darTZeel NHB-108/NHB-468: Curiously resp. strange Distortion Waveform in the Diagrams from Stereophile

[tiefbassuebertr]

Under
https://www.diyaudio.com/community/threads/dartzeel-amp-schematic-build-this.134362/
are describe various darTZeel clone versions but the diagram of the original models NHB-108 in fig 7 for distortion and noise waveform with fundamental notched out in the review under
https://www.stereophile.com/content/dartzeel-nhb-108-model-two-power-amplifier-measurements
so as the model NHB-468 in Fig. 9 under
https://www.stereophile.com/content/dartzeel-nhb-468-monoblock-power-amplifier-measurements
show an odd character of the distortion wave form;
only the lower half-wave has a sinusoidal character, while the upper half-wave is more like a triangular character (both mentioned diagrams also to see in the attached files No 1+2).
Every second half wave has the tip flattened (NHB-108) or rounded (NHB-468) in the upper area. Same diagram of attachment No 3-5 for the power amps Halcro DM38, Graaf GM200 and Pass Labs Aleph 4 don't show such an effect - lower and upper half-wave show identical character (clean H2 resp. H3 on Halcro's DM38).

I haven't heard the models NHB-108 and NHB-468 from darTZeel, but - as long as it is not a measurement error on the equipment from Stereophile - there is anywhere a deficiency in the circuit design which is also audible - so I think (go to the simplified schematic from attached file No 6 resp. to https://6moons.com/audioreviews/dartzeel/108.html) so as to to the various other schematics from attachment, which have a certain similarity in terms of the input stages and due the fact, that only local (serial) feedback (and no global NFB) exists regarded the "End Millenium" from LC-Audio (go for the circuit description to
https://www.octave-electronics.com/lcaudio/temil.shtml)

Where is to find a circuit description in detail for this topology from darTZeel ?

Thank you very much for an information.

 

[Rallyfinnen]

https://www.diyaudio.com/community/threads/dartzeel-amp-schematic-build-this.134362/

I have done some measurements on my clones, and as soon as I move the distortion measurement to the 'driver side' of the outputs base resistors, there is basically no distortion, so my conclusion is that almost all the distortion is generated by the output transistors.
In simulation, the diodes across these resistors rarely conduct, only at extreme currents.

 

[analog_sa]

Rather weird if there is much common behaviour between the original 108 and the subsequent amps. The original consisted of a NFB voltage amp followed by a pretty standard diamond buffer, the rest were all simple EFs with a slightly different front end. All of them with really high distortion.

 

[EdGr]

The strange shape implies that the output transistors are having difficulty turning off. The base resistors are very large, 27 ohms from the schematic referenced in post #2. There are no emitter resistors. I am putting this design in the "bad" category.
Ed

 

[Rallyfinnen]

I was interested in the amp for a long time because of the subjective vs objective stuff (a lot of people like the sound), just like tube amps. I built a cheap tube amp and found it kind of charming, so after that I decided to try the Dart clone as well. Oddly enough, I kind of liked the sound (mids & treble) from the clone, enough to start building another one..

Base resistors can be 'tuned' and lower values gives higher bias and lower output impedance. Depends on rail voltage how high it can go in class A. Measures pretty well until the outputs start switching, after that distortion is at -50-60dB depending on load, with higher orders tapering off slowly. Not that far off a decent tube amp, but probably more high order distortion than a tube amp though. The higher orders in class B seem to improve a bit with lower base resistor values, so that could indicate the outputs switch off better with lower values. Subjectively the sound changes, but I'm personally still not sure which I prefer..

Well, no reason to start a new thread, there is a lot to read in the previously linked thread.

 

[Rallyfinnen]

Rather weird if there is much common behaviour between the original 108 and the subsequent amps. The original consisted of a NFB voltage amp followed by a pretty standard diamond buffer, the rest were all simple EFs with a slightly different front end. All of them with really high distortion.

Not sure how to interpret the above? Do you think the clones are far off from the original? From what I have been able to find on the internet, is seems plausible that the clone with 1 pair of outputs is very close to the original first version released. After that things get a bit confusing with 'improved versions' with multiple pairs etc.

 

[EdGr]

Depends on rail voltage how high it can go in class A. Measures pretty well until the outputs start switching, after that distortion is at -50-60dB depending on load, with higher orders tapering off slowly.

That matches my expectations. Once the output transistors cut-off, the THD will be a few tenths of a percent.
Ed

 

[analog_sa]

Not sure how to interpret the above? Do you think the clones are far off from the original? From what I have been able to find on the internet, is seems plausible that the clone with 1 pair of outputs is very close to the original first version released. After that things get a bit confusing with 'improved versions' with multiple pairs etc.

Sorry, a misunderstanding.

By "the original" i meant the 108. It is the only one based on a diamond buffer, all subsequent Dartzeel amps (to the best of my knowledge) have been simple push pull followers with a Vbe multiplier. Not even an EF2.

I find it unlikely very dissimilar output stages to have an almost identical distortion quirk.

As for the clones, some were 100% true copies of the initial 108. Which itself received at least two mods - a servo and alternatively a supercapacitor in order to sort out the offset. The supercap modded one iirc was called model B. Much later came out the 108 model 2 which was a completely different amp.

 

[analog_sa]

Depends on rail voltage how high it can go in class A.

Absolutely. If you add outputs and run it in deep class A, the distortion is orders of magnitude lower. I played up to 4.5A 🙂