cmoy upgrade(s) - Page 3

^  Or a forum for people that can't stand when others offer a different opinion?   Suggesting something, spending the time to do so even if it doesn't agree with your ideal, tends to be the opposite of not giving a crap.  Is the ultimate goal to be an expert at op amps, or to end up with a good sounding headamp?

I just LOVE car analogies so here's one:  You can take an old GM car from the '80s, put stainless lines on it, rebuild the suspension with aluminum parts, put sports car rims and tires on, become VERY knowledgeable, even an expert on it, to have achieved perfection in the pursuit of making an '80s GM car the best it can be.  After doing so, it's still only a GM car from the '80s.  Can you apply some of what you learned towards perfecting another car?  Sure.  But you can't get back the time and probably didn't end up with what you were hoping for.

I think the point of starting out with a CMoy is to learn the concepts involved in amplifiers

without having to spend a lot of money or burn up a bunch of components. Sure, you

can make some pretty good sounding CMoys, but in the end you can only get so

much performance out of a chip that is not really design to be a headphone amplifier.

Because a CMoy is so simple, it is easy to see exactly what each component does.

Changing any one component can have a noticeable effect on the sound.

From the CMoy it is not too much of a step to move on to an A47 or add

a BUF634 (and variants) to the design.

Some of the most popular amp designs are basically a CMoy (the voltage gain stage) followed

by a buffer (the current gain stage). Wrapping the feedback loop around the buffer allows the

op amp to correct errors in the buffer and the buffer isolates the op amp from the load.

If you want to try a discrete design, then a Szekeres (as Nikongod mentioned) is a great place

to start.

The CMoy I use on a daily basis is a three channel design (I know, I know) with LT1363 on each

of the channels, a 9.6 volt rechargeable battery pack with low voltage indication and a double sided circuit

board with ground plane design. A 40 ohm output resistor on the L/R channels keeps the minimum

impedance seen by the op amp to something it is happy to drive. The gain is set at 5, the output resistors

give it an effective gain of 2.

It drives my Shure SE215 to more than loud levels...

Quote:

Originally Posted by drtturnip 

Oh that makes sense. Build something that doesn't work correctly but don't bother to figure out why. Maybe there should be a separate forum for those who know everything and don't give a crap about those that don't.

You have absolutely no idea how far off the mark that is. 

Its interesting that the people answering the question of "how do I fix this Cmoy" are always the same few. Its also interesting to see when the last time they built a Cmoy is (was?) and what they are working on now. So much for not caring about people who don't know much about electronics. 

Do you think that the best way to improve one's knowledge is to build the same thing over and over again, or to do something different? Doing the same thing over and over is just kind of repetitive and boring (IMO) where trying something new offers new challenges and ample opportunity to learn new things. Maybe you make some mistakes, or pick a project that just does not suit your tastes*, but mistakes are how we learn. 

The fact of the matter is that the Cmoy is an inherently limited circuit. Without resorting to a few specific op amps, that are kind of hard to use, the Cmoy only goes so far. 

*I would suggest deliberately building (or buying) something that you don't think you will like despite it being well regarded after you have a few projects under your belt. Force yourself to listen to it for a month or 2 - I found this to be a VERY rewarding experience & learned a lot from it. 

Quote:

Originally Posted by mono 

I just LOVE car analogies so here's one:  You can take an old GM car from the '80s, put stainless lines on it, rebuild the suspension with aluminum parts, put sports car rims and tires on, become VERY knowledgeable, even an expert on it, to have achieved perfection in the pursuit of making an '80s GM car the best it can be.  After doing so, it's still only a GM car from the '80s.  Can you apply some of what you learned towards perfecting another car?  Sure.  But you can't get back the time and probably didn't end up with what you were hoping for.

Why no love for the Buick Grand National?

Not perfect, but it does what it does well VERY well.

I was under the impression this was his first shot at building something. He calls himself a noob.  If not then my bad. I believe that a great way, perhaps the best way to learn, is to squeeze every bit of performance out even the most simple of circuits. Figuring out why things don't work is a fun way to learn. Triumph breeds satisfaction. Why move on until you have mastered something simple?

As far as what I have or have not done or am working on now, I have repaired,  built, modded, and and redesigned more guitar amps. effects and studio gear than I can remember without thinking harder than I care to. At the risk of sounding boastful I will leave it at that. I have had the great blessing of working in a very wide range in the world of electronics. I have always and will always learn something from the process, even if it is as simple as an op amp based impedance matching driver.

Quote:

Originally Posted by surajspai 

opamp circuit uses default 100k to gnd for the bias resistor,10k in the feedback loop,and 1k to gnd.and 47ohm in output

 

hiss is still present,mild though,with the casing on using a opa2132pa.i can check it up in college using a oscilloscope though.no change in opamp temperature(does not get warm or any such thing) .

is this because i am using iem's?will it go away when used with higher impedance headphones?i used the standard gain of 11(tangent's schematic),so i have to lower this?have no headphones as of now,as my sennheiser hd202 distort very badly.

 

A couple things to look at. 

It looks like the 47ohm resistors are IN the feedback loop. These resistors do occasionally help to improve stability & reduce noise, but when the op amp is driving a load they usually cause more trouble than they are worth. They make more sense in "pure" gain stages that are not driving the load (like a gain stage with a buffer after it). I suspect that these resistors are causing the distortion you are hearing. I would suggest reducing their value to less than 10ohms, or just jumpering them with a bit of wire. The noise may go up slightly after this, but the distortion you described with the HD202 should hopefully go down.

I'l get to really killing the noise in a minute but since were already talking about output resistors. 

The alternative to output resistors IN the feedback loop is output resistors OUT of the feedback loop. They do help with noise in very sensitive low impedance headphones, but they bring their own compromises too. With single driver headphones I often think the change is for the better with many headphones which makes the compromise an improvement to me, but with multi-driver headphones (IEMs) its just kind of sucky. Multi-driver headphones really need to be driven from an amp with as close to a 0-ohm output impedance as possible. Now I don't like the idea of using output resistors as a bandage to cover up the fact that the amp hiding behind them is all hissy, and it sounds like you actually need a low output impedance for the IEMs so just file this info away until later. Maybe experiment with it a bit and see if you like the changes.

So to really kill the hiss I would start my search with the gain of 11. Voltage gain of 11 is WAY HIGH for most headphones & sources. The real problem isn't so much gain, as the junk that lots of gain usually brings along for the ride. The problem is that ANY noise inherent to the circuit is multiplied by the gain - random junk IN the op amp its self, resistor noise, RF pickup, and the like. By reducing the gain we reduce how much this random junk is amplified and usually take a few steps towards a nice quiet amp. I'd try voltage gain of 3. 3 is still a little high for most systems*, but it offers you plenty of options for stable op amps that would not fly with less. When dealing with gain in hi-fi systems - less is more.

One of the compromises of the Cmoy is that it can be difficult to find an op amp that can drive a sensitive low impedance headphone with an adequately quiet/low output noise without resorting to output resistors. There are some op amps that can truly "do it all" but they are not really easy to use. Is it really a Cmoy when you have to add a dozen parts not on the original schematic to make the amp run stably? 

*personal definition of too much gain:

If you have the pot on the quiet side of the middle (about 25-30db of voltage overhead) at your preferred listening level with the most hungry headphones you have you have too much gain. This ideal leaves you plenty of extra pot for quiet recordings and gets even cheap pots into their best matched range but in practice it hard to actually achieve with headphones. It is also worth note that people freak out about spinning the knob up so high and equate spinning the knob that much that with lack of power (which is silly) so I would only apply it loosely. If you can get the pot 1/3 of the way up you are better off than most. 

Here is a pretty basic schematic for reference:

Nikongod that is great advice but he has already dropped the gain to 2.8 albeit with resistor values that help balance dcoffset. Some A hole took this thread sideways ( who me? ).

I guess I skipped that post. 

Just focus on the 47ohm resistor then. 

Those are pretty high resistor values. Higher = more noise. Search for a thread on the Pint amp, where resistors needed to be chosen to balance the input offsets on a BJT opamp. You would normally be forced into large values because of the 100k on the one input to ground. A solution was arrived at which allowed lower (much) resistor values, and it worked well. For a FET input opamp (like the OPA2132) balancing the input currents isn't really all that important. Also look at Tangent's page on working with cranky opamps, and his offset calculator.

I agree, but at that gain it doesn't seem to be a problem with a good op amp. The noise of resistors is really quite low compared to gain stage. I will admit that I have not worked with the OPA2132 so maybe I will bow out. I have worked with fet circuits and mosfets in the past and have a certain disdain for them. I think they have a very limited but useful purpose.

It sounds like you have pretty much solved your noise issue so concerning where to go next:

Quote:

Originally Posted by Avro_Arrow 

From the CMoy it is not too much of a step to move on to an A47 or add

a BUF634 (and variants) to the design.

I made almost exactly this my second build. I did it on protoboard with 22 guage wires, a star ground, and miscellanesous small caps to ensure stability. I used the OPA633 instead of the BUF634 because I had them handy.  I can say there is a pretty big difference in audio quality and an even bigger measurable difference.

Here's a link to the results:

http://kris.kalish.net/~kris/pub/amp_test1/Comparison.htm

I tested the amp with a real load, an old pair of Stanton DJ-Pro 2000's, and did my best to match the levels across all 3 headphone tests.  I had to compromise on a lower level than I wanted because the cmoy  and lineout couldn't drive them to absurd volumes. You can see that the buffered cmoy shows far better bass performance in the IMD graph and vastly superior crosstalk.

You mentioned that sourcing parts is a huge pain, so you might want to consider a simple discrete buffer like the "Monofied Sijosae Buffer." It's simple enough to learn how it works.  I'm sure it has an atrocious PSRR (power supply rejection ratio) but this isn't a huge problem on battery power. Anway, the point I was trying to get at is that you can probably source suitable transistors from a local place instead of ordering online.

-Kris