Inductance for tweeter OPT

[jnb]

I want to drive tweeters using a 2kHz first order active crossover into dedicated single-ended triode amps. I want a primary load of 6400kilohms and secondary of 5 ohms.

What is a good choice of primary inductance for the output transformers. I am thinking somewhere from 2.5H to 10H?

[d2134]

In an old book I found the formula:

Lp = (130*Ra)/fi

Where:

Lp is the primary inductance in H
Ra is the load in kOhm
fi is the lower frequence (3dB attenuation) in Hz

Lp = (130*6.4)/2000 = 0.416 H

[lt cdr data]

normally, you would want 4, 8 or 12 henries per 1000 ohms primary, but that looks a little different, so wouldn't know.

I would perhaps suggest as its crossing over at 2000 ohms, to say put the -3db on the transformer at 1000 ohms

thus it will be

6400= 2x pi x F x L

so L= 6400/2x pi x 1000

which gives 1 henry

actually, that formula above ain't far off, in fact, its spot on

so anything between 0.5H and 2H

[Bandersnatch]

hey-Hey!!!,
while those formulas deliver a specific number, let's examine what's going on. The -3 dB point is reached because of the shunt reactance between reflected load, and the primary L. I'd be far more comfortable with that elipticle load line happening far away from my cross over point. Something that would deliver marginal Lf performance when used full range, or more like 15 Hy.
cheers,
Douglas

[jnb]

I have been reading the tutorial at Geek's site http://geek.scorpiorising.ca/windingchokes.html

I originally came up with the 0H416 that d2134 came up with but wanted the effect to be 2 octaves lower than the crossover. Geek says the inductance should be 2-4 times higher to compensate for DC reducing the permeability of the core. I would like to use a large enough core to reduce this. I will want it to handle just a few watts. The photo shows a lamination taken from an old SS OPT, the stack for each channel will be just over 2.5cm thick. Will it be large enough?

One aim is to keep it small to improve the high frequency response therefore I don't want to go too high on the inductance if I can help it. Does a 3-2 primary/secondary interleave sound good or could I benefit from 4-3 or 5-4?

[lt cdr data]

15H would imo be complete overkill for a treble amplifier, whilst my knowledge is very limited, that's the value that's pretty common for about 90% of off the shelf transformers for regular amplifiers, for about a 3k ohm load.

It typcially requires around a 2-2.5" stack, but then dimensioning the core and getting good hi F is where I start to fall down, as of limited knowledge.

If its driving treble, obviously, you will want to wind the coil for best HF performance, too, I can't be much help there, but would suggest a more elaborate interleave than the 1st.

If you made it 2 henry, that wuold take your 3db point down to 500hz, which is 2 octaves below, as in the intial posting, 0.5-2H.

I would have thought that would be about right.

[316a]

Originally posted by jnb
I want to drive tweeters using a 2kHz first order active crossover into dedicated single-ended triode amps. I want a primary load of 6400kilohms and secondary of 5 ohms.

What is a good choice of primary inductance for the output transformers. I am thinking somewhere from 2.5H to 10H?

Is your tweeter a dome by chance ? You could probably get away with running a mains toroid instead of a gapped type if you go for a 2nd or 3rd order PLLXO input filter . Bandwidth would be better and the things are cheap , I had good results using a triode strapped A2134 this way running a ribbons from 5khz up

cheers

316a

[jnb]

Originally posted by lt cdr data
If you made it 2 henry,

Well, the geometric average between the largest and smallest values recommended in this thread so far, is 2.5H

I agree that interleaving has to be a priority. Geek suggested that too high a high frequency response may make the amp unstable and require NFB. I hope not, but I think I'm game

Originally posted by 316a
Is your tweeter a dome by chance ? You could probably get away with running a mains toroid instead of a gapped type if you go for a 2nd or 3rd order PLLXO input filter .

When I give my tweeters this particular first order crossover I get an almost perfect combined LR2 acoustic response so I really want to keep it that way if I can, but I appreciate the suggestion.

[Dave Cigna]

My two cents worth:

2H might be enough but 5H would not be difficult to achieve and might be worthwhile. I tend to worry a lot less than some people about LF phase shifts in an ordinary amp (i.e. 20Hz) but I might be more concerned about it in this application. Crossing over at 2kHz means there will still be output below 2k. It might be overkill, but designing for 200Hz should be relatively easy. That puts you at right about 5H.

Your core is much bigger than you actually need. A core is sized to handle the total flux (AC+DC.) The AC flux is inversely proportional to frequency; it goes down as frequency goes up. So, a 2kHz signal results in 1/100 the flux of a 20Hz signal at the same power level. Even if you design for full power at 200Hz you could still get away with a remarkably tiny core. It would only need to be 1/10 the size of a 2W OT designed for 20Hz... that is if you didn't have to contend with DC flux.

But, even the DC flux will be easier than with a 20Hz transformer. You will have comparatively few turns on the primary so the DC idle current will produce proportionately less DC flux in the core. With a core that big you can probably get away with a butt gap; no actual spacer between the E's and the I's, just stack them separately and butt them together. The butt gap makes it easy to get 5H with comparatively few turns of wire.

-- Dave

[jnb]

Originally posted by Bandersnatch
I'd be far more comfortable with that elipticle load line happening far away from my cross over point.

Agreed, and I might settle on Dave Cigna's suggestion with 5H being a factor of ten over the bare minimum.

Originally posted by Dave Cigna
Your core is much bigger than you actually need. A core is sized to handle the total flux (AC+DC.) The AC flux is inversely proportional to frequency; <snip>

This puts my mind at ease, very good news.

I went to see a motor rewinder. Not only is it good to buy something from someone who knows what Henrys and Ohms are , but it was much cheaper than buying magnet wire from a chain retailer.

Not only that but it seems to be top quality wire, double enamelled. I asked the guy if he had any solid core wire for wiring valve heaters and he said it's not too common these days but why not use the winding wire I bought for the secondaries. I said I want to twist two together using an electric drill and he said no problem, he could take that, put it on the bench and hit it with a hammer and still guarantee the insulation for kilovolts and 185 degrees Celsius. And who am I to argue with his decades of experience

There's around 200m of 0.28mm and 50m of 1.5mm.