The last two make a great combination for buiding a headphone amp, obviously . I took interest in the cmoy amp because of it's small size, relatively high quality, and low price. I'm looking to build one, but of course I want to follow some of the tweaks, because I want as good a quality as I can get. Of course, going down this road, I have some questions.

Just a note: I have done significant work with hardware, such as designing/building (relatively) simple circuits, and a lot of soldering work, but never before have I really done audio hardware work (except for replacing the cables on my headphones when they died). This amp will be my first endeavor into that field.

1. Op-amps. I came into possession of a few OPA2132PA's and a few OPA2277P's (don't ask, don't tell), all 8-pin DIP. There are tons of reviews everywhere for the 132's, and I'll probably try to get those to work first. However, I have had a bear of a time finding people's opinions on the 277. Anyone tried these out?

2. Power supply. I'll probably end up using batteries, simply because of all the horror stories I've found (here in particular) about wall-supplied sources and their noise levels. However, will one 9v battery be good enough for the amps listed above? Can I just swap in a second (in series) if I decide (in the future) to upgrade to a 627/637? Also, I'm not quite sure which virtual ground circuit to use. I was looking at this page, particularly at the TLE2426 design. However, it says that the output current is pretty limited. Will that be enough to supply my circuit (one of: the two above listed amps or a pair of 627/637's)? Should I go with another design?

I think that's all I'm really confused about right now - I might have more questions later, of course .

I'm trying to drive a pair of Grado SR-60's, btw.

thanks,
poly-p man

The OPA2132 is FET-input, whereas OPA2277 looks like bipolar input. That means the later one needs larger input bias current, which could cause too large voltage offset at the output of your amplifier.

> Power supply. I'll probably end up using batteries, simply because of all the horror stories ... about wall-supplied sources

9V should be OK. However, I don't think wall power is that dangerous. I have one off-the-shelf switch-mode wallwart, one off-the-shelf regulated linear wallwart and one simple "capacitance multiplier" filtered power supply I built myself. All of these work just fine for my audio needs. I am sure they do not provide the highest fidelity possible, but simple cmoy amp is not likely to be in that category either.

> ...which virtual ground circuit to use. I was looking at this page, particularly at the TLE2426 design.

Yes, TLE2426 alone might not be optimal solution for Grado load. A resistor divider followed with a bigger buffer than TLE2426 contains would probably be better option. These are discussed in that Tangent's article too. You might also want to check this thread, post #24 in particular.

---

[Quoting kvant]

> Op-amps. I came into possession of a few OPA2132PA's and a few OPA2277P's

The OPA2132 is FET-input, whereas OPA2277 looks like bipolar input. That means the later one needs larger input bias current, which could cause too large voltage offset at the output of your amplifier.

---

---

[Quote]

> Power supply. I'll probably end up using batteries, simply because of all the horror stories ... about wall-supplied sources

9V should be OK. However, I don't think wall power is that dangerous. I have one off-the-shelf switch-mode wallwart, one off-the-shelf regulated linear wallwart and one simple "capacitance multiplier" filtered power supply I built myself. All of these work just fine for my audio needs. I am sure they do not provide the highest fidelity possible, but simple cmoy amp is not likely to be in that category either.

---

---

[Quote]

> ...which virtual ground circuit to use. I was looking at this page, particularly at the TLE2426 design.

Yes, TLE2426 alone might not be optimal solution for Grado load. A resistor divider followed with a bigger buffer than TLE2426 contains would probably be better option. These are discussed in that Tangent's article too. You might also want to check this thread, post #24 in particular.

---

Wow, some of the topics in that thread are a little, um, advanced for me

Anyway, (checks some numbers), the amount of mA to drive my 32ohm headphones at .5v (arbitrary value derived from tangent's op-amp page) is... 15.6mA - which is pretty close for the TLE, especially since that doesn't account for other components. So maybe I should consider the BUF634 method? the op-amp method? Could I use the extra OPA2277's for that? The OPA2132's? Anything else?

Thanks so much!

poly-p man

Actually, it is over the limit, since you need to dump current from both channels. As for the BUF634, you will find that it is a bit expensive and also seems out of stock at many places right now. Therefore, it is probably better to go with the opamp method. However, according to datasheets for the parts you have, their maximum output current is 35 resp. 40mA, which does not provide that much headroom either. On the other hand, they are dual parts and there is a way to parallel both halves.

If I was building virtual ground like that, I would probably go with OPA551 as it can give you up to 200mA. There is a downside, though, as it might like more voltage than single 9V battery (absolute minimum for it is stated as 8V, and batteries discharge...). In fact, I have built an amp like that - OPA551 in the virtual ground as well as in L/R channels (I had bunch of those opamps). Works just fine with Grado, but I didn't try to go that low with voltage, I was feeding it some 20V+.

Of course, I haven't yet taken any formal electronics courses, and my knowledge of op-amps is a tad limited, but should this do what I want? (Please excuse the awful drawing - I did it basically freehand in inkscape.) Can I power the op-amps in series, or is what's shown the best I can do?

EDIT: Just realised why that's stupid - the dual part only has one voltage in/out

EDIT2: Fixed link above for new circuit.

Does it matter which op-amp I use?

Also, just so I understand the circuit a bit more - the two resistors (R1 and R2), are they there to do really basic voltage splitting? In my head, they do voltage splitting for the op-amp, which then powers the rest of the circuit, like a buffer. Correct?

BTW, if it makes sense, I can go for two batteries in series. Should I?

Thanks very much for your help!

poly-p man

[Edited by poly_poly-man on 02-21-2008 at 06:54 PM.]

> In my head, they do voltage splitting for the op-amp, which then powers the rest of the circuit, like a buffer. Correct?

Correct.

> BTW, if it makes sense, I can go for two batteries in series. Should I?

I think that higher voltage like that can only improve performance. You can always experiment and decide afterwards, what works better for you.

> So basically, don't use it.

No, basically you don't rely over-much on free advice found on forums, even from a sharp dude like kvant.... you look it up and figure it out yerself.

OPA277 datasheet http://focus.ti....ink/opa2277.pdf shows input bias current of 3 nanoAmps. One microAmp in 1 MegOhm is 1V which is a lot. But 3nA in 1Meg is 0.003V which is small. Most audio circuits can use bias resistors of 100K or less, now we are down to 0.000,3V or 0.3mV which is negligible. To this we must add up to 0.1mV of offset voltage error... still negligible.

In fact DC performance of OPA277 is frikkin great for most audio use.

What I see is modest high frequency performance. GBW is 1MHz, slew rate is 0.8V/uS. Same-as a '741. True, we used to use bag-fulls of '741 op-amps. For low gain and low voltage levels, they are still inoffensive. Might be "OK" in a headphone amp. But far better chips can be bought for just a few bucks.

Output drive is another question. OPA277 is rated 35mA into a short, not specified to work good for loads under 2K. (Again, it seems to be a '741 but with fabulous input DC performance.)

The '2277 would be a dandy DC servo amp.

FWIW: the old "FET vs BiPolar" line is fading. FETs can have startlingly high gate current at high temperature where BJT tends to decline (maybe not relevant for personal audio). The '177/'277 have a low intrinsic bias current (similar to '741, which echoes its '741-like AC performance) then adds a bias-current compensation technique which cancels 99%-101% of that bias current. (Some trim is done by lasering the die.) '741 bias current was rarely any issue in audio, and this is 100 times better. At the input... the output is still '741-like and we now prefer better devices for most audio.

OPA2132 has even lower bias current, but the '277 is already "plenty good". Bandwidth and slew at like 10 times better... if '741 was almost good enough for audio, these are fine audio devices. Performance is not specified for loads under 600 ohms, and max current is only 40mA: this is not an authoritative 32-ohm driver, though many people are perfectly happy.

FWIW: "FET input" '2131 has higher bias current than "BJT input" '277... when temperature is 100 deg C or above. Yes, we could not touch the knob while the amp was at boiling water temperature.

'2131 has amazingly lower THD specs than '277. The '277 numbers do not stink, and the '2131 zero-zero-zero specs look like a typo (but are probably genuine, under specific conditions). THD is a poor predictor of audio accuracy or pleasure.

Why don't you build the amp with OPA2132? It is very easy to use because I know it can withstand anything (other than electro-statics ... you know sweaters in dry-weather thing.)

DC offset stuff is very difficult to understand unless you play with the actual thing. I myself didn't believe it till I saw it on a multimeter. (But the calculation is fairly easy. Input offset current multiplied by bias impedance (sorta) )

This problem is rather insignificant if you are willing to bias the opamp with low resistance. I think it would cause you much problem unless you have TUBE CD player or TUBE preamp. (or a very, very special solid state preamp) In fact, 1k ohm seems to work good enough with my MP3 players, computers, and CDP. ... I have seen many CMOY-like amp with NJM4556 and NJM4580 in this manner. (Being Japanese, I see these more often.)

However, I vividly recall when I was a noob, I felt deathly uncomfortable with non-zero offset. So that is why I think you ought to go with OPA2132. On top of it, I am obsessive-compulsive so I think I spent a good part of a month once.

Being a Gentoo user yourself, I think you are also obsessive-compulsive. (Or I would usually call people like you "crazy". Debian seems more reasonable for me. ... Lately I wimped out more and went Ubuntu. Super-Noobified.) Heh.

Tomo

Yes, the dc offset scares me - I've seen reference that too much can make a good pair of headphones explode - I don't want that to happen.

I'm only a *bit* ocd - I mean, just because I spent 4 hours going through USE flags and one hour deciding on CFLAGS... and uncountable hours trying to figure out what circuit to use in this cmoy

I got a bit distracted from the project for a little bit, but I will be ordering the last few parts that I need, probably today.

poly-p man

You might consider building it on a breadboard (ie., plug parts in or pull out to replace) so you can try various values for components.

I built the Linkwitz-CMoy-Tangent crossfeed circuit (without the on'off switch) and I like it a lot with the CMoy. Tangent sells printed circuit boards.

There is an infinitude of alternatives.

F