|
Post by svart on Oct 21, 2014 7:03:37 GMT -6
Caps are pretty cheap and can do alot of good if added to a circuit where they have been omitted due to cost reasons. There are different approaches to this. Normally, it is good practice to stabilize the power rails at EACH integrated circuit in a design with a 0.1uF metal film and a 10uF or 22uF electro in parallel. You can even use up to 3x 0.1uF metal film caps for an IC, one for each power rail, and one between +Vs and -Vs. Douglas Self wrote something about this cap for NE5534 and NE5532 opamps...can't find it right now.... The low capacitance metal film caps should be placed as near to the IC as possible, hence solder it directly to the pin on the copper side. Use the next available ground pin or directly the nearest ground rail on the PCB for the other side of the capacitor. Maybe svart can chime in how he did it on his units, maybe he remembers? Try replacing the 10uF i see on the backside of the PCB with 22uF ones. I strongly guess they are something like the stabilizing caps for power rails, that will be in parallel to the small metal films we are going to solder in. You can easily check this, if one pin is + or - 15V and the other one is ground, try a cap with 22uF when recapping. Yes, you want 100nF from -V to ground and +V to ground at the opamp, at the very least. This is your local decoupling. It forms a mathematical low pass filter(for noise) and adds a current reservoir for the part so that when the opamp is asked to supply a fast transient, the current is available quickly from the local capacitors. A third 100nf capacitor between -V and +V can be added for PSRR (power supply rejection) which helps get rid of common mode noise, which is common between the power rails. This isn't necessarily required but it's easy to add in most cases. Metal films work here, but so do decent ceramic parts. The point is to supply the opamp with a current resevoir. If the type of decoupling cap makes an amp sound different, then you don't have enough capacitance on the power rails! 100nF isn't magical. you can use really anything from 10nF to 1uF as long as it's film or ceramic. 100nF is a decent trade off between cost and capacitance value, which is why it's the universal decoupling value. When you get into higher power and higher speed devices (RF and such) then things get different and you have to look at power density vs. size of the part package and things like that. Anyway, I wouldn't necessarily go swapping all values of electrolytic cap to higher values just yet. We do need to see a schematic preferably before that happens. If you are sure that the cap is part of the power system, then you can probably double the value safely, but if it's an audio signal electrolytic, then I'd hold off doubling the value and just stick with the same value. I'm not with the jim williams crowd that desires DC-to-ultrasonic audio chains. I think it's impractical and unnecessary, and overall almost impossible to do correctly by just swapping parts, unless you test it properly.
|
|
|
Post by svart on Oct 21, 2014 7:05:09 GMT -6
The low capacitance metal film cap lowers the ESR of the parallel electro extremely... Well, in fact, modern electros already have very low ESR, so this is not true anymore....just wanted to correct myself on this... This is correct. Most modern electrolytics have pretty decent ESR. It can't hurt to add a parallel cap, but honestly I think that it's a waste of a good film cap to do so.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Oct 21, 2014 7:13:10 GMT -6
Yepp, regarding upping 10uF to 22uF i therefore suggested to measure/check if the electros are between power rails before upping the value to not mess up things... Thanks for chiming in!
|
|
|
Post by svart on Oct 21, 2014 7:15:54 GMT -6
These caps also have a type of filter effect. The high capacitance cap (the electro) is kind of highpass filter. Upping the capacitance from 10uF to 22uF therefore *possibly* clears up the lowend response. It is *not* like more is better in this place generally. It depends on the voltage sag/current draw of the IC used. So, there are values commonly possible between 4.7uF and 47uF, depending on the type of IC. Sometimes, 10uF caps are used where 22uF performs better, because of cost reasons. Therefore, try. Sometimes, more is worse! Then go back to 10uF. But i wouldn't worry about this too much... The low capacitance metal film cap lowers the ESR of the parallel electro extremely and adds near nothing to the ESL. It is used to filter out high freqency noise. And, in proper designs, it is NOT optional, but mandatory. Therefore, many PCBs of well implemented designs using ICs look like they packed them full of caps. In higher than audio frequency application, you even commonly parallel another decoupling cap, like, say 10nF. Not necessary for audio circuits, though.... Just a short explanation why we do this stuff.... Any capacitor to ground forms a Low pass filter.. The larger the value, the lower the frequency of the cutoff. Larger caps like electrolytics are fairly slow to move charge in and out, so they are better for supplying slow currents to the opamps. Opamps attempting to amplify large and slow signals need more power than fast or small signals do because lower frequencies are much higher power than higher frequencies. In the case of low frequency signals, the charge stored in the small 100nF caps would be depleted quickly and current would be required to be drawn through the traces from the power supply. Long, thin, traces are not only resistive, but inductive too so both would resist the opamp drawing current through them. The remedy is local bulk decoupling, where a larger reserve of charge is stored in capacitors that are close to the opamps, so that when they require it, they can pull current from these bulk capacitors. That's the layman's version of why larger decoupling caps can help increase low frequency reproduction. So you have small and fast decoupling caps on the opamps for fast signals, and now you have larger slower bulk charge caps a short distance away from the opamp for the larger signals. That's the simple version of decoupling!
|
|
|
Post by mrholmes on Oct 21, 2014 8:27:29 GMT -6
I lost orientation now. Remember it is more than 25 years ago when I did ham radio stuff like this. For decoupling I use now: Is there a way that I can see an example drawing for decoupling. I do not want to do it the wrong way. Example picture?
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Oct 21, 2014 8:39:36 GMT -6
Yes, you explained it pretty well, svart.
Please forget the "kind of highpass" thing (mumble, facepalm to myself in the mirror), sometimes i should type slower and think faster... :-) Even if you generally know what you are doing, if you try to explain it, you can step into traps and get into difficulties to explain, like me. Not beeing native english writer makes this worse sometimes. Still fun, though, to re-think things once again... :-) Please continue to make remarks if i am writing technical BS, much appreciated. Still learning by making errors, hopefully....
Best regards, Martin
|
|
|
Post by Guitar on Aug 25, 2018 9:19:28 GMT -6
I just modified my MDX-1200 I made a blog post about it here: Dantone/DWA DIY blogMaybe some of you tech experts could tell me how I did. The main issue is that there is some distortion on a bass DI through the Behringer. Cranking up the "Enhance" knob really "Enhances" the amount of distortion big time, on certain settings! The attack and relase times seemingly have nothing to do with this distortion. It happens on transients. My DBX 560A here do not distort or click at all so I know it's not the patchbay or anything else in the chain. This seems to be common in poorly designed cheap compressors, my 3630 does the same thing. Other than that serious fault, it does sound pretty good on 50% of my bass playing where the notes are letting the compressor behave itself. This is probably my last "Mod" project. From here on I'll probably be building more high end DIY kits and stuff, like the Serpent SB4000 that's been waiting here.
|
|
|
Post by Guitar on Aug 25, 2018 11:10:11 GMT -6
I tried a bass synth since I could just hold a key down instead of finding the "wrong" notes on the bass.
Any attack setting below 10 ms results in crackly distortion. That seems to be the common thing going on, with the bass synth. Although the bass guitar test crackled at almost any setting.
Snare drum sound is excellent. Midas 512 and Autocom is getting a killer snare track.
I guess my Behringer racks are my "dirty compressors" just taking that for what it is, they do have uses in certain applications where you need an aggressive sound where a little clipping is actually a small feature rather than a complete defect.
|
|