[bg4533]

I am fairly new to DIY, but always have to try to figure out how things work. I don't know a whole lot about this stuff, but I pick up on things pretty quickly. I have built a Pimeta, Mint and a few CMOYs. I am currently in the process of modding my sound card.

I have a few questions, the first are just general for my own knowledge, the second about the Pimeta. This seems to be a somewhat complex topic, so I am looking for answers that are somewhat technical but still understandable. If you have any good papers or anything on this topic feel free to post them.

First, I have been reading a lot about opamp bypass capacitors here and at diyAudio. It seems most people here are not too concerned with opamp bypass capacitors. The people at diyAudio (or a few very vocal ones) are fanatical about them. It is my understanding that these caps are to provide quick power to the opamps and are more important as opamps get faster. Are there any other reasons? There seems to be 2 ways of bypassing, rail to rail and rail to ground. What are the advantages and disadvantages of each? I don't see how rail-rail bypassing would fit with my earlier definition of fast power, so what does it do? When bypassing rail to rail what rating capacitor would I need, greater than V or 2V?

Finally, the people at diyAudio mention that the OPA627 (I imagine the OPA637 even more so) needs a lot of close bypassing. I am using the 6.8uF Wimas for the Pimeta C4, but they are kind of far from the opamps, especially when you consider the browndog adaptor. Could the Pimeta benefit from a small bypass cap? Would rail to rail, rail to ground or both be better?

Thanks,
Brian

 

[llmobll]

You raise a very good question, one which I also am looking for the answer. Anyone?

 

[Earwax]

Quote:

Originally Posted by bg4533 First, I have been reading a lot about opamp bypass capacitors here and at diyAudio. It seems most people here are not too concerned with opamp bypass capacitors. The people at diyAudio (or a few very vocal ones) are fanatical about them. It is my understanding that these caps are to provide quick power to the opamps and are more important as opamps get faster. Are there any other reasons?

In various opamp datasheets I've found mention that the bypass caps are used in order to shunt power supply noise to ground. But when I bought this topic up on headwize, I was told that the caps are used to stabilize the opamps. /shrug

All I know is that they did indeed do good things for a few of the opamps I used in a couple of cmoys. (read the datasheets to pick what value to use for various opamps)

 

[tangent]

Quote:

In various opamp datasheets I've found mention that the bypass caps are used in order to shunt power supply noise to ground. But when I bought this topic up on headwize, I was told that the caps are used to stabilize the opamps.

Both statements are true.

The power pins of an op-amp can be thought of as inputs: anything that isn't DC is "signal", and some of it is going to appear at the op-amp's output. This is characterized by the PSRR curve in the datasheet. Also, keep in mind that the impedance of your power supply is not zero. Since the current draw on that supply is not steady, tell me what Ohm's law says. V=IR, right? Think about the consequence.

That's mainly a low-frequency argument. At higher frequencies, things like parasitic inductance come into play. Any wire has inductance. The thinner it is and the longer it is, the bigger the inductance. So, without bypass caps, you have the bonding wires from the actual chip to the package's pins, the pins themselves, and the traces all adding inductive impedance to the circuit before you get back to the rail caps. Since the impedance of inductance increases with frequency, this gets you in trouble with the voltage drop issues brought up above. Adding a high-frequency bypass cap right at the chip negates most of this effect, because the impedance of a capacitor decreases (generally) as frequency goes up. The idea of bypass design is to get the two effects to cancel.

 

[roadbuster]

Quote:

Originally Posted by bg4533 Finally, the people at diyAudio mention that the OPA627 (I imagine the OPA637 even more so) needs a lot of close bypassing. I am using the 6.8uF Wimas for the Pimeta C4, but they are kind of far from the opamps, especially when you consider the browndog adaptor. Could the Pimeta benefit from a small bypass cap? Would rail to rail, rail to ground or both be better?

Quote:

Originally Posted by Earwax In various opamp datasheets I've found mention that the bypass caps are used in order to shunt power supply noise to ground. But when I bought this topic up on headwize, I was told that the caps are used to stabilize the opamps. /shrug

Quick tutorial

Quite simply, if a capacitor is used to bypass some circuitry, it is called a
"bypass capacitor." This is not an intrinsic characteristic of the capacitor
itself; whether a cap is a bypass cap or not depends on how it is used in a
circuit.

The primary use of a bypass cap is to provide a low-impedance path of AC
signals to a power or ground rail. Capacitors have a property where their
conductivity increases with a rising frequency. Thus, a capacitor will have a
low resistance to high-frequency signals, and a high resistance to low-
frequency signals.

Why is this useful? Consider a circuit where you need to supply DC power to a
number of circuit elements. If the power supply has some AC noise on it, this
may negatively alter the performance of the circuit. One way to circumvent
this problem is to use bypass capacitors. By judiciously selecting the type and
value of a capacitor, it is possible to shunt these AC components to ground in
order to provide a cleaner, DC signal to the circuit elements.

 

[jamont]

Quote:

Originally Posted by bg4533 Finally, the people at diyAudio mention that the OPA627 (I imagine the OPA637 even more so) needs a lot of close bypassing. I am using the 6.8uF Wimas for the Pimeta C4, but they are kind of far from the opamps, especially when you consider the browndog adaptor. Could the Pimeta benefit from a small bypass cap? Would rail to rail, rail to ground or both be better?

I wonder about this myself. If I use 2xOPA627 on BrownDog in my Pimeta, they get very warm and I see ringing on the scope when looking at the square wave response, both observations suggesting that the amp is oscillating. I also am using the 6.8uf Wimas in C4, would an additional bypass cap cure the behavior I observe?

I might also note that when using an OPA2132, the amp runs cool and I see no ringing on the scope.

 

[bg4533]

Quote:

Originally Posted by jamont I wonder about this myself. If I use 2xOPA627 on BrownDog in my Pimeta, they get very warm and I see ringing on the scope when looking at the square wave response, both observations suggesting that the amp is oscillating. I also am using the 6.8uf Wimas in C4, would an additional bypass cap cure the behavior I observe? I might also note that when using an OPA2132, the amp runs cool and I see no ringing on the scope.

Unfortunately I don't have a scope to test mine, but it is probably similar. When I used the OPA637 it did sound a bit worse than the OPA627, so it might have been oscillating a bit more. I know my buffers (stacked 4X, 50ohm resistor) get pretty hot, but I haven't tried feeling my opamps. I will later today.

If I want to throw a few bypass caps from power to ground which ground do I want to use? Can anyone recommend a specific point to ground to?

At the very least I might throw a .01uF ceramic from rail to rail and see what it does.

edit:
I just felt my OPA627s after the amp has been playing music for about 30 minutes. They are pretty warm to the touch, not burning hot, but a little uncomfortable. I think the buffers are hot enough to burn be after a few seconds.

 

[tangent]

Quote:

which ground do I want to use?

Input ground.

 

[jamont]

Quote:

Originally Posted by tangent Input ground.

Thanks for the suggestion. Can you comment on the behavior I reported above? Have you observed this, or is there maybe some special problem with my sample?

 

[llmobll]

Quote:

Originally Posted by jamont Thanks for the suggestion. Can you comment on the behavior I reported above? Have you observed this, or is there maybe some special problem with my sample?

i've got my pimeta w/ 2 X OPA637's and it sounds fine. I'm only using a 1uF on my C4 though. However i don't get distortion or oscillations at all. I'm using a OPA627 on the ground.

 

[jamont]

Quote:

Originally Posted by llmobll i've got my pimeta w/ 2 X OPA637's and it sounds fine. I'm only using a 1uF on my C4 though. However i don't get distortion or oscillations at all. I'm using a OPA627 on the ground.

Do your chips get hot? I'm not sure I actually hear any problem, but there is certainly a difference in how hot the chips are, and I see ringing on the scope.

BTW, I put in an AD8620 last nite and that runs cool as well, it appears to be just the OPA627 that is causing problems in my unit.

 

[Voodoochile]

It seems to me that the larger caps are nicer for fast power reserve, yet smaller caps are better for addressing potential ringing. If you don't have room for both, then perhaps socket them and try both ways, seeing which performs better in a particular situation.

How bad is the ringing, John?

 

[jamont]

The ringing is clearly visible on the scope, and the OPA627's run noticeably warmer than the AD8620 or OPA2132, but I can't really say that there is an obvious effect on the sound. From an engineering point of view, it's a sign that something is wrong, my guess is that it's a high frequency oscillation.

Do you think replacing the 6.8uf Wima with something smaller would eliminate it? It might be possible to tack some 0.1uf NP0 ceramics near the power pins, or perhaps onto the BrownDog adapter itself.

 

[bg4533]

It seems soldering the caps directly to the opamp pins would be the best way, but you would need 4 caps. If you solder them to the Browndog you only need 2. Big ceramics seem pretty expensive. Would a polypropolyne work alright here? I have a few extra .022uF coming. I also have a few .01uF ceramics, but only 1 extra. I guess I could put it rail to rail.

 

[bg4533]

I just dropped an LT1364 in my Pimeta. It sounds really good from the start. Currently it is second only to the OPA627, but after a bit of burn in and listening it might be better. Problem is that the LT1364 is burning hot. I can hold my finger to it for about a second before my finger feels like it is burning. I know that can't be right...