ribbons.


Here is some resistance data for pure aluminum for 0.5 x 36" ribbon, where I assumed 42" of aluminum to account for the corrugations and connection ends.


microns ohms for 0.5"x42" AL
2..............1.0600
3............. 0.7067
4............. 0.5300
5............. 0.4240
6............. 0.3533
7............. 0.3029
8............. 0.2650
9............. 0.2356
10........... 0.2120
11........... 0.1927
12........... 0.1767
13........... 0.1631
14........... 0.1514
15........... 0.1413
16........... 0.1325
17........... 0.1247
18........... 0.1178
19........... 0.1116
20........... 0.1060
21........... 0.1010
22........... 0.0964
23........... 0.0922
24........... 0.0883
25........... 0.0848

[Stocker]

perhaps this is just a ribbon-noob question, but why not direct drive this low impedance from a low voltage, high current amp?

[CV]

Nice ribbons there!

I think the distortion may be the ferrite saturating or the transformer not presenting enough inductance. You need to consider 2 things:

1) A minimum number of turns to avoid saturation:

The equation is V = BANw
where
V is the peak primary voltage in volts
B is the flux in Tesla (probably want no more than about 0.5T for ferrite - check core spec)
A is the cross-sectional area of the core in *square metres*
N = number turns

30 turns is prob about the minimum for a 2cm^2 core area

2) The primary inductance - this will be equal to N^2 * the specific inductance of the core, usually spec'd in nH/sqrt(turns)
The transformer presents a load to the amp equal to the reflected load [ (ratio^2) * ribbon resistance] in parallel with the primary inductance.

For an 8 ohm load (say), you probably want the inductive component to be 5 or 10x greater at the lowest frequency of interest so that it is insignificant compared to the real load.

So if you are crossing over at 1kHz, you probably want an inductance of 6-12mH.

All very rushed and from memory so please double check!

Other things to consider are very low DCR secondaries and connections to the ribbons.

Cheers

[dhenryp]

Originally posted by tiroth
Aren't audio transformers usually made from steel laminations or even permalloy/amorphous cores? I know you only need ~1kHz and up, but ferrite has extremely low permeability....W composition is u=6, permalloy is around 20,000 I think. Primary inductance must be very low as a result.

http://hyperphysics.phy-astr.gsu.edu...ic/indtor.html

I also think the turns ratio can't be optimized for a normal amp. Your ratio is 3.7:1 yielding an impedance transformation of 13.69:1; I wouldn't think the dynamic impedance of the system was above 1 ohm + the capacitive reactance.

If it works, it works though. I'm just armchair quarterbacking. You must have a forgiving amp.

You might consider looking into those small PC mount Talema toroids that Digikey sells. I think they are only about $9-11. I have used them for audio and >1 kHz the distortion is -100dB down. The 1:2 voltage ratio model I used was flat 20Hz-20kHz small signal. I believe they have ratios up to 48:1 if you do series primary and use a single (or parallel) secondaries.

Another idea for cores would be to get the Edcor line transformers and remake them.

The "W" material has a permeability of 10000 and the "H" material has a permeability of 15000. I agree that the 3.7 turns ration is way too low but it's the best I could come up with last night with the wire I had. Here are the specs:

http://www.cwsbytemark.com/cws_catal...gproperty2.pdf

The F-150-W as an AL of 8800 mh per thousand turns so my input inductance should be 8800 * 37/1000 = 325 mh, if I understand the specs correctly.

I'll take a look at the transformers you mentioned. Maybe they are what I need and would not take any extra work.

Thanks,

Denis

[dhenryp]

Originally posted by Stocker
perhaps this is just a ribbon-noob question, but why not direct drive this low impedance from a low voltage, high current amp?

With a ribbon of .02 ohms, you would end up losing most of your power in what is usually insigificant resistances like speaker wire, interconnects etc.

[dhenryp]

Linesource - I'm using .0005 inch aluminium which (I think) works out to 12.7 micron. Based on your very handy chart, my resistance for a 7" strip (an 1.7 inch to account for corrugations etc.) should give me ribbon resistance of ~ .028 ohms - pretty close to my assumption.

Thanks for the info.

Denis