Line Transformers for Digital Audio

My latest blog upload chronicles my use of line transformers between a CD player or DAC. I was inspired to do this in the 90s after seeing an ad in the pages of MJ Audio Technology magazine.


I started out with this pair of surplus Tamura transformers wired 300:600.


Even if my DIY DAC has an all-tube analog output section, the use of these line transformers still make digital sources more enjoyable!

Here's a video as proof of concept.

 

I'm lazy, and I'm not going to click through to your blog or watch a YouTube video just to get the gist.

Can you give us the CliffsNotes on why this is desirable?

 

Since the 90s, Asian audio hobbyists have been using line transformers to tame the analog output section of DACs, which is usually populated by op-amps.

The line transformer can be used as a 600:600 shunt or wired with a mild step up in the primary. It's an easy DIY project. All you need is a pair of line transformers that has decent audio bandwidth.

 

You won't get the gist from his blog or the video anyway. The highlight of the blog is "I found my CD's a lot more enjoyable" and the video is one of those "record my stereo system playing music" type videos... so no real info there either.

 

Unless you need isolation or impedance matching (or some strange idea of gain regulation), I don't see any way they are beneficial in analog audio. Especially 1:1 transformers - only for isolation. Otherwise they just add noise, no matter how little if they are quality made.
Transformers in digital audio are pulse transformers, like in network switches. Those serve isolation purposes. They can't do reclocking or anything like that.

But hey if you enjoy it more, feel free to put whtever makes you happy in the chain, right?

I can see no reason why this would help in literally any way, unless you have to break a ground loop. Do you have any idea as to what the "taming of opamps" means? I'm not sure using transformers reactivity as lowpass filters works in this case (or at all). I'd say it's the case of "opamps are bad and digital, look I put this nice expensive quality analog component here to make it less digital and more analog". Maybe these folks had a bad case of ground loops.

 

The noise probably just adds color that sounds subjectively better to some people. It’s like the XLR->SE converters, but without the practical benefit of converting to SE and only the color. Seems unnecessary but I guess whatever floats your boat.

 

As someone who goes builds a lot of amps with line level transformers and has spent the better part of the last month measuring nanovolt levels of noise, I'll throw in my two cents.

The noise contribution of a transformer is basically that of the Johnson noise of the winding resistance. As a rule of thumb, at room temperature, that works out to 0.13*sqrt(R) nV/sqrt(Hz), where R is the winding resistance. The forum favorite CMLI-15/15B has a winding resistance of 1.75k, which works out to 5.4 nV/sqrt(Hz). That's about 770nV over a 20kHz bandwidth. Let's say your signal level is 2V: the SNR of this system would be 128dB. This is even better SINAD than a Topping amp!

The CMLI-15/15B has a fairly high DCR. Doing the math with the Lundahl LL1684 gives an SNR of 142dB. This is basically as close to noiselessness as it gets in audio.

I ran a quick measurement of the 15/15B to confirm this, using an APX515 and balanced inputs. Take note of the Y axis scale. The 60Hz noise spike is slightly lower with the transformer than with no transformer, owing to the part's very high CMRR rejecting the small amount of noise picked up by the balanced cable.

I can't speak to the OP's iron sounding good or bad, but even in fully balanced instrumentation, transformers can yield a measurable reduction in noise. With single ended gear, the benefit can be dramatic.

 

It's funny, before I had experience with transformers, I was intrinsically biased against them. I ignorantly assumed they coloured the sound and added noise but over the years I have experimented with various transformer coupled designs and they sound anything but mediocre. The best dynamics and detail from solid state designs and ultimately led me to choosing a transformer based preamp. I have a few small signal transformers lying about, I may have to explore a little more and have some fun.

 

Alright, so the noise is basically not a problem at all. And if the bandwidth is fine too, that means they should be completely transparent (I suspect they are not a huge source of nonlinear distortions). Good info, thanks!

But even so, other than isolation, in best case scenario, it's difficult for me to see what these might realistically add.

 

A transformer changes everything. Literally and figuratively.

Think about it electrically. You went from RC coupling to LRC coupling. It will change the total load on the output stage of your dac. In certain situations, this could make your dac run more linearly.

A transformer will have insertion loss, second order distortion, high frequency filtering, the list goes on an on.

 

If I'm not mistaken, due to transformer action, loads (and other values like inductances) are reflected on the sides of the transformer, and for a 1:1(done right), should remain the same. Almost as if it wasn't there. Insertion loss will just drop the voltage by a few db or so(at the worst).
Distortion can be generated in other ways, and less is almost always good while trying to reproduce, unless it pleases you. But I'm not sure a proper isolation transformer will add audible distortion, there are other ways of doing that that make more sense. High frequency filtering will probably do nothing, if it even occurs. Most if not all from that big list should do nothing much in terms of effect to audio. It'd be good to see a proper comprehensive measurement of one good (or several, why not) isolation transformers, to actually see if and what's going on. It does some things for sure, just not sure what a difference it actually makes in the end, probably nothing.
Now it's not the output stage of the opamp that's driving the whatever is next in chain, it's the collapse of the magnetic field in the coil, so that definitively has a more "audiophile" ring to it. I have a couple of transformers but no gear good enough to test them.
If someone wants, I can send you one to to test, but they are really not that expensive to get.
And, it's still the output of an dac's opamp that's the main driver, it's not like you are keeping everything analog like with vinyl. Da conversion probably matters more (to a degree to which even that can matter). As usual I might be wrong, I'm just a cat owner.

 

God I love this forum more and more every day. You have probably one of the nicest most respectful replies I have seen out of someone that isn't on a dedicated technical forum in years.

You are 100% correct from a purely theoretical point of view. And yes, it can be challenging to test a small signal transformer and see a result without dedicated equipment. However, you can very easily see a difference with a larger signal transformer with a scope that has an FFT function. An interstage transformer can have a similar effect to what is being described in the OP.

Insertion loss can be a huge deal depending on the circuit.

Distortion is indeed almost always a bad thing. But it's usually not the worst thing and it can be used to cover up other defects.

High frequency noise is a problem. It is almost impossible to build an amplifier that doesn't have some sort of circuit that is prone to oscillation of some kind or another. You can use all the snubbers you want, but sometimes a little extra filtering between stages really helps.

As far as your comments about the DA being more important, yeah, I would agree. The core circuitry will always matter more than some add on. I don't think anyone here would disagree with that. I don't think anyone is saying that a transformer ALWAYS sounds better. I think the OP is saying a transformer can make an improvement and people should be willing to give it a shot.

 

Fair enough, why not. It's a useful tool to have somewhere, if nothing else.
From personal experience, more affordable scopes with fft are good for about just the most basic of stuff. A proper hq spectrum or audio analyzer would probably be required for any serious work I'm afraid (would like to be wrong here).

Also, thanks for your observations!

This forum is cool, but it can bite you in the ass sometimes. Usually for good reason, sometimes due to misunderstandings. I wouldn't worry too much. It's important to stop the back and forth on time if you see it's going nowhere.

 

Hope its ok to rekindle this thread.

I need more gain in my system, and I have this thought/idea that I can get about 6db of gain if I use a transformer such as the CMLI-15 in between the XLR output of my Gungnir Multibit and the single-ended input on my tube amp. Right now, I am going from the modded SE output (XLR ground and XLR plus) of my Gungnir Multibit, into the SE input of my tube amp, as my amp does not have a balanced buffer and I don't want to add one. I measured it (Gungnir Multibit's XLR + to XLR ground) as a 2v out at 60hz (tone generator on REW set to 0db). So I assume the balanced (XLR + to XLR - ) would be somewhere around 4v. So, does this mean that with an isolation transformer I can have the Gungnir Multibit's 4v out into my (SE-only input) amp?

 

Connecting pins 2 and 3 without flipping the phase of one of them 180 degrees (what you're suggesting in the quote) would result in complete cancelation of your signal. I don't know the ins and outs of going from balanced to single ended with a transformer because I don't like the way transformers sound, so I have no interest - YMMV. If you're going to persue this maybe look at a pre-built solution. Also consider that the input of your tube amp could clip with a 4V signal so you may want to stick with 2 Volts.

Separately, on thread topic... it sounds like the OP likes what the transformers are doing to the sound of his DAC's output. It adds distortion (output not same as input) that improves his perception of his DAC. Good for him, but it ain't for me. Transformers, to my ears, roll off the high end to varying degrees and smooth it a bit as well as add a low mid bump and a little bass blur/mud. For a sharp/sterile/thin output (and the Gungnir Multibit to me has a lot of sterile/thin to it) this could be an improvement.

 

I'm not one to tell someone what to hear, but I think that's painting a category of passive components with a broad brush. My experience mirrors yours when discussing some old school steel core EI transformers or some high impedance transformers. Edcor, Hammond, and even Cinemag make some such transformers, and Cinemag even advertises them as such.

But high permeability cores like nickel alloys and amorphous steel/cobalt alloys do sound and measure differently, even when using the same coils. A frequent criticism of some amorphous is that it can be bass light or too punchy, if anything. Cinemag has a great app note comparing nickel to steel, with measured data using the same coil.

https://cinemag.biz/application_notes/PDF/AN-104.pdf

The winding arrangement matters as well. There's a little information in this Lundahl app note, but the short version is that leakage inductance (dictated by how well the windings are magnetically coupled) can make or break high frequency performance and may contribute to what you've heard. Many audio transformers aren't interleaved/sectioned at all.

https://www.erhard-audio.com/outp_typ.pdf

There are some truly dreadful sounding transformers out there. I own several. I'm not sure to what extent you've explored transformer coupled gear, but if you haven't already, I'd encourage you to get your ears on some of modern, well built offerings. You may be pleasantly surprised.

 

@Pabloaguas to answer your question, yes a Bal to SE transformer will get you the 4V. But like @peef said, the transformer you use matters.

 

I've listened to 3 different "Slagleformer" auto(trans)former volume controls in the last 6 months that all had rolled off/brassy treble and woody/boosted low mids to varying degrees. There was a definite price/performance correlation as the least expensive AVC made me think the guy's tweeters were broken until we switched in another preamp. I recently listened to a McIntosh MC462 on my Dynaudio Confidence 60's and it had the same sound. I would consider all of those to be modern. the Mac was better than my Bryston 4B^3 at least - that sounds like clean dark green neoprene to me. I have not heard an AVC with my Dyn's yet since they're still at the dealer (house remodel), but they are relentlessly and at times annoyingly transparent to what goes into them so I don't see my opinion changing.

The speakers I listened to the AVC's on (Tekton and Klipsch) were nowhere near the resolution of the Confidence 60's so the bass overhang was harder to hear.

I'm not here to kink shame people's taste in sound, if you like the sound of transformers go for it and be happy. That's how I judge audio components - does their sound please me? This is why I own a lot of Pass Labs gear, their house sound is pleasing to me. Many people don't like it, which is fine. Weirdly, most of it was *not* designed by Mr. Pass. I'm still looking for the right amp for the Confidence 60's, and I really hope it's not the Gryphon Apex though I fear that's where it's heading. At least it's cheaper than like a D'Agastino or Soulution amp. Time to start saving lol.

My only interest in transformers is on the recording end - their sound is very much part of rock and roll. You can still buy recording consoles and other gear that will give you a transformer output type choice (i.e. iron vs nickle vs steel core for example). Cinemag and Luhndahl are pretty popular in pro gear, especially mic pre's and isolation transformers for grounding issues. That transformer sound is great with guitars, bass, some vocalists, drums (except overheads), et. al. with EQ and compression. And tubes. Pro audio loooves tubes.

What is arguably the finest analog console available, the RND 5088, uses transformers out the wazoo because Rupert Neve (the RN in RND) started in radio in the 50's and that's how it was done and what he learned. He passed recently at the age of 95. A true giant of pro audio. The other contender for finest analog console is essentially a knock-off/update of Rupert's late 60's/early 70's 80 series consoles (Wunder Audio's Wunderbar console). Both companies are located here in Central Texas, which has a surprising number of audio companies, including Schiit.

All this (that's a lot of words up thar) to say I think transformers definitely have their place in audio but I disagree with many folks about where in the audio chain, from recording to playback, they belong.

 

Thank you.

I just need a few more Db of gain and right now I'm weighing the advantages/ disadvantages of going with the Cinemag 15/B or Lundahl 1948. The lundahl costs 3x more, so of course it would have better specs- the insertion loss of the Cinemag is 2.2db, but for the Lundahl I can't find the exact spec but I'm getting the sense that is probably around 0.5 Db. So, for the goal of getting 6db more of gain, (which was the whole point) the Cinemag really doesn't give me as much of the gain.

I recently upgraded my OPTs to Lundahl, and although they are in a different league as far as transparency than my previous Audio Note transformers, the Lundahls are slightly thinner sounding than the Audio Note trefos So, naturally I would consider this a consideration for a line transformer- would actually like to add some thickener if possible without loosing too much transparency.

The other options would be adding a Glassware Unbalancer or Glassware Broskie Cathode Follower as an alternative to the transformer, or to just add an additional mu-stage to my tube amp. Both of which are the cheaper solutions but would seemingly add significantly more noise and distortion than would the line transformer.

Would like to hear everyones thoughts on this, please.

@peef You seem to have a lot of experience with this if you would care to jump back in I would appreciate it so much.

 

I haven't heard that particular Lundahl, but the (cheaper) ones that I have heard sound at least three times better. Cinemag makes many products that I do like, but the CMLI-15/15b always sounded too soft in my systems.

It's probably a bit less. The CMLI-15/15b has about 5x the copper resistance of the Lundahl that you are considering. Copper resistance, along with secondary loading, are what ultimately define the insertion loss. I think Cinemag assumes a 15k load, but I don't see a spec for insertion loss on their datasheet. Most tube amps have an input impedance in excess of 50k; if this is the case for your amp, the insertion loss will be less.

In that case, I would look at Lundahl's nickel core input transformers instead of the LL1948. Most people find the nickel IPTs a bit warmer than the amorphous cobalt cored ones. As a bonus, they're cheaper than the 1948.

I like the LL790x series. They're the largest IPTs that they offer. They also have many options to configure the secondaries; the LL7901/7902/7903 can all be set up as 1:2 and 1:4 if you want to experiement with some extra gain. The 7901/02 can do 1:1 while the 7903 can do 1:8. I would bet on the LL7903 being the thickest sounding of the bunch, but I haven't had the opportunity to listen to the two side by side. Different winding configurations will likely sound a bit different as well, so you would have some ability to fine tune the sound.

I did find the LL1544A a bit thin. This is the smallest and most affordable of their amorphous input transformers, and in my experience not representative of the larger amorphous inputs like the LL1676. The 1545A puts the same windings as the 1544A on a nickel core, and sounds more natural to me, but I do think the 790x are worth the premium.