Ok, here's a new idea for my integrated amp...

OK, as some of you might recall, i'm working on an integrated power amp... now, the power amp will be ESP's updated 60w AB amp (modified to reduce power to arround 40w), and yesterday i had this idea: do you guys remember the page about the guy who did some research on the Lavardin "memory-free" amplification? He had a functional headphone amp schematic that got some great reviews, so i was thinking, why not base the preamp on that? any ideas? Suggestions? Thanks.

If you increase the gain of the power amp you will probably not need a preamp. Why pass the signal thru another gain stage if you don't need it. If the impedence of the volume control is low (10k or less) you could use a buffer in front of it.

The only reason I see for a preamp is if you plan to build a phono stage.


P.S. The buffer provided here (before the volume control) by smithy666 works well and is pretty transparent.


[Edited by jam on 07-07-2001 at 10:25 PM]

Not necessarily so, since most cd players and tuners have a 2 volt output.

But if you must have a preamp the memory distortionless circuit would be a good starting point though the input differential seems overly complicated. Just be careful and use a regulated power supply to power the preamp. Please let us know your progress.
Good luck.


I have been playing with elements of the design for sometime now and I feel there is some merit to parts of the design but I have problems with the complexity of the differential as a simpler differential works just as well. Also much that is claimed to reduce memory distortion is just good engineering and has been used in several top solid state designs for years.

Beware, the design is fraught with cascodes and current sources, not your favourite design elements.I have used similar ideas when I modified my Aleph 2's.

I also take issue with the author's need for huge amounts of feedback and the need for a current mirror on the differential. Implementing a current mirror in the Aleph's differential hurt the sound, despite lower distortion numbers.


I had suggested using the Lavardin design in a previous post by Dorkus regarding building a line stage among other circuits such as Pass Labs' DIY Discrete Op-Amp and Class-A designs using MAT02 & MAT03.

The Lavardin Low Memory Distortion topology no doubt looks a bit complex at the input stage but it might be worth the trouble. My idea was to retain the circuit till the buffer prior to the output stage, scrap the output PP stage and have a single ended class A stage to drive any load. This way, we avoid cross-over distortion and still have enough voltage swing, current and a low output impedance.

I have not yet tried the circuit practically but I have done some preliminary simulations using Electronics Work Bench. I used separate current sources for each stage consisting of a zener diode and three transistors. The simulation results are good indeed. However, I found that one can't play around too much with the feedback network and the differential input stage (FET) fails to get properly biased. Notwithstanding, the circuit is workable and worth the try.

Jam's link to a preamp design, is a reproduction of the Elektor published Battery Powered Preamplifier. The buffer stage looks good. However, the low supply voltage may hamper full benefits from being derived. No doubt, modern sources provide 2V at the output, but transients can well clip, leading to input clipping of the power amplifier. Take note that many commercial designs offer about 24V output voltage (Classe', among others), Cello perhaps, even higher output voltages and most commercial preamps have a range between 6 and 10V RMS at their output. High voltage swing plus low output impedance with a volume pot before the power amplifier, is a desirable setup. If the input impedance of the preamp is high enough, the extra input buffer stage can be ommited, and potential dividers used for each input source for level matching and minimising channel cross-talk.

All the best for your effort.
The Lavardins and the peufeu site reminded me of something
when I first came across them, and I've recently remembered
what it is. When Mission/Cyrus produced the Cyrus III
amplifier in the early 90s, they claimed that the circuitry
was designed to work in "constant power" mode, which they
claimed improved the sound quality, although with no mention
of "memory" effects. I wonder whether Mission/Cyrus have any
patents on this, and how their approach relates to the
circuits under discussion.


I have built the circuit( differential) as described by the author and replaced it with a cascoded current sourced jfet differential and have got better results sonically. Take note, the author has had problems in trying to duplicate the compound topology in voltage amplifier stage (stability).Sometimes complexity can be a double edged sword, his goal was to increase loop gain and hence increase the amount of feedback he had to play with. I have my doubts about the use of excessive amounts of feedback. But please build the circuit and let us know the results.

In regards to the buffer, it is nothing but a simple comp. cross-coupled buffer used mainly in video applications. The topology first appeared in integrated circuits,I believe, like the Elantec EL2001.

Levinson have used variations of it in their equipment for years.

If you look at the circuit carefully you will find that the circuit is easily scaleable to higher voltages, if clipping is an issue. I use a version that runs off +/- 24volts, but in Lisandro's case I doubt there would be a problem as the signal can swing pretty close to the rails and this is fine for most sources I know like cd players and tuners.


Cascoded current sources have been used for years, and is nothing new. I just happens that Lavardin found that it reduced a certain form of distortion, which they identified and hence part of the patent. I doubt it would hold up in a court of law. I have seen schematics of a Sony design from almost 30 ago that was pretty similar. There is not that much new in circuit topology. I would be great if we could get a copy of the Mission Cyrus schematic and compare the topology.


P.S. If I am not mistaken Nelson Pass used similar circuitry back when he was with Threshold.

[Edited by jam on 07-09-2001 at 09:14 PM]
Thanks for the feedback, i'll keep you briefed as soon as i start playing with the thing (in a month or so, have finals coming). Better start shopping for those JFETs!

Basically, i'm drawn to the concept because, despiste i'm not entirely convinced, it's backed by actual measurments and objective ideas, and, on top of that, it's said to sound great :) I'll see what turns out of this one!