Home |
Search |
Today's Posts |
#1
|
|||
|
|||
DELTA - Upgrades and Oscillations
Tweaking begins; expert guidance (or shoving in the right direction)
appreciated in advance. I've got a bunch of observations, but will keep this brief and on just one section for now. I have IC-socketed a few of my Delta input modules, and left the others alone (factory soldered TLO72s). In the pan buffer section one op amp section of a dual buffer feeds the left mix bus, while the other amp buffers the right. The pan pot is a 10K dual; the input signal is "Y"d so that it appears across both pot sections. Wipers direct feed the + inputs of the amps. Rotate the pot one way and one wiper is at ground while the other wiper is at the "top" of the pot. Here's what I've seen in this section: Buffer TLO72 OPA275 OP2134 Output noise: 2 mV 2 mV 2 mV (approx) - input 11V 0.5 MHz 22V 1 MHz 25V 1 MHz amplitude varies from zero to level noted. Peak is at exactly mid-rotation of pan pot. + Input 0-4 mV, varies with pan pot position Output noise spectra shows a strong component of the oscillation seen at the invert input. Note that *all* configurations do this (soldered, socketed, all three chip types), though the factory soldered TLO72 is at a lower freq and amplitude. This oscillation on the invert input would then appear to be "normal" (as there's no apparent oscillation at the output other than in the noise). Other conditions: bus and pan switches OFF so that nothing is going out of the module. 2.2K feeds the output to the invert input. Invert input also goes through 10K to ground. It seems as though I could bring the noise down a bit and perhaps reduce the oscillation level at the invert input by putting a small value cap around the 2.2K feedback resistor, and/or a small value cap across the 10K resistor to ground. I did try this, and got the ouput to start oscillating -- could have been leads too long or wrong kind of cap (470 nF/100V ceramic is what I had on hand -- should I try something else -- polystyrene perhaps?) It does appear that most of the op amp feedback loops in the Delta do not have BW limiting caps in the feedback network. I could be happy with a BW of, oh 100-200KHz, assuming cumulative BW-limited stages don't start impacting in-band performance. I also assume there's a simple calculation to set feedback cap value based on desired BW and invert input Z (in this case it appears to be 10K). Thanks in advance for suggestions; this is a bit of education and fun; I hope the postings are of value to others. Frank Stearns Mobile Audio -- |
#2
|
|||
|
|||
"Frank Stearns" wrote in message
... Here's what I've seen in this section: Buffer TLO72 OPA275 OP2134 Output noise: 2 mV 2 mV 2 mV (approx) - input 11V 0.5 MHz 22V 1 MHz 25V 1 MHz amplitude varies from zero to level noted. Peak is at exactly mid-rotation of pan pot. + Input 0-4 mV, varies with pan pot position Output noise spectra shows a strong component of the oscillation seen at the invert input. Let's see if I understand what's happening: when you stick your probe on the output of the IC, you get the oscillation component, even with the stock IC in place? Is it that way on the modules you haven't modified? Are you using a /10 probe? If not, try using one and see if you still see the oscillation at the IC output. It may be that the probe's capacitance is destabilizing the IC. It seems as though I could bring the noise down a bit and perhaps reduce the oscillation level at the invert input by putting a small value cap around the 2.2K feedback resistor, and/or a small value cap across the 10K resistor to ground. Definitely don't try the latter. Definitely do try the former, You won't reduce the audible noise, but you may get rid of the oscillation, if it really exists without the scope probe present. (Are you *sure* the 10k is from - in to ground and the 2.2k is from - in to out?) I did try this, and got the ouput to start oscillating -- could have been leads too long or wrong kind of cap (470 nF/100V ceramic is what I had on hand -- should I try something else -- polystyrene perhaps?) If you hung a cap on the input resistor, well, that's a good way to build an oscillator. Well, not a good way, but it'll do it. It does appear that most of the op amp feedback loops in the Delta do not have BW limiting caps in the feedback network. I could be happy with a BW of, oh 100-200KHz, assuming cumulative BW-limited stages don't start impacting in-band performance. I also assume there's a simple calculation to set feedback cap value based on desired BW and invert input Z (in this case it appears to be 10K). No, you want to calculate based on the feedback resistor (output to - input). The formula is: C = 1 / (2 * pi * R * f) where C is the capacitor in farads, R is the value of the feedback resistor in ohms, and f is in Hz, f being your desired rolloff frequency. Multiply your answer by 10^12 to get picofarads. Peace, Paul |
#3
|
|||
|
|||
Frank Stearns wrote: Tweaking begins; expert guidance (or shoving in the right direction) appreciated in advance. There should be no oscillation anywhere. Maybe you'd like to start your explanation at the beginning ? I have no idea at all why you would want to substitute different op-amps at the 'pan buffer' section now what it's meant to illustrate. Graham |
#4
|
|||
|
|||
That circuit is the patented Soundcraft "active pan pot". It includes
both negative and positive feedback to obtain a near perfect pan "law". Shove any scope probe into the inverting input of an opamp and the stray capacitance will cause phase margin to degrade. The important part is that the outputs are clean. Many opamps can be used in this circuit as long as some housecleaning is done. Place a 47 pf NPO mono ceramic cap across R54 and R 57, the 2.2 k feedback resistors. Place a .1 uf mono ceramic cap from pin 8 to ground, close to the pin of the opamp. Place another .1 mono cap from pin 4 to ground. Try to keep the lead lengths short, especially at the opamp pins. Install a machined 8 pin socket. For wide band opamps like the LM6172, connect a 100 ohm resistor to a 100 pf NPO cap. place this network across pins 2 and 3. Make another and place it to pins 5 and 6. This is a phase lead function that pushes the high freq. gain up out of the unstable region of unity gain. You can just about put any opamp in there after that. This pan mod applies to all Soundcrafts with this circuit like Venue, 6000, 3200, Sapphire, etc. Jim Williams Audio Upgrades Jim Williams Audio Upgrades |
#5
|
|||
|
|||
|