ack's system - end of round 1

[BMCG]

ack - you already acknowledged in post 234.

"Thank you Al for taking the time to write all this up so well, and I am glad you enjoyed it! Six hours went by really quick. And thanks BMCG! "

Still your courtesy is remarkable in this age...thank you!

But am curious as to when your seduction to the M3 commences?

You've definitely spurred my thoughts on the newer Spectral Amps...they are just non trivial to acquire in UK.

 

[ack]

ack - you already acknowledged in post 234.

"Thank you Al for taking the time to write all this up so well, and I am glad you enjoyed it! Six hours went by really quick. And thanks BMCG! "

Still your courtesy is remarkable in this age...thank you!

But am curious as to when your seduction to the M3 commences?

You've definitely spurred my thoughts on the newer Spectral Amps...they are just non trivial to acquire in UK.

LOL, I am not aging well It's definitely not good that Spectral pulled out of the UK market years ago; I believe there were posts here that More Music in NL is supporting the UK market??? And maybe selling into?? Not trivial for you folks, I understand.

If you've followed solid state as some folks like myself have over the last 30 years, you'd probably agree that Spectral are the most consistently linear and un-electronic solid state components throughout that period, and thankfully others are catching up. It's no coincidence that a lot of folks still use older designs like the DMA-200 from the 80's, ditto for the DMA-50/80/90, the newer DMA-180, the 360s, the 260, etc, etc. or continue to use the DMC-20 preamp as a phono stage (out of the tape buffer). They definitely have had their own sound over the years, with the newer designs sounding so incredibly real and transparent in so many aspects.

Last night I was at a live unamplified jazz gathering in a small venue, where a friend of mine plays the drums - she has a set at home as well, and I've played with it a number of times over the years. Last night, I had the pleasure to "test" again drums and alto sax. The dynamics and treble extension of cymbals from up close - as the microphone captures it - is just staggering. But instead of marveling at how good live is, I was marveling at how well Spectral does all this at home, and I don't have super-speakers. More than that, how good the typical tube recording electronics must be, the behavior of which [the device, not necessarily the circuits], as Al said before, Spectral is trying to mimic in the solid state domain - namely, extremely wide bandwidth, linearity, phase correctness, extension, incredible speed and settling times, un-electronic - but with a lot more power, current and control.

Regarding the M3 - I must admit I am still interested, but can't pull the trigger yet. For one thing, I am pretty sure the one time I heard them, cymbals sounded a little more timbrally correct than my system. I am actually listening to sax and cymbals right now as I write this, and I don't believe the sound is as accurate as from the M3's - but at the same time, I have to question the recording as well, as it wasn't the same as when I heard the M3s. Nonetheless, Magicos sound so incredibly close to electrostatics that they are just about the only option for me and my preferences, and though the older designs were not as good as current, it's the likeness to and path of progression towards electrostatics that has aroused my interest. I haven't said much about the M3s, because the one time I heard them driven by my electronics, the bass just wasn't there, and there was midrange harshness - not broken in, or set up correctly. So I need to go back in and listen.

 

[ack]

Going back to the original Alpha DAC modification with the 10nF Vishay 1837 black capacitor in parallel with the original 1nF in the output section, lowering the corner frequency to about 193kHz, I had mentioned a while ago that I may experiment with an even lower frequency, by adding another 10nF for roughly a 93kHz cutoff. Well, I had to try it as shown below, by adding the white cap:





Results? What a total disaster. I COULDN'T WAIT TO TAKE IT OUT FAST ENOUGH! The mid bass became so thin, the sound so dull, so undynamic and uninvolving. Unbelievable how easy it is to mess up the sound, just like how unbelievably successful the original mod was.

So I had to mention this, FWIW; any mod has to be carefully thought out and planned. All in all, the Alpha DAC was factory voiced with the Spectral 360 amps, and this one is now re-voiced with the incredible 400s. This adventure proves to me yet again how delicate voicing any component is, and why a component working well in one system may not work as well in another.

 

[opus112]

A thought has occurred to me about this output filter - how about the possibility of changing it from a first order type (series R, shunt C) to a second order (series L, shunt C)?

The drawback of a simple first order filter is that as you lower the cut-off frequency, the filter itself gets harder and harder to drive at higher frequencies because you're decreasing the shunt impedance (impedance to gnd). The 2nd order filter gets over this problem because the series arm turns into an inductor which has rising impedance at higher frequencies. Meaning the output buffer doesn't have to work so hard, so it generates less power supply noise.

If you're interested to try I could look for suitable inductors to give you a much lower cutoff frequency (say around 30 to 40kHz) which won't be likely to have adverse side-effects. Another advantage the LC filter has over the standard RC filter is that at lower frequencies (audio band) the output impedance is lower meaning better noise rejection. It does need a little bit of design work to ensure no ringing occurs though, but that's not hard.

 

[ack]

Thanks, sure, won't hurt to see what you can come up with. BTW, I forgot to mention that I revisited the heat issue we were discussing last... the regulator close to the ex C7 was just warm, with the transformer cores yet warmer than that. Contrast that with the hot-to-touch C6 and C7 capacitors, as I mentioned a while ago.

 

[opus112]

I ran a few sims and learned something - my previous understanding of a second-order filter was rather too simplistic. Seems the loading is lighter but only above a certain frequency, below that the loading's heavier than for a typical first-order filter due to the need for a damping network. So back to the drawing board to explore if there are any alternative ways to provide damping.

 

[Folsom]

I ran a few sims and learned something - my previous understanding of a second-order filter was rather too simplistic. Seems the loading is lighter but only above a certain frequency, below that the loading's heavier than for a typical first-order filter due to the need for a damping network. So back to the drawing board to explore if there are any alternative ways to provide damping.

Are you modelling with speakers or just output to another appliance? Tuning it with an RC that bypasses the subsequent inductance has benefits but you have to adjust by ear... the math doesn't always give the sound (good starting point).

 

[opus112]

Modelling with the details ack has given so far in his posts. Its a line-level circuit so nothing much to do with speakers.

Incidentally here are the three frequency responses - original (1nF) in green, 10nF (red) and 20nF (blue).



I rather suspect given the dreadfulness of the sound with 20nF that this amount of capacitance is inducing instability in the driving opamp. The give away for me is 'thin bass' - seems always to happen when there's some oscillation.

 

[Folsom]

Well the principles aren't different for using an RC without a speaker, they're just dynamic depending on the next device in the line (and arguable for importance). I guess you can anticipate the essential inductance of an average sized RCA cable, so it might be tunable within a light range of predictable. I wouldn't put in a 2nd order without trying this, but I can't tell you I've done it in this situation.

If the 20nf is giving you low bass then you probably do have a resonation, not necessarily oscillation. Restrictive current from the PSU will sound like the highs are low, not the bass, and resonaces will cause decreased bass impact. This might be fixable with a couple of resistors. However it could be challenging to get a couple inline with the output, and lower valued ones for the filter cap. The inductance of the line out might have been enough to prevent harmonics generated (that can resonate) from slew rate limits compared to the capacitors corner frequency, previously, but gets bypassed by the 20nf so they're being made now.

 

[opus112]

Here's what's achievable with a highly optimized LC filter - the passband is -3dB @40kHz and the roll-off is steeper than 2nd order due to the self-resonance of the inductors. Inductors' self-resonance is unavoidable so I've put it to good use here in placing it just where the DAC puts the first group of image frequencies (8X OS).

For reference, the blue plot is what ack has now - 10nF and 75ohms RC.

 

[ack]

Interesting discussion. Resonance is what came to mind as well, not oscillation, when I heard the result; but cannot be sure. I don't like steep filtering in general due to phase issues, so what it's place right now appears to be adequate. If nothing else, I would be more inclined to replace the caps with a variable one to experiment by ear, but if it ain't broken, then don't fix it. What might be worth doing is actually removing the original 1nF ceramic, because they don't necessarily sound as good, or replace with a 1nF film... Need to think about this...

 

[opus112]

There are different kinds of ceramics, most likely your 1nF is the benign kind (NP0 or C0G) as its a relatively low value. Larger values (nowadays into the hundreds of uF) might be based on ceramics which have piezo-electric properties (X7R for example) so they should be used with care due to their inherent microphony.

 

[ack]

I think I see markings "BC" on them, followed by something that looks like "101" - I had to use a loupe because they are so small. Do these mean anything to you?

 

[ack]

Here's another similar ceramic elsewhere in the analog section, with its marking clearly visible - BC 221

 

[ack]

So I found many capacitor decoders like this one http://grathio.com/assets/capacitor_tags.pdf, and "221" is 220pF, "101" would be 0.1nF (100pF), and "102" 1nF - so I think the marking on the one in question must be 102, not 101, but it's so hard to see in there. Let's see if we can figure out what "BC" stands for. For them to be used all over the place in the analog section, and it sounding so good, they probably are indeed high quality??!?!?

 

[Folsom]

If I were you I'd order a few resistors. Try .1ohm, .35ohm, .5ohm and 1ohm. Put one in series with the 20nf. The way I'd do it is start with .5ohm and listen, if it's working try .35ohm and if that's better try .1ohm. If .35ohm is worse than .5 ohm then try 1ohm. Essentially you'll be adding in a similar amount of resistance as the reactance near 50-200khz. BUT as frequency climbs so will the inductance inversely become greater than the resistance. The attenuation won't be as strong as an open 20nf, but it may not cause harmonics to be generated in the opamp - if that's the problem, either way it's likely to prevent the issue you have. If it doesn't work maybe the slew rate is much higher, in which case I'd forget about trying to increase the capacitance and only look toward using a 2nd order which is uh, well, we've been over that.

 

[ack]

Thanks. Meantime, I am thinking "BC" is probably BC Components, now a part of Vishay.

 

[microstrip]

So I found many capacitor decoders like this one http://grathio.com/assets/capacitor_tags.pdf, and "221" is 220pF, "101" would be 0.1nF (100pF), and "102" 1nF - so I think the marking on the one in question must be 102, not 101, but it's so hard to see in there. Let's see if we can figure out what "BC" stands for. For them to be used all over the place in the analog section, and it sounding so good, they probably are indeed high quality??!?!?

Some types of ceramic capacitors are the best type to absorb RF noise in spite of the low price - in many designs they are used in parallel with other capacitors with decoupling functions of supply.

 

[ack]

The effect of mass loading

The below is an additional simple mass-loading modification of the panel pillars - I'd been using 2 horizontal granite bars for years, bridging other vertical granite plates attached to the pillars; and now there are 6 horizontal bars, for another ~35 pounds of mass. I knew there ought to be an improvement, but I didn't know what to expect. Well, I now hear violins go higher, and 10-16kHz test tones are more easily discernible... interesting, those were slightly attenuated before, probably absorbed by micro-vibrations. But the more surprising and interesting sonic benefit came in the bass, with slightly more fuller and rounder character.

I guess this proves yet again that cabinet rigidity is a good thing; I honestly don't understand why people prefer lively cabinets. Anyway, nothing ground-breaking here, except that when I played the first CD - the Mahler 6 down below - the deep bass drums plus the hammer strikes were just jaw-dropping, with more tautness, power and jump factor - and thanks to whomever originally posted a while ago about this incredible Keith Johnson recording of the DSO!

 

[Al M.]

The below is an additional simple mass-loading modification of the panel pillars - I'd been using 2 horizontal granite bars for years, bridging other vertical granite plates attached to the pillars; and now there are 6 horizontal bars, for another ~35 pounds of mass. I knew there ought to be an improvement, but I didn't know what to expect. Well, I now hear violins go higher, and 10-16kHz test tones are more easily discernible... interesting, those were slightly attenuated before, probably absorbed by micro-vibrations. But the more surprising and interesting sonic benefit came in the bass, with slightly more fuller and rounder character.

Nice to hear you achieved yet another improvement.

I guess this proves yet again that cabinet rigidity is a good thing; I honestly don't understand why people prefer lively cabinets.

Yes, cabinet rigidity was one of the main criteria when it came to choosing my current speakers, after having heard what rigidity achieves with respect to timbral accuracy on Madfloyd's Magico M Project speakers.