eval telephone board produces lots of noise

Could it be the polarity of your phone line? The schematic looks to be sensitive to that.

Which part of it is sensitive to the polarity? The speech circuit is isolated with a transformer and I can't see how it can be sensitive...

....MM

Oh he just needs a charcoal toroid... one of those, and problem solved! :)

Look closer at the circuits involving diodes, I don't think they are sensitive to the polarity. The ring detector has a DC blocking capacitor and the loop current detector has two diodes in opposite direction...

/MM

Since I happen to know the OP I will answer. The signal in question is the "Line Audio" taken from pin 4 of the J6.

/MM

be

Telephone lines are polarized, the interface shown in the schematic contains diodes that would be sensitive to this. Many commercial phones have bridge rectifiers so that it doesn't matter which way the wires are connected, but the melabs device doesn't appear to do it that way. Therefore, you could try switching the two incoming wires, but a better method would be to check the polarity with a DMM and insure that the diodes are forward biased.

phones

are

capacitor and

You're right, I'm stupid. :-( I didn't look hard enough before and completely missed the back to back loop current detector optos.

Now, after looking at it some more, I wonder which "audio output" the OP is talking about. The one marked "line audio" is the one that appears to be the received audio. The one marked "voice output" on the schematic ("audio out" on the chip) is the playback pin of pre-recorded messages.

That's a pretty neat board.

value as

line,

than R33

The problem we are trying to solve is with the signal coming from the central office to the line out. The value of R33 is a bit strange, it is 1 kOm as you can see. We tried decreasing it to 680 Ohm. It improved things a little, but not dramatically. As the OP mentioned we also tried using a PC soundcard instead of a regular line. There is no noise when the card is used...

/MM

On Mon, 10 Jan 2005 15:05:43 -0500, "MM" wroth:

U9D is part of an active "hybrid network" similar to a "directional coupler" in RF circuits.

As you know, telephone lines carry audio in two directions at once. The "line audio" in intended to be only (as much as possible) the audio comming in on the line from the central office with the audio going out to the central office nulled out (again, as much as possible).

If you expect to hear the audio going *out* on the line, it will be very weak and distorted when compared to the audio comming *in*.

Could that be your problem?

If not, then the component values around U9D should be checked again to see if they are proper. R33 in particular is supposed to be the same value as the phone line impedance. If you're not connected to an ordinary phone line, the impedance of whatever you are connected to could be higher or lower than R33 and that could be a problem.

Jim

On Mon, 10 Jan 2005 23:16:22 -0500, "MM" wroth:

Sometimes the phone line impedance isn't purely resistive and the network that balances the line needs to use capacitors and/or inductors in parallel or series with R33 to make it look more like the line impedance for a better balance.

Try "building out" or padding the phone line impedance with two 500 Ohm resistors, one in series with each side of the line, to see if things get better or worse. A more sophisticated solution if you see some improvement would be an "H pad" to better normalize the phone line impedance. You'll lose some volume but you can make that back up with the amplifiers already on the board.

Jim

For some background information, try looking at the tons of information that the ham radio operators have compiled about "phone patches".

Rich,

It was the DSL!!! We should have thought of it ourselves! Thanks a lot!

/MM