advertisement


Speaker Cables, A Special Note (from ye olde Naim Audio)

Very little. It's a damped inductive isolation operating rather above the primary passband.

If anything you can argue that - by ensuring stability - it's a price well-worth paying.

(there are other ways to solve the problem, but generally they can be a bit tweaky/ less-generally-applicable - hence the universality of the approach)
 
No-one ever tested a Naim amp at the factory with a square wave, of that I'm 100% sure!
Now this may be anecdotal but I'm pretty sure that the wire-wound R22 in series with the output was considered to have enough inductance when added to the speaker cable to create enough of a Thiele network.

I've not aware of anything in the current range of amplifiers that suggest they're likely to be more stable than their forebears.
 
both NACA4 and Exposure Cable were among the cheaper options when first they came along; even cheaper if one went down the B&Q and grabbed some BICC outdoor lighting wire off the reel.
Can you still get yer akshul BICC lighting cable, ie the original and best? I mean without branding, snake oil, veils lifed, etc.

Most such units have a Zobel network. The missing inductor and resistor is sometimes called a Theile network and is there to add impedance at "ultrasonic" frequencies so that at say 50KHz+ where the problems are the amp cannot directly drive a capacitor but has to do so through the Theile network which will add maybe 3-6 Ohms ish in line at 50KHz.
Ah thanks Jez, so that's how it works.

Remind me, the simplest possible Thiele network is an inductor and a capacitor, in parallel, this being placed in series with the speaker load, is that correct? What are the typical values required to ensure that it only becomes effective at (say) 50kHz as you describe above?
 
Steve its usu of the order of 1-6uH, often with a modest resistor in parallel to damp the whole (ballpark of 10-15ohms, depending on implementation overall) and after that , a resistor with capacitor in series across the output (speaker + to -); values vary, but of the order of 100nF + 10 ohms.

MJS highlights one of the alternatives - Naim used 0.22ohms in series with speaker + output only. It can be enough, and effective but has to dissipate quite a bit of power (more so as a % of total output if you are driving a low-impedance speaker) and adds to output impedance (which may or may not matter).

There's a detailed study on the things in Doug Self's book on designing power amplifiers.

ETA: fewer words, more pictures - example diagram pinched from the same book to clarify - amplifier output stage is to the left:

zobel.jpg


(the 'if fitted' capacitor is for capacitor-coupled outputs - think Quad 303 (used 3300uF) and similar, often single-rail amplifiers)
 
Steve its usu of the order of 1-6uH, often with a modest resistor in parallel to damp the whole (ballpark of 10-15ohms, depending on implementation overall) and after that , a resistor with capacitor in series across the output (speaker + to -); values vary, but of the order of 100nF + 10 ohms.
Thanks Martin. That makes sense. The latter bit
and after that , a resistor with capacitor in series across the output (speaker + to -); values vary, but of the order of 100nF + 10 ohms.
is the Zobel network, AIUI, and I can see how that would act as a high-pass filter that just dissipates high frequency signals to earth via the series resistor. Does a Thiele network only work in conjunction with a Zobel network then?
 
Points taken, however, it would appear that, in Naim's case, they wished to ensure that cable inductance fell not much lower than 4uH (see Arkless' "3-6 Ohms ish in line at 50KHz", above), and parallel capacitance didn't much exceed 320pF. This based upon the published specifications, and minimum and maximum recommended lengths of the cables that they supplied/recommended; these in addition to the late MD's comments on the subject via his company forum and via occasional interviews.

Obviously, many thousands of Naim, Exposure and NVA customers have existed over the years, I doubt that many weren't made aware of the loudspeaker cable requirements at time of purchase. I've also never heard tell of a Naim amp that actually succumbed to incorrect cable choice, however, I've seen a few that were killed by errant shorting of outputs (in fact, I have one here). Perhaps we should all be naming and shaming those designs that aren't fully protected against dead shorts? Maybe @Arkless Electronics would like to cast the first stone here?

Regardless, my offer stands to run the specs of any alternative cables that Naim owners wish to consider. As @Gervais Cote found, the difference could be down to one's amp running circa 8ºC more than is typical. Those alternative cable users that prefer to run their's 24/7 should really consider having a closer look.

Two snags.

1) As I recall from working on amplifers at the time, the series of events largely came to light when low-inductance cables started to be used *after* the first Naim amps had been on sale for some time. Were the Naim speaker cables advertised from the same start time as their Amps were on sale? I don't recall that as being the case. But I do recall some dealers being worried by problems.

2) It isn't guaranteed to be sufficient to use cable to add specified levels of inductance or capacitance. Because those requirements then *vary with length AND choice of loudspeakers*. This is because the behaviour of a cable is 'distributed' not 'lumped'. It acts as a form of frequency-dependent 'transformer'. Hence if you simply rely on X meters of cable Y as the requirement someone may use a speaker that the cable transforms into an HF impedance that prompts the amp to oscillate *despite* a bench measurement showing the cable *in isolation* has an LF inductance of whatever.
 
Thanks Martin. That makes sense. The latter bit is the Zobel network, AIUI, and I can see how that would act as a high-pass filter that just dissipates high frequency signals to earth via the series resistor. Does a Thiele network only work in conjunction with a Zobel network then?

The zobel if anything is the more important bit as I understand it - it means at HF the amp 'sees' a resistive load. Good for stability. The Thiele network 'stands-off' parasitic capacitances (and RF) from the amp at HF, and so adds to stability from load interaction effects. Ironically, while you see amps often tested with a load of '8ohms in parallel with 2uF' for testing, it is smaller capacitances, tens of nF, that can really destabilise power amplifiers and provoke a riot above the audioband /meltdown resulting. The late Cyril Bateman published some analysis ('Cables, Amplifiers and Speaker interactions') using a DSelf 'Blameless' amp - and cooked the output stage once or twice in the process IIRC...
 
Thanks Martin,
So that being the case, AIUI the Arkless boxes only contain the LR part, the Thiele network itself. Does that mean that Naim amps have a Zobel as standard, so there was no need to add one?
 
Yep. Ideally one would open up the NAP250 (or whichever model) and remove the 0.22R resistor, replacing it with the Theile network.... You could call it something else then.... Avondale maybe:D

The 0.22R resistor wastes power especially with low impedance loads and reduces the damping factor to 16-18 when driving 4R speakers.
 
In light of your second comment, Jez, I wonder if the result of Naim's use of 0.22R resistors in place of Thiele networks had indirectly helped to sell the idea of their own brand fully active systems being the be-all/end-all upgrade?
 
As an aside, the Deltec DPA50s I use actually use a 0.1ohm wirewound in that position - no inductor- but that's inside the feedback loop (taken off the speaker terminals). Damping factor -a daft measure - er, 18000 (which Paul Miller in the original review noted as 'mildly excessive' ?!. )

The amp design itself has nothing to do with the 3-stage Lin, so incorporates all sorts of other choices/ routes to stability and compensation.
 
In light of your second comment, Jez, I wonder if the result of Naim's use of 0.22R resistors in place of Thiele networks had indirectly helped to sell the idea of their own brand fully active systems being the be-all/end-all upgrade?

I can't see the relevance of that really...

FWIW when I've tried adding 0.22R (yes also 0.1R and 0.33R etc) to amplifiers that don't need it for stability, in order to gauge the subjective effect, it is as expected a negative effect.... a downgrade... Resolution and transparency are reduced and it sounds generally muddier and less grippy. If there is any saving grace at all it is that the reduced resolution and transparency makes poor recordings maybe a little more palatable.
 
As an aside, the Deltec DPA50s I use actually use a 0.1ohm wirewound in that position - no inductor- but that's inside the feedback loop (taken off the speaker terminals). Damping factor -a daft measure - er, 18000 (which Paul Miller in the original review noted as 'mildly excessive' ?!. )

The amp design itself has nothing to do with the 3-stage Lin, so incorporates all sorts of other choices/ routes to stability and compensation.

Prima facie there would be no point in such a resistor if inside the loop but I'm sure there's a good reason in this case.
I'd be interested to see the schematic if you have it... but will fully understand if you only got it on condition of not sharing it.
 
In light of your second comment, Jez, I wonder if the result of Naim's use of 0.22R resistors in place of Thiele networks had indirectly helped to sell the idea of their own brand fully active systems being the be-all/end-all upgrade?
I can't see the relevance of that really...

FWIW when I've tried adding 0.22R (yes also 0.1R and 0.33R etc) to amplifiers that don't need it for stability, in order to gauge the subjective effect, it is as expected a negative effect.... a downgrade... Resolution and transparency are reduced and it sounds generally muddier and less grippy. If there is any saving grace at all it is that the reduced resolution and transparency makes poor recordings maybe a little more palatable.
You had mentioned 'wastes power' and 'reduces damping factor' which is what got me thinking of eliminating crossovers, going active, etc. It's just the way my none-EE qualification mind works. But thanks for explaining, regardless my inability to fully comprehend.
 
Yep. Ideally one would open up the NAP250 (or whichever model) and remove the 0.22R resistor, replacing it with the Theile network.... You could call it something else then.... Avondale maybe:D

The 0.22R resistor wastes power especially with low impedance loads and reduces the damping factor to 16-18 when driving 4R speakers.
+1 for the Avondale maybe :D
 
Were the Naim speaker cables advertised from the same start time as their Amps were on sale? I don't recall that as being the case.

Having owned my 160 for nearly 44 years (yikes is it that long) the original advice was as post #1 above. Fancy speaker cables hadn't been thought of then. If you were using anything thicker than bell wire you were doing pretty well then. My father and I had lengths of heavy stranded single core in red and another in black. NACA 4 or Linn K20 was a neater solution.
 
Having owned my 160 for nearly 44 years (yikes is it that long) the original advice was as post #1 above. Fancy speaker cables hadn't been thought of then. If you were using anything thicker than bell wire you were doing pretty well then. My father and I had lengths of heavy stranded single core in red and another in black. NACA 4 or Linn K20 was a neater solution.
Yes, I recall the local dealer twisting these up in the back room for each Naim NAP that was sold before NACA4 came along. Don't recall if it was the same 56-strand as RS used to supply; more likely bulk appliance hookup wire from a local supplier here in Canada.

What I don't recall is having seen/heard mention of there having been two recipes, 2.5mm^2 vs. 4mm^2; that is, until I stumbled upon the scanned letter featured in post #1.
 


advertisement


Back
Top