TVC preamps: good or bad?

Dgarretson, of course I did have to design for the real world, which includes not limiting myself to very low output impedance sources. In fact I do have one input from my input selector that is fed a high 6K output impedance coming from my phono stage. The thing you have to do to make that work is increase the number of winding turns and use that slightly oversized core of laminations.

That increases the LF inductance which pumps back up the LF extension. With that 6K input, the frequency response does reduce with LF -3dB cutoff at about 15 Hz. Also, there will be more FR deviations and some HF ringing as well as increased distortion. So I was forced to design accordingly.

That's why it's important to optimize the core size and the core material and the number of windings for these designs. For one thing, I believe I wound twice as many turns than Dave Slagle's AVC. More turns creates other problems, like possibly degrading performance for low impedance sources by increasing winding capacitance. The capacitance is what limits the HF cutoff.

I built a MC phono autotransformer step-up on the exact core form and material as the AVC and it turned out to be the best I've heard of any step-up as well (total serendipity for me). It's driven by a very low 4 ohm output MC cartridge (Koetsu Urushi) and is stepped up 1:10. Low impedance like that really drives down distortion. In fact the laws of physics says that nonlinear distortion approaches zero through a transformer as the winding resistance and driving impedance approaches zero.

But I think SuperPerm 80 is the good stuff responsible for making most of the low distortion and good sound. It's very good at low magnetization, and with copper windings it's slightly warm and sweet, but VERY slightly. Almost dead nuts neutral, but I don't think much of anything is truly dead nuts neutral. And the sound is extraordinarily detailed (transparent) and it actually gets more dynamic as you turn the volume down (increases turns ratio). I have an all Holco (the good old Holcos) 100K L-PAD passive step attenuator on the front end of a 27 line stage (now used as a haedphone amp). It's vastly better than the volume pots I had before, but inferior in every way to the AVC.

I've put the AVC in front of the 27 tube with the step attenuator at max position (0 series resistance and 100K shunt.) This is the best way to eliminate the sound of the step attenuator for trying out how the AVC would do as part of an active tube preamp. Well it wiped clean away the sound of the high quality Holco resistors at all levels. What I was left hearing was the sound of a very good tube, the 27, and its coupling caps. It was dissapointing really. How? Because it sounds so much better to just eliminate the 27 tube circuit and clean it up just that much more. I mean it sounded best standing alone, and it wasn't sounding too "sterile" either.

That's my story and I'm sticking to it.

Kurt

Any way you will make these or do a how to or sell directions? Im tired of pre amp swapping. Great work

The ASL Flora uses autoformers. Audiogon member Teajay wrote a review of his that you can look up here. It replaced a Bent Audio S&B TVC in his system (which he also wrote a review for).

Bent Audio is now using a custom version of the Slagle autoformers instead of the S&Bs. According to John Chapman the first few units are now in production and as soon as he stocks up on parts he'll be ready to go to full production.

Ktstrain, nice work on your AVC. I'm looking to have a custom one built for me that will include a phono stage (most likely the K&K Audio kit). Hopefully I'll have the details all worked out soon. If you have any advice for me on design feel free to email me.

That's interesting news. I bet those Slagles have been improved since the ones I owned. It really does sound better than S&B. I give it my recommendation without even hearing the newer version. Dave is good.

If you want to roll your own from my recipe, I looked around and found some info for for making these items. You too can have a $2K performing pair of AVC's and $3K performing pair of MC step-ups (judging with my ears against other products I heard on the market) for about $400 in parts, and have enough to do another pair of all of them for a buddy. But there's some difficulties to overcome.

First is finding a decent coil winder that works, one you can borrow if not own, capable of letting you handle 32 AWG wire if you have a careful smooth winder and a steady hand (and patience). This must have a turn counter because you will need to know the exact turn count all the time. Maybe you should stay with 24 AWG to prevent awkward motion from breaking the wire. That's still very good wire if it's Cardas, for example.

I recommend getting really quality magnet wire, preferably Cardas. It actually does sound smoother, more detailed and a little sweeter than Rat Shack type cheap magnet wire. And pure copper is always my favorite even if my Urushi cartridge uses silver clad copper (but so little of it).

You will need to buy about 500 ft of wire to get all these projects done, maybe 1000' to be safe if you're a klutzy winder just learning.

Now to source the hard pieces: The laminations - I can only recommend Magnetic Metals lams (www.magmet.com). Ask for a couple of lbs. of 21E1 laminations of grade 8406 and another couple of pounds of 21E1 lams of grade 8414. This will get you SuperPerm 80 (AKA HyPerm 80 in generic sense), but don't trust this to any other manufacturer because I cannot account for slightly different formulations that could be important.

For the bobbins, get them at Cosmo Corporation (www.cosmocorp.com). The part number to buy is 1-1254-0-000-00-0. Description for this is "for lams EI-21 x 5/8".

For the channel frames, go to Bahrs Die and Stamping Co., Inc. (www.bahrsdie.com). The part to order on their website is "1/2 x 5/8 channel frame", style 1. You might have to speak to a representative for these items. A lot of people want to get them by just asking for free samples. You need to be a corporation to do that and be nice following the rules.

I recommend a full single winding of 2000 turns. For -3 dB, you would send the input to the top of that coil, the 2000th turn, and then tap out on winding 0.707*2000=1414 turn. I recommend taps every 2.0-2.5 dB apart. Less than that and there's too many taps that will make a horrible rats nest out of it along with too many switch positions required. More dB apart means too coarse a series of volume steps for most people. You will need to do the proper math and have all windings taps pulled out and labelled appropriately so you don't lose where they go.

The taps can be free of splices by running a twisted loop out from the transformer at the tap position, and use a solder pot to melt the magnet wire insulation at the end.

Use a quality switch for these. I went to Michael Percy to buy a stereo ladder multiposition switch. The cost was a bit high, but so well worth it with a very high quality contact and smooth operation and lasting durability. I forgot the name of those switches he sells.

Plan ahead accordingly. I chose to wind up from the low taps on the inner core and moving out from there. The first tap was just one turn, or 1/2000th voltage out as voltage in. Amazingly excellent sound from that huge ratio. Wind end to end along the width of the bobbin and try to remain flat. The beauty of this simple winding scheme is, well, it's ease of construction, and you still get all the bandwidth. No fancy winding technique that amateurs can't try for themselves.

For the 1:10 MC step-up, use the same materials but wind 200 turns for the input (a tap out to the input) and then wind the rest of the 2000 turns for the output. Place these tranformers in a magnetic iron shielding box and maybe add a big extra layer of unmolested mu metal for a second magnetic shield. This is where microvolts of signal are happening and magnetic shielding is the challenge here.

Stacking the lams is another new trick to learn, so be patient on this. Again I recommend using 5 of the thin 8406 lams and then 1 of the thicker 8414 lams. Repeat this pattern all across the transformer.

You do not need to seal or varnish the lams after they are installed. The high nickel lams pretty much won't corrode in a home environment that steel lams could.

Good luck if you try this project.

Kurt

Dgarretson, I agree, it seems that TVCs have a number of advantages over straight passives. One of them is low output impedance, essential for interconnect control.

Kurt's approach optimizes the TVC approach, but at the price of his own admission, that you don't design for the real world. Having sources that have the ability to drive low impedances on their own really helps you with performance if you optimize the TVC for that.

Obviously, this approach is no good for production. In the 'real world' you have to accommodate the output impedance of a simple tube phono stage, which might be rather high. Its tricky to wind high impedances on the primary side while getting good bandwidth- in fact you can regard them as mutually exclusive.

On this account is why this is an on-going conversation that will likely never be solved. IMO its easier to build a high performance tube line section (we can reproduce a 10KHz square wave quite nicely too, and do a 20Hz square wave with no measurable tilt) than it is to get a transformer 'right', and still work with the real world.