Why are you choosing and recommending a time constant for the NFB loop that risks putting significant AC voltage across the electrolytic DC blocking capacitor?
I would recommend that if you insist on using 220uF for C9, that C13 be reduced below 600nF. 560nF or 470nF would do.
If you prefer to use a higher value for C13 then that requires one to select a higher value for C9 to avoid the risk of significant AC across the electrolytic.
For C13=1uF, C9 >=370uF, use 470uF
For C13=2u2F, C9>=800uF, use 1mF
In R13 I have tried MK132V and they're good, possibly MP9xx works even better but I didn't tried them...
Hi Andrew,
I'm not insisting on anything... C9=220uF, C13=1uF are the values specified by the designer, Penasa.
Maybe Penasa was wrong...but I don't have the knowledge to challenge Mauro's design consideration.
I have finished the second channel but this time had some problems. I've instaled in place of C9 two paraleled Nichicon KZ 100uf for testing but i had some dc on the output. After some solderind/desoldering and cheking i noticed the LM318 was dead and had 30v on the output.
After changed the Lm318 the channel started to work for seconds and triggered the DC protection.
After installed the Nichicon ES in place of C9 everithing seems to be ok, now it sings very nice without triggering the protection.
Any clue why this things happened ?
Tomorrow i need to go to buy some Lm318 to see finally how it sounds in stereo.
I already noticed it sounds warmish.
After changed the Lm318 the channel started to work for seconds and triggered the DC protection.
After installed the Nichicon ES in place of C9 everithing seems to be ok, now it sings very nice without triggering the protection.
Any clue why this things happened ?
Tomorrow i need to go to buy some Lm318 to see finally how it sounds in stereo.
I already noticed it sounds warmish.
I do recall that the original designer wanted input high pass time constant to be close to the high pass of the latter parts of the amp. If I recall correctly, 2Hz. I think if a 220uF cap non-electrolytic cap of small size could be found, we should use that.
Last edited:
I would rather keep my money in my pocket and sort out the input filters and NFB filtering effects before committing resources to a problem that disappears when designed for near zero voltage across the NFB DC blocker.
Sure, exotic components generally can be avoided. I'd spent more time on power supply since even general components can provide exotic results when used properly. The real difficult mindset is going back to the basic physics of these devices and figure out what really is going on in the circuit which SPICE models are not able to simulate. I actually have a normal chip amp on par with the MyRef, which I originally thought was not possible.
Electrolytics work great with DC across it. Without it, the bypass helps increase it's linearity, but still it is not best.
Electrolytics work great with DC across it. Without it, the bypass helps increase it's linearity, but still it is not best.
I've tested in position of C9 some oscons. One simple 270uf and two 470 back to back in series and definetly i can hear more details compared to Nichicon Muse ES that i have installed.
Also two 100uf Nichicon KZ paraleled sounded very nice but oscons have a lot of details. I hope this details to not become fatigueing.
So far i'm so excited with the sound of this project.
Mile grazie Mauro 🙂
Also two 100uf Nichicon KZ paraleled sounded very nice but oscons have a lot of details. I hope this details to not become fatigueing.
So far i'm so excited with the sound of this project.
Mile grazie Mauro 🙂
Last edited:
C7 seems to be just as critical for sonics as c13 is. any cap I put on there reveals itself in the music as if it's another input cap.
anything less than 100nf value hampers the dynamics and tonal balance.
mundorf supreme has something of a slowing down effect on music. It's hard to describe but I think it's the inductance that's causing it (siva you told me so!)
mundorf silver/gold brings the transients back to speed but it's bright sounding, as warned by the lettering 'silver'.
I have low reservation for russian caps because the three kinds i tried (teflon, teflon oil, pio) all fail my expectation in one way or another. they're all mid-fi solutions in my book...
the russian teflon though, while being unbearably hard and bright sounding, was incredibly transparent and lively.
I see vcaps on my doorstep in the near future... 😉
anything less than 100nf value hampers the dynamics and tonal balance.
mundorf supreme has something of a slowing down effect on music. It's hard to describe but I think it's the inductance that's causing it (siva you told me so!)
mundorf silver/gold brings the transients back to speed but it's bright sounding, as warned by the lettering 'silver'.
I have low reservation for russian caps because the three kinds i tried (teflon, teflon oil, pio) all fail my expectation in one way or another. they're all mid-fi solutions in my book...
the russian teflon though, while being unbearably hard and bright sounding, was incredibly transparent and lively.
I see vcaps on my doorstep in the near future... 😉
C9 is a critical position and finding the perfect cap is quite difficult.
Oscons sound quite fine but the best caps are audio grade ones (but not all)
The more balanced available cap so far is Nichicon FG.
The best choice, but out of production and costly, is Black Gate (FK better)
The circuit have an enormous potential, each step in parts quality gives two in performance improvement.
I agree C7 quality is always audible but C13 and C9 have far greater impact, IMHO.
Not in my experience.
Since quality is utterly important for C7 film/foil caps are simply the best you can use but in the 5mm size (critical for the bypass effectiveness) you can't go higher than 10nF.
Nevertheless I find a 10nF FKS2 the best choice with 10nF FKP2 a good second (less neutral).
Regarding C13 don't overlook the film/foils I'm suggesting (Mundorf Zn) or the possibly better Audyn True Copper, also film/foil.
dario, i already have mundorf sio for c13 😀
i'll be picking up a pair of them in .1uf for c7 today. they should cure the bloom issue i get with supremes. and sio doesn't sound bright at all, unlike silver/gold- though it may seem like it would be the other way around.
in fact, my amp sounds rather warmish right now. more so than symasym and possibly on par with a marantz 7200 i used to own, which's the the warmest sounding ss amp i ever had.
my justification for exotic spending is that $1000 is not that much for the best amplifier in the world, and that the myref very well could be that amp.
i'll be picking up a pair of them in .1uf for c7 today. they should cure the bloom issue i get with supremes. and sio doesn't sound bright at all, unlike silver/gold- though it may seem like it would be the other way around.
in fact, my amp sounds rather warmish right now. more so than symasym and possibly on par with a marantz 7200 i used to own, which's the the warmest sounding ss amp i ever had.
my justification for exotic spending is that $1000 is not that much for the best amplifier in the world, and that the myref very well could be that amp.
Last edited:
I doesn't mean nothing...
I'm not talking about price but about quality in a particular application.
SIOs are crossover caps for high level signals.
For low level signals (particularly for pulse ones) film/foil caps are vastly superior.
Film/foils (like Zn or Audyn True Copper, FKP2, FKS2, KP1830, PFR5, Orange Drop 716P, etc) caps are made with foils of metal (tin, alluminum, copper, etc) interleaved with dielectric foils (paper, various type of plastic, etc).
Metallized caps (like SIOs) are are formed only by the dielectric on which the tmetal is vacuum vaporized forming a thin conductive layer.
The foil is thicker (higher current capacity, lower resistance, lower noise), heavier (less microphonics) and metal crystals are way longer (think about OCC).
For low level signals such characteristics are higly desirable.
Using a phisically big cap for decoupling is non-sense.
You use a decoupling cap for a precise goal, offering a low impedance PS and filter noise.
Incidentally you also hear decoupling caps sound signature.
If you select the decoupling cap only as a function of its sound signature you're missing completely the point.
A big cap with long leads have (much) bigger inductance which nullifies its effectiveness as a decoupling cap...
It may even sound better but if that cap is necessary for stability you risk oscillation (and lesser real performance).
thanks for the informative outline, dario.
im gonna have to go with the sio caps, however 😛
i did develop my preference through an extensive cap swapping, after all. big axial caps may seem like they add to the sound but going back to wimas does feel like much more is taken away.
no cap is more transparent than leaving c7 empty (dynamics suffer too much tho) but to me the mundorf doesn't intrude itself any more than a through hole cap.
as for zn, id love to try it if i could borrow it 😉
im gonna have to go with the sio caps, however 😛
i did develop my preference through an extensive cap swapping, after all. big axial caps may seem like they add to the sound but going back to wimas does feel like much more is taken away.
no cap is more transparent than leaving c7 empty (dynamics suffer too much tho) but to me the mundorf doesn't intrude itself any more than a through hole cap.
as for zn, id love to try it if i could borrow it 😉
You're welcome
It's a free world...
I also rely on listening tests and I've tried a lot of caps but primary goal of parts can't be overlooked, IMHO.
Otherwise we're no more talking about electronics but woodoo... 😉
I suggest you to try Audyn True Copper (the hint come from Madisonear)...
They're rated, as crossover caps, by Tony Gee of HumbleHomeMade as good as Mundorf SIO and they're film/foils.
I suspect you'll never come back after testing them. 😀
Obviously I'm talking about C13...
A brief break from our regular programming, and a sneak peek at one of the projects I've been dabbling with for the last 3+ months:
Dual Howland (2x LM3886) Evo2 (derived from MyRef Rev C/E, 40W into 8 ohms, ~80W into 4 ohms) monoblock board, dimensions 127mm x 100mm (~30% larger than present V1.4).
The PCB is a fresh, ab-initio layout, and has not been validated yet for correctness or audible sonics. Nevertheless, I'm hoping that it's at least as good as the V1.4, due to improved shielding of high-current paths and/or shorter current loops at some locations, as well as better separation of small-signal and large signal sections. The input connector block is more convenient, and well-isolated from high-current or noisy sections.
The board will support most existing single-LM3886 MyRef Rev C/E BoMs also, just by omitting all components with a 3-digit 1xx designation and a few minor component changes. The LF01 will also be usable. Physically larger C13 will also be easier to fit (with some overhang above other components on the board).
Six prototype PCBs are in process today, and due in about 2 weeks. I hope to have some thing to report by mid- to end-May. If there are errors or corrections to be made, I'll use that opportunity to re-do the layout to include improved regulators and a few other mods - no predictable time-frame for completion at this moment.
Dual Howland (2x LM3886) Evo2 (derived from MyRef Rev C/E, 40W into 8 ohms, ~80W into 4 ohms) monoblock board, dimensions 127mm x 100mm (~30% larger than present V1.4).
The PCB is a fresh, ab-initio layout, and has not been validated yet for correctness or audible sonics. Nevertheless, I'm hoping that it's at least as good as the V1.4, due to improved shielding of high-current paths and/or shorter current loops at some locations, as well as better separation of small-signal and large signal sections. The input connector block is more convenient, and well-isolated from high-current or noisy sections.
The board will support most existing single-LM3886 MyRef Rev C/E BoMs also, just by omitting all components with a 3-digit 1xx designation and a few minor component changes. The LF01 will also be usable. Physically larger C13 will also be easier to fit (with some overhang above other components on the board).
Six prototype PCBs are in process today, and due in about 2 weeks. I hope to have some thing to report by mid- to end-May. If there are errors or corrections to be made, I'll use that opportunity to re-do the layout to include improved regulators and a few other mods - no predictable time-frame for completion at this moment.
Attachments
excellent! I'm definitely putting them on my to-buy list 😀
Siva, Please sign me up for a pair at the earliest opportunity.
Last edited:
Wow Siva! 🙂
Nice layout, it will be great for multiamp design. 😎
Didn't you found the space for the transistor shunt reg or it's a design choice?
How do you came to the decision of using only two more distant 330uF elcos?
I must correct myself.. Tony Gee rates them as good as SGO (which are better than SIO) 😉