Thank you, but the same component failing on both channels? I had sibilance on both channels and one is still working (and there is sibilance on it). I don't think cap swap was/is the problem, but I'm not sure how to be sure what is wrong, or shall i change the 3886 for a new one. All the resistors on board measured OK and they appear to be OK visually. Caps have no shorts (all checked with DMM). PSU voltages good.
Cheers
Cheers
The LM3886 is a very rugged chip - I have not had a single failure of an authentic NatSemi chip so far. A dry joint in the small-signal or Howland network is far more likely, though even that will not kill an LM318 that easily. Check the low-voltage regulator section carefully - zeners can fail open.
OK, up and running. Don't try to DC couple without 100K(R13) in place 
. Taking out R13 and DC coupling to a preamp (even though the preamp is having 1M shunt at the output preceded by mundorf supreme 2.2uF coupling cap, not leaking), created HUGE DC offset on one ch. (relay would not click in). Confusing part is that I matched every component by hand before soldering. Why one channel is showing huge DC offset without R 13 in place an the other is not?
1."Injecting signal to the c10 r47 junction without the 318 in the circuit" is a band aid approach and it works. I could clearly hear and see on the o.scope that 3886 is OK.
2. None of the 318 were dead. I made simple circuit to test them out of MyRefC (taken from national notes) but it is high BW circuit so wires and protoboard made it noisy and oscillating which led me to a wrong conclusion that parts are broken. Any of them play and o.scope OK once I put them in MyRefC.
3.Sibilance : I think I might need Zobel on the output.I changed several parts all around MyRefC, but sibilance stays, no matter dark or bright cap, carbon, film or comp., metal film etc. Tested with TI and National 318, no change, sibilance is there. Even enlarging C9 to 1000uF changed nothing ( in regard to sibilance). My speaker cables are 5m long (each), 4.5mm x 2.
Any ideas on the appropriate Zobel?.
. Taking out R13 and DC coupling to a preamp (even though the preamp is having 1M shunt at the output preceded by mundorf supreme 2.2uF coupling cap, not leaking), created HUGE DC offset on one ch. (relay would not click in). Confusing part is that I matched every component by hand before soldering. Why one channel is showing huge DC offset without R 13 in place an the other is not?1."Injecting signal to the c10 r47 junction without the 318 in the circuit" is a band aid approach and it works. I could clearly hear and see on the o.scope that 3886 is OK.
2. None of the 318 were dead. I made simple circuit to test them out of MyRefC (taken from national notes) but it is high BW circuit so wires and protoboard made it noisy and oscillating which led me to a wrong conclusion that parts are broken. Any of them play and o.scope OK once I put them in MyRefC.
3.Sibilance : I think I might need Zobel on the output.I changed several parts all around MyRefC, but sibilance stays, no matter dark or bright cap, carbon, film or comp., metal film etc. Tested with TI and National 318, no change, sibilance is there. Even enlarging C9 to 1000uF changed nothing ( in regard to sibilance). My speaker cables are 5m long (each), 4.5mm x 2.
Any ideas on the appropriate Zobel?.
Last edited:
I would go for a Thiele Network on the output rather than the stripped down version of a Zobel only.
National show the Full Thiele Network on the datasheet and yet many omit this from their builds. Some even omit the lonely Zobel. That is asking for trouble.
Sibilance can be a symptom of inadequate stability margins.
BTW, I have been listening to a sole mono channel of MyRefC for a couple of days now and I agree. There is an exaggeration of SSSS in the sound. I will need to look at following my own advice.
This omission may explain why the MyRefC and other implementations that do not have any of the Thiele Network components, are particularly susceptible to "sound" changes when swapping/adding/removing different types of components in the various locations around the two chips.
National show the Full Thiele Network on the datasheet and yet many omit this from their builds. Some even omit the lonely Zobel. That is asking for trouble.
Sibilance can be a symptom of inadequate stability margins.
BTW, I have been listening to a sole mono channel of MyRefC for a couple of days now and I agree. There is an exaggeration of SSSS in the sound. I will need to look at following my own advice.
This omission may explain why the MyRefC and other implementations that do not have any of the Thiele Network components, are particularly susceptible to "sound" changes when swapping/adding/removing different types of components in the various locations around the two chips.
Last edited:
Penasa discourage the use of Zobel networks with the My_Ref.
Can you post your BOM?
There are several positions that can exacerbate sibilance if not properly populated.
No guesswork... simply there are parts that perform better than others for certain tasks...
Resistors all PRP apart from:
R7 takman metal film
R1 generic carbon composite
R39 generic metal film
NO C21
C9 elna silmic II (220uF)
C1 and C2 nichicon muse bipolar (green can)
C6 C11 nichicon muse FW
Panasonic TSH for the 10kuF
Film caps wima and mica like in FE edition boom
BUT I changed several brand caps-resistors for C9 (panasonic, elna nichicon with values up to 1000uF and R 13 (carbon film and comp). The sibilance stays.
Sound changes, some combinations do sound better to my ears, but sibilance is there and it sounds like parametric eq is boosted with narrow Q around 4-6kHz. My tube buffer is out ruled as connecting the CD player directly to the poweramp produces the same results.
Thank you
R7 takman metal film
R1 generic carbon composite
R39 generic metal film
NO C21
C9 elna silmic II (220uF)
C1 and C2 nichicon muse bipolar (green can)
C6 C11 nichicon muse FW
Panasonic TSH for the 10kuF
Film caps wima and mica like in FE edition boom
BUT I changed several brand caps-resistors for C9 (panasonic, elna nichicon with values up to 1000uF and R 13 (carbon film and comp). The sibilance stays.
Sound changes, some combinations do sound better to my ears, but sibilance is there and it sounds like parametric eq is boosted with narrow Q around 4-6kHz. My tube buffer is out ruled as connecting the CD player directly to the poweramp produces the same results.
Thank you
Andrew, simply this is not what I did...
I didn't throw random components in the amp... I've selected them first (like for film/foil caps) and then evaluated subjectively their performance.
The same component I've selected to to test were tested also in different circuits to verify that their behavious was consistent.
My results are for sure subjective but consistent and repeatable, just try for yourself.
No guesswork... simply there are parts that perform better than others for certain tasks...
Absolutely, but silmic II is a dark cap, I use it in my equipment a lot, and never heard it being sibilant. I had R13,R12, R46 all carbon film (takman) or generic carbon comp in the circuit, but the same type of sibilance was there.
Everything else sounds great, apart from this high range that sounds over eq-ed.
On both ch.s and with every combination of parts I tried up yo now.
Last edited:
PRPs can be part of the problem, more later.
I would replace R7 with a plain KOA MF (or if you want to go boutique, Rikens)
TS-HA can be a bit bright but I don't think they're the problem.
For Wima you mean FKP2? Which voltage rating?
Rev C can easily sound harsh/sibilant if any element of the chain is not balanced.
I've had the same problem until I realized that my mod work on my Alphas crossover didn't taken in account the different ESR between film caps and bipolar elcos...
If you can rule out other parts of your chain I would suggest you to try FE's alternate compensation, sweeter than Rev C one.
"For Wima you mean FKP2? Which voltage rating?"
All Wima's and all film caps are like in you BOM . There are only two mica caps, again, like in your BOM . PRP i have in many builds and they can sound harsh but this amp is not sounding harsh, it is like I have EQ somewhere in the circuit boosting narrow, 4-6kHz range, so some singers have exaggerated SSSSSHHH. Upper highs have no problem very soft and smooth, same goes for all other freq. ranges.
I ordered most of your FE BOM already including Rikens, Susumus, Caddocks(R13)....
All other parts of my chain perform OK with different power amp (but MyRefC sounds better apart from sibilance thing).
" I would suggest you to try FE's alternate compensation, sweeter than Rev C one"
I was not aware that there is alternate compensation. I'll check it out.
Thank you.
All Wima's and all film caps are like in you BOM . There are only two mica caps, again, like in your BOM . PRP i have in many builds and they can sound harsh but this amp is not sounding harsh, it is like I have EQ somewhere in the circuit boosting narrow, 4-6kHz range, so some singers have exaggerated SSSSSHHH. Upper highs have no problem very soft and smooth, same goes for all other freq. ranges.
I ordered most of your FE BOM already including Rikens, Susumus, Caddocks(R13)....
All other parts of my chain perform OK with different power amp (but MyRefC sounds better apart from sibilance thing).
" I would suggest you to try FE's alternate compensation, sweeter than Rev C one"
I was not aware that there is alternate compensation. I'll check it out.
Thank you.
Last edited:
Fine, I've recently changed C7 with 100V FKP2, much more neutral and refined but phisically BIG, maybe you can give them a try.
Susumus? These are SMD parts... are you going to use them using adaptors?
Fine.
Exactly
You're welcome
Consider what I've PMd you.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.