Topaz Low-Capacitance Isolation Transformers - for Affordably Clean Power
Aug 8, 2017 at 4:38 PM Thread Starter Post #1 of 562
Joined
Apr 28, 2007
Posts
5,494
Likes
1,084
Have you been looking for a very effective, yet affordable solution for removing common-mode and normal-mode noise from your AC mains? How about a way to prevent noise from being injected back into the mains from your audio components' power supplies? How about a way to prevent ground loops between components and to discourage leakage currents? How about a way to remove any DC offset that might be coming in from the mains? And how about surge protection?

Several Head-Fi members are already aware of some other forums where people have done a lot of trailblazing with the use of "Series 30" Topaz Ultra-Isolator Isolation Transformers, such as those shown in the following PDF.

Note that only the "Series 30" models offer a 0.0005 pF inter-winding capacitance, which yields -146 dB attenuation of common-mode noise.

Other "Ultra-Isolator" models have higher capacitance ratings, with less noise attenuation.


http://www.pacificparts.com/vends/mge/images/t1.pdf

Here's my Topaz Ultra-Isolator 91095-32, rated at 120V, 50/60 Hz, 500VA, with a low inter-winding capacitance of only 0.0005 pF (recently purchased used, on eBay):

Topaz_91095-32_Ultra-Isolator_4_x960x2.jpg

Though long obsolete, the Topaz Ultra-Isolator "-31" and "-32" models ("30 Series") are readily available on the used market, at various VA ratings, as shown in the PDF at the link, above.

They are unique in that, unlike many other Isolation Transformers, including the Tripp-Lite is500, is1000, or is1800, which are still in production, the Topaz Ultra-Isolator -31 and -32 transformers have extremely low inter-winding capacitance, which equates to being far superior at removing common-mode noise from the treble frequencies.

The following page, shows additional specs and some interesting performance curves, for the exact same transformers, re-branded under the Daitron logo.

http://daitronglobal.com/products/power/power-supply-transformer.html

Greater than 146 dB (attenuation) throughout the audible spectrum, up to 20kHz, equates to a 20,000,000-to-1 reduction of common-mode noise. They also provide 65 dB (1780:1) reduction of normal-mode noise.

John Swenson (of Uptone Audio) attributes their ability to remove power line noise from the higher audio frequencies to the very low inter-winding capacitance of these "30 Series" Ultra-Isolator transformers (per several posts he has made at www.computeraudiophile.com).

This first chart shows that below 20 kHz, the attenuation of common-mode noise is even greater than -146 dB, going below -150 dB, at audio frequencies as high as 8000 Hz.

910_1.gif

Source: http://daitronglobal.com/products/power/power-supply-transformer.html

I'll add a follow-up later, but here's a (revised) guide to my current setup for power conditioning, using the Topaz 19095-32, pictured above, as well as some B&K Precision 1604A isolation transformers (currently in production), which play a different role in removing common-mode noise, as well as preventing any "backwash" noise (especially from DACs and any component having a SMPS, from getting back onto the mains to pollute other components).

Affordable_Power_Conditioning_with_Serial_Isolation_Transformers_Rev3.jpg


More later...

Mike

UPDATED this post on 27 Aug 2017, to withdraw my prior, misleading contention that isolation transformers with Floating-Neutral Secondaries attenuate Normal-Mode noise, instead of Common-Mode noise. After a great deal of research, I'm now convinced that Floating the Neutral of the Secondary neither prevents Common-Mode noise reduction, nor allows Normal-Mode noise reduction. In short: All isolation transformers attenuate Common-Mode noise, whether the Secondary's Neutral is Floating or Grounded. (Keep reading.)
 
Last edited:
Aug 8, 2017 at 7:53 PM Post #2 of 562
Prior to reading John Swenson's posts about the need for a low inter-winding capacitance, I had been very content with the lowering of my noise floor, as had with a modern Tripp-Lite is500 isolation transformer. But John pointed out that, unless the transformer is designed for low capacitance, as with these obsolete Topaz Ultra-Isolator transformers, common-mode noise that comes in from your mains will not be attenuated very well above 400 Hz.

I had assumed that sticking with the Tripp-Lite is500 was OK, however, because I've got a TRIAC dimmer switch in the dining room of my house, that when set to at about 50% of full brightness, generates enough noise for me to audibly raise my noise floor, when listening to unprotected, AC-powered Head-Fi gear that's plugged directly into the wall outlet. Inserting the Tripp-Lite is500 isolation transformer between the wall outlet and my AC-powered gear, was audibly lowering my noise floor, with that dimmer switch set to 50%. So... I was justifiably content, but erroneously concluded, "My noise floor can't get any lower than this, the Tripp-Lite is doing a great job!" Nope. It was doing only a mediocre job!

That's the insidious thing about noise: You don't know it's there, until you get rid of it!

With several people reporting that these old Topaz Ultra-Isolator transformers were opening up their soundstage, with more detail and separation in the treble, and with the encouragement of a friend, I decided to give the Topaz 19095-32 a try (pictured above), replacing my nearly two-year-old Tripp-Lite is500.

The difference made by going to a transformer with an extremely low inter-winding capacity was immediately discernible, most especially in the treble frequencies - just as John Swenson has reiterated. It was a real slap in the face, with my having stubbornly stuck to using my Tripp-Lite is500 for so long, but in the end, I'm very happy with the decision to get the Topaz, of course.

Now, the decays last much longer, as their energy descends towards a noise floor that is much deeper. The topology of the recording space becomes more tangible with those low-volume echoes and reverberations that can make everything sound more natural and less "reproduced." Micro details that were buried in the hash previously, are now revealed, better than ever.

It really makes me wonder just how much lower the noise floor could possibly go.


At this point, I believe its lower than what I'm getting from battery-powered portable gear, but that's probably just the difference in detail offered by my best desktop DAC (the Oppo Sonica DAC) and my best portable DAC (Oppo HA-2) - which is significant. If I could power my Sonica DAC with a DC battery pack, it would be a fair comparison. Still, even when I use a portable amp with the Sonica DAC, I can hear the noise floor drop when I power the DAC with the Topaz Ultra-Isolator 19095-32 instead of with the Tripp-Lite is500.

Note that, all along, for over a year now, I've also been using the B&K Precision 1604A isolation transformers, placed in between the Tripp-Lite is500 and my AC-powered digital components. I am continuing to do so, having replaced the Tripp-Lite with the Topaz.

So, the AC power reaches my DAC (and a DC power supply that provides 7.5V to an Uptone Audio UDB Regen), through this chain, but the DAC and the DC power supply, each have their own B&K PRecision 1604A transformers (see the chart above):

120V Wall Outlet > APC LE1200 Voltage Regulator, set to 107V > Topaz 19095-32 > B&K Precision 1604A

B&K_Precision_1604A.jpg

https://www.amazon.com/Precision-1604A-Single-Isolation-Transformer/dp/B000LDLF3M


Grounded-neutral isolation transformers (i.e. Topaz and Tripp-Lite) and floating-neutral isolation transformers (B&K Precision and some models of Hammond Mfg.) differ significantly, as shown in this diagram:

IsolationTransformer_Floating_vs_Grounded_Secondary_Rev2.jpg


UPDATED on 27 Aug 2017: The graphic, above, has been revised to correct my former, incorrect assertion that isolation transformers with Floating-Neutral secondaries attenuate Normal-Mode instead of Common-Mode noise.

If you were only going to buy one isolation transformer, get a low-capacitance Topaz Ultra-Isolator, selecting one that supplies at least twice the VA (volt-amperes) as the total of all of the loads you intend to power. Note that 500 VA is very roughly equivalent to 500 Watts (but only for purely resistive, non-reactive loads, blah, blah, blah...)

Again, refer to the linked PDF in the first post, to research the Topaz model number you need for the gear you intend to use and remember that for power amps (for speakers) you don't want to cut yourself short on selecting a big transformer. The load of an entire headphone system, on the other hand, can easily be handled with 500VA, typically. If in doubt, looking at the power specs of each component, you can always try measuring the load of each component, in VA or in Watts, while operating, by inserting a Kill-A-Watt between the power cord of the component and the wall outlet.

https://www.amazon.com/P3-P4400-Electricity-Usage-Monitor/dp/B00009MDBU

Note, also, that some of the Topaz Ultra-Isolators (model numbers ending with "-32") come with ready-to-use power plugs on the Primary side and power outlets on the Secondary side of the transformer, while others (the "-31" models) are meant to be hard-wired, with only conduit punchouts and recessed screw terminals. Note, too, that even my small, 500VA model, weighs 24 lbs. The larger ones get to be quite heavy (and expensive).

Back to discussing the chart, immediately above. John Swenson earnestly contends that there's no difference in common-mode noise attenuation between isolation transformers that have a grounded-neutral secondary (i.e. the Topaz Ultra-Isolators) and those which have a floating-neutral secondary (i.e. the B&K Precision 1604A) - at least not any differences that can be solely attributed to the the presence or absence of that shunt between the neutral side of the secondary and the safety ground. To be fair, I will add that he even proposed a test that can be performed with an oscilloscope, to show that attenuation of common-mode noise (voltage transients or spikes that occur between Ground and Neutral or between Ground and Hot) would be just as effective in a floating-neutral isolation transformer as in a grounded-neutral isolation transformer.

Please understand that my respect for John Swenson is immeasurable and let's not forget that he's a very accomplished EE, responsible for some very popular, state-of-the-art designs, but on this point, his contention is up against a large consensus of opinion I've seen in many other writings about isolation transformers and, in the end, it doesn't really matter whether a floating-neutral transformer can, as John says, reduce common-mode noise just as well as a grounded-neutral transformer, as long as you are using both, in series, as I do.

Note that now, after a LOT of reading, I am in complete agreement with John Swenson's position, and thus, I'm having to go back and edit many posts in which I've mislead readers. (The truth will set you free!)

We both agree the grounded-neutral secondary of the Topaz transformers is eliminating common-mode noise - that which is most harmful to our gear, in addition to being the most audible at consistently raising the noise floor. And in the case of the Topaz Ultra-Isolators, where the attenuation is 20,000,000:1 or better, right out to the end of the audible spectrum at 20kHz, thanks to their low capacitance, there's hardly any justification for seeking even more attenuation of common-mode noise by adding another transformer, in series.

But for careful examination of my first chart (in the first post, above), you might now be asking yourself why I would bother with inserting the B&K Precision 1604A floating-neutral transformers in between the Topaz and my digital components. Answer: Not only is there a consensus of [absolutely incorrect] writers who say floating-neutral transformers provide a 1000:1 attenuation of any remaining normal-mode noise coming from upstream [not true] - but I'm sure even John Swenson would agree that isolation transformers with floating-neutral secondaries can prevent "backwash" noise, which can emanate from the power supply of each component), from polluting other components that are plugged into the Topaz that resides upstream. They can also prevent ground loops. [These latter claims for transformers with floating-neutral secondaries are true.]

And as many people like to say, when it's all said and done, it's what I hear (or don't hear) which convinces me that there's no point in worrying about whether or not common-mode noise is attenuated by floating-neutral transformers just as well as with grounded-neutral transformers. I prefer the sound of the Topaz + B&K transformers, in series, supplying power to my Sonica DAC, to what I hear with only the Topaz - and I have the peace of mind of knowing that nothing else which is plugged into the Topaz can be polluted by any noise coming back from the Sonica DAC (or from the AC-powered DC power supply that I use to power an Uptone Audio USB Regen, plugged into the USB input of the DAC).

See the first chart, in the first post, above, for additional details, such as not using a surge protecting or circuit-breaker equipped power strip to give yourself more outlets on the output of the Topaz. As John Swenson has written, you want a simple power strip with as little wire between outlets and as short a cord as possible coming from the Topaz. By plugging all of your components into that one power strip, the impedance between components can be kept low, which helps to reduce leakage currents. (I hope I'm getting that correctly.)

Mike
 
Last edited:
Aug 9, 2017 at 1:00 PM Post #3 of 562
By the way, for those in countries with 240V power...

The Terminal style models of the Topaz Ultra-Isolator models are rated for 120 or 240, and have screw terminals that can be wired to the appropriate style of outlets - but please seek the advice of a skilled electrician (or hire one, outright) to do this work.

See the PDF at the link provided in the first post: http://www.pacificparts.com/vends/mge/images/t1.pdf
 
Aug 9, 2017 at 5:54 PM Post #4 of 562
I have my Topaz 91002-31 (2.5kVA, .0005pf) wired to provide balanced power (60 volts to Hot and Neutral relative to Ground). You might want to discuss the advantages of balanced power. The noise floor is supposed to be lowered even more (increase in dynamic range of 16dB to 20dB). Also, this Topaz had an audible hum that went away once wired for balanced power and it runs cooler.
 
Aug 10, 2017 at 6:51 PM Post #5 of 562
Hi Winders,

Thanks for joining in! I'd heard that some of the terminal-style Topaz Ultra-Isolators allow you to "center-tap" the Secondary, to very effectively create a balanced transformer, but mind doesn't allow that and I'm reluctant to experiment with balanced power - perhaps unjustifiably. One thing I've heard for sure is that you don't want to plug any kind of lamp or other appliance into the outlet of a balanced power transformer, if the switch for that lamp or appliance is a SPDT switch vs. a DPDT switch. In other words, if the switch only interrupts power on one side, the appliance will still see power coming in from the other half of the transformer.

Keep in mind that nearly every load we might ever plug into a balanced power transformer was not designed for balanced power. So, it would be best to just unplug the component from a balanced power transformer rather than using its own power switch to turn it off. Better still would be to install a DPDT switch between the balanced power transformer and the appliance.

I'm curious about your having referenced some source that says the noise floor is lowered even more (16 to 20 dB), but I'm content with the -146 dB attenuation had with my "unbalanced" transformer. :)

Update:

I found this article on the topic of balanced power. Keep in mind that in the UK, whenever they talk about an "isolation transformer," one can assume they are talking about a a 1:1 transformer that has a floating-neutral secondary. Just watch any YouTube video on isolation transformers, where the speaker has a British accent, and you'll see that they equate "isolation" with floating-neutral secondaries and they refer to grounded-neutral secondary transformers as something primarily seen in America.

Sorry about this long .jpg file, but I created it as a means of annotating the article that can otherwise be found at the link provided. (Open it in a new tab, then enlarge it.)

Crosstalk_ Readers' Writes - Balanced Mains - Annotated.jpg

Source: http://www.soundonsound.com/sound-advice/crosstalk-readers-writes

Quoting pp. 20-21 of http://www.surgex.com/pdf/PowerGround.pdf

NEC 647, which defines the requirements for balanced power systems, places some important restrictions on both their installation and use.

1. Conductors must be sized so that the IR drop does not exceed 1% of the line voltage under a load equal to 50% of the branch circuit current rating, and so that the combined IR drop of the feeders and the branch circuit wiring does not exceed 2%.

2. A dedicated Equipment Ground conductor must be run to all equipment and each receptacle.

3. All receptacles must be protected by a GFCI.

4. The neutral must be bonded per NEC 250, and must also be connected to the grounded conductor of the circuit that feeds the system.

5. Balanced power systems are restricted to industrial and commercial occupancies.

6. All outlets “shall have a unique configuration” and must be identified using specific language called out in NEC 647.7.

7. There must be a receptacle having a grounded circuit conductor (i.e., conventional unbalanced power) within 6 ft of each receptacle for the balanced power system.

8. All lighting fixtures connected to balanced power must be specifically rated for 60/120 VAC balanced power, must “have a disconnecting means that interrupts all ungrounded conductors,” and must be permanently installed.

9. Isolated ground receptacles are permitted.​

Balanced power systems are expensive, and their noise reduction capability is limited to about 10 dB. Isolated ground systems are generally a far more effective and less costly solution

Here's another reference:

Balanced_Power_Needs_GF CI.jpg
 
Last edited:
Aug 11, 2017 at 8:09 AM Post #6 of 562
Here's another PDF that may be of interest:

http://www.surplussales.com/Transformers/Pdfs/tp-91092-31_N.pdf

It has some interesting discussion about adding capacitors on the downstream side of a Topaz, as well as substituting the mains earth ground with one of your own making. Scary. :)

It also provides hookup instructions for the models that have terminal strips instead of ready-to-use, three-prong outlets.

UPDATE: It also provides (poorly written?) instructions for how to set the jumpers on the -32 (terminal style) models. Here are some helpful, related photos:

Topaz_Jumpers.jpg


The jumpers seen in the photo above are different from those found in some models (below):

Topaz_Jumper_plate.jpg


Wire can be used, instead, of course.
 
Last edited:
Aug 12, 2017 at 6:13 AM Post #10 of 562
Kind of thinking of making a diy one.. from the antek 1.5KV dual 58V secondary for balanced power.. and atl hifi has a pcb for dc blocking and soft startup..
 
Aug 12, 2017 at 6:33 AM Post #11 of 562
[1] Have you been looking for a very effective, yet affordable solution for removing common-mode and normal-mode noise from your AC mains? [2] How about a way to prevent noise from being injected back into the mains from your audio components' power supplies? [3] How about a way to prevent ground loops between components and to discourage leakage currents? [4] How about a way to remove any DC offset that might be coming in from the mains?

1. No. That's something I would expect even a cheap, mass produced bit of audio kit to deal with, let alone an audiophile bit of kit!
2. No. That's something I would expect even a cheap, mass produced bit of audio kit to deal with, let alone an audiophile bit of kit!
3. No. That's something I would expect even a cheap, mass produced bit of audio kit to deal with, let alone an audiophile bit of kit!
4. No. That's something I would expect even a cheap, mass produced bit of audio kit to deal with, let alone an audiophile bit of kit!
 
Aug 12, 2017 at 6:58 AM Post #12 of 562
hi gregorio I see from your profile you do a l a lot of pro audio stuff. You might want to read zilch0md's questions more carefully, most of those things he mentions are almost never implemented in normal audio gear and some like ac leakage current was only recently known about and you need either a battery supply, dual bank supercap supply etc to remove..
 
Aug 12, 2017 at 10:28 PM Post #13 of 562
Kind of thinking of making a diy one.. from the antek 1.5KV dual 58V secondary for balanced power.. and atl hifi has a pcb for dc blocking and soft startup..

You're a handier handyman than I am. Even soldering a new fuse holder into my Topaz was challenging, for me.

Still, though I congratulate your DIY skills, please consider that my Tripp-Lite is500 isolation transformer doesn't clean up the treble frequencies anywhere near as well as the Topaz 19095-32. They are schematically identical! The difference is in their construction, with the latter achieving the very low capacitance of 0.0005 pF.

In addition to John Swenson's "pot banging" on the need for this low inter-winding capacitance, when selecting an isolation transformer, there are some people posting to the same Computeraudiophile threads who have bought Topaz models with higher capacitance (i.e. 0.005pF), then deciding to buy and compare a 0.0005pF version, only to readily conclude they can hear the noise floor drop further still, in the higher frequencies.

What we are observing, when we get the noise floor down this low, is that it is tilted, not flat! It's easier to clean the lower frequencies, as evidenced by the first of four graphs, in my earlier post, above.

:)
 
Last edited:
Aug 13, 2017 at 7:46 PM Post #15 of 562
Mastec_01_Common-mode-Noise-Reduction.jpg

http://www.mastec.co.nz/JST/PK8.html

Quoting from the paragraph above: "With all transformers, stray capacitance, called inter-winding capacitance, can exist between the windings. It is via this inter-winding capacitance that [incoming] high frequency [common-mode] noise [from the mains] can couple through to the secondary winding. However, when a grounded [Faraday] shield is used to separate the primary and the secondary windings, the inter-winding capacitance is significantly reduced. This effectively increases the impedance (resistance) of the coupling path and therefore reduces the amount of high frequency [common-mode] noise that can couple through to the secondary winding."

Thus, the lower the inter-winding capacitance, the greater the reduction of high frequency common-mode noise.


Mastec_02_Common-mode-Noise-Converts_to_Normal-mode.jpg

http://www.mastec.co.nz/JST/PK8.html


Info about the sources of residential electrical noise:

Residential_Sources_of_Noise.jpg

PowerConditioners_cannot_not_do_what_ITs_do.jpg
 

Attachments

  • Mastec_03_Normal-mode-Noise-Reduction.jpg
    Mastec_03_Normal-mode-Noise-Reduction.jpg
    208.6 KB · Views: 0
Last edited:

Users who are viewing this thread

Back
Top