[lostspyder]

I'm planning on building a Pimeta, but as a poor college student, I want to keep the cost down. Can I create the the supply by doing this (see attached) with 2 wallwarts?

Also, why don't they use a bridge rectifier instead of D1 in the 'stock' Pimeta power supply - it would fix any incorrect polarities instead of just blocking them and destroying components.

 

[FallenAngel]

Pimeta runs off a single supply and creates its own virtual ground, just use a single wallwart (and build a TREAD). The reason that a single diode is used vs a bridge is because it's just there for protection and should never be actually used, a bridge would drop the voltage by about 0.7V and it's not something that anybody wants (more voltage = better performance).

 

[tangent]

Quote:

Originally Posted by FallenAngel /img/forum/go_quote.gif a bridge would drop the voltage by about 0.7V

1.4V.

 

[balou]

Quote:

'm planning on building a Pimeta, but as a poor college student, I want to keep the cost down. Can I create the the supply by doing this (see attached) with 2 wallwarts?

Should work. Just leave out the TLE2426 and wire the supply correctly to the board.

 

[lostspyder]

Decided to just go with the railsplitter because I can get samples of it, but now, Id like an alternative buffer solution that I can get samples of - does anyone know of one?

 

[error401]

Quote:

Originally Posted by lostspyder /img/forum/go_quote.gif Decided to just go with the railsplitter because I can get samples of it, but now, Id like an alternative buffer solution that I can get samples of - does anyone know of one?

There aren't really any drop-in replacements for the TLE. Your alternative is to buffer a resistor divider with a high-current opamp. It takes a lot more parts and board space, but achieves the same goal, and probably gets you higher output current and better regulation (assuming an appropriate opamp selection). Any voltage feedback high current opamp that is unity-gain stable is appropriate here; tangent recommends AD817 or LMH6642, but whatever you've got on hand (or can get easily) should work. Just use a high impedance divider (100's of Kohm) with equal resistors to the non-inverting input, connect the inverting input to the output and you're set.

 

[balou]

Do you want to use two wallwarts and a railsplitter? That would be a bad idea - you'd lose the advantages of a true dual power supply.

OPA551 would also be a suitable buffer opamp

 

[lostspyder]

Quote:

Do you want to use two wallwarts and a railsplitter? That would be a bad idea - you'd lose the advantages of a true dual power supply. OPA551 would also be a suitable buffer opamp 08-28-2007 01:02 AM

No, im just going to use a single wall wart and the rail splitter.

How do I use OPA551 as a buffer? What is a buffer anyhow - from what I understand its basically a opamp used to increase the amperage available to the load. If I have a low load on the amp, will I get much of a benefit from a buffer (32 ohms)? Can I just jumper the buffer out?

 

[balou]

32 ohm is a high load

U=R*I (voltage equals impedance times current)
an amp has a constant voltage output. if you have a low impedance (32 ohm), the current is going to be much higher than with a 600 ohm headphone

 

[lostspyder]

Quote:

Originally Posted by balou /img/forum/go_quote.gif 32 ohm is a high load U=R*I (voltage equals impedance times current) an amp has a constant voltage output. if you have a low impedance (32 ohm), the current is going to be much higher than with a 600 ohm headphone

Why are high-impedance phones like the HD600 said to be too quiet unamped then?

I always figured a 32 ohm 'phone' was easier to drive.

Anyhow, what do the buffers do?

 

[error401]

Quote:

Originally Posted by lostspyder /img/forum/go_quote.gif Why are high-impedance phones like the HD600 said to be too quiet unamped then? I always figured a 32 ohm 'phone' was easier to drive. Anyhow, what do the buffers do?

There are two definitions of 'hard' that are conflicting here. With high-impedance phones the amplifier needs to have a large voltage swing, and consequently needs to swing faster to provide the same output power. It doesn't have to provide a large amount of current for the same power, though (remember: P=IV - more voltage = less current). These sort of phones are too quiet unamped because most sources don't have high enough supply voltages to properly power them, and will put out maybe 2Vpp, which is not enough voltage to drive them - the current needs aren't the problem here. Note that this limitation can usually be solved just by upping the supply voltage to the amplifier.

Low impedance phones pose a different problem (and I suspect they've become popular due to the fact that they can run on low voltages), and that is current supply. Most opamps have a limit on the amount of dynamic current they can supply before they either overheat or go into current limiting. With low impedance phones, not much voltage is required, but the amp must be able to quickly source or sink a large amount of current. This is a more difficult issue because it's the device itself that's limiting things. A buffer can be the solution here. A buffer, by definition (in the analog domain) is a unity-gain amplifier; that is it's an amplifier with no gain. 1V in, 1V out. Such a device serves two purposes: first, it has a very high input impedance and isolates the source from the load, so that a load on the output won't affect the source. Second, it's specifically designed to drive current, and can often do a much better job of it than a standard opamp in unity-gain configuration, so rather than causing distortion by loading down your opamp, you can tack a buffer on and get the best of both worlds. It's very much the same concept as in a discrete amplifier where you would first have a voltage gain stage, followed by a current gain stage to drive the output, due to the limitations of each.

 

[balou]

High impedance phones need more voltage but less current to have the same amount of milliwatts going through them (again U=R*I and P=U*I).

But to complicate the whole thing: modern day cheap earphones are optimised for high efficiency, thus they provide both low voltage operation and low current. one example for an especially hard to drive 32 ohm headphone is the Grado series. In contrast to earbuds and cheap headphones they are optimised for sound quality, not low price and low current.

So if you look at non-audiophile low impedance headphones, it is true that they are easy to drive.


Ok, what do buffers do: basically they just pass the voltage on the input to the output without changing anything. But the input of a buffer has a very high impedance (ranging from hundreds of kiloohms to teraohms), and a low output impedance. once again, U=R*I: with a given voltage, low output impedance means high output current. so yes, they increase the amperage available.

but I just realised - my last paragraph was mostly based on audio output buffers - that's where they shine for low impedance outputs.
if you want to replace a TLE2426 with a discrete power supply, you most likely can't leave out a buffer for anything more complex than a simple cmoy.

...can't write anymore, gotta go. tangentsoft.net has a section about virtual ground circuits, the answer why you need buffers is in there. will probably write later some more stuff

 

[lostspyder]

Ok ,I understand buffers now

If I use OPA551 as a buffer, do I build it Similar to this?

I hobbled it together from tangents virtual ground article.

Also if OPA551 is capable of high current supply, why don't we use it as the primary opamp?

 

[lostspyder]

well I decided to go with HA3-5002 for my buffer. That's all the questions I have for now; everything is ordered and on it's way. I hope everything works well

Thanks for all the help

I wonder how many IC's I'm going to burn up.

 

[lostspyder]

I installed the buffers in a CMOY and I am shocked at the diffrence they made. It is night and day. I'm ecstatic I didn't leave them out.