ART T8 "Universal Eight Channel Transformer Isolator"

[Casey Butt]

I recently bought an ART T8 "Universal Eight Channel Transformer Isolator" specifically to experiment with using the transformers in it for "colour". I figured at such a low price for 8 transformers, they had to be "low-tech" steel transformers and not some virtually transparent nickel alloy high-end transformers.

So, to see what the ART T8 transformers are doing I ran some tests on the T8 using a Zoom R24 and ART Pro Channel II as interfaces (of which I also took baseline measurements for reference).

The first graph is the T8 through the Zoom R24. The R24 is driving the T8 with an output impedance of 1 kOhm and is loading the T8 with an input impedance of 50 kOhms. The red lines are the R24, the blue lines are the T8 running through the R24 and the green line is the T8 frequency response with the R24 response removed (i.e. the difference in their frequency responses). The flat-ish lines are obviously the frequency responses, the dashed lines are the phase responses, and the downward sloping lines on the bottom half of the graph are THD.



Judging from the linear-ish THD response, the T8 appears to have, as expected, nothing-fancy steel transformers.

Blowing up the frequency responses (which I won't attach here) shows that the T8 is down almost 2dB at 20Hz and almost 0.8 dB at 20kHz.

The next graph is the T8 through the ART Pro Channel II as an interface (with the Zoom R24 still sending the test signal and receiving (loading) the T8 at the other end). The ART has an output impedance of 300 ohms feeding the T8 (still loaded down at 50 kOhms on the Zoom R24 receiving end).



Here you can see a marked improvement in the frequency response, phase response and THD of the T8. Blowing up and centering the T8 frequency response (the next graph) shows the frequency response is now in the +/- 0.3dB range from 20Hz-30kHz. Much better.



I also took a look at some specific THD readings at 50Hz, 100Hz and 1kHz (the next three graphs). These were the minimum THD points I found while adjusting the level through the T8.







At 50Hz the whole "system" had a THD of 0.32%. The graph below is the THD @ 50Hz of the ART PC II through the Zoom R24 without the T8 in line, for reference.



Subtracting the THD of the PC II & R24 gives an approximation of the T8's THD @ 50Hz and this approx. dB level as 0.32-0.028 = 0.292%. Doing the same thing at 100Hz and 1kHz gives 0.102% and -0.007%. I didn't adjust the level when switching the T8 in and out of the line, so I suspect the negative THD value for the T8 at 1 kHz is coming from the fact that there's a slight dB drop going into the Zoom R24 when the T8 is in line. So there's some error involved, but these are, I believe fairly reliable ideas of what the T8 is doing a least to 2 decimal places. As expected the transformer distortion is most odd (3rd and 5th) order.

T8 minimum THD values:
50Hz: 0.29%
100Hz: 0.10%
1kHz: N/A

T8 Frequency Response:
20Hz-30kHz, +/-0.3dB

Phase Error:
<20 degrees, 20Hz~50kHz

...at least at these test conditions.

To put that at least in some sense of context, the 1176LN and LA-2A are both spec'ed at more than 0.35% THD in their "healthy" operating ranges and the LA-2A frequency is similar to what the T8 was showing here (the 1176 drops a bit more). Not saying that inserting a T8 will make your chain sound like a 1176 in line or anything , but putting it in context that it will affect the overall frequency response and THD levels similarly. In contrast, higher quality transformers would be practically transparent. As it is, these numbers aren't really that "dirty" for such a cheap piece of transformer gear, so unless you have very good ears you're probably not going to hear much of anything. On the other hand, this is a lot cheaper way of getting some transformer saturation into your chain than buying a Fatso or Warm Audio Tone Beast or something on that level.

I hope that helps someone who might be considering getting an ART T8, either for its intended purpose or as a transformer "colour box".

[Dave B]

Er .... thanks ... that all looks very interesting.

But just for those of us who are reading early in the morning before the brain has woken up, is there a 'capital letters and crayons' version of this?

[Dave B]

I recently bought an ART T8 "Universal Eight Channel Transformer Isolator" specifically to experiment with using the transformers in it for "colour".

... (snip) ....

I hope that helps someone who might be considering getting an ART T8, either for its intended purpose or as a transformer "colour box".

The T8's audio transformers have flat and wide frequency response and handle high signal levels while maintaining an isolated balanced output. This gives the T8 a very clean and neutral sound with a wide variety of signal sources

Just noting that Art think that it should not colour the sound. Your post seems to confirm that they do that quite well for a cheap product (if I read it right)

[ef37a]

All interesting stuff!
Just one point? 300 Ohms is a rather high output Z for modern gear? Even my venerable 2496 cards are only 100R.

I think most decent gear today has an OPZ nearer 100R and really classy kit around 50R. Some will even have an effective output resistance close to 1 Ohm or less.

I would be interested to see the distortion figures at these lower drive resistances since at zero R it should be bugger all! There will of course always be some primary resistance.

Dave.

[Hugh Robjohns]

But just for those of us who are reading early in the morning before the brain has woken up, is there a 'capital letters and crayons' version of this?

For the money, they are reasonably decent transformers that work well if driven from a low source impedance (which most stuff has, unlike the Zoom's 1K ohms used here!). Distortion becomes moderately significant at low frequencies, especially at high levels, but not unusably so.

H

[Casey Butt]

I was somewhat pleased to find that they fell right in line with ART's published specs, at least to the extent that I could test. As I said, I didn't want pristine clean transformers or there would be no point for me. But having said that, I really can't audibly detect a single T8 transformer in the line. Perhaps I can, or perhaps it's my mind. Maybe if I tried a perfect low-frequency sine wave I might notice something more obvious, but on everyday music, a single T8 transformer is practically transparent unless you have ears like a whale or something. I think a really trained ear might detect the transformer more than me.

One thing that drew me to the T8 for this purpose were some of the more expensive pieces of gear that use "cheap" steel transformers deliberately for colour also. I don't need another preamp or compressor, I just wanted the transformer vibe. So the T8 has eight steel transformers that distort at just the threshold of perception (if you have good ears) with most equipment. And since there are 8 channels, each with its own transformer, if you can't hear a single transformer's presence then you can chain up to eight of them together until you do get what you want.

I also considered taking a transformer or two out and putting them in their own separate drive circuit with an adjustable output level so I could crank the gain into them but keep the overall output level low... similar to the "tranny" circuit in Empirical Labs' Fatso. But the testing I did on the T8 makes me think that they can take a pretty hard hit before the distortion levels rise much. Driving impedance was the much more sensitive factor to play with there. At 50Hz, I really had to drive the gain hard before I got much in the way of distortion level rise, even then I couldn't get it to 0.5% with the ART Pro Channel II on blast (without clipping in the PCII itself).

Actually, the T8 turned out to be exactly what I expected and was after. The bigger surprise was the ART S8 (eight channel transformer-based mic splitter) I got at the same time. I did all the same tests on that, with the same equipment, and it showed significantly better technical performance. I couldn't get them to go above 0.1% THD @ 50Hz without going to a pretty severe gain range. For the practical range you'd send through them they stayed at 0.08% THD or below @ 50Hz. They were at ~0.03% THD @ 100Hz. Even the shape of the THD vs frequency curves makes me think there's something more in the transformer cores than just cheap steel. They have that negative exponential shape you get with some nickel in the transformer. It still isn't a top-of-the-line Jensen or anything but surprised me for something relatively cheap. The S8 costs 60% more than the T8 though, so I suspect the higher grade transformers are a large part of the reason for that. Physically, the S8 transformers look like the T8 transformers (same size, shell casing, etc) but the windings are wrapped in a different colour casing... probably to avoid mixing them up in the factory.

When I get some free time I'll post the results with the S8. Actually, when I get around to it, I'm going to take one of the S8 transformers and put it in the Pro Channel II preamp on the mic input. Then I'll be able to switch the transformer in and out of the circuit and adjust the load impedance and see what happens. Again, I'm looking for transformer "colour", but I don't think I'll get much at <0.08% THD @ 50Hz and 0.03% THD @ 100Hz.

[Casey Butt]

Just noting that Art think that it should not colour the sound. Your post seems to confirm that they do that quite well for a cheap product (if I read it right)

I think it comes down to the "real" world vs. the technical spec world. In terms of specs, there are much better grade transformers than those in the ART T8, but in terms of the human ear and typical music and equipment will you really hear it? Maybe not. Past a certain point, better specs become something of a moot point because better specs aren't necessary when something gets already "good enough". The T8 seems to be sitting right on the line that the average person wouldn't tell the difference... but an experienced engineer listening to a bassoon probably would. Plenty of consumer and even professional grade gear have more THD than we're talking here. So ART can make the claim that the T8 has a "very clean and neutral sound" because to a *reasonable* real-world human ear that's true. The high-end spec world is a different story.

[Jeraldo]

Thanks for doing this. Because the frequent mention here of Art's products with transformers, they've always been on the back-burner for me for potential problem solving, and should I need them, I'll have no hesitation in ordering.

[Casey Butt]

I would be interested to see the distortion figures at these lower drive resistances since at zero R it should be bugger all! There will of course always be some primary resistance.

Good idea! I ran it through 1 kOhm, 600 Ohms and 300 Ohms and saw the clear trend of improving response with each lower driving resistance, but never really though of going lower. I'll check the output impedances on some of my other gear and take some more responses at lower drive resistances.

[damoore]

Strictly speaking you can't subtract total distortion figures to get the distortion of an individual device, since the triangle inequality is working against you.

However very interesting and thanks for posting. I have been considering getting one of these for the troubleshooting porpoise.

When you tested the S8 I assume you did so at mic levels rather than line levels? I wonder if they are the same transformer.

[ef37a]

With THD figures to two decimal places being bandied I am minded to ask if there has been any recent research on our tolerance to THD?

You see my "head" is still back with Mullard and Peter Walker et al and "0.1%" being very good and 1.0% being JUST detectable.

I understand of course that it depend upon what and how you are measuring and with what but are there any new ball park numbers?

Dave.

[Casey Butt]

Strictly speaking you can't subtract total distortion figures to get the distortion of an individual device, since the triangle inequality is working against you.

That's why I was careful to say approximate.

When you tested the S8 I assume you did so at mic levels rather than line levels? I wonder if they are the same transformer.

I started off at mic levels, but then I was curious to see how the S8 transformers reacted to higher levels so I cranked it up. It really didn't make much difference to the responses so I thought they might be the same transformers as well... so I opened the units up and the S8 transformers are very similar to the T8 transformers in size and appearance, but with different colour winding wrapping.

But you've got me second guessing that, so to be absolutely sure I'm going to check the model numbers stamped on the transformer casings and run identical tests on both units. The S8's seem to be higher grade transformers than the T8's, but I'll make sure to confirm that.

[Casey Butt]

With THD figures to two decimal places being bandied I am minded to ask if there has been any recent research on our tolerance to THD?

You see my "head" is still back with Mullard and Peter Walker et al and "0.1%" being very good and 1.0% being JUST detectable.

I understand of course that it depend upon what and how you are measuring and with what but are there any new ball park numbers?

Dave.

I have to wonder about that too. I've seen some crazy low THD numbers from Jensen and others... far below what even high-grade gear specs out at it. I can see the use of that if you're adding to a circuit with already high-ish THD or are chaining a bunch of these things in the line, but for an individual piece of equipment I think it's largely a moot point. Today's junk is sometimes yesterday's high-end.

[Casey Butt]

I tested the T8 with some other equipment that has lower driving impedances. The first graph is through the preamp out of the ART Pro Channel II, which is spec'ed at 300 Ohms.



Here the low frequency response of the T8 actually slightly exceeds the baseline setup below about 60 Hz. It's up by about 0.4dB @ 20Hz.

The next is through a Rocktron Hush with deiving impedance spec'ed at <100 Ohms. Here the low frequency response of the T8 exceeds the baseline setup below about 100 Hz. It's up by about 2.5dB @ 20Hz.



The THD drops significantly with these lower driving impedances. The next two graphs are the THD of the T8 through the Hush @ 50Hz, and the system without the T8 in line. The THD of the T8 itself is down to roughly .184% @ 50Hz and is almost purely odd order harmonics (as it should be for a transformer).





Overall, I'm pleased with what the T8 is doing. From the shape of the THD response it's obviously a steel transformer (and the price gives that away), but driven by the appropriate equipment its THD levels are on the lowish side of what I think even a good ear would be able to reliably detect under most circumstances. And if I did want to detect the transformer "colour" (which I do), I'd just chain a few of the T8 channels in series until I got where I wanted.

[Hugh Robjohns]

And if I did want to detect the transformer "colour" (which I do), I'd just chain a few of the T8 channels in series until I got where I wanted.

And that is exactly what happened back in the day when transformers ruled the earth. In the days of valve consoles, every gain stage had transformers in and out, every tape recorder channel had transformers in and out, and every outboard processor had transformers in and out. So it was easy to end up with 10-20 transformers in the signal path without even trying.

The advent of transistor consoles moved the electronics inboard and a lot of transformers were dispensed with, but they still persisted on the console, tape machine and outboard I/Os.

It was only in the 80s that electronically balanced I/Os became more common -- if not the norm -- both for cost and quality reasons.

One of the fundamental reasons why digital recordings today sound so different from the analogue recordings of the 50s, 60s and 70s, is not the absent tape; it's the absent phase shifts and concatenated high-pass filtering imposed by all those transformers and audio interconnections between outboard, recorders, and consoles!

H

[Casey Butt]

I ran some tests to directly compare the ART T8 and S8 transformers through 200 Ohms driving impedance (which seems to be a happy medium for both the ART T8 and S8). I fed into the Zoom R24 1 kOhm balanced inputs.

These tests were done under exactly the same conditions. I simply switched cables from one unit to the other and left all settings the same.





As you can see, the S8 and the T8 are clearly not using the same transformers. Furthermore, even at these line levels, the S8 performs very well... even though it's designed for mic levels. At mic levels the S8 is surprisingly good. Here's a shot with the S8 input padded down -29 dB (for use to knock line levels down to mic levels).



This is almost stupid good for a cheapish piece of transformer gear, IMO. The frequency and phases responses are excellent and the THD is down to irrelevancy. I wouldn't try using the S8 for "colour" because these results are too clean and accurate for that, but it does show that the S8 transformers are certainly up to the task of pretty faithful signal passing at mic levels.

[Martin Walker]

the absent phase shifts and concatenated high-pass filtering

I LIKE that!

Martin