DIY line level transformers

Status
Not open for further replies.
Has anyone here attempted to build line-level transformers, or know of some resources for this?

My transformer-building experience has been for switching converters, which are not really conducive to hifi🙂. However, I have some feel for controlling leakage inductance, the use of interleaved or multi-filar windings, etc. However, I don't have any experience as to how any of this might be important to audio.

What I'm also not so sure about is things like proper selection of core materials, etc. I have access to a broad variety of ferrite cores in various shapes and sizes, mostly materials geared towards power conversion, but there are some signal type materials available. In addition, I know where a lot of tape-wound cut-C cores are collecting dust, but there is little or no info on their material type. Probably something geared for 20KHz or above. Most of the C-cores are oversized for audio signals, but perhaps this is an advantage?

Why am I doing this? I have a Tripath TA0104A eval board left over from a work project that I would like to make into an amp. I would like to have the option of balanced inputs. Originally, I planned to put together a INA103-based input stage to accomplish this, but then I thought that with all this core material laying around, maybe a DIY transformer would be the way to go. Maybe I'm losing it 🙂.

Any help would be greatly appreciated!

Thanks,
John
 
Why nickel?

Some of the ferrite cores are nickel-zinc-iron ferrites. As far as the tape-wound cores, some of them may be permalloy, which I believe is mostly nickel (this is from memory). If I'm lucky, there may be some Metglas cores floating around, but they are probably not in the junk pile.

John
 
John,

try to get some answers from Dave Slagle (he can be found on the SET Asylum). He customs-winds trannies and seems to have advanced knowledge, Gosh, wish he would join us here at DIYaudio.

Still new with transformer-coupled amps myself, i already made some observations, and one is that 9 out of 10 famed audio signal trannies are over sized C-cores, the fancy Tango (potted) and uttely ugly Lundahl (unpotted) among them. As long as the winding finds place on the former and does not rattle around, oversized core is a good idea IMO.

C-cores are close to toroids in their properties; DIY toroid signal transformers have proven to work exceptionally well but unfortunately noone manufactures them. Allen Wright used a power toroid trannie as cheapskate output transformer of his tube power amp and my friend Manfred Huber did the same with a transformer-coupled tube line stage.
Does not surprise me, the toroids are very effective magnetically, they have wide bandwidth, they have small coupling capacities because they use not as much copper, they have nearly no stray field.

Unfortunately they do not take any DC, they saturate at once. And this is where the C-core steps on the stage: being almost as good as the toroid, it can be gapped and so it can be adjusted to take any DC needed at the price of inductance. Moreover, the usually have two separate windings, and if no interleaving windings are used or if electrostatic shieled foil windings are used, the coupling capacity is close to zero. Near to impossible to realize shield foil windings on a toroid.

If your source of dust-collecting C-cores has more than you need, I would be interested in them as two buddies of mine already have wound trannies with good success and i would like to have them make me a special interstage trannie. if those C-cores are meant ot be for power applications, i would be interested too, as i need trannies with shieled winding for my directly heated PP amp's different heater supplies; C-cores would be very welcome!


Nickel increases the high frequency response.

Yes, Jocko,

and a bit too much Nickel brings the core's capability of handling imbalanced currents down to zero which in many cases is undesirable.

TMK; Lundahl uses Nickel for some of his small signal trannies such as the LL7901/7902/7903 and also for the LL1660 and 1635.

It seems to be very important in the audio realm that the sheetmetal used for the core is as thin as possible (Eddy currents already an important influence at those frequencies??); Simon Shilton, Per Lundahl and also the famous Tango brand uses 0.1mm thin core sheets.
 
Jocko, Bernhard,

Thanks for the info. I'll peruse the SET Asylum for a while to see what I can pick up, and see what meaningful question I can come up with.

It looks like it's time to head into the "tunnels" to get a closer look at the cut C-cores. It's been a while since I had a look. As I recall, they were not labeled very well, and I may have some difficulty finding out what materials they are.

What about ferrites? These are far easier for me to obtain, and I will know what I am getting. I realize that the permeabilities are probably not as high, and they saturate relatively easily, but any idea how I might expect them to sound?

Regards,
John
 
Ok, I did a little poking around. The C cores that I had come across some time ago had been mostly scrapped 🙁. However, I did get a few (3), and the person I got them from says they are most likely silicon steel, or perhaps Orthonol. The laminations appear to be 2 mil. I don't know that these are good candidates for anything, but I'll probably hang onto them for a while.

However, I did also get a few Permalloy 80 toroids, about 2.5-3cm diameter. I wound ten turns around one and measured 1.2mH @ 1kHz, for an Al about 12uH/turn^2.

I decided that I wanted about 100kohms worth of magnetizing inductance at 20Hz, which comes out to about 800H! This seems really high. This comes out to about 8140 turns for each winding! I see why these line input transformers all have more than a kilohm of resistance, and suddenly those Jensen transformers don't seem like such a bad deal after all! The idea of winding two 8000 turn windings on a toroid does not seem like fun.

I guess its time to look at purchasing transformers or revisiting the INA103 (or similar).

Any comments appreciated.

Regards,
John
 
Now you see why they cost $65 or so. A buddy has his custom made, and they run about $200.

Have you looked at the SSM-2142, etc. that AD makes? They are designed for balanced TX/RX applications.

Transformers do a good job of keeping out RF, which is essential.

Jocko
 
Status
Not open for further replies.