Thanks for the response. Capacitors which I have are screw mount GAA series rated at 100 VDC. Project is a amp. designed by Pavel Macura ( http://www.sound.westhost.com/project83.htm). The only change is to my psu, I will use lower voltage - 25 volts.
I have built several amps using output capacitors (The Pavel-Macura was I think the best sounding one). Usually they are low wattage amps so I thought maybe I could have a higly dynamic amp by using an over-rated output caps so I used not only massive capacitance but also higher voltage capacity. They are rated 150V to 450V (very big), Hitachi and Nichichon. My first impression was that the bass quality was undeniable, but I'm not very surprised because that was what I had expected from the system.
Later on, I tried to question the use of such an output capacitor. You must have known that many said that an output cap as well as output transformer in tube amps will degradate the signal. But you must have also known that it is easy just to say what other people have said. Bottom line, I changed the cap into normal quality cap. It was easy to recognize the change, but not at all an improvement! (Since then I used the caps in other projects without question)
Same capacitance, highly different voltage, different class of brand, different mounting. So should I have believed that the main reason was the voltage capacity? I thought, probably, that the mounting also played a role (you cannot mount the big caps on PCB!) This conclusion was affected by my other experiences combatting oscilation in the output stage. I have always omitted zobel network in amplifier output to increase dynamic, and only use it if I had to. The use of off-the-board output caps had helped a lot.
Later I will build another one using smaller darlington, 2SD1793, higher fT and not TIP
** Just Another Thought From A Different Kind Of Angle.
** If it is about electronics theories, I know you guys are better than me.
** If it is about ears, I know I'm one of the best.
** If it is about IQ, it helps me as much as it hurts.
Later on, I tried to question the use of such an output capacitor. You must have known that many said that an output cap as well as output transformer in tube amps will degradate the signal. But you must have also known that it is easy just to say what other people have said. Bottom line, I changed the cap into normal quality cap. It was easy to recognize the change, but not at all an improvement! (Since then I used the caps in other projects without question)
Same capacitance, highly different voltage, different class of brand, different mounting. So should I have believed that the main reason was the voltage capacity? I thought, probably, that the mounting also played a role (you cannot mount the big caps on PCB!) This conclusion was affected by my other experiences combatting oscilation in the output stage. I have always omitted zobel network in amplifier output to increase dynamic, and only use it if I had to. The use of off-the-board output caps had helped a lot.
Later I will build another one using smaller darlington, 2SD1793, higher fT and not TIP
** Just Another Thought From A Different Kind Of Angle.
** If it is about electronics theories, I know you guys are better than me.
** If it is about ears, I know I'm one of the best.
** If it is about IQ, it helps me as much as it hurts.
Oh boy! The darlington is the current source! 
No wonder a TIP could sound so good
** Just Another Thought From A Different Kind Of Angle.
** If it is about electronics theories, I know you guys are better than me.
** If it is about ears, I know I'm one of the best.
** If it is about IQ, it helps me as much as it hurts.
No wonder a TIP could sound so good ** Just Another Thought From A Different Kind Of Angle.
** If it is about electronics theories, I know you guys are better than me.
** If it is about ears, I know I'm one of the best.
** If it is about IQ, it helps me as much as it hurts.
Jay said:
By the way, it's worth noting that you don't have to take everything from the main output cap.
For example, if you've got a bi-wirable pair of speakers and the high-pass section has a series cap ahead of any shunt resistors or inductors, in most cases you can feed it straight from the amp's output and eliminate the main output cap from that part of the circuit.
se
PMA said:
You damm right! Due to my previous post, I took a special test/observation with this.
The capacitance usually determine the bass roll-off. Theoritically, several mF (3300 or so) is enough for 20HZ, but by listening you usually get more with higher capacitance, and it seemed like minimal distortion in the bass frequencies like PM said.
The bass is clean (deep) but it seemed something is missing or the signal is delayed. Paralleling with smaller cap does solve this problem. And guess what? I'm more convinced of my strategy to parallel higher capacitor with... silver mica! (I put the micas right in the legs of the bigger caps)
But it is the capacitor type or brand that matter more. In my previous post it is supposed to be Nippon-Chemicon, not Nichicon.
Re Silver-mica, I don't think it is so much that they sound better as that they have different functional properties. First off the range of values is fro 1pF up to less than .1uF and generally you don't find many with a value greater than 1000-2000pF. At these smaller values, audio frequencies are blocked almost like DC so the chances you will ever use one in the audible part of the signal path is small. There isn't much overlap with film except for the smallest film values and largest silvmica values. The most likely alternative is Silver mica vs. ceramic. Their biggest advantage is tight standard tolerancwes and high stability compared to just about anything else.
They are pretty expensive so in the case of power bypassing, they are probably overkill. Also they are larger than ceramics. The only place I use them is as complensation capacitors. That's still not in the audible signal patrh but it is close enough that useing them there gives me reassurance. Most likely ceramics would work as well.
I suppose that there *could* be some effect on what one hears in these applications if having very tight and stable values was important for a particular design.
They are pretty expensive so in the case of power bypassing, they are probably overkill. Also they are larger than ceramics. The only place I use them is as complensation capacitors. That's still not in the audible signal patrh but it is close enough that useing them there gives me reassurance. Most likely ceramics would work as well.
I suppose that there *could* be some effect on what one hears in these applications if having very tight and stable values was important for a particular design.
Jay,
I'm not sure I understand the the reference to electroliyics. If you mean why do you find the silver micas in parralel with electrolytics, I assume you mean when placed between the power rails and ground in parallel with an electrolytic. The principal reason is that the bandwidth of electrolytics is too low (perhaps 100kHz) to handle all the RF and other noise that might be catied by the power rails. A Silver Mica or ceramic in parallel extends the bandwidth to 100MHz or so. If electros were effective into the MHz spectrun, there would be no need for either Silver micas or cermic in this application.
Silver micas or ceramics when used for compensation are used for a somewhat related reason, namely to control HF signals that get affect the feedback loop (both local and global). Without going into the details, the objective is to supress high frequencies starting somewhere around 50kHz and extending up into the MHz region. Although, you cannnot hear these, the transistors in an audio amplifier will react to them and runnaway oscillations can result. The audible significance to you is great -- these capacitors if choosen correctly keep the amp from literally going up in smoke!
In most Amp designs the only capacitor that is *directly* in the signal path is the input decoupling capacitor. That's a completely different purpose entirely.
If someone want's to explain this better or correct my errors, please do so. I'm responding within the limits of my understanding.
I'm not sure I understand the the reference to electroliyics. If you mean why do you find the silver micas in parralel with electrolytics, I assume you mean when placed between the power rails and ground in parallel with an electrolytic. The principal reason is that the bandwidth of electrolytics is too low (perhaps 100kHz) to handle all the RF and other noise that might be catied by the power rails. A Silver Mica or ceramic in parallel extends the bandwidth to 100MHz or so. If electros were effective into the MHz spectrun, there would be no need for either Silver micas or cermic in this application.
Silver micas or ceramics when used for compensation are used for a somewhat related reason, namely to control HF signals that get affect the feedback loop (both local and global). Without going into the details, the objective is to supress high frequencies starting somewhere around 50kHz and extending up into the MHz region. Although, you cannnot hear these, the transistors in an audio amplifier will react to them and runnaway oscillations can result. The audible significance to you is great -- these capacitors if choosen correctly keep the amp from literally going up in smoke!
In most Amp designs the only capacitor that is *directly* in the signal path is the input decoupling capacitor. That's a completely different purpose entirely.
If someone want's to explain this better or correct my errors, please do so. I'm responding within the limits of my understanding.
An after thought: look at www.avxcorp.com. Somewhere on this site is a table that describes some of the charateristics or various kinds of capacitors.
sam9 said:
sam9, what do you mean that the later case is a different purpose entirely? Did you mean that your previous explanation did not apply? In fact, the use of silver mica *directly* in the signal path is what I had in mind. But I made a little "mistake" in my previous post, in that I myself had never tried it in the output stage of an amp like in the PM's MOSFET follower (but I will give it a try once I build such amplifier, with 1uF silmic).
Do you mean that the purpose of the input decoupling cap to block DC? Well, I have never cared enough to think about formulas or technical explanation behind these capacitors when I used them. That's because... it's just to complicated than doing a trial.
Using silmic to bypass elyt caps in the signal path had never come across my mind until I read that silmic is the best at RF. So I thought why not give it a try in the decoupling of Opamp (low pass filter) from DAC high frequency output. Later I consistently use silmic to parallel elyt caps in the signal path. In certain output decoupliing situation I can hear clear difference when I paralleled the silmic with high capacitance elyts, not the "standard" MKP.
For coupling (to ground) I still prefer paper or oil caps, and tantalum is just the worst. Well, there must be situations where tantalum is good (I used them once in RF), but in my audio experience, there are 2 types of caps that I found never been good enough for any applications: tantalum and the green small valued WIMA MKP. The latest are commonly used in audio gears (
)
To begin with I've never seen a 1uF silver mica, even in a catalog. Every table, including the website referenced above says the max value for Silver Mica is >10nF (i.e, 0.01uF). You can try that in the audio output if you insist but I doubt you will care for the result, all a capacitor that small will pass is RF. If it is in parallel with another cap of 1uF or greater you will get most of the audio because the audio is passing through the larger cap.
BTW, 1uF will attenuate base somewhat - you really nead at least 4.7uF to avoid this.
BTW, 1uF will attenuate base somewhat - you really nead at least 4.7uF to avoid this.
sam9 said:
1uF silmic doesn't exist? ??? I will check it again. The silmic is about 3cm wide. My 1nF silmic (in Aleph3) is as small as 5pF silmic (less than 1cm wide) so I think it makes sense if the much bigger one is 1uF. Or... I think I have to check my Aleph in case it is using 1pF instead of 1nF??? (if I'm not mistaken, 1000 is printed on the body).
What do you mean by attenuating base? Is this a bad thing? Can you explain because I use many of them in the audio path (in parallel). Did you mean that it will attenuate the low frequency if it is used individually (not in parallel with bigger cap)?
And where did the 4.7uF come from? Was that in refference with Pavel's design? He himself reccommended a 10000uF instead of 4700uF as stated in the documentation, in parallel with 1uF~10uF. IMO, the capacitance theoritically should be: C = 1/(2.pi.F.R). With F=20Hz and R=8Ohm, I think it is close to 4700uF. But yes, I mostly use a 1/1000 of the bigger cap, so it is about 4.7uF when the bigger cap is 4700uF. I don't know how to come up with such number, it is just the "formula" I found in many designs.
I didn't think that using less than 1uf or more than 10uF in parallel will create a problem in most situation (mainly as in the Pavel's amp). And because limited theories mostly cannot explain the phenomena around the use of the caps (and other areas of electronics too
), I have trusted my observations more than my knowledge. But if using 10pF in parallel with 4700uF will create a problem (e.g. in CDP output cap), I really want to know why.I think I know what you mean
I have a friend who is even worse. He not only tried different fuses in the transformer primary, even worse he tried different orientation of the fuse! I believe you will not try that. Me either.Cheers
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.