Rerurn to Romy the Cat's Site
In the Forum:
Melquiades Amplifier
In the Thread:
An ultimate transformer for narrow bandwidth?
Post Subject:
Re: Q: aircore transformer
Posted by
cv
on: 2/14/2005
amlodipin sandoz
amlodipin teva
inetapakistan.azurewebsites.net
symbicort inhaler
symbicort generic alternative
igliving.com
Alas, no time for anything at the moment, the listening valuation must await construction of the amps (and speakers...)
Also, it is a very custom application - it is designed to drive an array of ribbon tweeters. Normally these have matching transformers on the back of them, to give an 8 ohm load from the ribbon, but I'll be ditchng those and connecting the ribbon elements in series and driving direct.
So the transformer is something like a 3k:0.5 ohm device and I won't have any real reference point. Well, I suppose there's the one ribbon running with the mumetal OPT. Guess this conflicts with the Romy dictum of "thou shalt not fix what you are not sure is broken", but hell, my gut tells me this is the way to go.
Anyway, the rolloff at LF will probably just be 1st order (the equivalent circuit looks like a simple LR filter). For a 6C33 it's complicated by the fact that the winding resistance is comparable to the internal impedance of the tube (thinking about it, you would want a low DCR transformer for the 6c33 otherwise you might end up dropping 50V across the winding...) and that resistance is distributed throughout the winding - not immediately confident that the rolloff is still exactly 1st order.
What I will do when I get some time spare is take some measurements using a 100ohm source; this will then tell you what the response of a similar design in Melquiades might look like.
There is no reason to stop you running the OPT in its lower rolloff; actually, in that respect its better than a cored device. Distortion will increase as the tube sees a lower load (think of very low LF - all the tube sees as the load is the primary winding resistance), but unlike a cored device, there is no additional distortion from the core itself overloading. So as long as you have plenty of "power headroom" in the amp output stage, you should be alright. With my planned ribbon amps, this is a deliberate feature; the amp will be bandwidth limited by an LR filter at the input (actually a tiny little input special input transformer I wound myself on some incredible cobalt amorphous material, -3db is in the MHz range with just a single layer winding) - this will give the primary 1st order rolloff at 1.5k or so. Then, the OPT stage will have another rolloff at ~600Hz which will give the ribbons added protection. Because the 1.5k cut is before the output stage, this means the amp won't see much signal in the lower LF range, thus the tube won't be too stressed. Oh, the input transformer means I can hang the driver stage on a negative supply and direct drive the output valve. No coupling caps... I suspect you could use the same approach on Melquiades.
Now, as to why I think the aircore approach is interesting; in any cored transformer, you have eddy current losses. These increase with frequency; basically, the changing magnetic field induces currents in the core, which sap the signal.
This is why cores are laminated and (theoeretically) the thinner the better; this increases the resistivity of the core (the lamination have a thin layer of insulation applied). I say theoretically, because I know people who have listened to 2 almost identical devices differing only in lam thickness and preferred the thicker stuff.
Anyway, it's one big advantage of amorphous, and why very high frequency transformers are made of ferrite (lots of tiny particles glued together - high resistivity).
Thankfully, the losses also reduce with flux, which goes inversely with frequency, otherwise we'd get not much power out of transformers at all at HF.
Nonetheless, the losses are still there. Also the permeability of the core is a function of frequency, all of which combine to produce something not quite ideal. I dislike the idea of a loss that varies with signal.
With the aircore, the only "dynamic" loss mechanism is dielectric, but most of this can be mitigated by using a thin walled tube for a bobbin, or even having a stack of self supporting coils.
Further, there are less potential mechanisms for low level dynamic loss.
Anyway, must run, I will see if I can get those measurments sorted this week and also have a think about how one might optimise the aircore design to be more suitable for the 6C33. I for one am curious as to whether the existing top end rolloff of -1db at 20kHz (with 600 ohm source) will improve.
cheers
cv
PS I listened again to that ORTF Scherchen Mahler #5 you recommended and it suddenly clicked; dunno why I didn't quite get it first time. It's terrific- many thanks for the tip.
Rerurn to Romy the Cat's Site