PrimaLuna Technology Preview: Preserving the Sonics With High-Powered Amplifiers
(NOTE: This is the technical sidebar that will appear with my review of the PrimaLuna Seven in TAS #199. While there are undoubtedly valid reasons to refrain from posting the sidebar in advance of the review, I thought readers would appreciate a preview, at least in part, of this remarkable amplifier)
.jpg)
The design team at PrimaLuna faced a formidable challenge; namely, how to match (or exceed) the sonic signature of its DiaLogue Two in a set of monoblock amplifiers with twice the power. This was no small feat because the DiaLogue Two possessed a natural and highly engaging midrange, outstanding soundstaging, extended and articulate bass and highs, and a musicality reminiscent of a live performance. As many designers can attest, it’s quite easy to take a step backwards rather than forwards when creating a more powerful amplifier.
PrimaLuna’s Chief Engineer, Marcel Croese, was no stranger to design challenges. As Goldmund’s principal engineer for several years, he was credited with many innovative designs, including the Mimesis 29 and 28 power amplifiers, and the MM29.4 and MM28.4 monos, among others. He also developed PrimaLuna’s innovative, and highly effective, Adaptive AutoBias circuit.
While many may have used larger transformers to double the output power, Marcel ruled that approach out because big output transformers don’t behave as well or sound as good as their smaller brethren. He opted to use the same outstanding wide-bandwidth output transformers as in the DiaLogue Two (described in Issue 175), employed two discrete audio channels on one chassis, and paralleled them at the loudspeaker terminals so that the outputs would be summed, resulting in double the output power. He also added a 16-ohm tap to each transformer, which when paralleled resulted in an 8-ohm output.
However, the summing approach created other challenges, since a percentage of power from one free-running transformer was pushed into the other, due to output voltage differences caused by circuit tolerances and tube aging. To deal with this problem, Marcel developed a specific cross-coupled positive/negative feedback scheme to balance out both channels perfectly. It also enabled Marcel to offer a 2-ohm tap to drive very low sensitivity/low impedance loudspeakers, as well as some electrostatic loudspeakers.
Since less total gain was needed than in the integrated amplifier, Marcel was able change the pre-stages in a way that improved the drive capacity to the power tubes, resulting in even lower distortion, without the need for extra negative feedback. The circuit was inherently more stable and virtually immune to drive imbalance and its accompanying distortions with tube aging.
This innovative twin-channel-summing approach with cross-coupled positive/negative feedback not only resulted in surprisingly low harmonic distortion, but also matched (and improved upon) the sonic signature of the smaller DiaLogue Two. JH
Discussion Topics
- Speaker recommendation 2-channel, for paycheck to paycheck music lovers
- Magico V3 vs Magnepan 20.1
- Quad 2905
Speakers
- Bad Firmware Update to Oppo BDP-83
- Anyone bought the new Sony BDP-S350 BD player?
- Oppo BDP-83 Blu Ray Player
Comments
Jim,
Looking forward to your review in the next issue. What speaker did you mate this amplifier with? seems like speaker match will be important based on its design and power ratings.
Can be any full range speaker with 87db or more...
Sam,
I used them extensively with the Hyperion Sound Design HPS-968, as well as the Quad ESL-57 (restored by Wayne Picquet).
The Hyperions are 90dB efficient and the Quads present quite a difficult load. I've never heard the Quads produce such powerful and articulate bass, but they certainly don't go down to the depths. The Hyperions do go down into the low 20 Hz range and the bass was as good as or better than any tubed amp I've had in house.
These are definitely quite the bargain, in my opinion.
Enjoy!
Jim
Hi Jim,
It's great to see PrimaLuna's new monos - they look great!
I have a question about the design though: it seems like the decision to use the same transformers led to Marcel developing a cross-coupled feedback scheme. I read this as "more circuitry was added"... and this goes against the idea of design simplicity - which makes me think that maybe starting with a bigger tansformer might have been a better idea.
Now, I can see how using the same transformer would be good from a cost containment / ease of manufacturing perspective... it looks like the new design allowed the new amps to share not only the transformers but the chassis (and maybe other parts) as well. Do not get me wrong, I like a bargain just as much as the next guy, and because of their low cost PrimaLuna is quite high on my list...
Can you ellaborate on how the design achieves a better sound?
Feedback comes in sorts and varieties, but feedback is rarely ever extra circuitry. It's basically a wire which returns a portion of the signal from a certain part of the circuit to a location in that same circuit but closer to the inputs.
Mostly that wire includes a resistor and often a small stability capacitor, but that's all.
The basic idea behind feedback is that it enables you to modify the gain of the circuit.
When it concerns wideband gain change, it is used to optimise some design parameters like total gain, harmonic distortion and/or output impedance (damping factor of power amplifiers). Most amplifier circuits (not ours) posess way too much gain to be useful. Feedback is a great asset when you want to reduce gain to the required level.
When it is frequency dependent feedback we're talking about a filter, and this is how most active filters are realized. Even most RIAA amplifiers use this to get the required correction.
Bigger transformers on the DiaLogue integrateds would not be a good idea, they already are quite big and make for a very heavy amp already considering its modest output power. We're hovering around the optimum between the myriad of tradeoffs that any power output transformer inherently presents. Hence the decision to look for solutions in the electronics instead of the easy way. And yes, the manufacturing rationale was definitely a bonus.
Rgds
Marcel
Thank you for the clarificaiton, Marcel.
Thank you, Marcel, for clarifying this.
Nothing like having the designer discuss the actual design tradeoffs.
Thanks for clarifying this. I have 2 stereo amps (c-j) that can be paralleled to produce double the power. Some have said that this isn't a good idea due to "interactions" in the output stages, but they were unclear as to how bad the distortion was. It must not be too bad. Mine sound pretty good in parallel.
Output power is specified as power into a specific (mostly 8 Ohms) load. It is limited by the maximum voltage the amplifier can develop across that load. This does not change when you parallel amps. Consider two 1,5V remote control batteries, hook them up in parallel, you end up with...1,5V.
The only thing that's changed is the amount of current that can be supplied when asked for, that has doubled.
So actually what has happened is that you reduced the output impedance by half. Which means that you can drive speakers with half the impedance.
When the batteries are put in series, the voltage doubles to 3V.
If you do that to an amplifier, also here the output voltage doubles, which results in doubling the output current too provided that the load did not change, and suddenly the power has quadrupled (power is voltage x current). So specifically a doubling of output power is actually less easily achieved than you might think.
rgds
Marcel
Post new comment