Yamaha NS-1000M A Fresh Start

[Qwin]

A Fresh Start.

I just can't get away from the fact that I like so much about the NS-1000m.
The re-cap I completed, using more neutral capacitors helped the overall presentation, but I decided to make another attempt at sorting out my minor dislikes.

Over the years I have found more 2-way systems that I feel happy with than 3-way. I often find a resonant distortion exists around the crossover point on a 3-way, at the hand over between woofer/midrange. This can add a nasty edge to male vocals and fingers on strings etc.
While taking some measurements of capacitors, to satisfy my curiosity regarding Equivalent Series Resistance (ESR) values, a possible cause of this distortion came to mind.

The following ESR values were measured using Dayton Audios Test System (DATS).



I took these ESR measurements from capacitors of 30uF. This is in the 20 to 30uF bracket, not uncommon for the series cap in a low order filter, used for the midrange High Pass on a 3-Way. This is usually the biggest value on the crossover that is in the signal path.
The first thing I noticed, was contrary to what I have been reading on the forums, there is not a massive difference between the capacitor types. The Alcap Lytic, Mundorf Low Loss (LL) Lytic and Ansar Supersound PP caps all have remarkably similar ESR curves with only small differences. The second and most noticeable thing, was that the ESR values rise substantially below 500Hz. This is the area where the woofer/midrange crossover area lies on a 3-Way speaker, a 2-Way would cross at say 2kHz where the ESR levels are much lower. It might be a coincidence, but I wondered if this could in some way be linked to the resonance that hampers my listening experience on the NS-1000m and previously on the Celestion 66 Monitors.

I decided to take some measurements of lower value caps.



The 4uF Alcap Lytic and Sonicap PP capacitor, also show similar ESR curves, with the Lytic having only slightly higher values, 21.3ohm against 19.7ohm at 2kHz for instance. It would seem though, that the smaller the cap value, the higher its ESR. I measured a 0.2uF and it had a massive ESR, over 1kohm at 200Hz and going off the scale at lower frequencies.
To try and rule any of this out, in terms of having a detrimental effect on the sound, I decided to run an experiment, using a passive crossover that doesn't use a cap, running into a large ESR value, in the midrange signal path. I would create a line level active filter operating at 500Hz for the woofer/midrange handover and use the stock passive elements for just the mid/tweeter handover at 6kHz, where ESR values are much lower.





I had a spare pair of 3-Way active crossover boards from KMTech, I modified them to work as a 2-Way, crossing at 500Hz. Basic polyester caps, but good enough for a quick look-see. I borrowed the discrete op amps from another board.



This diagram shows how it would be set up as a system.



I cobbled together a set up, as per the diagram, by disconnecting a leg on the unused components in the stock passive crossover and using the modified active filter, just to see if there was any millage in the idea.

It works, and very well, with a much smoother hand over at woofer/midrange, without any noticeable resonance. My reasoning may have nothing to do with it, who knows, but the positive results from this little experiment, can't be denied. I bet your thinking, why not go fully active? The synergy between mid/tweeter in passive form is extremely good and it is not likely to be bettered by going active. Basically, if it ain't broke, don't fix it! The obvious reason though, is that this set up uses only four channels of amplification, instead of the six needed to go fully active. This arrangement has the signature sound from the midrange/tweeter that the passive filter delivers, with the benefits of a cleaner hand over at woofer/midrange and a separate Amp to drive the woofers, which has its own advantages.

While I had this lash up hooked up, I decided to put another issue to bed, digital or analogue active?

The Analogue filter above uses a LR 24dB/Octave slope at 500Hz. I set up a similar filter on my Hypex DLCP (DSP) and did a shoot out between them and the stock passive. My conclusions? The passive has the ability to get to the heart of the music and to express the finer detail in a natural way. Its down side is the edgy resonance around 400Hz on voices like Gregory Porters. The Semi-Digital Active set up is by far the smoothest, with very clean sounding crossover points, but somehow it doesn't involve me like the passive and feels less interesting to listen to. The Semi-Analogue Active is almost as clean as the Digital, but it grabs my attention and draws me in like the passive.

So Analogue Semi-Active it will be, this arrangement also works extremely well with my Turntable, where the digital filters do not.

Next, which Woofer, the Yamaha or the Skanspeak.

I set the two woofers up, one on top of the other and just swapped leads at the amp between the two, the passive x-over for the Mid/Tweeter and the analogue active for the woofers and of course, as a Pre-Filter for the passive. There was very little difference between the woofers, the Skanspeak seems to integrate just as well as the Yamaha and gives some extra depth when called for. As I already have the Skanspeak woofers, I may as well use them. They really are cracking drivers, as this near field response I measured shows.



Skanspeak 30W/4558T00 Bass driver.

Remarkably flat, +/- 1dB from 32 to 500Hz with an f3 of 29Hz in its 55L sealed cabinet, even better than predicted. Larger power handling and a massive xmax put it in another league to the venerable Yamaha Woofer I measured below.




I'm doing lots of listening to both bass drivers before making my final choice.

If I do choose the Skanspeak drivers, rather than bodge the Yamaha cabinets, a new pair will be made. Meanwhile, I'll work on the crossovers, both passive and active. I'll use Rod Elliotts active board, its the same basic filter circuit as the KMTech I've been using, but it has both Left/Right on the one PCB. It also has provision for balanced inputs and uses single channel attenuators on each output, which allows both amplifier and Left/Right channel matching. Also the board has the facility for output buffers, so all bases covered on these PCB's. The associated passive crossover, will be much simpler than the stock item, just two caps and two coils on each, I may as well make provision for a resistor network to replace the level controls. Again, I don't want to bodge the original boards, so a new x-over using good quality parts is in order.

I've summarized my findings and the work carried out, it actually took a couple of weeks to modify, try out and reach these conclusions.





I selected Jantzen Superior Z-Caps, as they are supposed to be (like the basic Mundorf Mcap MKP I tried) very neutral, but also superb at handling high frequencies. As only the Tweeter cap is in the signal path I think it's a sensible choice.





This is my layout.
As previously stated, its the stock Yamaha circuit and values, as used for the Mid/Tweeter crossover at 6kHz.
By using the line-level Active Pre-Filter, I've taken the big bank of (21uF) caps out of the signal path of the midrange driver.
The board is point to point and uses a star ground, decent caps and air core coils. I've taken the longer wire runs under and over on the board to stop them flapping about, sticking them down with hot melt or using cable ties doesn't look very tidy. I've used spade terminals for the input/output connections, once the development is completed, the final build will have the wires soldered at these connections.

I need to order the parts and build these crossovers as a first step.