AD815 beefy balanced line driver

[neutron7]

here is a circuit i have been working on. it is based on the app notes for the ad815.

so far i have just completed one side to test it. it is showing promise.

I just decided to try it because it is an interesting circuit, and i dont really like the DRV134 sound-wise and i need a line driver for my noisy computer room where i have balanced input powered monitors.

I compared the DRV134 input and output at unity with 20mv/div and there are quite a few strange spikey things visible even with a sine wave. when i tried the same thing with this circuit it was almost a flatline with just a little (much less) noise.

Here is the pic of the DRV134 in/+out comparison

[neutron7]

here is the schematic based on the analog devices app notes.

there is a bit of DC ofset on this circuit. i have not compared the 2 sides to see if it cancels, but somehow i think i may have to use a nulling circuit if i want to avoid the output capacitors.

[neutron7]

one side is done. its very fiddly with that small board!

the AD chips get warm! 54degC i am using 7815/7915 regulators
is there any way there could be oscilation i cant even see on the scope? its only a crappy 20mhz model.

[neutron7]

this is 20khz sine wave. very clean and nice.

as i sweep the oscillator through the audio range there is no (measurable by me) amplitude variation.

[neutron7]

i will finish the other channel on the weekend. then it will be time to try it with audio

[Limhes]

Looks nice...

Why not try to get rid of the noise even more, using two transformers for higher voltages? Don't know what your exact problems are, but it will reduce noise. It ofcourse might add some colouring, but that's to your taste (or not)...

[peranders]

Originally posted by neutron7
the AD chips get warm! 54degC i am using 7815/7915 regulators
is there any way there could be oscilation i cant even see on the scope? its only a crappy 20mhz model.

Have you consulted the datasheet in order to determine the power consumption?

[neutron7]

i would rather not use transformers except really good ones. i think this will work fine and be cheaper

[neutron7]

Originally posted by peranders

Have you consulted the datasheet in order to determine the power consumption?

The quiescent current is only supposed to be 55mA max. does that mean on each rail? if so thats about 1.6 watts total. what i do not know is if that amount of power can heat up the non heatsinked chip that much.

i do not think it is oscilating though because even if my scope can not see high frequencies. they make the trace blurred, and it seems fine. as an experiment i will try with and without 22PF rolloff caps across the feedback resistor.

anyways tomorrow is another day. it is a good start.

[Panelhead]

I have a dual chip running off a single +/-15 volt supply. It draws a total of 80 ma for the 2 chips. The CFM nulling circuit pulls another 5 ma on the positive rail.
An easy small heatsink is get a small pentium stick on sink. Then cut with a hacksaw to size. A regular one can yield 4 sinks for the AD815's.
Suspect you have a problem somewhere. Mine are barely warm, maybe 5 degrees C over room temp. The data sheet mentions local bypassing to prevent problems.
I was interested in building this circuit. I was told that the input impedance would need to be really low to get the offset down. Too low to drive with my sources.
Using output transformers was what I had in mind. Have an unused pair and was thinking of ways to use them with AD815's. Looked like it would require blocking caps and I did not want to go that route.
Let us know if you get the circuit from the first page of the app sheet working. I am interested in comparing it to the CFM circuit.
Would love to be able to use it to drive a transformer without caps or a nulling circuit. And still have a 25K input impedance. I do not want much.


George