High voltage capacitors in audio

I thing misunderstood the purpose of the "Z"s. They mark locations where the impedance will be non-zero.

The headphones hook to the GND on the right. There should be no "Z" in that path.

The only ground in the output section is the return path of the headphones. What you suggest here is effectively what I've done. The regulators may be bolted to the chassis forcing the issue somewhat on exactly how the wires go.

For mechanical reasons, the storage capacitors are likely to be some distance from the regulators. In a lot of ways, it would be best if they were nearer the driver stage than the output pair, but mechanically this doesn't make sense.

Yes, I was just making it clear.

You could use resistances if you had to. I would suggest lossy inductors, if cost is no object.

Working left to right.

The ones at the left side of the drawing, reduce the peak current in the bridge, and block RF. It is a serious bummer when your new amplifier circuit starts picking up the nearby radio transmitters. One path that radio signals can use to get into the circuit is up the power cord. For this reason, you want a more impedance at 1MHz than at DC.

The impedances after the bridge further suppress RF and also help to spread the current pulses from the bridge out. They force the RF to take the path through the capacitors on the bridge. The ground side of the bulk capacitors is likely hooked to the chassis ground and is certainly hooked to your signal source. You want to keep RF from going that way.

The impedances on the power to the regulator is to keep the high frequency currents of the output section's connections confined to the bulk capacitors. Since the two capacitors on the regulator are physically right at its body, there is a path from the input of the regulator through its 0.1u to its ground leg and then back up onto the regulated power by the 100u. Without the impedance in the way, this high frequency stuff could get to the front end power supply.

The Zs are just a way to consciously remember that every piece of conductor has resistance and inductance, and so, will produce voltage drop in response to current and changes in current, respectively.

Note that I said "ideally". I was suggesting "Z"s be placed on purpose. This can be just the way the wires go, small value resistors or even lossy inductors.

It doesn't matter much how they are done, it only matters that something like that be the case.

The ones over at the left are to keep RF out as well as to spread out the current in the bridge. They are likely to be actually resistors.

Sorry I misunderstood you. I thought you were just labeling the unavoidable impedance of those current paths. Have you designed audio or other equipment in which you actually added those components?

I have added a capacitor across the secondary of transformers to lower the resonant ringing frequency that occurs when the rectifiers snap off, but not small series impedances.

Yes, partly.

In one case: The impedances at the transformer connections where 0.22 Ohm resistors. The impedances in front of the regulators were type 77

- 0.25 inch toroids. The others were just the wiring.

The amplifier in question had a gain of about a million over the 850Hz to 4KHz band. Without the 0.22 Ohms in the transformer winding, the harmonics near 2KHz caused troubles. Spreading the current pulse out decreased them enough.

In another case, we had several LM78XX regulators. The transformer and bridge were in another chassis. Here in the SanFransisco area, the AM radio station KGO seems to like to get into stuff. Keeping it, the mains harmonics, the noise from the data computer and clatter from other parts of the system out of the high gain electronics required that there be EMI filters on everything coming and going and several

The recovery time of the rectifiers is the main way that the bridge invents new frequencies.

Could someone tell me the function of the 22uF and 10uF capacitors around the gain potmeter? (After the first opamp)

I want to replace the gain potmeter by a gain switch, with 3 positions (that is, two resistors in series where the potmeter is now; the switch selects one of the points).

There should be a capacitor to reduce switching clicks (150nF or so), but where should it be placed?

I want to add a volume potmeter at the input of the amplifier, after the 10uF. But it would affect the filter (with the 1nF), wouldn't it?

As I suggested, the one feeding the pot serves no real purpose. The one after the pot makes it so that the 100K to ground is the only DC path to the second op-amp.

Keep the capacitor and 100K to ground after the switch. This way the volume goes to zero if the switch connections go open. Letting the input of an opamp float is a bad thing to do.

BTW: Why only 3 positions?

If this is a real mechanical switch, you are doomed to have switch clicks. If there is a signal on one side of a switch and none on the other, you get a click when it is operated just because the voltage steps from zero to where the waveform currently is.

Why not leave the volume control where it is?

Yes.

The 'volume' control is actually a gain control, it doesn't offer volume 0%-100%. As the author says "The gain control being used to pre- set the gain so that the pre-amp's gain control is normally used for setting the listening level.". Gain might start from 100%, maybe less, but not 0% (silence).

I don't use a preamp, so I have to add a volume potmeter. I don't see sense in having two potmeters (one volume, one gain) do about the same thing, so I figure gain could be a 3-position switch which makes a lot more sense (I could also use a 2 position or 4 or 5... but 3 I think will be enough).

I think the volume should be in the first stage before the first opamp. However it would certainly affect the filter around the 1nF in some way, I'm afraid.

Yes.

The 'volume' control is actually a gain control, it doesn't offer volume 0%-100%. As the author says "The gain control being used to pre- set the gain so that the pre-amp's gain control is normally used for setting the listening level.". Gain might start from 100%, maybe less, but not 0% (silence).

I don't use a preamp, so I have to add a volume potmeter. I don't see sense in having two potmeters (one volume, one gain) do about the same thing, so I figure gain could be a 3-position switch which makes a lot more sense (I could also use a 2 position or 4 or 5... but 3 I think will be enough).

I think the volume should be in the first stage before the first opamp. However it would certainly affect the filter around the 1nF in some way, I'm afraid.

It looks like google lost my reply.

The pot should he on a high amplitude low impedance signal. Putting it after the first op-amp is the right place to have it.

The circuit can be reduced to a simplified model that looks like this:

In ---- 1M ! --/\\/\\--- \\ 100K ! ! /---+-+---- Vout ! GND-----!+/ ! ! ! --------------------------

A cute trick is to make a spice model of this, apply 1V DC at the input and step the pot setting. This way you can make a graph of the gain vs pot setting. I think you will find that it follows a curve very like the one you want for a volume control.

You could also plug the math into a spreadsheet.

The math looks a bit like this:

Let X be the pot setting:

Vth(wiper) = Vin * X + Vout * (1-X) R(wiper) = 100K * X * (1-X)

Vout = -1M / (R(wiper) + 100K)

I'm too lazy to do the subst and solve for Vout but it will increase much more rapidly at the top of the dial than at the bottom.

This is quite nice for a panpot, a bit of positive feedback works wonders, but I never got around to trying it

martin

That looks like too much positive feedback. Are you sure you transcibed the values correctly?

(I also lost my post.. Luckily I know how to trust crappy Google groops, I saved my text message locally before sending, so here it is again)

===

Funny you mention spice just now: I went to simulate the whole circuit (except for the transistor output stage) in PSpice 9.1. I took the exact circuit with all component values as in the design, but I took the output of the last opamp as output for the feedback (because I left out the transistors as I said).

There's nothing above simulation I think... because I found out I was wrong! The gain potmeter, as is, *does* provide a volume setting

I put 300mV at the input ('VAC' in PSpice) (over the 220k resistor), and get the following voltages at the last opamp's output (potmeter %)

The bandwidth of the whole thing is about 5Hz to 40kHz (flat).

I h> A cute trick is to make a spice model of this, apply 1V DC at the

I'm not sure yet how to do this... it would be useful, yes, because now I manually adjusted the potmeter & generate a graph each time.

Thanks, E.

Yep, it's screen capture of a service manual (soundcraft), and apparently patented,sorry, lost the references

martin

On Jun 29, 8:06 pm, ectoplasm wrote: [...]

I use LTSpice. It works quite well for this sort of thing. It has hacks that were intended to make DC-DC converter stuff work better. I think they also help to make class B output stages converge.

This is a nice curve for a volume pot. Now you just need to make a nice little dial for the pot with numbers up to 11 and you will have a very good headphone amplifier.

Manual net list:

• A POT is two resistors R1 Node1 Node2 {100E3*(1-X)} R2 Node2 Node3 {100E3*X}

*Note resistors can never be zero ohms

• Do an operating point computation .op

In LTSpice, you can enter this on the schematic.

Thanks, MooseFET. I am trying to figure it out in PSpice. I didn't succeed yet with potmeter variations for plotting frequency response, or for any attribute. Luckily there are numerous tutorials available for PSpice. It's fun!

Just to report back:

I built this headphone amplifier; finished it about a month ago now. In terms of noise: it is dead silent... and I really mean DEAD SILENT. It has incredibly good sound, and it goes incredibly, incredibly loud; and that's nice because I also like loud music styles. There is no audible distortion. I use Sennheiser HD480-II and Koolsound HD-627 headphones.

Basically, it is this design:

I just made a few changes / additions, with the help of everyone in this thread; thanks! (excluding the few audio-phoolery bashing trolls.. it's all in *your* minds :-D ) Here are the changes and additions I made:

- Added a bass enhance stage, an extra 5534; so I used 5532's. The

- Added a 2nd order rumble filter in the input stage, fT under 10Hz. This is useful when using a record player esp. when bass enhance is on; DC coupling capacitors 10 and 100uF at input stage became unnecessary.

- The 470pF at the opamp input is omitted

- Output transistors, I used the Toshiba 2SC5171 (NPN) and 2SA1930 (PNP)

- Additional 100nF capacitors at each opamp IC's supply voltage pins, to ground (for both pos + neg voltages)

- Attenuation switch, an optional -24dB. Useful with my low impedance / high dB/mW headphones, you use more volume pot range which is handy.

- Additional elco's around the 7815/7915 that I saw recommended in one of these regulators' data sheet