Oscillation on VU meter feed, and a missing buildout resistor...

[leigh]

The situation:

Modding the group modules in a Trident Series 65. Switched out a bunch of ICs from TL072 to LME49710. Power the board back up, and suddenly the VU meters for the groups are all floating at -7dB, without any signal present.

I get out the oscilloscope and see what's going on: there's an oscillation on the meter feed line, at about 7MHz. It swings to about a 0.6V peak, + and -.

If I disconnect the meter line, the IC feeding it (IC2, for those reading along on the schematic) stops oscillating.

I could just swap IC2 back to a TL071, and the oscillation stops. Unfortunately, that IC is not just functioning as a meter driver, it is also in the main signal path, so I'd prefer to keep it as an LME49710.

Two strange twists:

1. The output oscillation is at 7MHz, but IC2 has been bandwidth limited to 300KHz (negative feedback path consists of a 16k resistor in parallel with a 33 pico cap).

2. On the original Trident schematic, a buildout resistor R93 is shown in series with the meter feed, with a value of 3k6. However, in the actual console, although that resistor marking appears on the PCB, it has been replaced by a wire jumper.

The group VU meters are Componex 2LX. I can't find any specs on those - is it possible they opted for a model of VU meter that contains that standard 3k6 resistor internally?

Also, I should note, in testing if I add a 100ohm buildout resistor in series with the meter feed, the oscillation stops. The difference in meter readings with and without that 100ohm resistor is only about 0.1VU, so that may be an acceptable solution.

 

[JohnRoberts]

A common technique to isolate capacitance loads from fast op amps is to add a small resistance in series with the output. Keep the feedback capacitor directly across the op amp output to minus input, but move the feedback resistor connection from the op amp output to the far side of the 100 ohm resistor (the new effective output).

Try that, it should stop the oscillation and get your .1 dB back.

Note the 300k bandwidth limit is only for signals coming into the input. Using the output isolation topology I suggest will cause some overshoot at very HF but probably well above audio band with your feedback values.

JR

 

[leigh]

A common technique to isolate capacitance loads from fast op amps is to add a small resistance in series with the output. Keep the feedback capacitor directly across the op amp output to minus input, but move the feedback resistor connection from the op amp output to the far side of the 100 ohm resistor (the new effective output).

Try that, it should stop the oscillation and get your .1 dB back.

OK, that makes sense. That's what I'll do.

Still, I wonder about the "missing" build-out resistor, spec'd for 3k6, that doesn't exist on the PCB. They must have changed which VU meters they were going to use, or something.

 

[JohnRoberts]

That value sounds about right for the R used with old school VU meters. There are diodes involved too, perhaps inside the old meters. 

JR

 

[gridcurrent]

Still, I wonder about the "missing" build-out resistor, spec'd for 3k6, that doesn't exist on the PCB. They must have changed which VU meters they were going to use, or something.

measuerments of  two Componex model 2LX "vu" meters:

sn 850122.
approx. 30,000 ohms DCR

from 0 dBu, 1000 Hz, no series resistor
meter indicates -4 VU.
introduces 0.55 % thd from a 600 ohm source.

from +4 dBu, 1000 Hz, no series resistor
meter indicates 0 VU
introduces 0.75 % thd from a 600 ohm source

from +7.9 dBu, 1000 Hz, 3600 ohms series resistor
meter indicates 0 VU
introduces 0.14 % thd from a 600 ohm source



sn 851113.
approx. 37,000 ohms DCR

from 0 dBu, 1000 Hz, no series resistor
meter indicates -4 VU.
introduces 0.58 % thd from a 600 ohm source.

from +4 dBu, 1000 Hz, no series resistor
meter indicates 0 VU
introduces 0.75 % thd from a 600 ohm source

from +7.9 dBu, 1000 Hz, 3600 ohms series resistor
meter indicates 0 VU
introduces 0.16 % thd from a 600 ohm source


The Componex 2LX meters do not conform to sensitivity or induced distortion parameters associated with historically accepted VU meters, including models from Weston, Assembly Products, Simpson, and Triplett.