Small la2a transformer challenge

The following were put on the AP today..

The Lundahl model is brought to you by the courtesy of our very own SSLTECH.






what we have in the photos from top to bottom.

Lundahls (approx $220 for a set)
Sowters (approx $160 for a set)
and UTC's (HA100X re-issue and an A-24) (approx $250 for a set off ebay)

all sweeps were ran with no T4b modules in the units. With the meter switch set to "G.R."

The following photos are of the sweeps that were performed. I'll group them as A, B and C. In no particular order.

you will notice some dramatic differences. I should mention that the tests remained the same for all three compressors.



Batch A



THD+N over Amplitude



THD+N & Freq resp over Amplitude (there were two ran here. one set into 600 ohms and the other set into 100k ohms)



oops I should mention that you need to look at the cursors in the comparision sweeps for THD+N


Batch B






Batch C






So onto the question... based solely on what you see int he measurments.. which one would you pick?

A
B
or C

Very interesting... I'm pretty new at the whole pro audio thing, but I'll go out on a limb.

I'm not sure what you've plotted on the THD+N vs. Amplitude. I'm guessing the three traces are different frequencies, yes? Looking at the other plots which seem to fit in my brain better... C looks the most well behaved (UTC?). A has some wacky stuff going on in the low end and higher distortion overall (Sowter due to smaller core on 4383 input?). B looks cleanest except for the low end driving 600ohms (blue trace?) where it loses it pretty badly (Lundahl with the smallest output tranny core?). Or am I full of it?

Oh, which would I want? Uh, I built with Sowters, but maybe I want UTCs if my guesses are correct. Do I win a free Bloo kit? Pleeeease? :green:

:guinness:
Analog Packrat

p.s. Out of curiosity, did you use the same tube sets in each test?

ok let me clarify...

the 3 images that each have 3 sweeps on them..

each sweep represents one of the boxes. the sweep is THD+N over Amplitude.

The reason there are three of them is that I put the cursors on the appropriate sweep so that you could see the exact amount of THD+N (the cursors are the little marks you see on the sweep with the black boxes showing the amount of gain and THD+N value.

so we have three sweeps on the plot..

the top one shows the following...

at +17.976 dbu we have .75307 THD+N

the next line down shows..
at +17.976 dbu we have .49661 THD+N

and the bottom line shows
at +17.976 dbu we have .34252 THD+N

I put the appropriatly marked one with the correct group (A, B and C)

Just a silly question:

I`m looking at the pics of the units & with the Lundahl & the Sowter its plain to see that the input transformer in each case is in the bottom right hand corner.

Is the UTC HA100x input transformer really that much bigger than the A24 ??? I`ve only seen an A10 in the flesh so I`m amazed at the size difference.

A-10 and A-24 = T4B-sized.

HA-100 and HA-100X = Encyclopaedia Britannica-sized.

They weigh a ton as well, making your unit very back-heavy!

...I want to see what CJ's guesses are... :green:

Keith

Cool test Steve and Keith.

I guess I'd pick C based on the measurements alone, takes more input level, smoother onset of saturation and a smoother HF.

I'm guessing this is the UTC?

No idea whatsoever but my guesses would be:

A - Sowter
B - Lundahl
C - UTC

Any audio clips too?
-Tom

Here's an attempt to clarify what the data represents:

BRAND A:

THD+N over Amplitude:

The three curves in this graph represent the three transformers fed with a 1kHz tone at increasing level. As the level rises, so does the distortion. All three transformers show steadily increasing rates of distortion, and all three exhibit a 'break point' at which the distortion begins to rise much more rapidly. (load = 100kΩ)

Brand 'A' is highlighted with the "cross hair" cursors, and consistently has the highest overall distortion, with the break point at around +18dB.

THD+N over Frequency/Amplitude over Frequency:

The upper red and magenta lines represent the distortion at rising frequency. One at 600Ω load, one at 100kΩ load.
The lower yellow and green lines represent amplitude response at rising frequency at 600Ω (green) and 100kΩ (yellow) loads.

The distortion frequency response doesn't appear to change much with varying load. Nor does the amplitude frequency response; just a slight drop in overall level, consistent across the band.



BRAND B:

THD+N over Amplitude:

The three curves in this graph represent the three transformers fed with a 1kHz tone at increasing level. As the level rises, so does the distortion. All three transformers show steadily increasing rates of distortion, and all three exhibit a 'break point' at which the distortion begins to rise much more rapidly. (load = 100kΩ)

Brand 'B' is highlighted with the "cross hair" cursors, and has by far the lowest distortion at low signal levels. However, it reaches the 'break point earlier than the other two, and above that point, the distortion rises ore steeply and rapidly.

THD+N over Frequency/Amplitude over Frequency:

The upper green and blue lines represent the distortion at rising frequency. One at 600Ω load, one at 100kΩ load.
The lower yellow and green lines represent amplitude response at rising frequency at 600Ω (blue) and 100kΩ (red) loads.

Brand B appears to exhibit significantly greater low frequency distortion when the output feeds 600Ω. Amplitude response seems to be consistent across the band. Note the HF response smoothness and extended response compared to "Brand A".


BRAND C:

The three curves in this graph represent the three transformers fed with a 1kHz tone at increasing level. As the level rises, so does the distortion. All three transformers show steadily increasing rates of distortion, and all three exhibit a 'break point' at which the distortion begins to rise much more rapidly. (load = 100kΩ)

Brand 'C' is highlighted with the "cross hair" cursors, and initially has distortion levels somewhere between the two others, but as the signal level rises and the other two begin to distort more severely, Brand C becomes the best-performing at high signal levels. It's break point is the highest of the three and the slope above the break point is the gentlest.

THD+N over Frequency/Amplitude over Frequency:

The upper red and yellow lines represent the distortion at rising frequency. One at 600Ω load, one at 100kΩ load.
The lower teal and green lines represent amplitude response at rising frequency at 600Ω and 100kΩ loads.

"Brand C" exhibits the smallest level difference with varying load. It exhibits the smallest overall HF distortion rise, erspecially so at 100kΩ load, and more comparable at 600Ω load.

Hopefully that makes it easier to read the data as I see it, I think a couple of folks might have been confused?

Keith

Scenaria and Keith,

Thanks for the clarification on the THD+N vs. Amplitude plots. Very interesting. I'm sticking by my first guess:

A Sowter
B Lundahl
C UTC (preferred)

Any chance of adding THD+N vs. Amplitude plots at, say, 30 Hz, 200Hz, and 5KHz?

Analog Packrat

Okay, just to get a handle on this, the THD versus level chart is the the same for all three right. It's actually all three cases on one graph. And each one is highlighted with the cursors to illustrate a point.

Brand A is the high magenta curve.

Brand B is the low magenta curve.

Brand C is the middle blue curve.

Seems like brand A saturates much earlier.

Now, for fun, try the tests at 100Hz...I'll bet brand A packs it in very early. and brand C does really well. Bigger cores will do better at LF.

Now, for fun, how about showing us what the distortion spectrum is at 1kHz. While THD is an interesting parameter, I think it's better to know what the spectrum of the distortion is....is it primarily pleasant 2nd or 3rd harmonic, or is it evil 6th harmonic!

Also, what is the pulse response of these transformers? Check the squarewaves for ringing. Perhaps there are some zobel networks or other terminations that need tuning.

Cheers,

Kris

guessing two :grin:

A Lundahl
B UTC
C Sowter


Kind regards

Peter

What the..........!

[quote author="CJ"]What the..........![/quote]

What's the matter Colonel Sanders?

... "Chicken???"

:green:

Keef

Hats off to you guys. Thats a lot of work right there!

It is a little harder to figure out exactly why the curves are different, as you are not just testing one transformer by itself, but two transformers as they interact with the line amp and Audio Precision. But, you can still figure some stuff out.

I realize that with the AP, you can not super impose the three frequency charts from the three different transformers on the same graph without doing it with a pen and paper, which is a pain.


Bass Response:

The bass distortion for the Lundahl compressor is certainly not from the 7903 used as output, as it measured better than most for bass response over here. Most likely it is from the input transformer. The distortion specs are cool, but you can learn a lot by comparing scope shots at the low end. Most of the stuff I test has different waveforms at 20 hertz when the core starts to protest.

The Sowter probably uses a high perm-low weight core, so that is why you see the big difference in bass distortion when you load it. The lower weight core means less core loss on the lighter signals, but less db in bass handling.

The UTC's probably have a bit less lively core material than the other two brands, thus, the smoother low end distortion specs.

Remember that the UTC's have been around a while, so there might be some influences from aging, like the core. It could have some lams that are not seated together as tightly as when they were first assembled, which I have seen with almost every UTC that I have yanked apart, and the core material could have been "broken in" due to many years of audio going thru it, although I still do not know exactly what is going on here, but that's alright, because neither does anyone else!

Midrange:

At 1 khz and above, the 0.014 mil lams start to drop out of the transformer puzzle pretty quickly. So I believe that most of the distortion at 1 khz is probably dependent on the coils, not the core. But the shape of the coils is depends on the "shape" of the core, so in a way, the core still has an influence, albeit indirectly. So different figures for leakage inductance due to different coil/core geometry is probably responsible for the different distortion specs at 1 khz.

It's kind of neat that you are comparing three different lam styles here:

Lundahl: DU
UTD: UI
Sowter: EI

A 1 khz ring test would be cool to see.


The 7903 I tested started to draw a lot of ac current starting at about 1 khz, peaking at about 10 khz, dropping into a valley at about 15 khz, and peaking again at about 20 khz


High Range:

The different cores mean that the Sowter uses one coil, and the Lundahl and UTC use two coils. The single coil structure seems to exibit the lowest distortion on the sweep test, but has the biggest resonant peak.

Here are some of the important dimensions that are used a lot by transformer geeks. It is a standardized system, so if you tell a guy that the D, E, F, and G numbers are a certain value, they will most likely know what you are talking about:




The ratio of G to F can influence the cou[pling, leakage and HF response. A 3 to 1 ratio is a good balance for all of these. The Lundahl and UTC cores have a higher G to F ratio than the Sowters, which might be another reason for the reduced peaking in the HF.

OK, my mind is fried from the river and waiting for the shuttle to land last night. They have a Nasa/Ames building a few blocks from my house, so I am going back down tonight. They let the public in to watch Mission Control on a theatre size screen, and the place is full of eigineers that did the heat modeling on the computers to predict the saftey of the turbulance due to re-entry, so it is a cool place to hang out. Darn Steve was on the Hookah all last night, making those low clouds that prevented landinf last night, so I have to stay up again tonight!
:razz:

see ya,

cj

CJ,

Cool information... -VERY cool indeed! -In fact I think that's the sort of information you should copy, paste and post on a Vacuumbrain page, dontcha think? :thumb:

All I think that we're prepared to say at the moment is that nobody has guessed the correct transformers yet.

Okay this should be the last day or so, then maybe we'll reveal the contestants' true identities! :wink:

Analog Packrat, Tom Waterman and Dr. Frankencopter all included reasons why they made their guesses... I think that's really cool and very fascinating. -I think it's be great if everyone can make a guess and include a quick hint as to why they think "brand 'X' is the Jensen" or whatever...

:thumb:

Keep going everyone!

:thumb:

Keef

common guys!

take a stab at it..

we would like to reveal the trude identity :green:

CJ seems to always have a few more secrets to reveal, doesn't he? Very interesing, indeed.

A: Lundahl
B: Sowter
C: UTC

Why? CJ indicated that the single vs dual coil structure would cause reduced peaking in the HF response. Sowter (single coil) should have the highest peaking hence B. B also has the most trouble driving 600ohms at bass freqs and CJ hinted that the Sowter output iron might be high perm leading to early saturation. He also says single coil structure leads to best midrange distortion specs. This all lines up with B.

Smoother overall ("less lively core material" as CJ put it) is definitely C. Good old UTC.

That leaves Lundahl for A. Low end distortion is similar for 10k and 600ohm loads, so must be casused by the input iron. Interesting.

Thanks to Scenaria, Keef, and CJ for this educational experience. I might actually start to understand some of this transformer stuff one day...

Analog Packrat

Hi Keith....actually, I didn't take a stab at it, but now I feel compelled. Actually, I didn't realize we were playing "name that transformer".

Okay, maybe this will spill the beans, maybe it won't, but a little digging on the manufacturers websites (okay and CJ's site) reveals the following information:

For the input transformers, max level is:
HA-100X = +15dBm (dBm....I'll assume a 600 ohm load and call it +15 dBu)
Sowter = +16 dBu
Lundahl=+26 dBu (wow!)

Now, I'm going to assume that each one reads about 1% distortion at these values (true for the Lundahl, but others I'm not sure about).

Now, I will further assume that the La2a's under test were run at unity gain. Otherwise, we can't tell wether the input or output is likely to be the cause of distortion. My guess right now is that input trannys are distorting owing to their small core sizes.

So, with the lundahl at 1% distortion at +26dBu, that corresponds very well to the brand C curve of distortion vs. level.

The Sowter has the next highest max rating, which could correspond well to the brand B curve, though the knee is at more than +16 dBu. Well, it's kind of hard to tell the difference in the curves between A and B based on when disttortion rises suddenly. But, the Sowter should be a cleaner transformer than the older UTCs based on the way the sowters are wound. Curve B tends to sit lower than curve A throughout the range where the core isn't saturated...again leading me to think that B is Sowter. Brand B has a harder time driving the 600 ohm load than brand A, so seeing as the UTC were from the time when 600 ohm gear was the rule rather than the exception I will guess that brand A is the UTC, and brand B is the Sowter.

So, going against almost all the post above,

A: UTC
B: Sowter,
C: Lundahl

Cheers,

Kris

Edit: fixed some typos....probably left a few in there though

So far,

Buttachunk's first try and Peter Simondsen have both been:
A=Lundahl
B=UTC
C=Sowter

Analog Packrat's first try and Tom Waterman have both been:
A=Sowter
B=Lundahl
C=UTC

Buttachunk's second try, Rodabod and Kris (Frankencopter) have all gone for:
A=UTC
B=Sowter
C=Lundahl

Analog Packrat's second try and Butterylicious' vote is:
A=Lundahl
B=Sowter
C=UTC

drpat's guess is:
A=Sowter
B=UTC
C=Lundahl

So who's next?

-I'm not telling if anyone's got it yet...

Keef

Oh, OK, I am brain dead. I thought that Steve's A B C listing in post 1 was the order of the transformers. In other words, I did not know that you are supposed to guess which is which. Cool!

Don't tell us yet. Let me have a few more hours on this so I can guess.

...