Recapping ala Jim Williams

[paulrichards7]

So I came across these words of wisdom for Jim Williams:

Another problem with leaving in older caps is for one they are usually too small in value. Designers typically use the smallest and cheapest parts they can so low end roll off are set to 20 hz or so. That will create phase shift beginning at 200 hz.
To remove it, roll-offs must be set to 2 hz so the phase shift begins at 20 hz or lower = not audible. Soundcrafts and others have that problem as they use 47 uf/25v caps everywhere, the use of 220uf caps does wonders. Those are still quite cheap, a 1000 piece bag of Panasonic FM 220 uf 25 v caps costs only $80. That's 8 cents each. There is no reason to buy lesser caps when these are so cheap.

and wondered how they might apply to the caps on the audio path of my mixer channel
http://i390.photobucket.com/albums/oo34 ... h-caps.jpg

They have different capacitance and voltages and wondered if this is specific to that design or due to Studiomaster pricing
Thanks
Paul

[argonautlabs]

It depends on the not just the capacitance but also how the output is loaded. You could calculate the phase response of every step, but why don't you just measure the in/out phase difference at 40hz or something versus 1k and then decide if you need to worry about it. You would get the answer much quicker. The phase response may even be in the manual.

"Designers typically use the smallest and cheapest parts". This statement is not necessarily true or provable.

[Jim Williams]

Use the standard 2xpi/rc formulas to determine the -3 db roll-off points.
Most manufacturers select a roll-off point regardless of the phase response. A 20 hz low end is not sufficiant as phase shift begins a decade above. One or two stages is not serious, but run through an entire console and these roll-offs stack up. Consider each el cap roll-off a single pole of a high pass filter. Run through 10 or 20 stages of high pass filter poles all set to 20 hz and you get the idea. That will create a bit of group delay, or phase shift vs frequency. That will re-orient harmonics of a waveform. Set the roll-offs at 2 hz and phase shift begins below 20 hz.

[paulrichards7]

Thanks for that Jim
Could you elaborate a liitle bit more on the math of
Use the standard 2xpi/rc formulas to determine the -3 db roll-off points.
Thanks
Paul

[nclayton]

I'd say it's a pretty huge exagerration to imply that console designers regularly or even ever set single stage 3dB points at 20Hz. Those guys kind of knew about how those poles stack up, and you can almost always expect the individual stages to have 3dB points in the low single digits at the highest. For example, just at a glance, most of the stages in the studiomaster schematic posted have 3dB points at around 1Hz or lower.

The one weak point on the studiomaster schematic is the gain control DC blocking cap. For almost all gain settings, that size of cap is fine, but at the VERY highest gain it really does have a 3B point PRETTY close to 20Hz. Pegged it would be 15Hz to be exact. You could upgrade that cap to a higher capacitance. Of course, that's if you're worried about have a 15Hz rolloff at max gain, which probably wouldn't bother most people. Hell, I've even known some people to intentionally use an HPF!

Otherwise, even on this budget console, none of the caps are anywhere near making a 20Hz rolloff, even when cascaded through the strip. At sort of "normal" gain settings (like with the gain knob at maybe 3 o'clock or lower) the LF rolloff from end to end of this entire circuit here is probably at about 3-4Hz at the highest.

Ned

[nclayton]

It's hilarious to imply there's something intrinsically wrong with a 47uF cap. Soundcraft used a lot of them, yes, but they also used a shitload of 10K ohm resistors. What's the 3dB point of a 47uf cap loaded by a 10K resistor? It's about 0.3 Hz.

[argonautlabs]

I just don't get how people can go ripping their working and decent gear apart and replacing things without even taking the effort to actually evaluate if what they have is unacceptable. I don't know, i would think people would spend more time recording a good record than worrying if their board is 20 degrees out of phase at 20hz. I would even debate whether it matters if it was. And I'm a full-time electronics engineer. You would think I should be the one worrying about this stuff. My RAMSA 8616 is probably worse than that in terms of phase in both the low and high frequencies but it sounds great, and as long as im not running only one mic off a stereo setup into it or something it isn't going to matter much to me. And there are also electrolytics in the audio path but i doubt i would notice if I replaced them with "better suited" caps. I just think people should take more time to TRULY understand what the end result of their "upgrades" are before they flag them as necessary. I would not regard someones advice off the internet as understanding the situation.

[chris harris]

Isn't the assertion in the original post that phase shift begins at 200Hz? Seems that that would be something worth being concerned about.

[argonautlabs]

"begins" doesn't really mean much. There are pro boards that people take a second mortgage for that have a couple degrees or more at 200. Their overall 3dB point (for frequency) can still be less than 20hz. Not everything in life is a flat line, and a flat line won't necessarily sound better. I personally am not concerned about 10 degrees at 40Hz either. There are much more important things to worry about in a mixer than a couple degrees of phase response. I care more about the armrest.

[paulrichards7]

Here is a High pass corner frequency calculator
http://www.opamplabs.com/rfc.htm
Paul

[xhavepatiencex]

I may try this on a compressor i am going to recap, if there is no room on the underside of the pcb due to the way its mounted to the case is it fine to leave some longer leads on the electrolytics and wedging the bypass caps under there on the top of the board?

[paulrichards7]

From what i understand, its best to keep legs as short as possible
Otherwise there is more resistance, i think?
the others will clarify this though
paul

[paulrichards7]

Just out of curiosity, what sets the high frequency roll off
thanks
paul

[themagicmanmdt]

Just out of curiosity, what sets the high frequency roll off
thanks
paul

this would most likely be the bandwidth limitation setting of the IC circuit, either left to it's own operating conditions, or set to roll of at a specific frequency to prefent oscillations -

correct?

i can't say that i feel HF rolloff in ANY old or modern preamp has even been a concern for me -

LF rolloff, however...

if it's only 1db down at 40hz, then a little EQ does the trick...!

does it not?

or am i being simple minded towards the premise of simply designing a better audio path?

[paulrichards7]

hi themagicmanmdt,
thanks for that
paul