[jeffreyj]

Quote:

Originally posted by meat01 My MZ-N707 is less than a mV. Does that mean I can lose the input caps on my CMoy?

Definitely! Oops... As long as your Cmoy does not run off of a single-ended power supply!! If you aren't using two 9Vs in series with the centertap as ground, or a 9V split by a TLE2426, etc., then you NEED input and ouput caps. 'Course, there's a whole 'nother philosophical argument about whether you are getting any real bang for the buck when caps are required, but let's leave that for another day...

 

[jiggler]

Quote:

As long as your Cmoy does not run off of a single-ended power supply!! If you aren't using two 9Vs in series with the centertap as ground, or a 9V split by a TLE2426, etc., then you NEED input and ouput caps.

Jefferyj,

Would you please explain this comment. I have never heard or read anything like this before and it does not make any sense to me.

 

[yejun]

Quote:

Originally posted by jeffreyj Definitely! Oops... As long as your Cmoy does not run off of a single-ended power supply!! If you aren't using two 9Vs in series with the centertap as ground, or a 9V split by a TLE2426, etc., then you NEED input and ouput caps. 'Course, there's a whole 'nother philosophical argument about whether you are getting any real bang for the buck when caps are required, but let's leave that for another day...

As long as the power is isolated, most power supply, it doesn't matter at all.

 

[tangent]

Before I post some data, some comments:

1. For 1/8" jacks, you can measure DC offset between the "sleeve" part (the long bit closest to the part you grip) and either the tip of the plug (left channel) or the "ring" that's between them (right channel).

2. I propose that people give both left and right channel offset. As JeffreyJ's pointed out, it is almost always different. You can give absolute values for both voltages -- the magnitide of the offset matters much more than its polarity.

3. DC offset can change for a bunch of different reasons. Different units of the same productin run will even have different offsets, and they'll all diverge from each other to different degrees based on temperature, operating voltage, load, etc.

4. Jiggler, I hope you understand that no matter how many people post here with low DC offset values, it won't prove that DC blocking caps are unnecessary. Assuming that your design will work DC coupled when all of the components are working properly, all this thread will prove is that you don't need a blocking cap as long as everything continues to work correctly. A DC blocking cap is a safety issue in case of a failed component more than anything else.

5. Jeffreyj, it's true that you need input and output caps with a single-ended supply, but only if you don't create a virtual ground, as so many of the amps discussed here do. If you have a virtual ground, it doesn't even have to be stable for DC offset to remain low, because the supply is isolated, so everything just floats nicely around that virtual ground -- everything "believes the lie" of the virtual ground, no matter where it lies between the battery's terminal voltages, within one reasonable limit. The limit is, if the supply voltage becomes very low (like 1V or so) all op-amps I've tested for this will lose control and DC offset will begin to rise quickly. But, that's not a common occurrence with the high battery voltages we typically deal with. It only happens when you do something stupid like leave a battery-powered amp on all night with the headphones plugged in.



Now, my data, all measured with a Fluke 189, rounded to the nearest tenth of a millivolt:

Panasonic SL-CT570 PCDP: 0.5mV on both channels

Pioneer DV-444 DVD player: 0.5 and 0.2mV

HanGo PJB100: 0mV on both channels

Audiotron TV-101: 0mV on both channels (behaves (and sounds!) like it has an output capacitor)

Music Hall CD-25: 0mV on both channels (probably not an output cap on this one -- DC servo, maybe, or just very good components)

 

[jiggler]

Tangent,

How do you like that Music Hall player?

 

[tangent]

Jiggler, you're not supposed to topic-drift in your own thread!

I like it fine, but since I have heard nothing else in its class or higher, my opinion of it isn't very reliable.

 

[jiggler]

Your right Tangent - Sorry about that.

This next question is on topic - Don't the Corda headphone amps lack input capacitors? I thought I read somewhere that this was so.

 

[jeffreyj]

Quote:

Originally posted by jiggler Definitely! Oops... As long as your Cmoy does not run off of a single-ended power supply!! If you aren't using two 9Vs in series with the centertap as ground, or a 9V split by a TLE2426, etc., then you NEED input and ouput caps. Jeffreyj, Would you please explain this comment. I have never heard or read anything like this before and it does not make any sense to me.

Sure... Mind you, this is a nearly ubiquitous design, just not one you often see in reasonable hi-fi equipment.

To reproduce the positive and negative portions of an audio signal you can either run your amplifier from positive and negative supplies (with respect to a centerpoint... ground) or you can refer one input of your amplifier circuit to the mid-point of your power supply. If, e.g., your power supply is a 9v battery, you would refer one input to a voltage divider splitting the supply in half, or 4.5v. That means that with no signal present, 4.5v will be present at the output. Not to many speakers, and no headphones, I wager, will appreciate this, so a capacitor must be inserted between amplifier and load to block this DC. Now, only varying signals will pass through, but with a phase shift proportional to the capacitance and the frequency, and a signal loss proportional to the dissipation factor, dielectric absorption, ESR and ESL (equivalent series Resistance and Inductance, respectively). That explains the output, but what about the input? Well, without going into a long dissertation here, let's just say that the two inputs want to be at identical levels when nothing is going on, so if you bias one input to 4.5v, guess what... the other input will assume that value as well. Oops - bet that source farther up the chain will hate that! So, a DC blocking capacitor is required on the input as well. Fortunately, because the size of the capacitor necessary is proportional to the load impedance, and the input impedance of most amplifiers is high, said capacitor won't need to be so large in value, so high quality polypropylene film units can usually be used, instead of unbelievably bad electrolytics. But you still get the phase shift with frequency, and that is not good.

Regards,
Jeffrey

 

[tangent]

I was teasing you, Jiggler, not chiding you.

And to your second item, it is true.

 

[jeffreyj]

Quote:

Originally posted by tangent 5. Jeffreyj, it's true that you need input and output caps with a single-ended supply, but only if you don't create a virtual ground, as so many of the amps discussed here do. If you have a virtual ground, it doesn't even have to be stable for DC offset to remain low, because the supply is isolated, so everything just floats nicely around that virtual ground -- everything "believes the lie" of the virtual ground, no matter where it lies between the battery's terminal voltages, within one reasonable limit. The limit is, if the supply voltage becomes very low (like 1V or so) all op-amps I've tested for this will lose control and DC offset will begin to rise quickly. But, that's not a common occurrence with the high battery voltages we typically deal with. It only happens when you do something stupid like leave a battery-powered amp on all night with the headphones plugged in.

It is true that slow changes in overall supply voltage will not affect the quiescent point of the output, but fast changes will pass through owing to the less than infinite PSRR of any amplifier circuit (op-amp, discrete, whatever). Asymmetrical loading of the power rails can also lead to some bouncing around of the virtual ground, which no amount of PSRR will fix - changes here will appear as a valid input signal!! But this all argues for DC blocking capacitors, and I'm supposed to be arguing against them!? Sheesh... Agreed, then, that the input blocking capacitor is a necessary evil... alas.

However, end users never do anything so stupid as to leave the amp turned on with the headphones plugged in until the batteries run out, so this is not a valid reason for including an output capacitor

 

[tangent]

If anything, Jeffrey, that's just an argument that virtual grounds are imperfect. Yeah, we know.

 

[ofb]

all right, this is for grins. i'm using a bottom end digital meter measuring nasty old gear. all readings had some fluctuation on the 200mV scale, and the 2000mV scale wasn't much better.

how sad is the meter?

leads together, 00.0, stable.
waving leads in the air, 00.5, fluctuating.
held by fingers, 07.0, fluctuating.

so...

nikko nd-590II (tape deck), fluctuating wildly, R -1.6 +1.8, L -0.4 +0.6.

technics sa-200 (receiver amp & monitor stand), R 69.4 L 77.9. wow.

technics sl-xp7 (1985 'portable' cd), line-out R 9.4 L 0.4, headphone R -5.0 L 0.7.

sony prd-150 (early cdrom discman), line-out R 0.0 L 0.0, headphone R 0.7 L 0.0.

sony wm-af54 (classic sports wm), ... all over the place. no idea what's going on here. anywhere from R -80 L +80 down to R +9 L +7, depending what minute you test. seemed to be slowly climbing and descending. often R & L were the same except - & +.

mongrel music computer with some SB card, R 0 to 80, fluctuating wildly, L 0 to 4 also fuctuating more than normal, but not as wild.

 

[tangent]

Quote:

how sad is the meter? leads together, 00.0, stable. waving leads in the air, 00.5, fluctuating. held by fingers, 07.0, fluctuating.

That's normal for a digital meter.

Quote:

fluctuating wildly

Did you measure with music playing, or without? Most any meter will fluctuate with music playing, but it should be steady (if the DC offset is steady, of courrse) when the player is paused or stopped.

 

[ofb]

Quote:

Originally posted by tangent: Did you measure with music playing, or without? Most any meter will fluctuate with music playing, but it should be steady (if the DC offset is steady, of courrse) when the player is paused or stopped.

doh!

ah, well, some things changed and other didn't.

nikko nd-590II, fluctuating , R 0.4 to 0.7, L 0.4 to 0.7.

technics sa-200, R 66.0 L 79.0. stable.

technics sl-xp7, line-out R 0.0 L 0.0, headphone R -5.1 L 0.6, stable.

sony prd-150, line-out R 0.4 L 0.2, headphone R 0.4 L 0.0, stable.

sony wm-af54, ... starting at -340mv and slowly decending, both channels negative this time, and decending in sync. note, this unit doesn't have a 'pause'. it's technically 'off'.

mongrel pc with SB card, R -0.2 to +0.7, L 0.2 to 1.1. merely fluctuating, not wildly.

bummer about the sa-200; i still use it. just how dangerous is that for my headphones?

 

[tangent]

Quote:

just how dangerous is that for my headphones?

It depends on the headphones, and on the amp that the offset is travelling through.

If the amp strips DC, then there is no problem. If not, then you have to consider what the amp does with it. If the amp adds no DC offset of its own (unlikely), it will simply magnify the offset by its gain. You have to consider the entire system gain, including the attenuation from the volume control. For a line level source, the overall system gain might be 1 or less, so the DC offset in that situation would be the same or lower.

Now, in all likelihood, the amp will have its own small DC offsets. If it has a negative DC offset and you run a positive offset through it, they'll cancel out, and leave you with a lower overall offset. But, they can also add, if they're in the same direction, leading to a greater overall offset.

There can be other wildcards in here. For instance, some bass boost circuits will also boost DC voltages (i.e. they have no low-frequency cutoff) so DC can be boosted disproportionately to the majority of the signal. Or conversely, there can be various filtering stages in the amp that reduce DC offset without eliminating it.

Now, to the core of your question: what happens if you put 60 mV into headphones? Think about how headphones work: alternating currents (AC) move the diaphragm back and forth in some analog of the output signal of the amplifier, and this moves air, which you perceive as sound. A DC offset is nothing more than telling the headphones' diaphraghm to go one direction and stay there. The magnitude of the offset tells the diaphragm how far to move. A large DC offset tries to move the diaphragm a longer distance than a small offset, and at some point, the diaphragm can't take the stress, and it breaks down.

That's all for a pure DC offset. If there's an AC component riding on top of the DC, then what you get is that the diaphragm doesn't return completely to its resting point: it's always extending one direction or the other. If the DC offset is high enough, you get something very much like clipping, where the diaphragm can't extend one direction far enough to complete the wave the amp is telling it to try for, and so you get these harsh clicky/crunchy sounds as the diaphragm continuously tries and fails to do what the amp is telling it to do.

Try it with some throwaway headphones some time. It's educational.

How bad all this is depends on the headphones. For Grados, 60mV is a very high fraction of their normal ~300-500mV full-scale signal level. (I mean that for most people, this is their loudest listening level with these headphones.) For some headphones, it takes 2V to make the headphones put out a full rockin' volume level, so 60mV is less significant.