My last super release (of the 7/8 released
in the last 15 years) of the NAIM H140 clone:- Power supply soft-start circuit,
- Power supply circuit with two-stage RF suppressor,
- Dual-mono stabilised power supply stage for "differential input stage" and "pilot stage",
- BJT linear power supply stage of low impedance dual-mono type with 4 fast double-diodes for each channel and total filtering capacity 84,000 uF,
- 600VA audio-grade toroidal transformer,
- circuit type in Class "AB" with "almost complementary" Darlington (Rail positive) + Sziklai (Rail negative) configuration,
- DC voltage protection on the output speaker terminals,
- Continuous power output x 2: 110 Watt on 8 Ohm - 190 Watt on 4 Ohm
- Maximum power output on 10 mS burst and loads up to a minimum of 2 Ohm: 320 Watt per channel
- Bandwidth: 120,000 Hertz / -3 dB
- Input impedance: 18K Ohm
(Detail of amplification PCBs mounted under the heatsink)
And now thanks to the good and esteemed René - @rensli
who managed to get me 4 PCBs the now unobtainable H140 clones, I'm already working on a four-channel stereo power amplifier (low + high) with digital crossover and active room correction DSP system, Dirac Live® integrated.Although at the moment I am still undecided whether to use the original ZTX652/653 transistors (which have given me some stability problems in the past) for the VAS stage or to continue using the reliable, stable and musical triptych of Toshiba 2SA1145/2SC2705/2SC2240 transistors.
Last edited:
Thanks,
Nice looking amp, i want to hear it in every music program! Maybe someday in the future
I would definitely stay with these if you use H-140 PCB boards, better play with diodes, in baxandall circuit.
ZTX might not work(can bring fatigue to some extent in some programs, or you might loose the front-end signature and amp becomes a dead-sound amp) and i never tried them on H-140.. was no point cuz it was OK with 1145/2705, the front end signature was there.
If you still want to try ZTX series, then i strongly suggest to install transistor Sockets. This gives u an opportunity to try them with various transistor LEAD dimensions. Starting from short lead all the way to the longest. I guess H-140 PCB was never designed for ZTX series.
U must also keep the leads in perfect condition, no bending, no random solder and so on. Keep everything perfect is important.
Nice looking amp, i want to hear it in every music program! Maybe someday in the future
I would definitely stay with these if you use H-140 PCB boards, better play with diodes, in baxandall circuit.
ZTX might not work(can bring fatigue to some extent in some programs, or you might loose the front-end signature and amp becomes a dead-sound amp) and i never tried them on H-140.. was no point cuz it was OK with 1145/2705, the front end signature was there.
If you still want to try ZTX series, then i strongly suggest to install transistor Sockets. This gives u an opportunity to try them with various transistor LEAD dimensions. Starting from short lead all the way to the longest. I guess H-140 PCB was never designed for ZTX series.
U must also keep the leads in perfect condition, no bending, no random solder and so on. Keep everything perfect is important.
Last edited:
Spice is an essential tool. It's great for bad assumption discovery and also for trying extreme, real-world impractical things. I have also used it to gradually add imperfections to "idealised" components to see what effect this has.SPICE is a good, often free and precise tool for audio design. Audio fanatics and newbies who dismiss it as some evil, mechanistic science or imagine that it leads to inhuman, unmusical results are being an ostrich (i.e. head in the sand). What surprised me about me about it's usefulness, was how many assumptions I had based my design mods on, were actually false and how many more solutions existed than basic textbooks suggested. I just hadn't bothered to check them out properly because of the time and expense for reading, obtaining parts and making the changes in order to test them
Spice has limitations that will catch you out if you are not aware. Garbage in, garbage out. If you don't model everything, don't expect it to behave exactly as reality; trace impedances and RF interference, for instance. Not all of the transistor models are accurate either...I made a big circuit mistake when I trusted a mfr's MOSFET model whose Cdg was 10x too small. I think it was Zetex or Fairchild. Besides, in reality there will be significant differences among same numbered parts.
So I use Spice to confirm simple theories but I use my imagination to predict and bench tools to see what's really going on.
Spice can also lie. Well, not exactly lie, but it allows you to fool yourself because it always performs a small signal analysis by assuming all components are linear. This is not wrong in itself but is a simplification. If this is not the case, the model analysis may not accurately predict the circuit reality. This is a common mistake made by experts too.
I find one of the great things about spice is that it enables you to quickly "construct" a circuit and test it.
As you say - you can get situations where the circuit works in spice but simply doesn't work when applied to a breadboard.
However I think it can give you a basic starting point (depending on circuit complexity of course!).
As you say - you can get situations where the circuit works in spice but simply doesn't work when applied to a breadboard.
However I think it can give you a basic starting point (depending on circuit complexity of course!).
Something that isn't mentioned when we discuss those VAS and current source transistors, is their actual bias current. It likely needs to be different to operate the ZTX653/753 or 2SC2705/2SA1145 correctly in a similar part of their operating range. The NAP140 and similar series amplifiers all used the same transistors with a high 12mA VAS bias. As shown on the H140 schematic above, it is set by the emitter resistor of Q3 (68R) and will also vary with supply voltage so we need to be careful to compare apples with apples when rolling semis there.
I haven't played with this recently but I think it's likely you can change a lot of sound qualities with the VAS bias level, as long as its operating point is essentially in the linear range.
Yes, bias is one of the many things...it is dependable.
But, taking only bias current into discussion is not enough,
Unfortunately, it is never enough
, one must understand this circuit as a whole.Hmm, input pair transistors, there is an idea.
But VAS, i mean...loading can be adjusted via phase shift networks and transistor types are not that critical here... there is also burn-in time ofc, so if Transistor change is applied, results can be expected after some heavy loading or after a slightly longer use of the amplifier.
And PCB layout is what i am into.
Yep, the sound will move around in the 24h period too, it is important that it will remain fatigue-less through all the time the amp is switched on.
I still wouldn't suggest any mods if the concept is unclear to the user.
I understand why you say that but my experience with "burn-in" is often repeated with each listening session - particularly after an interval of a few days or more.. That is, it's my ears - my concentration, mood and even the wax in my ears etc that's "burning in" rather than the components in the amplifier. It's a predictable effect over a short period but less so over several days. I think that's the key to distinguishing between real and subjective sound effects better.
The point I am making though, is that small and fast VAS transistors like 2SC2705 are not much more than small-signal transistors. They are typically biased with about 5-8 mA current to get them in their linear range. The ZTX653/753 transistors are obviously bigger chips in a very small package. That makes them tough and capable drivers too, as in they are in NAP90, NAIT1,2,3 and other UK brands of the 1980s. Think of them as BD139 size chips in a TO92 case, like BC639. They still need more bias current to keep them operating in their linear range over the full voltage swing from almost rail to rail. https://www.electronics-tutorials.ws/amplifier/transistor-biasing.html
So, for using the small Japanese driver transistors, it just means measuring the VAS bias current and experimenting with less by say, 15, 30% etc. It may be difficult and inconsistent to analyse by ear but should be easier with an oscilloscope or a sound card audio analysis program and a preamp for your PC, to find the "sweet spot".
The point I am making though, is that small and fast VAS transistors like 2SC2705 are not much more than small-signal transistors. They are typically biased with about 5-8 mA current to get them in their linear range. The ZTX653/753 transistors are obviously bigger chips in a very small package. That makes them tough and capable drivers too, as in they are in NAP90, NAIT1,2,3 and other UK brands of the 1980s. Think of them as BD139 size chips in a TO92 case, like BC639. They still need more bias current to keep them operating in their linear range over the full voltage swing from almost rail to rail. https://www.electronics-tutorials.ws/amplifier/transistor-biasing.html
So, for using the small Japanese driver transistors, it just means measuring the VAS bias current and experimenting with less by say, 15, 30% etc. It may be difficult and inconsistent to analyse by ear but should be easier with an oscilloscope or a sound card audio analysis program and a preamp for your PC, to find the "sweet spot".
I have a STK4182MK2 based amplifier, outputs are quasi complementary, like naim circuit.
Comparison moment is always available to me. Mood, wax or something alike does not contribute to the fatigueness.
Its like that: the more your healthy, the better all sounds will become.
Sound of this topology do change in a 24h, its not constantly changing but from a cold start it will drift for sure
Yes, that is what i am going to do next, no point in describing the sound, we need statements, proofs.
Comparison moment is always available to me. Mood, wax or something alike does not contribute to the fatigueness.
Its like that: the more your healthy, the better all sounds will become.
Sound of this topology do change in a 24h, its not constantly changing but from a cold start it will drift for sure
Yes, that is what i am going to do next, no point in describing the sound, we need statements, proofs.
Personally, as transistors in the driver stages of my many creations of the NAP140 / H140 clone, I have experimented with the TIP41/42 pair (which is a transistor born for other applications but not really for audio constructions) that gives a sound characterized by a clean and dry bass but without refinement and detail in the mid and treble.
The pair MJE243/253 instead is characterized by a more marked and convincing musicality.
But for 4/5 years now, for my NAIM H140 clone creations, I have been using the more robust MJE15030/15031 pair (without heatsinks) with great satisfaction, which together with the extraordinary refinement of the MJE243/253 transistors (perhaps with an even better timbre chiseling of the midrange), also guarantees me a more balanced and quiet thermal management but also a more authoritative driving ability of the 2SC5200 power transistors.
Last edited:
Does not matter as long as rest of the circuit is tuned accordingly
Lol, you are partially right. Treble and mid with no detail happens only if you "just swap drivers for drivers" and hope it will play differently and better lol.
Its like an equalization circuit...
But so many blablabla, and still no proof or anything similar.
Last edited: