The AI-1 Dreadnought Project Pt.1
[Actually, the Dreadnought was the beginning of my fifth SDA honeymoon: (1st)stock-->(2nd)new xovers-->(3rd)new tweeters-->(4th)new SDA interconnect. Sorry...I lost track.:)]
Originally Posted by DarqueKnight
I recently completed an implementation of the AI-1 interface. The AI-1 is an isolation transformer that allows some SDA models to be used with non-common ground amplifiers. The Parasound Halo JC 1 monoblock amplifiers I currently use with my SDA SRS 1.2TL's do not require an AI-1 since they are strappable (i.e. the negative terminals of each monoblock can be wired together to make a common ground connection). My motivation for building an AI-1 was that one would be required if I decided to move up significantly in amplifier power. The JC 1's are very conservatively rated at 400 watts per channel into 8 ohms [footnote 1]. As I went up in amplifier quality and power from 200 to 300 to 400 wpc, I realized a larger, heavier, and more detailed sound stage. All of the over 400 wpc amplifiers I am interested in are bridged designs where the positive and negative outputs are "live". Therefore, they are not common ground and the negative output terminals cannot be tied together.
Figure 1. The AI-1 Dreadnought brought unexpected improvements to an already
expansive SDA sound stage.:D
Figure 2. Cardas CCGR binding posts...to match those on my 1.2TL's.
Figure 3. AI-1 Dreadnought transformer enclosure and interconnect cables.
Figure 4. Dreadnought rear panel.
Figure 5. Not much to see inside...just a big 800 VA isolation transformer.
The Dreadnought comprises a large 800 VA toroidal transformer with four ten foot 9 AWG OFC copper cables and a custom black anodized aluminium (ah-LOO-min-ee-um) enclosure.
Figure 6. The Dreadnought's interconnect cables are constructed from Monster
Cable Z2 Reference speaker cable.
Four 10 foot lengths of Monster Cable Z2 Reference cable were used to make the positive and negative conductors for each side. The cables were terminated by Monster Lock pins, which allow the interchanging of ends (bananas and different size spades). Monster Lock Medium Angled Spades are used at both ends of each cable. They are a perfect fit for the Cardas CCGR binding posts used on the 1.2TL's and on the Dreadnought. The wire gauge of each conductor in the Z2 cables is 12. Soldering those conductors together at both ends made a cable with an effective wire gauge of 9, which is the same as the effective wire gauge of my speaker cables. This also effectively cut the resistance nearly in half since 12 gauge cable has a resistance of 1.588 ohms per 1000 feet and 9 gauge cable has a resistance of 0.7921 ohms per 1000 feet.
Figure 7. Finished AI-1 Dreadnought cables.
Figure 8. AI-1 Dreadnought cable terminations.
A couple of years ago, Matthew Polk provided the following specifications for a toroidal transformer based AI-1:
"1. DC Resistance of primary and secondary should be less than 0.8 ohms and preferably less than 0.5 ohms. The lower the better.
2. Inductance of both primary and secondary should be at least 10mH. Ideal is around 14mH.
3. Current capacity to saturation should be 5 Amps minimum.
4. Of course the turns ratio is 1:1."
I sent those specs to transformer manufacturer Avel-Lindberg and they recommended the following two toroidal transformers:
Avel Part No: Y236856, 5.4" diameter, 2.6" high, 10.8 pounds weight.
Avel Part No. Y236906, 6.3" diameter, 2.2" high, 13.2 pounds weight.
I chose the Y236906 since it was the more robust of the two and its specifications exceeded the requirements. Listening evaluations confirmed that The Y236906 works exceptionally well.
DC Resistance: 0.275 ohms primary winding, 0.360 ohms secondary winding. The dual primary and secondary windings were wired in parallel as follows:
Right Pin = Red/Yellow
Right Blade (Negative Speaker Terminal) = Black/Orange
Left Pin = Grey/Brown
Left Blade (Negative Speaker Terminal) = Blue/Violet
Inductance : 153.3 mH primary, 168.8 mH secondary.
Output Current Capacity: 6.95 Amps
Output Current Capacity To Saturation: 6.32 Amps.
Figure 9. Dreadnought prototype trials. Everything turned out much, much better
Figure 10. After listening trials were completed, the prototype was bolted to a smaller board and tucked behind one of the 1.2TL's until the aluminium
The cost of the Avel-Lindberg Y236906 isolation transformer, inclusive of shipping, was $114.92. Of course, other high quality isolation transformers are available from other vendors such as Plitron and Antek. Avel-Lindberg was more forthcoming with technical information about their transformers. There are many options for enclosures, binding posts, cable, etc. This project can be completed for under $200...unless you decide to go upscale for associated parts.;)
Results with the original AI-1 interface cable have been mixed. Some found it made no audible difference while others said it diminished detail a small amount. There were thrills and disappointments immediately after the
Dreadnought was installed. Thrills due to a small increase in detail and to my sound stage growing a foot on each side, coming forward a foot, and extending a foot further back. Disappointments due to some upper midrange graininess and some slight "softness" and slowness in the bass. The grain was similar to what was heard when I first installed the RD0198 silk dome tweeters. Fortunately, as with the RD0198's, the graininess went away after 10 hours. The bass aberrations had also melted away after 10 hours. I assumed the transformer coils and cables needed a little burn-in time. From the 11th hour onward, the sound with the Dreadnought was better than with the strapped configuration. I was pleasantly surprised to find that the bass definition, sound stage holography, and image weight continued to improve over the next few days.
A couple examples of increased detail:
At 1:50 on the "Happy Home" track of Boney James "Backbone" CD, I heard some intermittent subtle "popping" percussion sounds in the background just to the left of the right speaker that I had never noticed before. The pops were surrounded by a reverberant envelope. I reinstalled the strapped configuration to see if I could hear the pops. They were there, but they were almost buried by the sound of other percussion sounds around them. The reverberant envelope around the pops was absent. With the Dreadnought, the popping sounds were a foot further back and slightly above the other percussion sounds.
With Russ Freeman's electric guitar solo beginning at 1:57 on the "Girl with the Indigo Eyes" track of The Rippingtons "Sahara" CD, I heard some twists, turns, and grace notes that I had not noticed before. There were also more distinct reverberant envelopes surrounding each note. Listening to the same passage with the strapped configuration, those details were there, but not as finely rendered. The reverberant envelopes around the individual notes were also diminished.
I wondered if sound quality would deteriorate if I cranked the volume. I went up to 110 dB-C without even a hint of stress, stain, grit or grain from the speakers. Of course, at that level, I was listening from the next room. I stayed in the room up to 105 dB-C, then I had to go. As the volume went up from a comfortable listening level of 85 dB-C, the image weight and tactile sensations also went up, which was pleasing on one hand because of the increased realism and displeasing on the other hand due to my eardrums not enjoying the sensation of being beaten with hammers.
Figure 11. Enjoy your retirement. Former SDA SRS 1.2TL custom 9 AWG SDA
interconnect and 9 AWG amplifier strap cable.
My original intention was to construct the Dreadnought so that it would be on hand for high power amplifier evaluations. I did not intend for it to replace the JC 1's ground straps. However, I had no choice because the Dreadnought brought Such Good Sound!
[Footnote 1] From Stereophile June 2003: "With continuous drive, the Parasound clipped at 545W into 8 ohms—way above the specified 400W. With a low duty cycle 1kHz tone burst more representative of music, the
Halo was a powerhouse. Its clipping power increased by 0.3dB into 8 ohms, reaching 586.5W at 1% THD, with 1154W available into 4 ohms, 2255W into 2 ohms, and no less than 4.2kW into 1 ohm."
The AI-1 Dreadnought Project Pt.2
After 100 hours of operation and prior to the aluminium case arriving, I used a Tektronix TDS 2012 oscilloscope to take waveform and noise spectrum measurements at 60, 61 and 1000 Hz at the left speaker's SDA cable inputs. All measurements were taken with the preamp's volume setting at 24. Noise measurements with the strapped configuration and the Dreadnought were compared first, then the Dreadnought's Z2 cables were replaced with Audioquest Volcano and Monster XP. The red noise at the bottom of the spectral plots looks like a raging fire, but the average RMS magnitude of the noise is actually in tenths of a millivolt whereas the magnitude of the signal is in hundreds of millivolts. However, just a little bit of noise can go a long way toward obscuring musical detail and spatial properties. The Fast Fourier Transform (FFT) plots of the signal and noise spectrum are given in figures 12-15, 17 and 18.
Figure 12. 61 Hz tone measured at speaker's SDA inputs with the strap.
Figure 13. 61 Hz tone measured at speaker's SDA inputs with the
In figure 12 the RMS magnitude of the SDA signal using the strap is 382 mV and the average RMS magnitude of the noise is 0.72 mv. In figure 13 the RMS magnitude of the SDA signal using the Dreadnought is 360 mV and the average RMS magnitude of the noise is 0.36 mv. At 61 Hz, the noise content of the SDA signal was significantly lower in both amplitude and density with the Dreadnought.
Figure 14. 1000 Hz tone measured at speaker's SDA inputs with the
Figure 15. 1000 Hz tone measured at speaker's SDA inputs with the
Dreadnought using Z2 cables.
In figure 14 the RMS magnitude of the SDA signal using the strap is 445 mV and the average RMS magnitude of the noise is 0.57 mv. In figure 15 the RMS magnitude of the SDA signal using the Dreadnought is 417 mV and the average RMS magnitude of the noise is 0.4 mv. At 1000 Hz, the noise magnitude in the SDA signal was much lower with the Dreadnought. The noise density was about the same as with the strap.
As expected, the signal to noise ratio was significantly better when the monoblock amps were isolated from each other.
Figure 16. Just for grins, the Dreadnought's Monster Z2 Reference cables were replaced with Audioquest Volcano and then Monster XP.
Figure 17. 1000 Hz tone measured at speaker's SDA inputs with
Audioquest Volcano speaker cables on the Dreadnought.
Figure 18. 1000 Hz tone measured at speaker's SDA inputs with
Monster XP speaker cables on the Dreadnought.
It is easier to see the differences in the 1000 Hz FFT plots (figs. 14, 15, 17, 18) if you save them and then click through them. While it is easy to see the decrease in noise as you go from the strap to the Dreadnought/Z2 to the Dreadnought/Volcanoes, the Dreadnought/Z2 and Dreadnought/XP plots (figures 15 and 18) appear very similar at first glance. However, the signal's DC component was 6 dB higher with the XP cables. The XP cables also had significantly more noise in the 2 kHz - 6 kHz region.
Figure 19. 61 Hz tone measured at speaker's SDA inputs with Monster
Z2 Reference speaker cables on the Dreadnought.
Figure 20. 61 Hz tone measured at speaker's SDA inputs with Monster
Z2 Reference speaker cables in the strapped configuration.
Figure 21. 60 Hz tone measured at speaker's SDA inputs with Audioquest
Volcano speaker cables on the Dreadnought.
Figure 22. 60 Hz tone measured at speaker's SDA inputs with Monster
XP speaker cables on the Dreadnought. Nasty. Sounded nasty too.
I learned a while ago that the quality and gauge of the wire carrying the SDA signal makes a big difference in the overall sound quality of the speaker and not just in the imaging and sound staging. I replaced the Dreadnought's Monster Z2 Reference cables ($3.78/foot) with Audioquest Volcano speaker cables ($206.25/foot) and then with Monster XP speaker cables ($1.00/foot). The XP was unpleasant to listen to. It was as if a bed sheet had been draped over the speakers. The bass was muddied a bit and the sound stage shrank by about three feet all around. The sound volume was also apparently lower. In addition to being a lower grade of copper, the 16 gauge conductors in the XP had 5.3X the resistance of the effective 9 gauge conductors of the Volcano and Z2 Reference cables.
I did not hear a difference between the Z2 Reference and Volcano cables, even though the Volcanoes had a lower average noise spectrum density and magnitude. This seems to indicate that either there is an upper bound on improvement in sound quality with improvement in SDA cable quality or my equipment could not convey the difference or that my ears could not discern the difference.
While I did not hear a difference or improvement with the use of a pair of $3300 cables on the Dreadnought, I did hear a moderate improvement when the Dreadnought was placed in its $200 aluminium enclosure. This was a pleasant surprise.
The Aluminium Clad Dreadnought
Figure 23. The Dreadnought sits vertically behind one of the 1.2TL's.
There was no space for the Dreadnought in my equipment cabinet and I didn't want to sit it on top. The only other option was to orient it vertically behind one of the 1.2TL's. I was advised (by Avel-Lindberg) not to over tighten the transformer's retaining nut because the primary and secondary coil insulation could be pierced and cause a short. Hand tightening only was advised. To brace against the 13 pound transformer creeping downward over time, a small wood block was placed between it and the enclosure side panel below it.
I had some concerns about electromagnetic interference since the Dreadnought would be sitting right next to a big fat (but well shielded) power cable and a big fat power amp with its accompanying big fat electromagnetic field. The Dreadnought, while still attached to the wood platform, was measured and listened to while it was placed behind the speaker (right next to a power cord), again when placed next to one of the JC 1 amps, and again when placed on top of the audio cabinet. I neither heard nor measured a difference at any location.
Immediately after the Dreadnought was placed in its aluminium enclosure, I was hearing more. More detail. More bass. More liquidity. More velvety smooth inky black background behind the music. The dead (quiet) spaces between songs on a CD began to call attention to themselves like a pregnant pause.
I dug out the oscilloscope once again to have a look at what I was hearing. I already knew that aluminium, and other non-ferrous metal enclosures, provide a measure of shielding against electromagnetic interference. This was the first time I had heard and measured the effects of such shielding.
Figure 24. FFT plot of 61 Hz tone measured at speaker's SDA inputs
with the aluminium clad Dreadnought/Z2 cable.
Figure 25. Sine wave plot of 61 Hz tone measured at speaker's SDA
inputs with aluminium clad Dreadnought/Z2 cable.
Figure 26. 1000 Hz tone measured at speaker's SDA inputs with
aluminium clad Dreadnought/Z2 cable.
Comparing figure 24 to figure 13 and figure 26 to figure 15, a much lower and overall flatter average noise characteristic is seen with the aluminium clad Dreadnought. Comparing figure 25 to figure 19, the 61 Hz sine wave output of the aluminium clad Dreadnought is much smoother in the peaks. [Comparison of the figures is much easier if you save them to your hard drive.]
The AI-1 Dreadnought Project Pt.3
Quantitative Results - Continued
The Monster XP and Audioquest Volcano cables were measured with the aluminium clad Dreadnought. Sine wave and FFT plots for the XP cable are given in figures 27 - 29. FFT plots for the Volcano cable are given in figures 30 and 31.
Figure 27. Sine wave plot of 61 Hz tone measured at speaker's SDA
inputs with aluminium clad Dreadnought/XP cable.
Figure 28. FFT plot of 61 Hz tone measured at speaker's SDA inputs
with aluminium clad Dreadnought/XP cable.
Figure 29. FFT plot of 1000 Hz tone measured at speaker's SDA inputs
with aluminium clad Dreadnought/XP cable.
The XP cable had a much better output signal when connected to the aluminium clad Dreadnought. Compare the sine wave plots in figures 22 and 27 and the 1000 Hz FFT plots in figures 18 and 29. The aluminium clad Dreadnought/XP measured better than the wood bound version, but sounded only a little better. Images didn't clump in the areas between and just in front of the speakers this time, but the sound stage dimensions were still greatly diminished. There was only a small difference between the sound stage dimensions with no SDA interconnect and the sound stage dimensions with the Dreadnought/XP. Clarity-wise, the sound was much better without the Dreadnought/XP.
Figure 30. FFT plot of 61 Hz tone measured at speaker's SDA inputs
with aluminium clad Dreadnought/Volcano cable.
Comparing the 61 Hz FFT plots of the aluminium clad Dreadnought/Z2 (figure 24) and Dreadnought/Volcano(figure 30) shows the Volcano's moderate reduction in both noise magnitude and density.
Figure 31. FFT plot of 1000 Hz tone measured at speaker's SDA inputs
with aluminium clad Dreadnought/Volcano cable.
Comparing figures 17 and 31, the Volcano's FFT plots at 1000 Hz show a drastic reduction in noise between the wood bound prototype and the aluminium clad Dreadnought. Comparing the 1000 Hz FFT plots of the aluminium clad Dreadnought with Z2 cables (figure 26) and Volcano cables (figure 31) shows virtually no difference between the noise magnitude. The noise density of the Volcano cable is slightly lower. The noise density of the XP at 1000 Hz is similar to that of the Volcano, although a hint of the "bump" in the 2 kHz to 6 kHz region is still there (figure 29).
There was also no discernible sonic difference between the aluminium clad Dreadnought with either Z2 or Volcano cables.
AI-1 Dreadnought vs. Original Polk AI-1
I was able to borrow an original AI-1 from a fellow SDA cultist [footnote 2]. I expected it's sonic performance with non-common ground amps to be comparable to the performance of the stock pin/blade interconnect with common ground amps. Since a heavier gauge, better shielded interconnect cable far surpassed the sonic performance of the stock interconnect, I expected the Avel-Lindberg Y236906 transformer with heavy gauge interconnects to surpass the performance of the original AI-1. It didn't disappoint.
Figure 32. Original Polk Audio AI-1 Non-Common Gound Interface cable.
The AI-1 comprises a small isolation transformer with a 3 foot cable attached to one side of the plastic enclosure and a 17 foot cable attached to the other. The cable is the same 16 gauge two conductor SJT M/E type cable that the standard blade/blade SDA interconnects are made of. The transformer enclosure measures 3-5/16" long by 2-1/2" wide by 2-1/8" high. The entire unit weighs 2.4 pounds, of which 1.1 pounds is the cable (a standard 20 foot blade/blade cable weighs 1.1 pounds).
I began my listening evaluation with the following music selections:
Track 10, "You Don't Have To Go Home", Boney James, "Pure" CD.
Track 4, "B'wana He No Home", Michael Franks, "Sleeping Gypsy", CD.
Track 1, "The Picnic", Dan Siegel, "Another Time, Another Place", LP.
Listening levels ranged from 85 dB-C to 105 dB-C.
The first thing I noticed was a shortening of depth. Images previously projected far behind and far in front of the speaker plane were pushed back closer to the plane. For example, Boney James' saxophone was pushed back to the speaker plane rather than being projected three feet in front of it as usual. At the beginning of the track there are background sounds of people conversing and cups, glasses, and dishes tinkling...as if the song were recorded in a small club. The crowd sounds were not as clear with the AI-1.
The AI-1 provided a three dimensional sound stage, just not as big and defined as either the Dreadnought or the strap. Soundstage width was unchanged, but images at the far sides of the stage were not as defined. Sound stage height was a little lower.
As I went up in volume, the bass became heavier but less defined. Horns took on a rough edge (glare). At 105 dB-C, the overall sound was veiled. Unplugging the AI-1 relieved the congestion. I assume the loss of detail was caused by mainly by the transformer saturating at higher volume levels and to a lesser extent by a interconnect cable noise and environmental noise.
I noted that the sound of the Dreadnought improved when its transformer was placed in an aluminium enclosure. I wrapped the AI-1 in 5 6" x 9-1/2" sheets of aluminium foil and heard an improvement in detail and depth. I added an additional five sheets of foil and heard additional improvements in detail and depth. Readers using AI-1's may realize some benefits by wrapping their transformer enclosures in aluminium foil or by placing it in an unvented aluminium enclosure. I did not measure or hear further improvements beyond adding 10 sheets of foil. The amount of foil required to reduce noise to an acceptable level will probably vary depending on your environment and associated equipment.
Figure 33. Dont laugh. Wrapping the AI-1 in sheets of aluminium foil resulted in audible and measurable improvements.
Figure 34. FFT plot of 1000 Hz test tone taken at left speaker SDA terminals.
Figure 35. FFT plot of 1000 Hz test tone taken at left speaker SDA terminals with AI-1 transformer enclosure wrapped in 5 sheets of aluminium foil.
Figure 36. FFT plot of 1000 Hz test tone taken at left speaker SDA terminals with AI-1 transformer enclosure wrapped in 10 sheets of aluminium foil.
Figures 34-36 show significant reductions in noise magnitude and density as foil insulation is added. Measurements were taken from the binding posts that the Dreadnought is normally attached to. It is easier to see the differences in the figures if you save and print them or save and step through them with an imaging program.
Figure 37. FFT plot of 60 Hz test tone taken at left speaker SDA terminals.
Figure 38. FFT plot of 60 Hz test tone taken at left speaker SDA terminals with AI-1 transformer enclosure wrapped in 5 sheets of aluminium foil.
Figure 39. FFT plot of 60 Hz test tone taken at left speaker SDA terminals with AI-1 transformer enclosure wrapped in 10 sheets of aluminium foil.
Figures 37-39 show measurements taken at 60 Hz. At this frequency, adding 5 sheets of foil significantly reduces noise magnitude and frequency. Figure 39 shows that adding twice as much foil does not produce much further noise reduction at this frequency.
The 1.2TL's original pin/blade cable was reinstalled along with a 10 gauge strap cable with PVC cladding. The sound with the pin/blade and cable/strap combination was more detailed and holographic than with the non-shielded AI-1. I could actually tolerate the pin/blade strap combination. I did not like the sound of the unshielded AI-1 at all [footnote 3].
As I listened to more discs, the pin/blade-strap combination was slightly better than the aluminium foil shielded AI-1 on some music. On other music, I could not hear a difference between the pin/blade-strap and shielded AI-1 .
[Footnote 2] Thanks to Jesse (F1nut) for the loan of an original Polk AI-1.:)
[Footnote 3] But that didn't stop me from buying one. Thanks to Sal (Lasareath) for his assistance in locating an original AI-1 for purchase.:)
The AI-1 Dreadnought Project Pt.4
This exercise ends on some ironic notes. I started all this because I wanted a heavy duty AI-1 interface to use when evaluating more powerful non-common ground amplifiers. After living with the Dreadnought for over a month, I find my desire for an amp upgrade diminishing. It already sounds like I have new, more powerful amps.
I had avoided auditioning an AI-1 in my system for some time due to concerns about diminished sound quality. While those concerns were shown to be valid for the original AI-1 in my system and for the kind of music I listen to, an enhanced version of the AI-1 brought higher levels of three-dimensionality, clarity, detail, and Good Sound.
The biggest irony of all is the requirement for ultra high powered amplifiers, most of which are non-common ground designs, to "work" the 1.2TL's. The SDA SRS 1.2TL's are rated with a power handling capability of 1000 watts. I wonder if an amplifier with that level of power output is required to take them to the highest level, or if a significant part of that 1000 watt rating is just for "headroom". When I say highest level, I don't mean sheer volume (loudness). I mean the best possible detail retrieval, sound staging and imaging.
This quote from a SDA SRS review in the November 1985 issue of Stereo Review provides some insight:
"Not surprisingly in view of its large driver complement and rated power-handling of 1,000 watts or more, the Polk SDA-SRS thrives on power levels that would vaporize some lesser designs. We ran out of power from our amplifier (one of the most powerful available for home use) when we attempted to reach the speaker's limits with 1,000 cycle tone bursts. At 100 Hz the amplifier clipped at 1,650 watts into 6 ohms, and at 1,000 Hz it clipped at 765 watts into 20 ohms. Only at 10,000 Hz, where presumably only one driver was in use, did we reach the limits of the speaker and the amplifier at about the same time-1,265 watts into 5 ohms."
The monoliths are nervous about the prospect of
no new amps for Christmas.
The JC 1's are thrilled to no longer be conjoined twins.
Pass Laboratories X0.2 Preamplifier
Pass Laboratories Xono Phono Preamplifier
Parasound Halo JC 1 Monoblock Amplifiers
Teres Audio Model 255 Turntable
Ortofon MC Windfeld Cartridge
Graham Phantom B-44 Tonearm
Cary Audio CD 306 SACD Player
Polk Audio SDA SRS 1.2TL Speakers
Audioquest LeoPard Tonearm Cable
Audioquest Sky XLR Interconnects
Audioquest Everest Speaker Cables
PS Audio Power Plant Premier Power Regenerator For Source Components
AI-1 Built. I have an issue.
Improved SDA Interconnect Cables
AI-1 Dreadnought - Preliminary Results