My "audiophile" LM3886 approach

Feedback cap

I listened to JosephK to long, and went with a low voltage model. He is using a 180 ufd 6.3 volt Oscon.
I did not have one, so used a 220 ufd 6.3 BG N series. So far so good. The measured voltage is 0.1 volts idling. Have no idea what it is on power up/down.
Must have asked him a half dozen times about longevity of the caps in his amp first. He has several months on his Oscons.
Must be either bravery or stupidity, but the caps are holding up fine so far in my amp. Hope the relay saves my drivers if the BG blows.
If you really are brave/stupid try the BG's there. They are the best quality caps that will fit on the board in a 200 ufd or so value.


George
 
I think Mauro put alot of thought and testing into the values of the components, so all thats left is to find a balance of components that works together pleasantly... actualy looking at Russes new amp boards reminded me I got some Wimas in the mean time that could replace some of the ceramics on my boards.

I have a suspicion that changeing the electrolytics decoupling and driving the amp chips (where they 220uf?) to audio grade caps may also be a solid idea.

Unfortunatly this amp doesn't lend itself to working well with inferior parts.
 
The Rubycon ZA 35V, 220 ufd works well as C9. It is an ultra low impedance cap, having an impedance of 0.026 ohms max. at 20C and 100kHz. It is on the expensive side at AU $13.00 (US $9.75) each, from Farnell in One.

I do not know if it needs bypassing, but nevertheless I have it bypassed with a Vishay Roederstein 0.1 ufd, 160V MKP cap, at AU $4.76 (US $ 3.90) each, also from Farnell. I mounted it on the bottom of the board, as it was too large to fit on the component side.

Audie.
 
He is using a 180 ufd 6.3 volt Oscon.
Can value be changed for C9??

By chance, I'm waiting for 220uF/10V Oscon's :)
ERS about 0.03, I think.
Please inform us if they kaput :xfingers:

The Rubycon ZA 35V, 220 ufd works well as C9. It is an ultra low impedance cap, having an impedance of 0.026 ohms max. at 20C and 100kHz.

I saw Rubicon ZLH 220uF/50V at Farnell, for GBP 0.56 (if I recall well). It has "only" 0.042 ohms ERS but life espectancy is 10000hrs at 105º!!!

Regards
M
 
Value shoudl stay at 220uF. 10V should be fine for the feedback, though 25V was the original recommendation (I think, hard to remember).

Remember to that Mauro spent a lot of time designing this amp to not rely too heavily on any particular component, so that inexpensive parts would produce an extraordinary result. I would definitely try it first with the "stock cap" for comparison, or one channel each way. I would also not get too caught up in tweaking before you spend some time listening... may decide you don;t need to do anything. ;)
 
More feedback cap options

I have built four different My_References. Used a different feedback cap in all of them. Most likely have screwed around with these caps much more than necessary.
1. 220 ufd 25 v ZL
2. 220 ufd 25 v ZL, Then a 0.33 ufd bypass, then someone said a 470 would deliver a better lowend, so a 470 ufd FM, bypassed.
3. 220 ufd 25 v ZL, with a 0.33 ufd Siemens stacked film and foil.
4. Mono, ordered a handfull OF 270 ufd, 25 v ZLH. George (JosephK) talked me into trying the Oscons instead. Since I was not sure about using them as a coupling cap, went with the 220 ufd N, by passed with a 47 ufd NHX, 6.3 volt BG.

From my experimenting, there are benefits to be gained by using a low impedance electro, or a standard and a film bypass. Using a larger value does not SEEM to make any difference. My speakers are only flat to about 45 Hz so it may be more evident with speakers that go down another octave.
I am sure the caps in the kit are plenty good enough. It is a real case of diminishing returns. Whether it is a 50 cent cap or a 15.00 dollar one.
Now time to put on my asbestos suit. The Rubycon ZA,ZL, and ZLH series caps all have steel leads. Components with steel leads have no business in the signal path. Input or feedback. Even the LM3886 has tinned copper leads. So does the AD815AY.
The Rubycons are my faves for decoupling though. Steels leads or not. I suspect it is environmental reasons why these have steel leads. Meeting the European standards is causing changes in all components construction.


George
 
Re: More feedback cap options

The Rubycon ZA,ZL, and ZLH series caps all have steel leads.
Tinned copper clad soft steel. All modern electros has new electrolyte formulation, and no way to use copper leads due to internal galvanic/corrosion problem, afaik.
A lot of film caps also builded with steel leads, to prevent internal overheating when soldered. They much more reliable.

Components with steel leads have no business in the signal path.
Well, each tube have a pins maked from covar (Fe+Cu). Almost steel. Same history for all semiconductors in metal-glass packages (like TO-3).

Even the LM3886 has tinned copper leads. So does the AD815AY.
That's right to all IC in plastic/ceramic packages or epoxy sealed metal can packages.

Back to electros.
I want to say, that lead material is not the main reason for good or poor sounding. It can be explained at next way: newest cap has better perfomance in terms ESR, leakage, ripple current, size bla-bla-bla, i.e. closer to ideal capasitor. It's a complex function of the many things - formulation, construction, materials, etc. On second side, it's distortion signature can be better or worse, with no any warranty from manufacturer, it's never measured and putted to spec.

Personally I preffer BG in audio circuits, also happy with FC/FM when I need big values or for cost reason. Not played with ZA/ZL Not impressed from some audio caps from Elna. Don't want to search "best cap in the universe" anymore.
In my version of My_Ref Rev.C (original layout was slightly changed to fit my stock components) I use 220u FM and amp sound great Tried same value BG Standard, 100u BG-F - sound is very similar, and I'm too lazy to search smallest differencies (if yes).
 
Thanks Vidalgo for an informative post. Personally, I think the question about whether partly-steel component leads are detremental to good sound is a non-issue, if the component sounds good.

Mauro originally specified >35V for C9.

Using a Variac to power up/down when using 6.3V - 10V for C9 would protect them from switch on/off surges.

Audie.
 
RevC board dimensions posted

Here is something that needs to be in the public domain: the board dimensions for latest RevC PCB, including height. The tallest components are the two 10,000 uF capacitors. Their data sheet says they are 50 mm tall, but this is incorrect, the capacitors are more like 52 mm tall.

The acutal board measures 83 mm wide by 104 mm long. Adding in the capacitors, without the supplied 1/2" standoffs, the height is 54 mm. See my attached photo.

For people choosing or manufacturing cases, every mm counts.:)
 

Attachments

  • revc.jpg
    revc.jpg
    59.8 KB · Views: 1,133
Thanks Dcibel, I am too tall ignoring the standoffs ie the board & caps make me about 2mm too high, I was going to space the board from the case by as little as 2-3mm with washers. the internal height of my case is 51 - 52mm..... depending on how much case flex i find acceptable. ;)
 
D1GGY said:
Ohh..... Bother or words to that effect, that makes it JUST (by about 2mm :( )to tall to fit my case, am i able to lay the capacitors on their side?

Why not try shorter standoffs? Standoffs come in various lengths. I am going to use 0.25" tall standoffs instead of the supplied 0.5" standoffs to get my board to fit in the case. You could probably mount the capacitors on their sides too, but shorter standoffs seem like a better solution if that is an option.
 
D1GGY said:
Thanks Dcibel, I am too tall ignoring the standoffs ie the board & caps make me about 2mm too high, I was going to space the board from the case by as little as 2-3mm with washers. the internal height of my case is 51 - 52mm..... depending on how much case flex i find acceptable. ;)


Thanks for your reply,
I am not looking to use the standoffs included in the kit. I appreciate that it will be shorter I use shorter standoffs.

Does anyone know if i can get away with mounting the capacitors on their sides?

My attention is however turning too the ATi cases which look yummy!!!! and will be the best solution if a little pricey.......
 
They are Snap-In caps, so you don't have any lead length to play with. There is also not a lot of room around them on the boards.

It's possible to mount them on their sides, but it will be tricky - probably need to use wires to do it, and mount them off-board. I personally wouldn't recommend it.

Shorter standoffs or a different case would be a better solution.