[Spasticteapot]

As a college student, I am, not surprisingly, pretty broke. As such, I'm designing an amplifier that uses a tiny PCB, can be built entirely using through-hole parts, and is comprised entirely of ICs that our good friends at National Semiconductor will give us completely free.

The amplifier is pretty simple - it's the reference design from the datasheet of the LME49600, a BUF634 clone manufactured by NatSemi. It's a basic "op amp with a buffer in the feedback loop" design, with the addition of a servo to eliminate voltage bias (which I think should be bypassable by removing a few parts and adding a jumper - am I correct in this assumption?)

http://www.national.com/an/AN/AN-1768.pdf

I've laid out the schematic in EAGLE, and would appreciate it if someone could take a look and check to see if I've done something stupid. It's functionally the same as the one in the schematic, though a few parts (mostly power supply caps) have been swapped. Also, the two dual op-amps have been swapped with a quad op-amp from the same series.

RapidShare: 1-CLICK Web hosting - Easy Filehosting

EDIT:
Here's a picture:
Here's a picture:

 

[Spasticteapot]

PCB design is going...slowly. This is my first PCB with more than a few components, but I think I've got the hang of it. The best way to describe laying out a circuit board is that it's like the hardest puzzle game you've ever played, only harder. Also, if you screw up, you get an electric shock.

On the plus side, I think I can squish the whole thing on a 50mm x 50mm PCB. This means that I can get a batch of 10 made up for the princely sum of $3 each thanks to Adafruit's Propaganda service.

 

[FallenAngel]

If you have the buffer in the feedback loop of the opamp, why do you need the DC Servo?

 

[Spasticteapot]

Quote:

Originally Posted by FallenAngel /img/forum/go_quote.gif If you have the buffer in the feedback loop of the opamp, why do you need the DC Servo?

The design has no DC-blocking caps - without it, DC in would result in DC out.

 

[bidoux]

Quote:

Originally Posted by Spasticteapot /img/forum/go_quote.gif The design has no DC-blocking caps - without it, DC in would result in DC out.

For starving students, an input cap looks much simpler than a DC servo.

 

[MASantos]

And how are you getting free opamps?

If it is with a sample program, they are not for this purpose...

 

[Juaquin]

Quote:

Originally Posted by Spasticteapot /img/forum/go_quote.gif I'm designing an amplifier that uses a tiny PCB, can be built entirely using through-hole parts, and is comprised entirely of ICs that our good friends at National Semiconductor will give us completely free.

That's a BIG can of worms you just opened. Many people feel that sample programs are for the purpose of sampling - deciding whether or not to use that part. If you like your sample, are you going to buy 100+ more? If the answer is no, that's not a sample.

 

[Juaquin]

Also, I think you mean Seeed Studio's Propaganda service. Adafruit doesn't make PCBs.

 

[Spasticteapot]

I think I've got a handle on the component layout, though Imageshack seems to be broken and preventing me from uploading it. Next step: Figuring out how the heck I'm going to ground everything. 2-layer board design is hard!

Edit:
Success!


Quote:

Originally Posted by Juaquin /img/forum/go_quote.gif That's a BIG can of worms you just opened. Many people feel that sample programs are for the purpose of sampling - deciding whether or not to use that part. If you like your sample, are you going to buy 100+ more? If the answer is no, that's not a sample.

They give 'em away to anyone from a an accredited university. National Semiconductor knows full well that no student is going to be mass-producing anything, but they also know that the students of today are the purchasing agents of tomorrow.

Incidentally, you can buy both the LME49740 and LME49600 for a net cost of $25 at Digikey. Even if you do buy the components instead of getting 'em free, you can still build the whole amp for less than $50, including PCB. That's cheap! However, this amplifier was designed specifically for use around parts that the nice people at NatSemi give to students for free - hence the name. Also, from the numbers, it should perform really very well - the LME49740 is a popular upgrade for pro audio equipment and is used in some high-end headphone amps, and the BUF634 is generally considered to be an excellent (if pricey) buffer.


Quote:

Originally Posted by Juaquin /img/forum/go_quote.gif Also, I think you mean Seeed Studio's Propaganda service. Adafruit doesn't make PCBs.

Ooops. My bad.

 

[-=Germania=-]

Here is a simplified, tiny Pimeta variant with a digital potentiometer made by Maxim. Everything is through hole and only two layers, so you can make the PCB yourself at home. The decoupling needs to be done on the back of the board because of space and all I/O connections are done off-board. Put it together today for my little brother to make as I have all the parts on hand. There is no charging circuit onboard for battery operation (ones I have come with built in charging circuits). Yeah, it is meant to make as small as possible, not as clean as possible. I am a novice when it comes to PCB design.

3x BUF634 or subsitute
1x Opamp that runs on +-5V or +-3V (9V battery or 2x 3.3V Li-Ion)
1x DIP DS1802

I have it set for everything socketed, so that if wanted it could be a nicer CMoy (using jumpers instead of the buffer chips) if needed for opamp experiments.

Bunch of caps and resistors of your choice.

Have fun.

Download simpletinypimeta.pcb from Sendspace.com - send big files the easy way

Also if you see any circuit errors, let me know.

 

[diditmyself]

Spasticteapot, unless you're planning on using batteries like in the AN-1768 document, you don't have a proper virtual ground. Using batteries for virtual ground will keep this amp from being a smashing amp.

I don't understand why you want a DC-servo. It adds complexity noise and distortion. Are you afraid of DC offset from your source?

Depending on your source, the gain might be a little low.

With this gain, LME497xx will peak in the MHz region and there's a risk of instability. You should make room for some compensation caps.

 

[rds]

Quote:

Originally Posted by Spasticteapot /img/forum/go_quote.gif They give 'em away to anyone from a an accredited university. National Semiconductor knows full well that no student is going to be mass-producing anything, but they also know that the students of today are the purchasing agents of tomorrow.

The idea is that some of these parts will go towards innovative projects and that the people building these products will want to keep using company x's parts when they start selling product.
The thing is companies can't know who will be the next steve jobs and who's legitimately using the samples, so they err on the side of caution.

But if you plan to distribute pcbs designed to take advantage of the sample program that is asking for rampant abuse.

Recently a "helpful" user posted a comment on a message board telling people that any university student can sign up to IEEE for around $20 and thus gain access to microsoft software provided free to IEEE members (specifically windows 7). The program is not meant to be available to any university student, this was basically a loophole (it's an engineering society).
Anyways, guess what happened? Windows 7 is no longer available to anyone in IEEE, legitimate or otherwise. Not so helpful in the end.

Similarly, you will notice that the TI op-amps that are generally "sampled" by users here are no longer available in the sample program.

 

[tomb]

Quote:

Originally Posted by rds /img/forum/go_quote.gif The idea is that some of these parts will go towards innovative projects and that the people building these products will want to keep using company x's parts when they start selling product. The thing is companies can't know who will be the next steve jobs and who's legitimately using the samples, so they err on the side of caution. But if you plan to distribute pcbs designed to take advantage of the sample program that is asking for rampant abuse. Recently a "helpful" user posted a comment on a message board telling people that any university student can sign up to IEEE for around $20 and thus gain access to microsoft software provided free to IEEE members (specifically windows 7). The program is not meant to be available to any university student, this was basically a loophole (it's an engineering society). Anyways, guess what happened? Windows 7 is no longer available to anyone in IEEE, legitimate or otherwise. Not so helpful in the end. Similarly, you will notice that the TI op-amps that are generally "sampled" by users here are no longer available in the sample program.

Agreed. As someone who often orders samples for legitimate purposes, it appears to me that TI has stopped the sample program altogether except for new chips. In some cases, there are older varieties available as samples, but only if they are new production versions/revisions. Maybe that's just for the stuff I've been looking at, but seems like there's already been a drawdown in the sample program.

I keep telling them from a customer viewpoint, the sample program is very important, but I'm not sure it drowns out the numbers of abuses.

 

[Spasticteapot]

Quote:

Originally Posted by diditmyself /img/forum/go_quote.gif Spasticteapot, unless you're planning on using batteries like in the AN-1768 document, you don't have a proper virtual ground. Using batteries for virtual ground will keep this amp from being a smashing amp.

The LME49600 is a power hog, so I'm running this off of a split-rail regulated power supply.

Quote:

Originally Posted by diditmyself /img/forum/go_quote.gif I don't understand why you want a DC-servo. It adds complexity noise and distortion. Are you afraid of DC offset from your source?

Yes.
Incidentally, even with the DC servo, noise and distortion specs are VERY low.


Quote:

Originally Posted by diditmyself /img/forum/go_quote.gif With this gain, LME497xx will peak in the MHz region and there's a risk of instability. You should make room for some compensation caps.

I'm not so sure of this - it's the reference design, after all. How would you suggest modifying the schematic?

Quote:

Originally Posted by rds /img/forum/go_quote.gif The idea is that some of these parts will go towards innovative projects and that the people building these products will want to keep using company x's parts when they start selling product. The thing is companies can't know who will be the next steve jobs and who's legitimately using the samples, so they err on the side of caution.

I'm suggesting that STUDENTS take advantage of the sample program. I'm suggesting that everyone else buy them from Digikey or Mouser or whatever. The name of the amplifier is intended to be extremely literal.

 

[diditmyself]

Quote:

Originally Posted by Spasticteapot /img/forum/go_quote.gif The LME49600 is a power hog, so I'm running this off of a split-rail regulated power supply.

You mean something like LM317/337 or L78xx/L79xx? Why not a single supply and a railsplitter made from LME49xxx/LME49600?

Quote:

I'm not so sure of this - it's the reference design, after all. How would you suggest modifying the schematic?

You can either increase the gain to about 10x or you could add a capacitor in parallel with the feedback resistor like in the example below. This is from LTSpice, but I've had the same experience with LME49720 in real life, peaking without proper compensation.

Generally there's good practice to use a base stopping resistor on the opamps input and maybe an input filter. LME49xxx is sensitive to EMI (at least I've had this problem with it).