Yeah I found a transformer study that used the 111C as one of the subjects and saw the distortion curve. It was pretty impressive.
And yes I understand your sentiment about people saying they can hear stuff that maybe they can't.. it's a plague in the pro audio field. Guys will buy 10k converters that no average listener could hear the difference. Or using high sample rates vs 44.1, or dither, or silver wires vs standard. I get it.
I have been mixing audio for at least 15yrs and I try to stay as objective as possible. One thing I can say is there definitely is a difference running the signal through the 111C even at +22dbu. Will the average listener hear it or care? No. But I can hear it and I like it and I want to hear what happens when I push it harder.
Also, Im not looking for a heavily colored transformer.. just a little analog goop to blur the edges. And you can kind of test to see if the transformer is doing anything by running a null test on a before and after file (after level matching them as good as possible). I haven't done that yet but I am definitely down to do so and post the results. My prediction is that the two files will not null perfectly and will have some peaks crackling through and possibly some sound stage discrepancies due to the non-linearity of the 111C.
Hey I'm totally down for a less expensive solution. I am a deer in headlights with this stuff so far. I'm trying to find someone that can help me design/build this, for a fee. Nobody here has shown any interest, and I'm not surprised... People have lives and barely time to do their own projects let alone help a stranger build theirs, so I'm trying to find people that do it for a living.
I reached out to Jim Williams whos somewhat of a legend in the pro Audio world for modifying peo gear. He's too busy though. Gotta call some more people.
Check the continuous output current rating. Consider what you need to drive +36dBu into 600 ohms. (Hint: 32 watts). So now you need an output buffer, and more complexity the OP probably won't want to deal with. 600 ohms because you'd want to terminate the 111C.
That's why I stuck with the 990, at a max of +31dBu (and, of course, I have them). You can actually do that.
Or use a power amp, and build out the output with 600 ohms.
That's an excellent point about the current output needed. I completely overlooked that; I was too focused on getting the voltage swing.
What do you mean by use a power amp? Doesn't s power amp only feed speakers? I know that power amps are usually 2 - 8 ohm output so are you suggesting that I change the output to 600ohms then convert the speaker output terminals to XLR? Does that mean using a power amp would mean I'd have to send an unbalanced signal out? Or are there power amps already configured for 600ohm/balanced output?
Btw when I called Jim Williams and explained what I was looking for he immediately said I need a power amp, based on my desire for a max output over +30dbu.
Oops, should have double checked and not assumed. 38.6Vrms into a 600R load is only a little over 64mA. Well within the limits of the OPA552.
The load will be a lot higher than 600 ohms.
The easiest way to increase the voltage. (x2)
https://www.jensen-transformers.com/transformers/line-output/
The jt-123-blcf can handle 33dbu (only .02% HD at 20hz).
The easiest way to increase the voltage. (x2)
https://www.jensen-transformers.com/transformers/line-output/
The jt-123-blcf can handle 33dbu (only .02% HD at 20hz).
I actually was looking at those. The SSL compressor clone I have has 1:1 600ohm Jensen transformers with a max output of +22dbu. I thought that was the bottleneck in my system because the output gain the SSL was only at 10db of 20db max going into the 111C. So I bypassed the Jensens and went straight into the 111C from the SSL and the signal still crapped out at the same level as with the Jensens. I know it's not the 111C busting up too because I removed the 111C from the path and the signal still broke up at the same point. I had a -30dbu pad before the ADC so I know it wasn't the input converters clipping either.
I wonder what part of the SSL is distorting before the transformers... If it was easily replaceable I might be able to use the full 20db of gain on the SSL, but I doubt it's just one component after hearing from you all that it really would take something like a power amp to get the max dbu I'm looking for.
Good suggestion.
I would point out that the OP has no design experience, but has soldered cables and is looking for a "kit". Making an LM3886 work without oscillating isn't exactly entry-level. Mounting SMDs like an OPA552 is also not at all basic soldering.
A packaged solid-state power amp would work though. Something sitting around like a Crown D75, or D60, if it's still functioning, is a good 111C driver, if you run it bridged it can go over +31dBu, with no issues driving a 600 ohm load.
Remember that +20-21 is the max for any sigle ended stage with +/-15V power supply. Upgrading would be difficult. You'd have either put in a differential output amp (gain 6dB) or and/or add a higher voltage PSU just for the output buffer or go to a different output transformer. It's not pretty any way you look at it.
But still....question why you need to go higher, and if there is not some other way to get the goop you want. It's just nonlinear distortion anyway. Tube amp simulators might even be better since they're often richer in even-order harmonic distortion. Just a mindless thought....
I wanted to run the null test to shore up this point, and short answer is yes, there is a clear difference between using the 111C and not.
I printed a section of a mix I am working on, bare. I made sure the signal did not clip the output converter. Then I ran that printed signal through the 111C. After that I brought back in both prints, and put a time alignment plugin on both tracks, since adding plugins increases the delay compensation (which could affect the null test if it was only on one track).
I inverted the phase of one of the tracks (via the time align plugin) 180 degrees and started adding delay to the bare track (using the smallest denomination possible, us) until the two signals got as quiet as they possibly could be. Then I adjusted the gain on the bare track to account for any signal loss from the 111C.
The resulting adjustment values are that I reduced the bare track by -0.38db, and delayed the bare signal by 1.13 samples. These are extremely small values in terms of delay and level. If the files were exactly the same, then the combined signal would be silent. What actually happened is there was a full frequency spectrum range of signal difference. This is quite unusual, as these tests almost always ONLY leave mid and high range frequencies since low frequency waves are so large (they are much easier to reduce in level with phase cancellation). There was about -60dbfs peak level of different audio information across the frequency spectrum, between the two files. I attached a screenshot of the response.
One strange thing I noticed was the Prominent lack of 4k cycles information in the resulting null response. That means that the 111C that I have is stunningly accurate in that narrow band. I do see a big null centered in the 300hz region but it is wide and that is just because the song is deficient in that region, and not a result of transformer affect.
This transformer absolutely has a sound. Now maybe that's because its 80 years old and is not operating as intended. I don't know. But I assure you it's doing something. I want to see what this thing does when REALLY pushed.
Attachments
Any active differential/balanced output will get you that 6dB too. The transformer would be a step-up.
Well...not exactly. The data sheet for the 123BLCF doesn't open, but the one for the 123BMCF does. It shows the typical rising THD with falling frequecy, so 20Hz at +28 is more like 0.8%. So there's no way the BLCF does .02% at 20Hz at +33dBu, probably just under 1%.
I'd like to reiterate that I don't have a particular sound I'm after... I actually already like what the 111C does at the current levels. But knowing that it can accept more gain, and that more gain means a different affect, I just want to hear for myself what that affect is. I'm a a very curious and experimental person, and will spend time and effort just to satisfy that curiosity.
As I've said before, the result might be crap. I might not like what I hear after pushing more level into it, but I will have learned a lot in the process, and can say for sure that I know what happens in that situation, and not go based on what I read on the internet. If I am able to get more gain into these trannys, and it sounds like crap, I will be perfectly fine with it. I will also have closure that I did the work to find the answer to my question.
We need to scale the results to reality though. A null to -60 is really an excellent null, not much audible difference there at all, if any. But also realize that the frequency responpse of a 111C is not perfectly flat, nor is the phase response, both of which will strongly limit the null depth. The phase response is way below audibility, and the frequency response is in the "highly questionable" range. Nothing shown is validation of any "transformer sound", though.
What did you use for the test signal? Music? Or a flat chirp? Music alone would display that erratic null response. To see the null you need to test with a sweep, 20Hz to 20KHz, at some level a tad below 0dBFS, so the test signal is equal at all frequencies. Only then would you see a good picture of the null response.
Even more important, you need to start testing using a double-blind methodology or your results will definitely be skewed. That form of testing eliminates many uncontrolled biases, which unfortunately will strongly affect your results.
The concept is to present choice A, and compare to choice B without knowing which choice has the transformers. Then, compare each to choice C, which will be either A or B, and see if you can determine if it matches A or B. Choice C should be randomly assigned to A or B. It's not easy, and the simpler switching methods require the help of an assistant to remain fully blind, but just comparing the known transformer path to non-transformer path is loaded with expectation bias. Once you really clobber the 111C with hot levels, you'll no doubt pick it out correctly. But at normal, reasonable levels, the differences are very small and influenced by biases.
This is one rason the whole transformer color thing is a mess. Nobody tests it right.
I agree that we are getting into territory of whether the difference would be significant to the average listener, and my opinion would be no, that at these levels it does not. However I would disagree with your original notion that the difference is immeasurable. While the difference in signal was low, it was clearly there, and to a trained engineer ear it is audible. And although it is music, the information in both files was the same, I believe the general perception of tone difference is something that can be reliably perceived, albeit maybe not by the average listener.
I do believe that sending more level to these units would increase the affect, as we both understand that distortion increases as level is increased. It is that distortion change I am interested in hearing.