EC 2A3

Discussion in 'Eddie Current' started by nachocheese70, May 7, 2018.

  1. nachocheese70

    nachocheese70 Facebook Friend

    Contributor
    Joined:
    May 10, 2017
    Likes Received:
    112
    Trophy Points:
    43
    Location:
    US and China
    I figured it’s time to post in a new thread, since this is a different amp than the Aficionado. Although I’ve been playing longer with pure RCA input versus XLR transformer coupled to RCA, I’ll kick this off with a more recent mod. Based on all of the great/near universal positive feedback of swapping out the interstage caps with Jupiter Copper Foil caps 0.1 uf. The stock ones on my 2A3 were Audyn Plus 0.1 uf.

    Just got it done an hour ago. Took maybe 1.5 hours total but that includes setting all my soldering stuff up and taking the case bottom off. I took my time double checking all the traces to make sure I swapped out the right caps and soldered the new Jupiter in correctly.

    So far, with just the 0.1 uf caps the sound is rich with strong bass, while others have suggested upping it to 0.22 uf. I trust Marv when he says the 0.1 uf is sufficient, though I am using Abyss that probably could cover any slight low end difficiencies. There may be just a slight bloom that requires a long burn in, but could just be the extra punch that I’m getting playing tricks with my ears. Right away there is more clarity in the upper mids to treble, not that the stock with RCA 2A3 had any issues with being veiled. Will need more time to listen, but there’s no easy way to go back and forth.

    [​IMG]
     
  2. nachocheese70

    nachocheese70 Facebook Friend

    Contributor
    Joined:
    May 10, 2017
    Likes Received:
    112
    Trophy Points:
    43
    Location:
    US and China
    Transformers Adventure

    Since the EC 2A3 only has RCA input, I wanted to play around with different options of connect my Gungnir Multibit USB Gen 5. I used EC RCA input 1 to connect to Gungnir Multibit RCA out, and then RCA input 3 to connect to Gungnir Multibit XLR via various transformer. I ordered a bunch of different transformers just to see how they would sound compared to all RCA.

    Note, I used basic all copper wires and basic gold plated copper connectors. I used Oyaide solder that contains tin, with some silver and copper, with an eutectic temperature around 217 C (I think). Maybe better wires, connectors, and or solder could have made a bigger difference, but my main goal was to focus on the transformers.

    Epic Fail

    I’ll start off with the most embarrassing incident. I got a pair of Neutrik NA2F-D0B at a good price. It said 600ohm/10kohm on the spec sheet, but the device itself is 1:1 200R:200R. I believe this is actually 200ohm. Either case, the Gungnir Multibit had no hope of driving these properly. Any bass heavy note and it’s distortion city. One dead giveaway that it’s overdriving the current output is the RCA will also sound distorted too if the Neutrik is connected to Gungnir Multibit but disconnected from EC.

    Anyone can just google an image of this transformer so I’ll skip posting.

    Round 2

    I found a Taobao store that sells semi-DIY transformer kits. They also ship inter nation too. To avoid overdriving the Gungnir Multibit, next up was a 10kohm:10kohm transformer. It is enclosed in a metal cage, should be a faraday cage. I wired it up as normal (XLR pin 2 positive terminal pin 3 negative, pin 1 ground, RCA tip positive, sleeve ground) but did not do anything special with grounding the faraday shield.

    The sound was very bright compared to RCA, though getting past the harshness there was some extra macro detail. There was a low end roll off too to compound the very bright sound, but the bass control felt tighter. Overall I did not like this but not sure if the issue is the 10k:10k or the lack of grounding of the faraday cage. Either case I kept going.

    Here is what I used. Note, the cables/connectors were cut off (quicker than desoldering) and reused for the next transformer.
    [​IMG]

    Round 3

    Next I bought a more traditional transformer next, with a lower impedance of 2.4kohm:2.4kohm. The PC is about the same as before, so just soldered on the same connector as before and was up and running quick.

    This produced the best overall sound so far. It did not have the bright lean of the above 10k, though it did still have a slight slight bass roll off (maybe slightly less roll off). There overall plankton and microdynamics was slightly improved, as well as macrodetail. Small improvements.

    I suspect the majority of the improvement is coming from the balanced versus RCA out of the Gungnir Multibit. Not sure what Schiit does to get single ended output out of their multibit megaburrito DAC, but this transformer seems to be able to top it slightly. At about $20 not a bad deal.

    Below is a picture. Connectors still attached (for now).

    [​IMG]


    To be continued
     
  3. nachocheese70

    nachocheese70 Facebook Friend

    Contributor
    Joined:
    May 10, 2017
    Likes Received:
    112
    Trophy Points:
    43
    Location:
    US and China
    I can’t resist posting this, the Jupiter Cap upgrade is wow. I’m just chilling and unwinding right now, listening to So What from Miles Davis Kind of Blue. I just heard it last night before the cap upgrade (and it’s a regular listen for me).

    The cymbal strikes are just to die for. This difference is bigger than any of the tube rolling with the 2A3 or Mjolnir 2 (though I understand that amp responds less than normal to tube rolling). With a top notch transformer (will post on this later) and some modding to the Abyss Phi, this is incredible. Btw at Freddie Freeloader now, and still wow.
     
  4. nachocheese70

    nachocheese70 Facebook Friend

    Contributor
    Joined:
    May 10, 2017
    Likes Received:
    112
    Trophy Points:
    43
    Location:
    US and China
    Btw, I should have posted this first for people not aware of this. XLR out is usually at a higher voltage than RCA out, up to +4 dB versus -10 dB. Traditional boxes that convert XLR to RCA will add resistors (and capacitors) to level the line.

    In all of my setup, I did not bother with that as IMHO more components means more chances to alter the sound. I did check with Purr1n that the EC 2A3 should be okay taking full +4 dB signal from the RCA inputs, and so far so good. But don’t assume this is the case with all EC and definitely all other amps.

    Also, since I can’t edit the already posted, I will add an addendum here.

    All the above transformer increased the volume about 2-6 o’clock worth on the amp. Meaning if I’m using all RCA input and for a specific song/passage I’m at 3 o’clock on the volume knob, then switching to my transformer means I have to reduce the volume to between 9 to 1 o’clock. This is by my ear so highly subjective.

    The 10k:10k is hardest to judge, since its skewed so bright I had to turn it down the most. I’m maybe slightly more than 3 o’clock reduction but it’s meaningless.

    The Neutrik did not need to be turned down that much, around less than 2 o’clock. It’s probably due to the fact the Gungnir Multibit can’t hit the voltage swings needed due to too low impedance (and really badly distorting).

    The 2.4k:2.4k are about 3 o’clock.
     
  5. Garns

    Garns Friend

    Pyrate
    Joined:
    Jul 9, 2016
    Likes Received:
    2,483
    Trophy Points:
    93
    Location:
    Sydney, AUS
  6. nachocheese70

    nachocheese70 Facebook Friend

    Contributor
    Joined:
    May 10, 2017
    Likes Received:
    112
    Trophy Points:
    43
    Location:
    US and China
    I don’t recall seeing a 10k:2.5k or 2.5k:10k (I assume one can always reverse primary and secondary in such low power situations with no harm), but I’ll check around. Good thing EC handles the extra volts no problem.

    I’ll see if any Carnhill transformers pop up on Taobao. It’s worth the shot just to try it out. I see some UK based Vigortorix that’s 1:2 winding and 600:2.4k at about the same price, but specs only so so (30-30 kHz but no mention at what dB). The other local transformers are about $8-12 shipped, pretty low cost to play around (much cheaper than tube rolling, which I am not doing as the cost is crazy).
     
  7. nachocheese70

    nachocheese70 Facebook Friend

    Contributor
    Joined:
    May 10, 2017
    Likes Received:
    112
    Trophy Points:
    43
    Location:
    US and China
    Round 4

    So the same store above suggested their ready to use balanced XLR to unbalanced RCA, 600:10k 1:4 winding. I would have thought it backwards so it’ll naturally lower the voltage, but since they didn’t t post a schematic, maybe this has a resistor attenuator. Either case, this definitely boosted the volume. I had to lower the volume knob almost half a turn to match the all RCA.

    I was more concerned with 600 ohm on the primary that it wouldn’t work with the Gungnir Multibit, but it worked fine with mine (built Oct 17). Bass sounded a little more rolled off than the 2.4:2.4, but better clarity. It’s slightly brighter as well. Overall a nice looking all in one that works. I’d say this is an improvement over RCA. Some rolled off bass but actually more mid bass punch and slightly more forward mids. I liked the sound.

    Overall if looks mattered I’d take this. For pure sound quality looks be damned, I’d lean towards the 2.4k:2.4k. Plus this is about $28 each ($56 a pair), while the 2.4k:2.4k is $8 each or $16 a pair. I’ve paid more for connectors and wires than that.

    [​IMG]

    Round 5

    Just a few days ago, I ran into a Chinese transformer company called Raphaelite. They cost a bit more ($45 each) but they look physically bigger (the case is 2.5x but can’t see inside), is heavier (more than double), and have incredible specs if you can believe them. The 600:600 claims 9-100 kHz at -1 dB, so according to this should there should be no audible roll off at all.

    I wired in the final connectors, so female XLR to connect directly with Gungnir Multibit and male RCA to EC, with the RCA cable as short as possible so let the balanced handle the long run, but since we’re talking only 3’ it’s probably irrelevant.

    Overall, I really liked this. Wow it’s got a heck of a bass slam and I can’t tell any roll off versus all RCA. 1 dB down at 9 Hz is meaningless other than good marketing trick, but if it can do 20 Hz with less than 1 dB loss then that’s more than enough.

    Of all the transformers this has the best bottom end (and I like big bottom ends and I can not lie - sorry had to say this). The mids are not as forward as the 600:10k, but still very clear (clearer than RCA). Treble is brighter and more detailed than RCA, similar to other transformers except 10k:10k. Volume impact is maybe 2-3 o’clock. I’d say this was well worth the money. I’ve been listening with this since wiring it in. I did the Jupiter cap upgrade just 2 days ago. With all the extra treble gains, I was able to mod the Abyss to tame the slight treble harshness, but still end up with higher clarity and macrodetail than when I started (unmodded Abyss, Audyn Cap, all RCA).

    [​IMG]
     
  8. nachocheese70

    nachocheese70 Facebook Friend

    Contributor
    Joined:
    May 10, 2017
    Likes Received:
    112
    Trophy Points:
    43
    Location:
    US and China
    Reflection

    When I started my transformer journey, it was a jump in head first and let me listen to the results. I made no attempt to understand any theoretical/practical basis, and Chinese transformer suppliers give practically no detailed specs to warrant any understanding. Plus this is line level voltage stuff so it’s not like I’m going to blow anything up

    That said, I’ve been brushing up on the practical aspects that can affect sound. Since SBAF has people much more knowledgeable, I wanted to bounce these thoughts to make sure I’m understanding correctly:

    1) Transformers don’t just roll off on the bottom end like regular circuits (high pass filter for example). Low frequency current can saturate the core, and generally the lower the frequency the lower the current needed to saturate. Once the core approaches saturation, it affects all of the signal not just the low frequency, and the effect is highly complex. The saturation conditions itself is also highly complex. This is why Cinmag has all those THD plots. I think Jensen has some. Neutrik and all the Chinese transformer manufacturer show none of this info.

    In the case, I assume a bigger transformer using similar core material, winding, build quality, should result in less issues (lower THD)? Part of a transformer “coloring” the sound comes from this THD versus direct coupling.

    2) All the Gungnir Multibit output is driven by the same chip, so if one single output is overdrawing current causing higher distortion then all outputs show same issue.

    3) Before the Gungnir Multibit starts clipping caused by too low impedance load forcing excessive current draw, there is a range where distortion is abnormally/excessively high. In other words, if you start with anhigh impedance load the Gungnir Multibit will have very low THD, and as you drop impedance it would remain the same. But there will be a point as impedance load gets low where THD starts taking off and may even get very high BUT not clip.

    Sorry if the above is obvious to most EE or experienced audio folks. I really didn’t think or consider any of above when I started, and was really looking at just clipping. BTW maybe I should rename this thread to just Transformer Journey.
     
  9. Priidik

    Priidik MOT: Estelon

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    2,153
    Trophy Points:
    113
    Location:
    Estonia
    Transformer core saturation occurs when voltage is driven too high or when frequency below the rated bandwidth is used with same signal level. When secondary starts drawing a lot of current then opposite to saturation happens, total flux in core goes down.

    Yes, you are correct that for same power output low freq signal effectively requires more core cross section than for high freq signal. In case signal transformers loading is minimal, thus not much current flows.
    Basically then what you need to know for signal transformers is rated voltage input (volts, dBu etc) and sometimes it's winding impedance matters, like for matching 600 ohms or smth.

    Don't worry, many EE courses only scratch the surface wrt transformer inner life.

    There certainly isn't one superior transformer to every application. And bigger does not mean automatically better for signal transformers. For power transformers disregarding electrical bill impact, yes bigger is better for sound.

    When you increase core size you get more core losses (hysteresis losses and Facoult current losses), which are obviously going to degrade sound. Increasing core size can mean more wire --> again more losses. When transformer impedance needs to be low, this becomes a real problem, that's why silver is sometimes used and here it's an engineering choice not bling or snake oil.
    For signal transformers leakage flux is a thing, it's often easier to contain it with smaller design.
    So, really the perfect transformer for a special need is still a compromise and a highly optimised one.
    What really makes one competent transformer obviously better sounding than another is probably down to core material.
    Construction plays part too, although based on what I have seen, it's more or less the same for different manufacturers.
     
    Last edited: May 12, 2018
  10. nachocheese70

    nachocheese70 Facebook Friend

    Contributor
    Joined:
    May 10, 2017
    Likes Received:
    112
    Trophy Points:
    43
    Location:
    US and China
    @Priidik thank you for the info.

    Sorry I have a follow up question. It took a lot of reading to finally find info that transformers actually have impedance (all the basic info assumed none).

    Is the impedance, or at part of it, due to the winding of the wire around the core, with more winding/bigger core necessary for higher impedance (assuming all material used and design/construction same)? And this portion of the impedance is really wire resistance in a DC world?

    The reason I ask this is because I just bought a new line level signal transformer with 10kohm input impedance 1:1 winding that’s huge at 0.7 kg. They have a 600ohm version that’s much smaller at 0.2 kg. Both have same general spec (600 actually wider bandwidth) and power handling (5W max with secondary can take 0.03 mA).

    The reason I’m looking at higher impedance is Purr1n warning on the Gungnir Multibit out. While I hear the obvious clipping only when using Neutrik transformer, I believe I may be hearing the higher distortion from Gungnir Multibit due to too low impedance. Purr1n said to basically get as large impedance transformer as possible.

    Using my Fluke 107 series dmm, I measure a resistance of just 45 ohms with Neutrik transformer. With the 2.4k:2.4k I get 145 ohm. With the 10k:10k reviewed above I get 451 ohm. The Raphaelite 600:600 I get 215 ohm. The new big transformer i get 1003 ohm. I assume this is DC resistance since my Fluke doesn’t have the fancy AC generating feature to measure impedance.

    Can I assume bigger DC resistance is moving in the right direction of putting more overall load on Gungnir Multibit output? The bigger transformer needed to get high impedance is tradeoff to potentially worse core and leakage flux.
     
  11. nachocheese70

    nachocheese70 Facebook Friend

    Contributor
    Joined:
    May 10, 2017
    Likes Received:
    112
    Trophy Points:
    43
    Location:
    US and China
    Round 6

    I spent a lot of time really listening to this latest transformers. Unlike all the other transformers, this one does not appear to be available in the US (there appears to be a smaller version sold on Amazon marketplace but now discontinued). It’s huge at 0.7 kg (1.6 lbs) each like an OPT with a 5W rating, but it’s wound specifically for input signal. The seller put in a little more tech detail than most others, but being in China I’d still take it with a grain of salt. It uses a Z11 EI48 core where each laminate is 0.35mm thick. The spec is 1000-1025 ohm resistance, support up to 18000-19000 gauss flux, uses 5N OCC copper wires for all winding, and rated 20-65 kHz at -0.1 dB. It uses a complex winding process with forwards and backwards winding, with 6 clip/clamp and 4 windings (not sure what any of this means). At about $60 USD for a matched pair, I had nothing to lose to try these out.

    Anyways on to my impression. I spent time specifically comparing these to the Raphaelite 600:600, and to Gungnir Multibit RCA out with XLR unplugged. As I noted in my Reflection post, I am convinced some of the lower impedance transformers if left plugged in was also affecting the Gungnir Multibit RCA output.

    The biggest difference I noticed with these 10K is a big clarity improvement and a smidge brighter versus RCA output. It is not anywhere as bright as the small 10K:10K but some songs with lots of cymbals or top hats does fatigue my hearing quicker than just RCA output. Plus the Abyss Phi doesn’t help. The bass is also noticeably tighter than with Raphaelite 600:600 but overall less bass sound (less warm feel). Compared to RCA out, the overall amount and control of the bass was better. The mids of this transformer was clearer than the Raphaelite and about same as RCA out.

    Note, I believe the Raphaelite was causing issues for the Gungnir Multibit. Plugging it in did affect the RCA outs, causing even the RCA to sound warmer by boosting the bass while putting a veil on the mid to treble. It’s a subtle but noticeable effect. This 10K transformer did not affect the RCA at all.

    Overall, I’d say this is the transformer I’m most happy with, and I prefer it over the Gungnir Multibit RCA out. I would say the sonic impact between Gungnir Multibit XLR out+transformers vs Gungnir Multibit RCA out is much bigger than I get with NOS RCA 2A3 vs PSVane WE275. I spent a grand total of about $220 for 5 different Chinese transformers (not counting Neutrik) and learned quite a bit, so I feel well worth it.

    Finally, here is a pic of the transformer next to the 2.4K to show how big these are
    [​IMG]
     
  12. Priidik

    Priidik MOT: Estelon

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    2,153
    Trophy Points:
    113
    Location:
    Estonia
    Ahh, somehow forgot to answer for so long.

    Transformer impedance usually means voltage drop relative to ideal transformer at rated power. It is often also called transformer regulation or 'transformer voltage drop'.
    That is different to impedance of individual windings. For your exploration winding impedance is important.
    What you are measuring is DC resistance and it should be diminutive relative to the reactance that the inductive character of winding provides at any audio frequency. Inductance means rising impedance with rising frequency, so in treble it's really high, also very very light load to your Gungnir Multibit. Bass could very well be problem zone.
    To nail that one you'd need at least a LCR meter to measure inductance. If you feel rebel you could get by with your dmm and some sine gen, too.
    Winding impedance reaches close to it's resistance only in overload conditions, or close to dc signal, otherwise it's much higher.

    A perfect scenario (in most cases) would be such that winding impedance would be infinite, so that it's resistance is zero and inductance infinite. Reactance means no lost power, no signal degradation. Resistance of any kind is lost power, usually also degrades signal.
    Note that core losses act like resistance, only selective of frequency. It's not trivial to understand and it's not like impedance or reactance despite it's frequency dependency. Core losses dissipate real power, real heat is generated. And conservation of energy still holds despite flat earth conspiracy and 'free energy' theory.

    Real life windings have significant resistance (most of the time seen as bad thing) and inductance is not infinite either (also bad).
    Worst part is that in quest of getting more inductance resistance increases; by trying to lower resistance and maintaining inductance core window must increase to allow for thicker wire --> more core losses.
    That is why pre-s with OPTs have small OPTs, and why most input transformers are tiny.
    Going bigger has the benefit of better linearity down low, though. That might be what you have confirmed with your experimentation.

    Also called bifilar winding, industry standard.

    This one looks to be silicon steel, probably not grain oriented.
    I don't want to encourage you to spend any more money on doubtful gains in sound for some more transformers; but.. silicon steel is not really ideal for signal transformer.
    It has significant hysteresis losses (magnet domain switching, expressed in coercive force) versus something made of amorphous metals (metal glass) or other fancy alloys (Supermendur, Permalloy, etc).
     
    Last edited: May 23, 2018

Share This Page