There are a few solutions to fight the lack of banwidth in final transistors, but that is the easiest and cleanest way to get very low distortions .Check Kenwood L-A1 quadrive system.
Just catching up by reading all the links that mention Blomley amplifiers on the forum and I found this comment from Allen Wright, wherein he mentions the very amplifier under discussion here. Bromley Amp
Last edited:
What causes the magic?
Unfortunately, my technical knowledge is not enough to support the following with electronic arguments, but I am increasingly convinced that Blomley's or Rowan McCombe's amplifiers sound so good not because of the proper handling of crossover distortion, but to look elsewhere for the cause of the magic.
In my opinion, the magic is to be found in the grounded based VAS. To simplify things, what do we see in both amplifiers? (we could also list Mr. Hartsuiker's derivative here). Two CCS' facing each other, which are controlled by the input stage through their emitters. This topology is very fast and linear, and most importantly eliminates subtle "time errors" due to many difficult-to-determine capacitances of transistors (problems due to Miller modulation and many other variable capacitances). I built three such amps this year, and even though their inputs and outputs are different, they have almost the same sound. All three produce the same "presence" that Mr. Dermot Herron writes about so enthusiastically in his article. My other amps, no matter how good they sound anyway, unable to do this. The sound of these three are so vivid, clean and smooth that they can really be mistaken with live voices (my wife and daughters have confirmed this several times).
Summarizing the above, I am already 110% sure that the unpleasant dryness and sterility that comes from solid state amplifiers mainly comes from the unwanted capacities of VAS, eliminating or reducing them will result in an amplifier with orders of magnitude better sound. Our ears seem to be much more sensitive to these errors than to THD or TIM.
I know the problem is much more complex than that, but I find it worthwhile and important to mention the above as well.
Unfortunately, my technical knowledge is not enough to support the following with electronic arguments, but I am increasingly convinced that Blomley's or Rowan McCombe's amplifiers sound so good not because of the proper handling of crossover distortion, but to look elsewhere for the cause of the magic.
In my opinion, the magic is to be found in the grounded based VAS. To simplify things, what do we see in both amplifiers? (we could also list Mr. Hartsuiker's derivative here). Two CCS' facing each other, which are controlled by the input stage through their emitters. This topology is very fast and linear, and most importantly eliminates subtle "time errors" due to many difficult-to-determine capacitances of transistors (problems due to Miller modulation and many other variable capacitances). I built three such amps this year, and even though their inputs and outputs are different, they have almost the same sound. All three produce the same "presence" that Mr. Dermot Herron writes about so enthusiastically in his article. My other amps, no matter how good they sound anyway, unable to do this. The sound of these three are so vivid, clean and smooth that they can really be mistaken with live voices (my wife and daughters have confirmed this several times).
Summarizing the above, I am already 110% sure that the unpleasant dryness and sterility that comes from solid state amplifiers mainly comes from the unwanted capacities of VAS, eliminating or reducing them will result in an amplifier with orders of magnitude better sound. Our ears seem to be much more sensitive to these errors than to THD or TIM.
I know the problem is much more complex than that, but I find it worthwhile and important to mention the above as well.
Last edited:
I don’t want to deal with this any more, I just mentioned my experience and thoughts with the Blomley (and similar) topology. Moreover, this explanation is not taken seriously by the majority of members. 🙂
Last edited:
Looking at the schematic it's amazing someone designed that in the early '70s. (I was 15 - struggling to understand the schematic of my SWTPC "plastic Tiger"). Without SPICE running so easy-peasy on a PC. At my aged EE, I can understand the language of the topology, but I couldnt create the poetry on a blank sheet.
That's an interesting theory, and I had similar thoughts about the grounded base VAS.
The long-tail pair can be seen in a lot of very clinical and dry transistor amps as well so that certainly isn't it.
Allen Wright's implementation uses a very special matched transistor pair in a single can called the LM194, which hasn't been available for many years. I did manage to acquire ten of the AS194 equivalents recently from a seller in Poland. He also used these in a preamp design around the same time as the Magic Amp, and I recall listening to the Doobie Brothers on headphones and being completely amazed at the sound.
The use of that part necessitated inverting the whole power amp circuit, since Rowan's original design used PNPs for the input pair. Allen sought - and received - Rowan's blessing to do that for the production prototype that was constructed on a PCB.
Last edited:
egra is very welcome here, he made a Bromley PCB available and I've just had a set of ten made in China. 🙂
These guys were above my head at the time too, I would have been 22 when Rowan did this circuit. I guess he would have been about 40 and Allen around 30 at the time. I also knew the guys who invented the Fairlight synth, now that was indeed a work of black magic. I couldn't afford to live in Sydney these days, but was very fortunate to have spent the seventies there amongst such people.
Below I've attached a copy of the LTSpice file for this amplifier as shown in Allen Wright's Tube Amp Cookbook. There are very few minor changes to the original.
1. The value of R12 was not shown. I've chosen 1k5 as a reasonable value.
2. The identities of the transistors in the output stage were taken from the original EW+WW article. This is known to be wrong for the version I have seen photos of, but that amplifier wasn't a 30W version.
3. The current source on the long-tailed pair hasn't been expanded out in full. It should be the same circuit as shown attached to the emitter of Q3, but with resistor values chosen for 2mA.
4. All transistors up to and including the common base VAS stage were chosen from parts available in LTSpice. The originals were BC series parts, and I doubt it will make much difference.
1. The value of R12 was not shown. I've chosen 1k5 as a reasonable value.
2. The identities of the transistors in the output stage were taken from the original EW+WW article. This is known to be wrong for the version I have seen photos of, but that amplifier wasn't a 30W version.
3. The current source on the long-tailed pair hasn't been expanded out in full. It should be the same circuit as shown attached to the emitter of Q3, but with resistor values chosen for 2mA.
4. All transistors up to and including the common base VAS stage were chosen from parts available in LTSpice. The originals were BC series parts, and I doubt it will make much difference.
You will quickly discover that the upper and lower gain pots run out of adjustment. Suggested change is to use 1k trimmers in these locations. Settings for the lowest THD are as follows:
1. DC Offset: 0.2622
2. Upper gain: 0.655
3. Lower gain: 0.038
These values result in a THD of 0.04089&
According to the Build Notes at bottom right, increasing the value of R23 so that the diode current is around 700uA reduces distortion. With a 47K resistor I get 0.034438% at a diode current around 756uA. That's about as good as the circuit gets without further modifications.
All above measurements were taken with a 0.1V 1kHz input signal.
1. DC Offset: 0.2622
2. Upper gain: 0.655
3. Lower gain: 0.038
These values result in a THD of 0.04089&
According to the Build Notes at bottom right, increasing the value of R23 so that the diode current is around 700uA reduces distortion. With a 47K resistor I get 0.034438% at a diode current around 756uA. That's about as good as the circuit gets without further modifications.
All above measurements were taken with a 0.1V 1kHz input signal.
Last edited:
The value of 1K5 turned out to be too high. Use 1K instead for 15V regulation dropper resistor R12. If you build this a 1/4W resistor will get too hot, so use 0.4W metal film in this location and space it off the board.
The best setting for the Crossover Distortion trimmer seems to be zero, where I get a THD of 0.03216%
Fortunately changing this isn't interactive with other adjustments if done last.
Changing D9 for a Schottky (or Germanium) diode gets us down to 0.03%
Fortunately changing this isn't interactive with other adjustments if done last.
Changing D9 for a Schottky (or Germanium) diode gets us down to 0.03%
Last edited:
All this tells me that the current source is set far too high. Changing R30 to 150 ohms for 4.5mA through Q3 and readjusting the DC Offset to 0.2715 gives us 0.01516% THD. That's getting to be more like it.
At this point we have a nice little 1975 vintage 30W amplifier. THD at 30W is 0.112% and the harmonic content looks pretty good. Any further tweaking will need a different topology. It should be noted here that no-one in their right mind would build their own current sources these days, I'd use Semitec FET current sources. Also, 2mA and 4.5mA are stock values.
Sensitivity for full power is 1V p-p.
Sensitivity for full power is 1V p-p.
Last edited:
Last edited:
Some people prefer to see a schematic.
I tried to attach it here, but my LTSpice (I do not use it, but TINA),
failed to open symbol "potentiometer". Somebody else may be able
to do it for all.
A netlist alone is of no educational value ..
I tried to attach it here, but my LTSpice (I do not use it, but TINA),
failed to open symbol "potentiometer". Somebody else may be able
to do it for all.
A netlist alone is of no educational value ..