ANN: LF01 Discrete/Hybrid Opamp Module

[linuxguru]

This is an announcement of the LF01 discrete/hybrid module that uses a monolithic SOIC-8 single opamp, some BJTs and passives and plugs into a DIP-8 single-opamp socket. The intent is to bias the output stage of the opamp into Class-A, the discrete output stage into Class-A/AB and bypass the current limit of the monolithic opamp (more details will follow later).

Schematic, Layout and picture of a prototype with an LM318 installed - this is intended for the MyRef Rev C:

[linuxguru]

Preliminary BoM for the LF01/LM318M prototype (for use in the MyRef RevC):

R1: 1k/0.125w/5% CFR
R2: 10r/0.125w/5% MFR
R3: 4.7r/0.25w/5% CFR
R4: 33k/0.25w/1% MFR
R5: 150r/0.125w/5% CFR
C1: 10nF X7R or C0G ceramic plate-type
D1: 1n4148
Q1, Q3: 2sc1815Y
Q2: 2sa1015Y
Q4: 2sc1845
IC1: NatSemi LM318M

[madisonears]

A very interesting development.

Are any other mods to the amp required? Must your Ref C boards be used? Will this be stable with the standard PS regulation of the current Ref C?

Boards available when?

Peace,
Tom E

[linuxguru]

Originally Posted by madisonears

Are any other mods to the amp required? Must your Ref C boards be used? Will this be stable with the standard PS regulation of the current Ref C?

Boards available when?

Thanks - the first tests were actually done with a V1.2 board. There may be mechanical interference with R3 and C6. A Futaba MPC74 at R3 and low-profile electrolytics (say Black Gate PK) at C6/C11 solves the clearance issue on V1.2 Twisted Pear boards.

It should also fit and work fine with my V1.3/1.4 boards, but I haven't yet tried it out. It appears to be stable and sounds fine with the components listed, but some further tweaking of resistor values may be necessary for optimizing the sonics.

The bare boards are available now; kits and assembled boards may have to await some tweaking of the passives - say 2-4 weeks.

[PreSapian]

so it doesn't fit in 1.2 board if say silmics are installed at c6/11? how about running it through a wire off board like this?



p.s. sign me up for one regardless :thumbsup:

[PreSapian]

so what are the sonic benefits of this module? i bet everything's better but specifically, does it cure that last bit of hardness which they say is due to the op-amp?

[QSerraTico_Tico]

why not omit the opamp all together?

[AndrewT]

the opamp was chosen by the designer to complement the unusual topology of the MyRefC.
My RefC was then compensated with the chosen opamp in place.
Swapping opamps is discouraged because that will almost certainly require the compensation of MyRefC to be altered.

It appears Linux is using the characteristics of the same opamp and it's own compensation to avoid changing the compensation of the MyRefC.
Admirable criteria.

[linuxguru]

so what are the sonic benefits of this module? i bet everything's better but specifically, does it cure that last bit of hardness which they say is due to the op-amp?

I'm actually not sure if the slight residual edginess is due to the opamp, the chipamp, the topology, the quality of the passives or any/all of the above. I'm hoping that the LF01 reduces the contribution from the opamp output stage, with some tweaking on the LF01 alone if necessary.

The first impressions are that the imaging is tighter and the soundstage is wider, but I'll defer on this claim until some other experienced listeners have auditioned it on a wider range of tracks.

It appears Linux is using the characteristics of the same opamp and it's own compensation to avoid changing the compensation of the MyRefC.

Yup, exactly. It may still require a few components to be changed on some boards, but that will be due to clearance/physical size issues.

[linuxguru]

Originally Posted by QSerraTico_Tico

why not omit the opamp all together?

The discrete output stage of the LF01 is only a unity-gain buffer/voltage follower - you'll still need an input stage and a VAS. This could be made fully-discrete, but the module will be larger and may have no advantage over many high-performance opamps (LT1028, OPA627, OPA1611, etc.).

I sought to combine the advantages of high-performance monolithic opamps (low noise, high PSRR, high CMRR, low distortion, small physical footprint, relatively low quiescent current, etc.) with the advantages of a discrete output stage (Class-A operation, no current limit). The main intent was to drive low-impedance loads, including headphones, without clipping. It turns out that it also works pretty well for most opamp applications (except micro-power, precision offset, and low-voltage/rail-to-rail applications).