| Solid Snake |
| My latest amp uses no caps in the signal path, it's pretty much a DC amplifier. It doesn't use ground as a signal ground so I can't use more than one channel per power supply. Is there any real advantage to not using caps? |
|
|
| PanzerLord |
| nope, you are just adding a risk... |
|
|
| Petter |
I disagree with the prior post on sound quality, but agree on risk.
What you should do if you are going capless is to make measurements to ensure that the output resides at the right DC level before the cap ..... If not, you can't go this way without making modifications.
Petter |
|
|
| Solid Snake |
| Well after all the attention given to passive crossovers and why they are bad (phase shift) I thought I would leave caps out of my amp altogether. I have built a working amp without caps but because ground references are different on every channel, I need a separate power supply for each channel. Sound quality is the reason I did this in the first place so if there's no benefit to excluding caps, I may as well add them. The thing that I was thinking about is that most devices (computers, cd players, dvd players) have a dc blocking cap on their output. If I had a cap on my amp's input, it would just be a part of the CD player's DC blocking cap (2 caps in series) so my efforts seem to be in vain. |
|
|
| Peter Daniel |
It all depends on you. If you hear the difference the coupling caps make, you may try to avoid them. If you hear no difference, it's better to leave them in a circuit. After all the good drivers are expensive.;)
PS: I hear the difference the coupling caps make ( especially the cheap ones) and if possible, I'd rather avoid them. |
|
|
| fdegrove |
Hi,
| quote: | | The thing that I was thinking about is that most devices (computers, cd players, dvd players) have a dc blocking cap on their output. If I had a cap on my amp's input, it would just be a part of the CD player's DC blocking cap (2 caps in series) so my efforts seem to be in vain. |
Yes, most PCs, CDPs, DVDs do have DC blocking caps at their output. Thank God for that...
So if you're confident with that there's no need for a cap at the inputstage of the preamp or amplifier unless you have a conventional volumecontrol there and need to shield it from DC on its wiper.
Now, have you considered what happens if two coupling caps are put in series?
Is that amp you talk about DC coupled throughout because you leave out a coupling cap at the input?
Are you sure the PS caps aren't in the signal path?
Knowing where the DC goes is one thing, understanding where the AC goes is the key I'm afraid.
Cheers,;) |
|
|
| Christer |
Bear in mind that while you may be able to control the DC
offset generated by an amplifier so no output cap is needed
(provided you have failure protection circuitry, or take the
risk of failure), you cannot control what goes into the amplifier.
What if something goes wrong earlier in the signal chain so
you get several volts of DC into the amplifier? If you can still
regulate the output offset to a reasonable value, then fine,
otherwise it seem wise to have an input coupling cap, which
anyway is usually not of higher value than you can use a good
film cap.
Then, for those who prefer to risk burning their speakers to
avoid a coupling cap, fine, just be aware of what might happen
if a DC servo earlier in the chain goes weirdo.
Edit: Just saw Franks parallel post. Since output caps must
usually be of much larger value than input caps, making it
hard to avoid lytics in many cases, it might, perhaps, be more
sensible to use an input blocking cap, and remove the output
one in previous equipment, provided the design of that equipment
allows this. Just my :2c: (US or Euro ones? Pick your choose). |
|
|
| fdegrove |
Hi,
I guess driving a Volvo with two sets of safety belts per occupant does make it a safer car??
Cheers,;) |
|
|
| Peter Daniel |
| I had my drivers burned twice ($500 loss each time), and it was never because the input cap wasn't there. |
|
|
| Christer |
| quote: | Originally posted by Peter Daniel
I had my drivers burned twice ($500 loss each time), and it was never because the input cap wasn't there. |
I believe you, but if you leave it out and the preceeding
piece of equipment is also DC-coupled at the output, you
have two pieces of equipment that can fail so they might
burn the speakers. Is is all about statistics. |
|
|
| Christer |
| quote: | Originally posted by fdegrove
Hi,
I guess driving a Volvo with two sets of safety belts per occupant does make it a safer car??
Cheers,;) |
Not sure whose claim or post that referred to, but the answer
is not, ar at least that it is such a marginal improvement that
it won't be worth the cost and awkwardness. |
|
|
| Peter Daniel |
| quote: | Originally posted by Christer
I believe you, but if you leave it out and the preceeding
piece of equipment is also DC-coupled at the output, you
have two pieces of equipment that can fail so they might
burn the speakers. Is is all about statistics. |
It is true. And since my amp is not protected by a cap, I always check the source for DC offset, before I connect it to the amp. Let's face it, if your woofer is $500 to replace, you simply have to do it. And I got 4 of them;) |
|
|
| fdegrove |
Hi,
| quote: | | Is is all about statistics. |
If we have two caps in series the second one isn't biased at all as it doesn't carry any DC.
Assuming the one in front of it can fail than, statistically spoken the one behind it can fail just the same.
And since we have two caps in series the total value is cut in halve and for a give F/C we'd need twice their size to maintain the same F/C as before.
Now, I wonder how that's going to help the sound if you know that bigger caps are inherently slower to release their charge?
Cheers,;) |
|
|
| fdegrove |
Hi,
| quote: | Since output caps must usually be of much larger value than input caps, making it hard to avoid lytics in many cases, it might, perhaps, be more sensible to use an input blocking cap, and remove the output one in previous equipment, provided the design of that equipment allows this. Just my (US or Euro ones? Pick your choose).
|
I'll go for the Euros today, thank you.
But it seems you don't get the point, Christer.
The input and output cap in an amplifier do have totally different functions; leaving the ouput cap of the amp out at the risk of speaker killing DC is silly and I never count on DC servos to save my speakers.
Having two caps in series, one at the output of the source and one at the input of the amp is plain redundance and also has an impact on bass performance of the both of them as they're in series.
The Volvo analogy should have triggered that already, or so I'd hoped.
Cheers,;) |
|
|
| Solid Snake |
| There are no caps ANYWHERE on the signal path, not input, not output. As long as you apply 1VDC on its input, you'll get 50VDC on its output. What I was asking is that a cd player with a coupling cap will cause a 90 degree phase shift no matter what. If I add a capacitor to my amp's input, it will effectively join itself to the cd player's cap, so it won't add any more phase shift, correct? My amp doesn't use output caps so there's no need for that. I should be able to use electrolytics on the input because they'll always have a positive voltage across them. Should I opt for a different type of cap? Input impedance is high enough so that I won't need a high value, 10mF or so should be fine. |
|
|
| Christer |
| quote: | Originally posted by Peter Daniel
It is true. And since my amp is not protected by a cap, I always check the source for DC offset, before I connect it to the amp. Let's face it, if your woofer is $500 to replace, you simply have to do it. And I got 4 of them;) |
Sure, but the previous equipment may still fail after you have
measured it. You do get an increased probability of failure
the more such DC-coupled equipment to chain up. |
|
|
| Christer |
| quote: | Originally posted by fdegrove
Hi,
If we have two caps in series the second one isn't biased at all as it doesn't carry any DC.
Assuming the one in front of it can fail than, statistically spoken the one behind it can fail just the same.
And since we have two caps in series the total value is cut in halve and for a give F/C we'd need twice their size to maintain the same F/C as before.
Now, I wonder how that's going to help the sound if you know that bigger caps are inherently slower to release their charge?
Cheers,;) |
I have a feeling you must have misunderstood me. I didin't
advocate two caps in series. I said, get rid of output caps and
use input caps instead. since you are picky about cap sound
I am sure you prefer a cap small enough that you can get
a good polypropylene one. |
|
|
| Christer |
| quote: | Originally posted by fdegrove
The input and output cap in an amplifier do have totally different functions; leaving the ouput cap of the amp out at the risk of speaker killing DC is silly and I never count on DC servos to save my speakers.
|
I did say that you should have proper protection against DC
on the output if you leave out the output cap and care about
your speakers. If you don't like relays, maybe a crowbar might
still be satisfactory to you. If you don't like the sound of caps,
especially big lytics, I cannot see how you can stand a power
amp output cap, but maybe they are small enough to be
film type since you
go for those glowy things. Don't know enough about that stuff. |
|
|
| fdegrove |
Hi,
| quote: | | What I was asking is that a cd player with a coupling cap will cause a 90 degree phase shift no matter what. |
And where did you read that?
| quote: | | If I add a capacitor to my amp's input, it will effectively join itself to the cd player's cap, so it won't add any more phase shift, correct? |
Phase shift won't change but it's not a shift, more like a group delay which is NOT the same.
Add the same value cap as the one on the output of the source and tell me you don't hear the difference.
While you're at it justify the use of that cap in the first place for I just don't see the logic.
Any decent CDP, DVD and what have you should have no DC at the output whatsoever if it has to hit the market.
Cheers,;) |
|
|
| Christer |
| quote: | Originally posted by fdegrove
Any decent CDP, DVD and what have you should have no DC at the output whatsoever if it has to hit the market.
|
Sure, but they do so at the cost of having lytic output cap,
unless they use a DC servo, that might fail. |
|
|
| fdegrove |
Hi,
[QUOTE]If you don't like the sound of caps, especially big lytics, I cannot see how you can stand a power amp output cap, but maybe they are small enough to be
film type since you go for those glowy things. Don't know enough about that stuff.
QUOTE]
Sorry, Christer one's got nothing to do with the other.
Whether interfacing a speaker with semi-conductors or a glowy things as you call it has got nothing to do with it as the value of the cap will still be the same for the same Z of that speaker.
Once again, that was not the question.
Cheers,;) |
|
|
| Christer |
| quote: | Originally posted by fdegrove
Hi,
[QUOTE]If you don't like the sound of caps, especially big lytics, I cannot see how you can stand a power amp output cap, but maybe they are small enough to be
film type since you go for those glowy things. Don't know enough about that stuff.
QUOTE]
Sorry, Christer one's got nothing to do with the other.
Whether interfacing a speaker with semi-conductors or a glowy things as you call it has got nothing to do with it as the value of the cap will still be the same for the same Z of that speaker.
Once again, that was not the question.
Cheers,;) |
Then I must be missing your point entirely. What did I get
wrong?? |
|
|
| fdegrove |
Hi,
[QUOTE]Sure, but they do so at the cost of having lytic output cap, unless they use a DC servo, that might fail.QUOTE]
And DC servos are fail proof you think? No more or less than any cap I'd say.
Now, would you put a DC servo at the output of that CDP and another on the input of the amp?
Cheers,;) |
|
|
| Christer |
| quote: | Originally posted by fdegrove
Hi,
[QUOTE]Sure, but they do so at the cost of having lytic output cap, unless they use a DC servo, that might fail.QUOTE]
And DC servos are fail proof you think? No more or less than any cap I'd say.
Now, would you put a DC servo at the output of that CDP and another on the input of the amp?
Cheers,;) |
Hm, I did assume capacitors are much less likely to fail, since
they are one single simple component, which if run well within
limits should be very reliable. A DC servo on the other hands
contains a lot of components that might fail, including one
compelx one, the op amp. I find it hard to believe the occasional
failure of a film cap should be even close to the failure rate
of DC servos, but I may be wrong there.
The interesting thing is that despite your usual "allergy" to most
capacitors they suddenly seem not to bother you. However,
it is late, so maybe I'd better try reread what you have said
tomorrow. |
|
|
| fdegrove |
Hi,
| quote: | | Then I must be missing your point entirely. What did I get wrong? |
It's late for me too but the point was the usefulness (securitywise) of having a DC-blocking cap at the source side (CDP out) and at same time having a DC blocking cap at the input of the amp.
My point is that there's no point of having DC blocking at the amps input as there should be nothing to block and if there's DC present from the source it should be blocked, choose your means, directly at the source.
Putting the two blocking means in series by connecting one to the other is not a good idea and I pointed to the effect of two caps in series too as here C1+C2= C1/2 given equal values.
Cheers,;) |
|
|
| sam9 |
Think of the total signal path starting with the microphone. Then add the various amplification stages, mixers, EQ and what not. How many caps do you estimate trhe signal has gone through before it gets to your amp? I would bet that some of them were electrolytics, even.
Analogous to the same worry about op-amps by those dedicated to 100% discrete audio. |
|
|
| fdegrove |
Hi,
| quote: | | The interesting thing is that despite your usual "allergy" to most |
Unless you want to go "iron" all the way it's hard to avoid them altogether.
I've been trying to explain on hands and knees how to avoid yet another cap and you're thinking I'm wanting to add one???
| quote: | | Analogous to the same worry about op-amps by those dedicated to 100% discrete audio. |
Oh yeah, let's "Make things better". Why bother with audio anyway?
They messed it it up big time already, right?
Cheers,;) |
|
|
| Solid Snake |
| I'm one of those dedicated to discrete components but I do see your point. Using an input cap would greatly simplify my design. I still see no reason for an output cap as I can always get the offset within a few mV of 0v. Besides, it would have to be huge to allow low frequencies through a 2 ohm subwoofer. Something tells me an output cap placed between a 2 ohm sub and a 200w amp would fail rather quickly. |
|
|
| fdegrove |
Hi,
| quote: | | Using an input cap would greatly simplify my design. |
How?
The only justification for a cap at the input of an amp is the assumption of DC at the source?
| quote: | | I still see no reason for an output cap as I can always get the offset within a few mV of 0v. |
Fine, no problem. I do have the same arrangement with my tubed OTL amps and haven't had any trouble with DC on the output for twenty years.
| quote: | | Something tells me an output cap placed between a 2 ohm sub and a 200w amp would fail rather quickly. |
Not with the right cap for the job but 2 ohm subs??
That's very low impedance calling for alot of current from the amp.
Cheers,;) |
|
|
| Solid Snake |
| quote: | How?
The only justification for a cap at the input of an amp is the assumption of DC at the source? |
Because I had to add more circuitry to make sure that 0v on the input always = 0v on the output regardless of the output impedance of the source. If you short circuit the input terminals on the amp, the output will still stay at 0v. Adding a cap would allow me to get rid of those pots.
| quote: | | That's very low impedance calling for alot of current from the amp. |
Yes, It lets my amp make the most of its +/-20v rails. It seems rock solid down to 1 ohm but a load like that makes my power supply very sad. The TIP35/TIP35 output devices can pass a continuous 25A so there's no problem there.
Even if I use a 6800mf cap on the output, it would still absorb around 30% power at 40hz. I don't think the cap would hold up too well, especially since half the wave charges it backwards. |
|
|
| fdegrove |
Hi,
| quote: | | Adding a cap would allow me to get rid of those pots. |
Assuming the pots are there for DC balance I can only say that the cap is going to be in the direct signal path whereas the pots are much less so.
Normally, once adjusted for DC balance they should stay that way for ages unless the circuit uses a fishy topology and the devices aren't matched and age differently because of that imbalance.
Up to you,;) |
|
|
| Solid Snake |
| One pot is responsible for setting the "no signal" offset. Another is responsible for setting the signal "ground" which in reality is about 0.95v. The problem is that every channel has it's own set of pots. If a device was to be connected that has more than one channel, its signal grounds will likely be tied together (probably to its chassis). On my amp the signal ground of my left channel may be .93v and the right channel may be .97v. If you connect a CD player's L and R channels, these two pots would be shorted together and as a result, you would get huge DC currents on the outputs. |
|
|
| Peter Daniel |
There is not too many SS designs that use a cap at the amp's output. One that comes to mind is a Zen amp from Nelson Pass.
So if you care about DC, you use a cap either at the input of your amp, or at the output of your source. There is no reason to use a cap in both locations. |
|
|
| fdegrove |
Hi,
| quote: | | One pot is responsible for setting the "no signal" offset. Another is responsible for setting the signal "ground" which in reality is about 0.95v. The problem is that every channel has it's own set of pots. |
Yes, of course that's going to happen. What did you expect?
Volts disappearing into the digital null?
You run a floating supply and expect it to somehow magically balance??
The problem isn't that every channel has its' own pots, its just that your source is already tying them to ground at the source.
It just ain't that simple my friend,;) |
|
|
| Solid Snake |
| quote: | | The problem isn't that every channel has its' own pots, its just that your source is already tying them to ground at the source. |
No. The amp's ground is not earth ground. The transformer's center tap is not tied to earth. The problem is that the pots are being tied to each other. |
|
|
| fdegrove |
Hi,
| quote: | | The problem is that the pots are being tied to each other. |
And to the source ground once you connect it?
Without a circuit diagram it's going to be very hard to discuss what's going on.
Can you produce one?
Cheers,;) |
|
|
| Solid Snake |
| No I have no diagram. The source ground is isolated from the amp ground, but since the source's L and R ground are the same, they cause problems. |
|
|
| fdegrove |
Hi,
| quote: | | The source ground is isolated from the amp ground, but since the source's L and R ground are the same, they cause problems. |
Does that surprise you?
Cheers,;) |
|
|
| Solid Snake |
| No it doesn't suprise me. I can run one channel with very little offset with almost any device, so it may be practical for a subwoofer amp. |
|
|
| Christer |
Now it is early (hm, well, for me at least) in the morning instead
of late at night, but let's see if I can clear up what I meant.
If we have full control of the whole signal chain, and know we
won't swap any equipment, it won't matter much where we
put the caps, since we know what the input impedance of
the next piece of the equipment is. Franks resaoning makes
some sense, blocking DC at the output, but since we have
full control of chain it doesn't really matter at which end we
put the cap IMHO.
If we don't have this full control, we cannot know if a preceeding
piece of equipment can be trusted to have zero DC out, and
we cannot know what the input impedance of the following
piece of equipment is. As far as I can see, this means two
things, 1) an input cap is useful, and 2) putting the cap at the
output means we must use a larger one than necessary, possibly
a lytic, just to be sure we don't filter out to much bass even
with somewhat lowish input impedance following.
That's all from the armchair engineering department for the
moment. |
|
|
| Petter |
Contrary to popular belief, you don't actually get 90 degrees phase shift when passing signal through a capacitor ...
If you don't have servo and is worried about the result, why don't you make two separate inputs: 1 with, 1 without and try both without speakers on? If OK, figure out if you can hear a difference worth the effort.
Petter |
|
|
|
