I don't understand Apheared's 47 output resistors

[doctorkelsey]

I have tried to find a good explanation to why the apheared 47 requires the 47 ohm output resistors. Since the second opamp is acting like a buffer, it seems like having the extra output impedence goes against what PPL always says about ouput resistors on buffers. Is this just a short-circuit protection for the opamps (when you unplug the headphones), or is there something I am missing??-- such as a really good understanding of opamps

Thanks!

 

[MisterX]

The "buffer" is paralleled and without the resistors the two opamps will end up fighting and one of them will eventually loose.

 

[doctorkelsey]

So, buffers don't need output resistors because there is only one thing supplying the output... the buffer. I guess my follow up question is, is it always required to use output resistors when stacking multiple buffers... since, again in this scenario, more than one component is supplying the output?.

 

[Porksoda]

With open loop buffers, no. It is necessary with op-amps because their closed loop output impedence is on the order of microohms. With an output impedence that low, only a few millivolts of offset can lead to huge currents between op-amps in paralell. With buffers, they are usually open-loop so their output impedence is a couple ohms. In this case, a few millivolts of offset results in a milliamp or so of current between them, not too big a deal.

 

[moeburn]

Since the gain of the A47 was high enough for me, I used 280 ohm resistors, four of em.

 

[doctorkelsey]

Got it. Thanks for clearing that up!!

That brings to mind another relavent question. I have recently expermented and realized that it is not possible to "stack" TLE2426 rail splitters to attain higher power supply currents, due to basically the same reason opamps need output resistors when in parallel: the slight differences in output voltages creates a relatively huge and wasteful current. My question is could you double up on the rail splitters if you separated the output by a few ohm resistor and still get significant increases in total output current capability? I attached (tried to at least) an image of what I mean. Would this work, and be a reasonable (or at least cheaper) option to supply the 50 or so mA required to power many A47s and cmoys, instead of using a BUF634. Say the TLE has a current limit of 25mA or so, would this design still supply roughly 50 mA?

 

[moeburn]

Why do you need so much current, are you experiencing clipping? In my experience, if it isn't clipping at the loudest volume you can tolerate, you don't need more current for better sound quality, you need different opamps.

 

[doctorkelsey]

moeburn
just brainstorming here. tangent talks about how a rail splitter is a good upgrade for a cmoy but it may not have enough current. I am wondering if this (previous post) is a viable option for poor people. In my small amount of testing it seems like 50 mA can comfortably supply these smaller amps. My question is whether or not it would work. I'm not a real guru on basic circuits.

 

[amb]

doctorkelsey, using a resistor on only one TLE2426 but not the other creates an unequal load situation, so basically the one without the resistor will still bear the brunt of the current, making the "stacking" less than effective. Putting resistors on all of them is not a great idea either, because that increases the output impedance of each, and since the TLE2426 contains an open-loop buffer that does not employ negative feedback correction, the virtual ground voltage will fluctuate with the load current due to the added resistance. Obviously this is bad.

Simply put, the TLE2426 is good for low-current applications. If you want to have a ground splitter that could handle more current, you'll need a different design.

 

[Porksoda]

Also, I would point out that the PPA actually DOES use three TLEs in paralell, and people seem to think it works great...

My recommendation would be to set an opamp to unity gain and use it as a buffer for the ground channel, or use a BUF634. If you were to use a 4134 op-amp, you could do the whole thing with one chip, though it wouldn't fit on the standard CMoy footprint.

 

[moeburn]

I'm using an unbuffered rail splitter on my pocket amp, if you have buffers they're much better off on the outputs.

 

[amb]

Quote:

Originally Posted by Porksoda Also, I would point out that the PPA actually DOES use three TLEs in paralell, and people seem to think it works great...

The PPA has three TLE2426s driving a common ground, but each is off of separate JFET-isolated rails. The JFETs provide some "slack" for the three TLE2426s' tolerances. So, in a sense, they are not truly "in parallel".

 

[mono]

Regarding the TLE railsplitter, it is not providing current for the amp. It is only providing a virtual reference point to the BUF634, which itself provides vGND. You would have absolutely zero benefit from stacking them even if it were possible.

Why do without BUF634? If you really must, you could use an opamp instead... but you're not reall building an A47 at this point, it can't really be considering one if all you're doing similarly is pairing up two opamps in pseudo-parallel.

 

[moeburn]

Quote:

Originally Posted by mono Regarding the TLE railsplitter, it is not providing current for the amp. It is only providing a virtual reference point to the BUF634, which itself provides vGND. You would have absolutely zero benefit from stacking them even if it were possible. Why do without BUF634? If you really must, you could use an opamp instead... but you're not reall building an A47 at this point, it can't really be considering one if all you're doing similarly is pairing up two opamps in pseudo-parallel.

I left out the buffered vGND in my A47, and it sounded just the same. I really don't see the point in buffers unless you're only using one amp per channel, or something stupid like an LM386.

 

[doctorkelsey]

mono,

I should clarify... I am questioning whether or not you can 'stack' tle2426s when they are the primary current route. (when not using a BUF634) Tangent talks about this option. The drawback is the low current rating of the TLE... hence, I was wondering about stacking posibilities-- as an upgrade for a cmoy. The reason I thought this question was relavant is because of the similarities in some of the problems that arise when you 1) try to connect two opamps in parrallel without output resistors or 2) try to run two TLEs in parrallel without some sort of output resistor. The problem is: the small fluctuations in output voltage in either of these devices causes big wasted current flow because the outputs of the opamps and TLEs have very low impedance. I agree, there is no reason to "stack" when using a buffer. Also, as it turns out, there isn't a good way to stack them in true parrallel in any situation (according to amb).

"Why do without a BUF634?" Just a space issue in a cmoy I made. Also, to save $8. (I'm not really a doctor yet)

Quote:

but you're not reall building an A47 at this point, it can't really be considering one if all you're doing similarly is pairing up two opamps in pseudo-parallel.

I'm not sure what you're refering to.


moeburn,

I think (correct me if I am wrong) the TLEs are actually rated at 30mA maybe even 20mA. My A47 with two opa2134s gain 2 at full volume into my grado sr-80s sucks about 45-50 mA and a no load current of 20mA. So, I think eventually if I didn't have a buffer, I would fry a TLE. Now with a cmoy, much less current is needed, so I can see maybe using a single unbuffered TLE for that- but it still isn't ideal current output.