[Stephonovich]

Yet again, I tried re-building my CMoy. Well, during testing, (with alligator clips running amok) I had sound, albeit scratchy, through both channels. I passed it off as the poor connections. Today, finished wiring everything up to panel components (in a lovely Hammond case, I might add), and no sound.

So, I first checked the - and + on my opamp sockets. One the 1st, I'm getting 1.8v. On the second, I'm getting 4.8v. (oh yeah, this is an original CMoy, with two mono chips) This has happened to me before, and I've been told to check for solder bridges, and whatnot. There are no bridges.

Now, in previous iterations, there's always been sound from at least one channel. Not amplified, but at least it'd pass it through. The lack of any this time leads me to believe I have the I/O wired up wrong. I can check this tomorrow, I suppose.

If this doesn't work, is there anyone willing who I can ship this to for repair? How much would it be? I'm dang well fed up with fixing it. Screw learning. I just want an amp!

(-:Stephonovich

 

[tangent]

Quote:

I had sound, albeit scratchy, through both channels. I passed it off as the poor connections.

Long leads are more likely to lead to buzzes and such than that scratchy sound. I know the sound you mean.

Quote:

One the 1st, I'm getting 1.8v. On the second, I'm getting 4.8v.

What's the supply voltage supposed to be? If it's a 9V battery and it's supposed to be new or at least not nearly dead, you're having severe voltage depression due to shorting the battery out. Something will be getting hot if this is the case. Otherwise, why are you trying to run your first CMoy from just 6.6V? It can be done, but it takes careful work. Not something you should attempt for your first amp.

Also, test with the headphones plugged in and with them out, and with the chips in the sockets and with the sockets open. And if you can, tell us how much current the amp is drawing with the chips in vs. out.

You say you've rebuilt it 4 times now. If it's always been with the same parts, maybe something is broken? Rebuilding it again won't help without replacing parts.

 

[Stephonovich]

First, sorry it took so long for a reply. I was busy with crap when I posted this, and when I was done, I had forgotten. Since then, every time I'd remember to check, I was away from the computer. Anyway...

Well, the alligator clips had long leads (over a foot, I'd say) on them. I suppose that'd do it.

I have one 9v, and it's putting out 8.4v. I tried feeling various components for heat, but nothing got even warm.

Today, when testing, I got 6.7 volts out of one, and almost none out of the other. And I tried with stuff plugged in, with it not, the pot at various locations, (and yes, I also tried directly connecting the wires, bypassing the pot) nothing. No sound whatsoever.

The first two tries were with mostly Digi-Key parts, with Ratshack caps, due to me forgetting to order them from Digi-Key. The third and fourth were with different caps and wiring, from Mouser. Same values, but different brand. And the wire is Alpha PVC. Also, during a past troubleshooting session, my dad made a huge version of the power supply with the Radioshack parts, soldering everything together in one big wire loop. No PCB. Oddly enough, it worked perfectly, with voltages split exactly.

Methinks my skills (or lack thereof) just suck. What I really want at this point is an amp. Screw learning; so far, there's been about $100 put into this thing. It's currently a big joke. I've made a $100 light. (yay, the LED lights up)

Which brings me to the question... Is the META42 any easier, since it's on a pre-drilled PCB? I mean, how hard can dropping components in be?

(-:Stephonovich

 

[Stephonovich]

I've been reading about using railsplitters, and it seems like a good idea to me. Am I right in thinking that you get a perfect +- split with one of these? Can you buy these from Ratshack?

(-:Stephonovich

 

[jeffreyj]

Unfortunately, no. Radio Shack carries only a token selection of components and seems far more interested in trying to sell you a satellite system or cellular service these days.

A rail splitter IC - like the TLE2426 from TI - is merely an op-amp configured as a unity gain follower with it's non-inverting input fed from a high resistance voltage divider. You could make one yourself out of a spare op-amp and a couple of 1M resistors - if the op-amp has a high enough output current capability. The LM324, available from Ratshack, could do the job if you put all four op-amps in parallel (with 10 ohm resistors in series with each output). Try it out: you'll probably be the first person here who's used one for that purpose.

 

[tangent]

Or, you could go cheap and use a 741. Try this:

If you use a JFET-input op-amp (like the TL082, also carried by Radio Shack) you can make the 10K feedback resistor a jumper. Also if you go with the TL082, you can band the two channels into one.

 

[jeffreyj]

The "only" drawback -

- to using the 741 (or the TL082, for that matter) is the feeble output current capability. However, a healthy dose of reservoir capacitance will cure most of that problem ('til you get to the sub 100Hz range...)

Ahem... better add the bypass capacitance to that schematic of yours, tangent

 

[tangent]

I chose that op-amp because it's cheap and available, and it has the same output ability as a TLE2426, which can be used on its own. It isn't perfect, but it's okay.

Bypass caps...foo. I lashed it up on a solderless breadboard and called it good.

 

[Stephonovich]

Umkay, let me get all this straight...

If I were to build one of these, would this replace the two 220uf caps and resistor mess that currently comprises the PS?

As for cheap, seeing as how the LM324 mentioned is $1.39, I don't see a price problem.

Also, a couple new questions... does it matter if the grounds from all the components (pot, jacks) are all connected to a central wire that's connected to the virtual ground? Also, does the virtual ground need to be connected to the case (Hammond) itself? I checked with a multimeter on continuity, and the case apparently wasn't conducting electricity. Do you want to keep the PCB isolated from the case, or grounded?

(-:Stephonovich

 

[tangent]

Quote:

would this replace the two 220uf caps and resistor mess that currently comprises the PS?

Leave the 220uF caps in. The 22K resistors you see in my schematic replace the 4.7Ks CMoy used in his design to split the voltage. The 741 with the 10K feedback resistor buffers the 22K divider, giving better performance than CMoy's unbuffered design.

You might also put some small ceramics or films in parallel with the 220uF caps. 0.01 to 0.1uF or so. These are the bypass caps jeffreyj spoke of.

Quote:

seeing as how the LM324 mentioned is $1.39, I don't see a price problem.

The LM324 path is more complicated.

EDIT: I think you'd set it up like I drew in my schematic, but the midpoint of the divider goes to each of the four +INs on the LM324 and you need 4 10K feedback resistors, one from each op-amp channel's OUT to -IN. Then you tie the four 10Ks together, giving a big combined buffer that's more capable than a single 741. Kind of a mess, but it should perform better than the 741.

Quote:

does it matter if the grounds from all the components (pot, jacks) are all connected to a central wire that's connected to the virtual ground?

Much better to run separate wires from each item needing grounding back to a central grounding point. The ground bus you see in my CMoy tutorial is a compromise between these two extremes, and it does work. How anal do you want to be?

Quote:

Also, does the virtual ground need to be connected to the case (Hammond) itself?

If nothing else is "fighting" for the chassis connection, this will work fine. But, add a metal DC power jack, and you're likely to have problems. See this:

http://tangentsoft.net/audio/pancomp.html

Quote:

the case apparently wasn't conducting electricity

It's anodized, which means there's a layer of oxide covering the aluminum. Scratch the aluminum or probe on the edge of a drilled hole, and you'll get conductivity.

Quote:

Do you want to keep the PCB isolated from the case, or grounded?

You don't want any of the metal bits on the PCB (leads, traces) to touch the metal case. Use standoffs, or line the case with something insulating, or glue the board to the case with hot glue balls so it stands up off the bottom, or....

 

[jeffreyj]

Stephonovich,

I'm going to have to go with tangent on this one. Paralleling the 4 op-amps in the LM324 will give higher output current capability, but the circuit will be a real pain to wire and not really worth the effort/board space to do.

I do suggest a couple of modifications to tangent's circuit, though, that will improve performance into highly capacitive loads (i.e. - the reservoir capacitors!). Wire a 100pF capacitor directly from the output pin to the inverting input pin. Put a 10 ohm resistor in series with the output pin, then connect the 10k feedback resistor.

 

[tangent]

I'm not sure a feedback cap is helpful with a slow chip like the 741. Also, I've found* that you don't really need the 10K feedback resistor for balancing the circuit -- the 10 ohm resistor jeffreyj wants here seems to be enough. I suppose putting a 10K from the Vgnd node to -IN and a small cap in parallel with that might make some tiny measureable difference, but it doesn't seem to be worth chasing.

* = I simulated the circuit and also tried it on a solderless breadboard, and it seems to work this way. I haven't actually made this circuit drive an amp yet.

 

[aos]

You're not seriously going to use ua741? Surely not even the weakest PCDP would sound worse than that?

 

[Stephonovich]

Umkay, this looks good to me. Who sells the 741? Ratshack apprently doesn't, as least according to their site. Ah wait, I see Mouser does. I'd be needing some .01 uf's anyhoo. $.33 a piece from Mouser.

So, can the 741 run on higher voltages? As long as it can handle what the OPA134 can, I'm fine.

Quote:

But, add a metal DC power jack, and you're likely to have problems.

I am seriously considering adding one. I have a 1.5-12 volt adapter from Ratshack, and I did have a 3-way jack, so's I could have batteries and a wallwart hooked up simulataneously. It disappeared recently. But I'm not sure how noisy it's going to be. If it's going to be really bad, then I'd rather just run it off batteries.

So, do you have to ground the virtual ground to a case, or what?

(-:Stephonovich

 

[tangent]

Quote:

You're not seriously going to use ua741? Surely not even the weakest PCDP would sound worse than that?

For stabilizing the virtual ground in a CMoy amp? Why not? The op-amp in the TLE2426 might actually be worse than a 741: it must be BJT all the way through*, I notice that the bandwidth of the device isn't specified, and the queiescent current on it is miniscule.

* = If it were a CMOS op-amp, it wouldn't have the voltage tolerance it does, and JFET inputs would be pointless since you've got that resistor divider right in front.

Quote:

Who sells the 741? Ratshack apprently doesn'

"741" is the generic part number. Lots of companies make this part, and they call it slightly different things. Radio Shack's is apparently made by National Semiconductor, so it has an LM prefix. Part number 276-007. The original 741 was Fairchild's uA741, which is why I put it that way on my schematic.

Quote:

I see Mouser does.

Yes, but if you're going to place a Mouser order, just get a TLE2426 from them and be done with it. It's a whole lot simpler to apply than this hack we're making with the op-amps. We're only giving you these circuit ideas because you said you didn't want to mail order a rail splitter.

Quote:

can the 741 run on higher voltages?

Up to +/-18V in the commercial grade.

Quote:

I'm not sure how noisy it's going to be.

It won't be the jack's fault, it'll be the power supply's. Since you can switch to a different power supply if you have noise problems, I don't see why you'd hesitate on adding the jack.

Quote:

do you have to ground the virtual ground to a case, or what?

You don't have to ground the case, but you may end up with it grounded whether you like it or not! What I'm trying to get across to you is that if one of your panel components has a connection to virtual ground that is tied to the metal shell of the component, and another metal-shelled component has a connection to another voltage level (like V-) that connects to that shell, you will have a conflict.