The new "My Ref" Rev C thread

[ra7]

Apart from the heatsinks, there is little difference in cost between the F5 and the MyRef. In fact, the F5 has fewer components than the MyRef. Yes, it is probably more complex to setup. But after setup, it does not require any more tinkering. It can deliver up to 54W in class AB. Also, low voltage gain is a desirable property in a power amp in terms of noise.

This is no disrespect to the MyRef though. Mauro's design is unique and of course sounds wonderful in its own way.

[udailey]

25W

[ra7]

umm... no, the F5 does 25W class A and then upto 54W class AB. As far as I know, the MyRef does about 40W. Anyway, I don't mean to start a war of words. This amp has some strong following. The F5 has its own set of followers and a thread that is 10,000+ strong and still going. Just wanted to say that it is worth finding out for yourself what the difference is between this amp and the F5. PCBs and transistor kits are readily available.

[AndrewT]

Originally Posted by ra7

umm... no, the F5 does 25W class A and then upto 54W class AB. As far as I know, the MyRef does about 40W. Anyway,

you forgot to specify the load impedance/resistance.

The F5 is a 25W into 8r0 ClassA amplifier.
It is also a 50W into 4r0 amplifier, but only the first 12.5W is in ClassA at this reduced load resistance.

The MyRef is 50W into 8r0 and also 50W into 4r0. But the PSU must be built/adjusted to suit the two different load resistances to get that maximum power capability for each.

[benb]

With mention of the F5, I'm wondering about another possible "improvement" on the LM3886, forced Class A. I've seen Walt Jung do it with op-amps at the line level, but I don't recall such a mod on a power amp. It seems a 100mA current source or sink connected to the chip's output would make the "first watt" Class A, and not create too much extra heat.

[AndrewT]

Potentially the forced single ended ClassA operation of the output stage could/would make for a good ClassA amplifier of low output.
But one cannot release that potential because the front end circuit to which we have no access falls woefully short of the performance standards of what we would expect of a low power ClassA amplifier.
The ClassA version would certainly not make a good headphone amplifier.

[soongsc]

It sounded better than the class A I compared against. More detail, better image, similar depth, better bass control. Theoretically, the way it operates will deliver performance just as good as a class A.

[Milos P.]

Originally Posted by AndrewT

I use a formula that I found in this Forum which relates the input RC to the NFB RC to specifically limit the AC voltage across the NFB cap. The reason why we might want to do this is to minimise the distortion that the electrolytic will pass to the -IN amplifier terminal. Self and Cordell and many others tell us that the NFB cap will distort the amplifier output if it is made too small, i.e. develops significant AC voltage across it.
The formula that gives some meaning to Self and Cordell is

NFB cap >= sqrt(2) * Input cap * Rin / Rfeedback.
For the MyRefC 1uF would require >=362uF for Nfb cap
1u5F requires >=540uF and 2u2F requires >=725uF.
MyRefC has 220uF installed.

That's the theory behind the RC ratio and why .

I have been using input RC~90ms and NFB RC>=140ms for some years now and I have found it to sound good to my ears. It's nice when theory explains why it sounds good. It's even nicer when Theory allows one to predict the component values that might be close to optimum for good sound.

Now back to the input filters.
If you want the 0u68F to give extended bass response then increase Rin to 130k. RC = 88ms.
NFB cap >= 1.414 * 0.68 *100k / 390 >=247uF 220uF is probably close enough to satisfy the no AC voltage criteria for low distortion.
If you want to be sure that the NFB is not creating avoidable distortion then try increasing it to 330uF, or maybe even as high as 470uF
As an aside, I intend trying NFBcap = 220u//220u as an easy to carry out experiment on the effect it has on sound.

Very interesting, If I understood well C9 should be increased to 725uF (or more) for c13=2.2uF and R13=100k?