I would think having current response constant was the goal, especially when using them to drive the drivers in active XO design. But driving normal speakers with passive XO build in might have different requirements.
This is going to be an interesting study.
you can measure phase shift in respect to the previous stage or another system's signal. Or even between two CDs in which one of the signal is shifted. From a CD, what is the reference, the original non-phase-shifted signal?
Remember the golden rule: Phase is not an absolute value, but a value compared to a reference signal.
Remember the golden rule: Phase is not an absolute value, but a value compared to a reference signal.
That was not result of measurement, but rather the author's preferred polarity in the author's system. What was mentioned was each brand had quite consistent trends with some exceptions. My experience is the same as on that chart as well, this is why I have a polarity switch in my active speaker for convenient switching.
I'm sure there are many threads that contain lengthy discussions about the importance of absolute phase, but perhaps we should drop the subject here. Those who claim they can detect it will never agree with those who claim it doesn't matter. Religious argument.
Some other interesting points being made, but I have to wonder how much all this nitpicking and circuit manipulation really improves or could improve an already superb amplifier. I think you can change the sound quite easily, but will it really be better? I bought the parts to change compensation, but now I wonder why I should tinker with it. Part of the beauty of the MyRef is its simplicity and the fact that Mauro designed it BY LISTENING TO IT. No simulation or graph is a substitute for what our ears tell us.
Glad that the Caddock at R3 worked out for you. I would hesitate to change many other resistors in this or any other amp to the same type. There can be too much of a good thing. Want to walk on the other side of the street for a while? Throw a Shinkoh tantalum into the mix at R10 or R12. Sweeet.
Peace,
Tom E
Some other interesting points being made, but I have to wonder how much all this nitpicking and circuit manipulation really improves or could improve an already superb amplifier. I think you can change the sound quite easily, but will it really be better? I bought the parts to change compensation, but now I wonder why I should tinker with it. Part of the beauty of the MyRef is its simplicity and the fact that Mauro designed it BY LISTENING TO IT. No simulation or graph is a substitute for what our ears tell us.
Glad that the Caddock at R3 worked out for you. I would hesitate to change many other resistors in this or any other amp to the same type. There can be too much of a good thing. Want to walk on the other side of the street for a while? Throw a Shinkoh tantalum into the mix at R10 or R12. Sweeet.
Peace,
Tom E
Hi Tom,
until now I've kept reserved the alternate compensation also for this reason and as you saw I'm ready to recognize when, maybe, I was wrong
But for sure that alternate compensation has an important impact on sound that obviously should be carefully evaluated and measured.
It would be interesting to read your opionions on such changes since you're a very good listener.
I think we should be finished with the "polarity" thing. Mainly I felt since someone had some questions as to what it meant, it should be explained. It is just part of listening evaluation.
I think whenever we design something, there are some basic criteria that are related with more accurate reproduction of sound. These should be measured. However, whenever we hear something wrong, it is necessary to figure out what the exact problem is that does not show up in the basic measurements. This is more technology involving, and usually not easy. At least this was what I was facing when trying to find a solution for interconnecting cables (just as example, further cable discussions will be ignored). The amplifier I use in my active speaker is an LM3886 chip amp, and it currently sounds better than the MyRef I have, but due to all the impedance compensation an equalization necessary, the overall power efficiency is not so good. The current pump seemed like a good way to equalize and transmit power to the driver more efficiently which is what attracted me.
Generally, I use the most common components to start with for the basic performance evaluation. The Caddocks were appealing mainly because they were smaller than what I was using, and I'm a sucker for getting things small without degrading performance. Since the Caddocks improved the resolution and high volume playback on another amp, I decided it would be best to use it as a baseline to detect deficiencies in listening tests. Personally, I think if they can increase the cross section of signal path with the wattage rating, it would perform better.
Last edited:
Just changed the R3 in the other channel to the Caddock 30W resistor. Interesting experience. The alternate compensation values sound more damped, but somewhat lacking in emotion. The MyRef C TP with the Caddock R3 seems more live, but with some coloration in sound.
Music Fidelity A370 seems to use a different form of compensation.
Last edited:
Since you've reported changes with first Caddock burn-in maybe it would be clearer after some hours of use...
Let us see how is going.
Sure but remember that in the A370 the current pump is discrete... MyRef's LM318 compensation is specific to make work the LM3386 pump...
The basic sonic coloration in the MyRef C, which seems to be in the mid/lower range below 1KHz, is the same regardless which R3 I use. The Caddocks reveal more detail, a bit edgy when new which smooths out as time goes on.
A370 uses a different compensation, not just different values. I wonder, probably the limiting factor is the LM3886, but they work great in the voltage source operating mode.
A370 uses a different compensation, not just different values. I wonder, probably the limiting factor is the LM3886, but they work great in the voltage source operating mode.
I do wish we had someone doing some spice simulations. But it seems nobody is doing it. I guess it's just like going to the bathroom, if it needs to be done, best do it myself in private.
It's difficult to simulate the My_Ref without having complete models of both the LM318 and LM3886.
Unfortunately no one has developed such models until now...
Mauro did some sims in the initial phase of development, when compensation pins of the LM318 were not used.
LinuxGuru has managed to do a partial simulation using a LM1875 (if I remember correctly) model instead of the LM3886 and a customized LM318 model.
MyRef C Caddock R3
An externally hosted image should be here but it was not working when we last tested it.
MyRef Dario Caddock R3 old C32 value
An externally hosted image should be here but it was not working when we last tested it.
MyRef C seems to damp better in the lower mid frequencies,
Yup, that's correct. Here's the LTSpice simulation schematic with those models, and the preliminary THD20 FFT plots at ~15V amplitude output into 8R. The FFT plots for Rev C (green) and Rev E (blue) are shown.
As always, numerous caveats apply:
1) This is not the schematic of the original MyRef - this uses the LF01/LM318 discrete hybrid module, and some other minor modifications to keep LTSpice happy.
2) The models of both the LM318 and the LM1875 may be inaccurate.
3) The THD20 plots may be completely irrelevant to the audible sonics.
4) There may be various simulation artifacts that may not be reflected in the real MyRef behaviour, and vice versa.
With that disclaimer out of the way, my interpretation of the simulation results is as follows: Rev E has higher distortion (larger C34 contributes to higher distortion, while larger C32 contributes to lower distortion - but the net effect in Rev E is higher distortion) compared to Rev C, but it's almost entirely higher H2 and even harmonics. The odd harmonics quickly drop below that of Rev C. Moreover, the cascade of harmonics in Rev E is closer to the classical Class-A linear drop-off, which may contribute to better audible sonics. The net audible effect of Rev E is likely to be greater warmth, and smoother presentation, offset against higher distortion ('coloration').
Attachments
Last edited:
MyRef C Caddock R3 with LM3886 grounded at power GND
An externally hosted image should be here but it was not working when we last tested it.
Those are "reverse-driven" tests (both in simulations and real-world measurements).
Ground the input (better through a resistor of a value similar to the source output impedance) and drive the output. That is, add a voltage source in series with the load.
The results of this test tells you about the amplifier-load interface behavior, which is the most important thing behind Mauro's designs.
Last edited: