Green Pre capacitors - signal path

[DaxLiniere]

Greetings all!
I've finished 2 channels of Green pre, one with ordinary panasonic electros in the signal path and one with Elna Stargets.

WOW man, what a difference!

I do have a little question though - What would be the audible effect of using 100uF caps instead of 47uF caps in the signal path?

Also, are the 63v ratings on those electros completely necessary? (phantom power related?)

Thanks all!

 

[clintrubber]

[quote author="DaxLiniere"]Greetings all!
I've finished 2 channels of Green pre, one with ordinary panasonic electros in the signal path and one with Elna Stargets.

WOW man, what a difference![/quote]
That's interesting (no pun) ! Could you please describe the difference in sound using at least five terms from the list below ?

I do have a little question though - What would be the audible effect of using 100uF caps instead of 47uF caps in the signal path?

Probably nothing, but if the circuit originally skimped on uF's then there might be some less phase shift in the lows. I won't say anything on how noticable this would be.

At the other side of the spectrum: if you'd be using even bigger values & depending on types etc you might start to notice less highs because of the big 'lytic, but I guess it won't be happening here yet.

Also, are the 63v ratings on those electros completely necessary? (phantom power related?)

Do the math ! You might be fine with 50V types or less, but why risk it ?


Bye,

Peter

 

[DaxLiniere]

"Five terms from the list below" ?? Is this a joke? I'll give you a list! *Incessantly shakes hands in the air*

Here's the email I sent to PeterC:

An interesting little side note; I had some Elna Starget electros sitting around and since I was short on parts, I decided to build one channel with those and the other channel with the Panasonic caps. The difference is absolutely amazing. On my first couple listens, the Starget channel has higher frequency response, much cleaner transient response, lower distortion/more headroom and an audibly lower noise-floor. I will do some objective tests on this soon.

I'm certain this sonic difference is solely due to the caps. When I was building, I tested EVERY resistor and was able to get exact values (or at least identical sets) for every position on both channels.

My question to you; the Stargets I had lying around were 100uF, not 47uF. (the 22uFs before the ferrites are the correct values).
Is the better performance due to the higher quality or the higher capacitance value?

Also, the only 100uF Stargets are 35v not 63v. I assumed this was related to the phantom power, so I haven't used phantom on the Starget channel yet.

After writing that, I considered that the appearance of more highs could be an auditory trick of a LACK OF LOWS. To test my theory, I did some frequency plots.

http://www.puzzlefactory.com.au/Green+Stargets.gif

But as you can see, this theory was (thankfully) incorrect.
Overall, the Starget's frequency response is a lot smoother.
Understand that this is not a scientific test using the same sample, it's me singing the part twice. (But it's a peak graph over 30 seconds)

I'm a little reluctant to post samples because they're of ME singing.

Incidentally, I am very aware that 50 is close to 48. :wink:
The reason I asked the question is because I'm wondering how the 48v is "getting so far into the circuit." Shouldn't the first set of caps decouple the whole circuit? I'll be the first to say I'm no electronics mastermind.

After researching today, seems that Stargets are discontinued. (No mention anywhere on Elna's site) They seem to have been replaced with Cerafine, SILMIC & SILMIC II. I am yet to obtain some of these.
WES Components has them, but they're far too expensive (au$3.95ea!)

The other problem is the limited range of available values.
http://www.audio-cube.nl/elnaform.htm

 

[clintrubber]

[quote author="DaxLiniere"]"Five terms from the list below" ?? Is this a joke? I'll give you a list! *Incessantly shakes hands in the air*[/quote]
I was & am seriously interested in the differences in sound - the list was just a lame joke.

Incidentally, I am very aware that 50 is close to 48. :wink:
The reason I asked the question is because I'm wondering how the 48v is "getting so far into the circuit." Shouldn't the first set of caps decouple the whole circuit? I'll be the first to say I'm no electronics mastermind.

Wasn't joking, the 50V will be OK but if 63V or 100V parts are available & space permitting, these would be preferred. But 50 V should be fine of course, just somewhat less margin. In addition, during typical phantom-connected & used operation the full 48V doesn't develop on those caps, but because of the non typical moments that come by...
Yes, like you said, the first phantom-blocking caps do the total job. So if the rest of the circuit would be powered from say +/-1.5V (one-dot-five)(just to illustrate the point), then 4V 'lytics for all others would be fine. And even less than 4V for signal-coupling caps since full supply won't... etc

After researching today, seems that Stargets are discontinued. (No mention anywhere on Elna's site) They seem to have been replaced with Cerafine, SILMIC & SILMIC II. I am yet to obtain some of these.
WES Components has them, but they're far too expensive (au$3.95ea!)

I've always bought them from RS. Are they around at your side ?

The other problem is the limited range of available values.
http://www.audio-cube.nl/elnaform.htm

Recognize that from RS...

Bye,

Peter

 

[DaxLiniere]

the first phantom-blocking caps do the total job.

Hmmm... Why didn't PeterC stipulate 63v ratings for almost all signal path caps? Even the 22uF caps preceding them have 63v ratings - and they're after the zeners!
Maybe a capacitor with a high working voltage rating has inherently lower ESR or distortion? maybe? maybe?

Also, what would happen if you tried for higher gain, using a 0 ohm link in the gain stage? Would it oscillate or something?

 

[peterc]

I wouldnt go for less than 63v for the Phantom blocking caps. I'm sure that there will be a spike or 2 when switching on the phantom.

The rest of the caps can be a lower voltage, minimum of 25v for the decoupling electrolytics, the signal caps can be 16v .

There should be NO DC after the Phantom blocking caps, if there is, they are leaky.

Peter

 

[clintrubber]

[quote author="DaxLiniere"]

the first phantom-blocking caps do the total job.

Hmmm... Why didn't PeterC stipulate 63v ratings for almost all signal path caps? Even the 22uF caps preceding them have 63v ratings - and they're after the zeners![/quote]
Sorry, I don't understand. The Zeners are (IIRIC) not ground referenced, but I may be wrong. During correct circuit operation the Zeners should be 'absent'. And with Phantom (& no input-TX)you defin. need a cap with say a 50V rating.
Could type more but I better look at the schematic first.
But about those Volt-ratings: the story is simply that the first caps should be able to withstand the phantom, and the rest not. As said, don't have the schematic here, so I don't recall whether the first caps are the 22u or 47u or...

Maybe a capacitor with a high working voltage rating has inherently lower ESR or distortion? maybe? maybe?

Yes, at least for those I saw this was indeed the case.

Also, what would happen if you tried for higher gain, using a 0 ohm link in the gain stage? Would it oscillate or something?

Would be limited by the ESR of the biiig cap, and I don't know how linear that resistance is. Apart from that, don't know how much the stability margin is. But the smallest resistor is already very small, so I doubt you get much more gain since the invisible resistances will join more & more.

For more gain perhaps modify some values later on in the circuit, but this was just wondering, right ? As in you don't need more gain but was just curious ?

 

[DaxLiniere]

Yeah, we have RS here.
They're only got the Stargets, no mention of the new Elna audio products.

I sent PeterC a link when I emailed him, but noticed today it's a bad link.

Here's the list of available Starget caps:
16V - 470uF, 1000uF, 2200uF
35v - 10, 22, 47, 100, 470uF
50v - 4.7, 10, 22, 100uF
That is all.

I guess I'll have to go to the expensive place to get the 2 63v caps/ch

 

[clintrubber]

I just remember my two channels of Green are also Starget all over, except for the first caps (indeed because of the voltage rating)

 

[DaxLiniere]

but this was just wondering, right ? As in you don't need more gain but was just curious ?

Yeah, exactly. I mean, if it were as simple as adding a 0ohm setting for more gain, I'd do it for my ribbon mics.

But it's not necessary, might change my BC550s for BC550BPs though - no idea what gain range on the current transistors is.

 

[clintrubber]

[quote author="DaxLiniere"]

but this was just wondering, right ? As in you don't need more gain but was just curious ?

Yeah, exactly. I mean, if it were as simple as adding a 0ohm setting for more gain, I'd do it for my ribbon mics.

But it's not necessary, might change my BC550s for BC550BPs though - no idea what gain range on the current transistors is.[/quote]
IIRIC that 'BP' was because of the drawing-program. BC550C are recomm.

 

[DaxLiniere]

HEY, HEY! I just had an idea.

What if I put 2x 22uF 35v Stargets in parellel as a replacement for the 47uF 63v?

My first thought was that this would degrade the audio, passing it through 2 components instead of just one.
BUT, then I realised parallelling them should halve the resistance and that it's still passing through 44/47uF of capacitance, whether it's 2 caps or 1 shouldn't make a big difference........ I think

On second thoughts, as long as you guys can tell me if won't make a negative difference to the sound, I'll go with 2x47uF (35v)

The space issue doesn't really bother me, I can find a way to make 'em fit!

 

[DaxLiniere]

IIRIC that 'BP' was because of the drawing-program. BC550C are recomm.

Ahh, I see.
Damn, I wish the BOM had been more accurate

 

[clintrubber]

Would work fine. Dunno where in the circuit it is - is the 35V rating OK enough ? That'll be a yes as long these are NOT the phantom-blocking caps.

 

[DaxLiniere]

Yes actually, I was talking about the phantom blockers.

I figured parallelling 2x 35v caps would give me a 70v rating, but it IS 1:30am here, so maybe it should be 2x 47uF 35v in series for 70v rating.

 

[clintrubber]

Caps in parallel double the uF's but keep the same voltage rating.
Caps in series double the voltage rating (if the voltage spreads evenly, you might put high-valued R in //...) and halve the uF's.
But you knew all that when fresh awake :wink:

 

[DaxLiniere]

I didn't know about the series capacitance halving asleep OR awake!

How is that possible though, I don't understand. If the signal travels through, say, 20 layers of dielectric/electrolyte in the first cap, producing a capacitance of 47uF and then through another 20 layers of dielectric/electrolyte, how can the capacitance DEcrease?

 

[clintrubber]

Capacitance halving: the most elegant way to explain this escapes me for now, but I guess this will be explained in one of the META's. Or have a look for other sites about this.

With caps it's like resistors, but then the other way around w.r.t. doubling & halving. You could understand that, but I think it won't feel yet right, yes ?

 

[DaxLiniere]

Okay, I found a website that explained why my theory is wrong. Not explicitly*(see note), but enough to make me trust it :wink:


For anyone else looking in on this thread, here's that website:

http://www.play-hookey.com/dc_theory/series_capacitors.html


So there's no way to increase the voltage handling of a capacitor by using double parts.
How come when in // they don't each 'take on half the work'?
Is it because 'work' is current, not speed of electron flow (voltage)?


*
To summarise the theory from that website: high capacitance is where the 2 plates are close and low capacitance is where they are farther apart (as in 2 series caps).
So doesn't that tell you that tiny little ceramic caps should have the highest values, not the lowest...? :?

 

[clintrubber]

So there's no way to increase the voltage handling of a capacitor by using double parts.

Sure there is. To make a higher voltage part from caps with smaller ratings simply put them in series, but you need to use caps with double the uF's each.

Or as an example: you find a bag with 47uF/50V caps but need a 100V rating. Just use four to get one cap of 47uF/100V: put two times two in // and put these two pairs in series.

BTW: coils behave like resistors w.r.t. the series & // thing:
(value of) L_total = L_1 +L_2