Where did you see the one that grounded to signal ground? My TP revC board grounds to power ground.
I haven't checked the earliest versions of Mauro's schematics, but IIRC the LM3886 ground has always been connected to the signal ground in Rev C (there is a good case for connecting it to the ground junction between C1 & C2, but AFAIK it hasn't been implemented on any Rev C PCB).
The Speaker return has always gone to the relay contacts and then to PGND in all versions of the schematic and PCBs, including mine.
As LinuxGuru well said all MyRef original versions and his versions have Lm3886 gnd tied to signal ground.
The Evolution Rev A and my My_Ref Fremen Edition have Lm3886 gnd tied to power ground.
Since I'm in the design phase I've yet totry it.
It's not an oversight but a precise design choice, also the first version of the Evolution has the LM3886 ground tied to signal ground.
I guess it makes sense if you consider the both amps as a whole.
I really don't know, where do you measured this DC?
I can check it on mine this evening.
There's very little current on the LM3886 ground - maybe a few mA. It's mostly a reference for the mute circuit and similar. I don't think it will contribute much to the DC offset. It could contribute to raising the noise floor a bit, by contaminating the signal ground, though.
There is one side benefit from having it connected to the signal ground (probably unintentional): For some kinds of user-related faults (i.e. interchange AC1 and PGND connections, etc.) sufficient current will flow through the LM3886 ground, through the 1-ohm R11 resistor to PGND, that R11 will fuse and blow open. After that, for some miraculous reason, the LM3886 has never blown in my experience.
In other words, there is a protective function in keeping the LM3886 ground connected to the signal ground.
I just measured at the RCA input connector.
I think I will study what effects this difference in ground connection of the LM3886 has.
At one time I tried to connect RCA ground connection to power ground instead of signal ground, there was large DC swing at the output and the speaker drivers would bottom.
Last edited:
I've measured mine...
On RCA connectors I've no DC offset (0,0mv) while I've some (12mV) after the coupling cap.
I suppose it has something to do with the polarization of LM318 Mauro wrote me about, maybe....
On the output I've 1.2mV DC offset.
Also for me, 100mV dc on the input of the lm318 - very very much off.
I have -3.7mV right on the input (after the coupling cap) on both channels..
And 7mV DC offset on the output, on both channels..
That much dc on the chip input throws off completely the linearity of the input long tailed pair - it is a completely different amp in this way. Sounded very bad to me, when I had committed a similar error in the past.
Many things can be the cause: RF oscillations are often the culprit for strange dc shifts.
Ciao, George
I have -3.7mV right on the input (after the coupling cap) on both channels..
And 7mV DC offset on the output, on both channels..
That much dc on the chip input throws off completely the linearity of the input long tailed pair - it is a completely different amp in this way. Sounded very bad to me, when I had committed a similar error in the past.
Many things can be the cause: RF oscillations are often the culprit for strange dc shifts.
Ciao, George
100mV is around the ballpark input offset voltage value for all BJT LTP input stages.
1uA of input bias current across 10k of Rin gives 10mV.
This would be unusually low.
5uA across 51k gives 255mV. This is unusually high.
expect anywhere between 10mV and 500mV at the -IN & +IN pins.
Much more importantly, the voltage across the -IN & +IN pins determines the output offset. That's why we try to match the resistances seen by these input PINs. That resistance matching lowers the output offset.
1uA of input bias current across 10k of Rin gives 10mV.
This would be unusually low.
5uA across 51k gives 255mV. This is unusually high.
expect anywhere between 10mV and 500mV at the -IN & +IN pins.
Much more importantly, the voltage across the -IN & +IN pins determines the output offset. That's why we try to match the resistances seen by these input PINs. That resistance matching lowers the output offset.
Hmm,measuring the Lm318 pins seem quite normal at around 12mV can't explain what is causing the measurement at the RCA yet, so I'm going to leave it like that. So the next step is the input cap and the cement resistor. These in my experience are not good. The resistor I will try to look for a 25W aluminum case type with thicker leads, since I have some 0.75Ohm on hand, I am temped to give them a try. The 1uF is going to be a tough decision since I really would like to keep them in the case I have now.
We settled on Caddock MP915 / MP930 0R5.
I wouldn't raise too much the value of this resistor since is critical and higher the value lower the output power...
I'm referring only to R3.
Every single person thay buyed from the Ultimate BOM MyRef GB received one of those Caddocks.
The first person to test them was Madisonears, I've made the comparison too (Xicon Cement, Mills MR5, Caddocks MP915, Fukushima/Futaba MPC) and other people settled with the same choice.
I like the nice small package. As a matter of fact, I'm going to put them into another amp based on the LM3886 to see how it changes the sound.
Running this baby in, it's starting to show better performance. The sound stage is becoming better and opening up. The timbre is sounding more natural.
100A/V gain is pretty big, I find myself turning the volume to very low levels, not something I particularly like for digital systems or analog systems. Really would like to be able to use the max output of players without having the amp clipping.
Running this baby in, it's starting to show better performance. The sound stage is becoming better and opening up. The timbre is sounding more natural.
100A/V gain is pretty big, I find myself turning the volume to very low levels, not something I particularly like for digital systems or analog systems. Really would like to be able to use the max output of players without having the amp clipping.
This might be what I would eventually do, but some additional tweaks and measurements will be necessary to further understand what is going on. Changing the gains might effect phase and damping, and I am almost sure that phase relationship will change with the loading characteristics, but I'd like to see just how sensitive it is with varying loads. Might need to clean of the workbench to spend a long time on this.
Last edited:
Got the Caddock MP930 resistors! Amazingly fast from Digikey placed order Sunday here, get delivery notice Monday midnight, get devices before Thursday Noon! I was beginning to get tired of the sound and switched back to my active speakers which are more clean and transparent.
The initial measurements start today.
The initial measurements start today.
Last edited: