If i remember correctly (im not at home now) its c32. Did i do something wrong?
C32 is 150pF with both Rev C compensation and my alternative one.
An old version of the alternate compensation used 330pF.
I suggest you to revert to 150pF
Exactly, red ones are NOS.
Red ones sound slightly fuller and a bit less harsh.
Nichicon FGs are too great caps for C6/C11, more balanced Vs Cerafines, IMHO.
At first i used an 160pf ERO KP1834 , but when i tried a wima FKC 330pf i was very impressed by the depth and width of the soundstage.
I also replaced R1,R4 with LM317/LM337+resistor. It is a fantastic tweak Dario. Thanks for sharing
I know, that value was selected for that reason but after carefull listening and thanks to Soongsc analysis I've determined that sound with 150pF is more correct.
BTW the KP1834 is a more refined (and performing) cap than the FKC2.
You're welcome
It improves things quite a lot, particularly if you use Caddock MP915 resistors.
Last edited:
C0G/NP0 will work OK at C10, C30 and C34. However, most people report better sonics with Silver Mica at C10, C34 and FKP2 at almost any location. As Dario stated, KP1834/1835 are a viable option for C32 (check the voltage rating - 63V is OK, 100 or 200V is better). My Rev E compensation values are 330pF for C32, 100pF for C34 and 27pF for C10 - mainly because I like the immersive soundstage and palpable/percussive mid-bass that it provides, even at the cost of higher THD20.
I'll try Dario's present recommendation of 150pF, 27pF and 33pF in a future build - it will probably have lower THD20, but may not be as immersive.
It's the CCS mod for the MyRef. You can replace R1 and R4 - each with an LM317 + ~56..68R resistor Constant-Current Source. It improves the PSRR for the LM318 section, by reducing the ripple on the LM3886 rails feeding through to the LM318 rails.
PreSapin, IMHO - the values from D were a great blend of all the sonic qualities mentioned in the last few posts. I described it as "tube like" balancing of sound and stage.
linuxguru, I had planned to ask you if any of those changes would impact negatively on your Ultimate BOM and/or the LFO1. It sounds like your saying there won't be a conflict.
Also, are there pictures, diagrams or posts that I missed clarifying the LM317 CSS mod? I may well have not caught the significance if there was an earlier discussion. Are you incorporating that into your new design?
linuxguru, I had planned to ask you if any of those changes would impact negatively on your Ultimate BOM and/or the LFO1. It sounds like your saying there won't be a conflict.
Also, are there pictures, diagrams or posts that I missed clarifying the LM317 CSS mod? I may well have not caught the significance if there was an earlier discussion. Are you incorporating that into your new design?
Last edited:
The broad gist is this:
C10 - feed-forward element: larger values reduce distortion, but also contribute to instability, which may show up as edginess/brightness. Can be omitted in some compensation setups - practical values range from 0 to 33 pF.
C34 - dominant pole for LM318: larger values increase stability, but also reduce the loop gain and hence increase THD20. At least 10pF is needed for stability in just about any compensation setup, except Rev A. If C10 is increased, C34 generally also has to be increased (disproportionately) to maintain stability.
C32 - GNFB integrator element: Larger values tend to flatten out the phase response of the loop gain with frequency. Audibly, this seems to make it sound aggressive, forward, percussive and immersive. Either too little (<100 pF) or too much (>470 pF) can make the GNFB loop unstable. Practical values range from 150pF to 330pF for stability as well as sonics.
I haven't evaluated all the combinations of compensation changes along with the LF01 and/or regulator mods, so it's too early to say. Broadly, the LF01 uses the existing compensation on the main MyRef board, so it should not affect the stability, timbre or presentation significantly. Its impact is mainly to make the audible sonics smoother, less fatiguing, more airy - i.e. making the LM318 sound more Class-A. With the Rev C compensation, it's mainly smoother icing on the cake; but with the rougher and more aggressive Rev E, the effect of the LF01 is more audible.
The LM317 CCS mod is Dario's baby - I've only played with it in simulation. It's not yet in the MyRef_X2 prototype layout, though it can be kludged in as a mod during assembly. I'll incorporate an improved version of it (after validation in a prototype build) in a future revision of the MyRef_X2 layout.
thanks again, siva. i can only understand half of that atm. but i'll wrap my head around it eventually
and dario, big thanks to you for your mod. I just got done trying out the full package and it's no small upgrade. the improvements are significant! most noticeable is soundstage depth but there's also more air, speed, bass, dynamics, everything. everything's better!
I can't say the same about the lm317 ccs, however. the bass firms up but at the expense of harmonics. it was sounding too much like ss so i had it taken out. and as i've noticed before, it adds a bit of grain structure to the sound. but the good news for me is that major portion of the improvements come from the shunt regulator mod.
i used fairchild+caddocks for ccs, btw.
also i notice much louder transformer noise from one channel after the mod. i can even hear the cycle noise through the midwoofer. bad bc6xx perhaps?
and dario, big thanks to you for your mod. I just got done trying out the full package and it's no small upgrade. the improvements are significant! most noticeable is soundstage depth but there's also more air, speed, bass, dynamics, everything. everything's better!
I can't say the same about the lm317 ccs, however. the bass firms up but at the expense of harmonics. it was sounding too much like ss so i had it taken out. and as i've noticed before, it adds a bit of grain structure to the sound. but the good news for me is that major portion of the improvements come from the shunt regulator mod.
i used fairchild+caddocks for ccs, btw.
also i notice much louder transformer noise from one channel after the mod. i can even hear the cycle noise through the midwoofer. bad bc6xx perhaps?
Last edited:
You're welcome
You're tasting part of the Fremen Edition performance, why not partecipate to the Release Candidate GB and get the full thing?
Fairchild LM3x7?
For optimal performance you should try OnSemi LM3x7 (they sound more full and rich)
Last edited:
thanks again, siva. i can only understand half of that atm. but i'll wrap my head around it eventually
and dario, big thanks to you for your mod. I just got done trying out the full package and it's no small upgrade. the improvements are significant! most noticeable is soundstage depth but there's also more air, speed, bass, dynamics, everything. everything's better!
To wich mod are you referring Presapian ?
I supposed the same but this sentence was saying something else: I can't say the same about the lm317 ccs, however. the bass firms up but at the expense of harmonics. it was sounding too much like ss so i had it taken out. and as i've noticed before, it adds a bit of grain structure to the sound.
Good news anyway because i was planing to build a better PS for LM318 but if it's so sensitive to the components i must wait until i find some Caddoks for example.
Good news anyway
because i was planing to build a better PS for LM318 but if it's so sensitive to the components i must wait until i find some Caddoks for example.
Last edited:
Yea, Me too - which one? More detail and or pictures would help clear some confusion that's floating around.
BTW, I hope I'm not being preemptive but if your not watching the XY thread, Here is a fun and useful project from Andrew T. http://www.diyaudio.com/forums/chip-amps/207971-bought-xy-lm3886-kit-30.html#post3018905
I've got seven LM3886 chips that I don't know if they are good or bad. A BrianGT PCB , or other simple board and some parts from my spares bin will make a good test bed.
BTW, I hope I'm not being preemptive but if your not watching the XY thread, Here is a fun and useful project from Andrew T. http://www.diyaudio.com/forums/chip-amps/207971-bought-xy-lm3886-kit-30.html#post3018905
I've got seven LM3886 chips that I don't know if they are good or bad. A BrianGT PCB , or other simple board and some parts from my spares bin will make a good test bed.
Last edited:
hi atupi, i used the full network that clave described here. http://www.diyaudio.com/forums/chip-amps/54571-my-audiophile-lm3886-approach-373.html#post2968535
as you can see, both zeners and resistors are replaced with bc639+zener+resistor and lm317+resistor respectably. for me, the bc639 bundle stays but the lm317 bundle comes out, leaving room for future iterations that hopefully uses mosfets?
Clave it seems that the ones i used were on semi LM after all.
and I'm honored by your invitation. I've been meaning to ask you, though- can siva's lf01 modules be adapted into your boards? i know it's smd 318 but if the tracks are adequate perhaps i can punch holes, scrape the copper... do what i always do- hack it up
as you can see, both zeners and resistors are replaced with bc639+zener+resistor and lm317+resistor respectably. for me, the bc639 bundle stays but the lm317 bundle comes out, leaving room for future iterations that hopefully uses mosfets?
Clave it seems that the ones i used were on semi LM after all.
You're tasting part of the Fremen Edition performance, why not partecipate to the Release Candidate GB and get the full thing?
and I'm honored by your invitation. I've been meaning to ask you, though- can siva's lf01 modules be adapted into your boards? i know it's smd 318 but if the tracks are adequate perhaps i can punch holes, scrape the copper... do what i always do- hack it up
Last edited: