Bryston article on transformers - Part 2


The current Bryston newsletter has "Part 2" of their article on transformer design, and compares "EI" transformers to toroids. The two articles are well written without being overly technical. Here is the link to "Part 2":

http://www.bryston.ca/newsletters/85_files/vol8is5.html
sdcampbell
Where's "part 1" at? I'd like to read that before reading part 2. I somehow missed it if it was previously posted. Sean
>
Hi, Sean:

"Part 1" of the article was published in the previous edition of Bryston's online newsletter. You can access all of the previous issues of the newsletter from the Bryston home page (www.bryston.ca), or simply use this link to go directly to the previous issue:

http://www.bryston.ca/newsletters/84_files/vol8is4.html

Best regards,

Scott C-
Hmmm... Let's see here. We have an article comparing EI type transformers to toroidal transformers. The article just so happens to be written by a guy that works for a company that designs and manufactures toroids. On top of that, the party presenting the article is a company that uses toroidal transformers in their products. Not too hard to figure out which transformer is going to win this comparison.

Only problem is, the author forgot to mention that toroids have FAR greater coupling between the primary and secondary windings due to stray capacitance. This reduces the isolation between the mains and the component power supply by a drastic margin. As such, far more stray noise and interference is allowed to enter the power supply circuit, both raising the noise floor and potentially playing games with the calibration / linearity of the circuitry.

On top of that, the mutual coupling between the primary and the secondary works both ways. That is, noise that is generated within the component on the secondary side of the transformer is coupled back into the primary side and allowed to exit back into the mains at the same rate that noise is allowed to enter into the system. Devices using switching supplies and digital / RF based circuitry "leak" back into the AC system, which can then re-enter other components. This is why some PLC's make use of individual feeds for each component i.e. to minimize inter-system AC contamination.

As far as saturation goes, toroids of a given power rating tend to saturate faster than an EI of the equivalent power rating. That's because manufacturers tend to use less core material due to the ferrite's slightly higher efficiency (8% according to this article ). Like any other type of manufacturing, these companies are looking to keep production costs down and profit margins up. Since most devices don't pull anywhere near their rated power and are basically steady state loads, they can get away with this under most circumstances. The only problem is that music is dynamic by nature and power required is anything but steady state. As such, large dynamic swings will tend to pull as much or more than the "slightly under-sized" toroidal core can handle. When this happens, you run into non-linear hysterisis distortions as a result. The result is loss of bass control, severe treble grain and glare and a general loss of liquidity. Granted, this part of my argument takes for granted that the toroid is slightly undersized and the EI is "adequately sized", but my experience says that this is not unrealistic or off-base. That's because companies that are willing to invest in the greater cost of the EI transformer to begin with are actually concerned about quality. They therefore spec and buy components that are up to the task at hand.

Continuing on with core saturation, distortion and noise, i'm sure that most of you have seen more than a few posts about toroidals humming, buzzing and vibrating away in various pieces of gear. Evidently, toroids aren't as good as the author thinks, there's a lot more DC on the lines than he's aware of, manufacturers don't know how to install toroids correctly, etc... or some of this article has been idealized with little respect for real world situations.

If you doubt any of the above, try looking for "hospital grade" or "lab grade" isolation transformers and / or power line regulation systems. Chances are, you won't find anything that uses a toroidal transformer at the heart of the circuitry. That's because EE's know that when performance counts as the bottom line ( regardless of cost, weight or size ), you simply can't compare an EI type transformer to that of a toroid. Audio manufacturers and parts suppliers may tell you something different, but when those that make their living with precision test and manufacturing equipment that could be the difference between life & death and maintaining repeatable levels of a given performance standard choose a completely different design, that should tell you something. Sean
>

PS... As i mentioned in another thread, even though John Curl has designed several pieces for Parasound that utilize toroidal transformers, he has publicly stated that EI's are far superior to toroidals when it comes to getting rid of "grunge" in a power supply circuit. I can't point you to a specific thread to verify this, but if you dig in the AA archives, i'm sure that you'll find it.
Sean- Interesting post. Have you replaced any toroidal transformers in your equipment with EIs? If so, what were
the result and what brand of EI do you recommend?
Kana: It is rather difficult to replace a toroid with an EI. First of all, most toroids are relatively smaller transformers, so an EI of the proper rating won't fit in the alloted space. As mentioned, the core size of a toroid is typically measurably smaller even though it is only a few percent more efficient than an EI of similar power ratings.

On top of that, the added weight due to the increased mass of the larger IE core may overload the chassis, causing physical damage. This is not to mention that the use of an EI requires more attention to circuit design and component lay-out within the chassis. As such, this type of modification isn't a "drop in" in most cases.

What i have done is selected amps that work best in given situations i.e. optimized the components for the jobs at hand. That is, amps that are typically used to drive "tough loads" with large dynamic loads make use of EI type transformers and / or toroidals that make are unusually large in size.

Regardless of the type of transformers used, all the components have been bolstered with greater quantity of power supply reserve with increased filtration. I've slowly been going through them component by component and replacing the stock rectifiers / diodes with higher quality devices, which further lowers the noise floor and reduces transient hash.

In order to help bolster the performance of all of the components in any of my given systems, they are all fed from a central source that is filtered by 400 lbs worth of EI isolation transformers. These transformers are rated at -145 dB's of filtration. On top of that, certain components are further isolated individually in the systems with smaller EI isolation transformers. This is done in order to minimize the "nasty leakage" that these components tend to feed back into the AC system, which could further contaminate the other components after the main filtering was already achieved.

Taking that a step further, the power cords used are of a design that is of a low inductance characteristic with a geometry that is naturally resistant to RF and EM interference. Since power cords are wire and wire can act as both an incoming and outgoing antenna, i didn't want to have to worry about that either.

What one can do to get around the lack of space inside of a toroidally based product, especially in an amp, is to "stack" another identical toroid on top of the original. The amplifier can be converted to "dual mono" ( if it isn't already ), which increases the core capacity, reduces the potential for saturation and lowers the impedance of the power supply on the whole. All of this adds up less draw on the transformer, which also reduces transformer induced noise and allows more clean power as needed. By separating the channels, crosstalk between channels is improved and dynamics should improve markedly.

We've used this approach on the four amps that my brother runs, which are toroidally based and had smaller core sizes. This has resulted in better performance in every aspect, so it was worth it as far as he's concerned. Sean
>