Limiting bandwidth of non-inverting opamp buffer

[.Mike]

Hey guys... I'm working out the final bugs on my multi-boost, and running into some new issues for me.

Not surprising, cascading three boosters full blast causes parasitic oscillations. I tracked it to the high impedance input wire for the non-crackling SHO variant I used. Moving the wire caused a change in volume and tuning. I removed some of the ringing by replacing the high impedance input wire with a shielded wire, shield grounded on one side. Sticking a 1n cap from the high impedance input wire to ground got rid of the rest without much impact on the sound.

Now my new problem. I have a buffer before and after a two-knob tonestack to avoid interaction between the boosters, the tonestack, and the master volume. When I crank the treble control, I think I am forcing too much treble, or more accurately too much treble that is outside the audible range, into the non-inverting opamp buffers I am using. I think this errant treble is slamming against the power rails, causing distortion and splattiness on the initial attack of the notes.

The buffers I am using are like these: http://www.singlecoil.com/tb-strip/buffer.pdf ...except I'm using 1u film for the caps, running it off of a +/-9ish bipolar (1044-based), and have a 1M reference to ground hanging off the non-inverting inputs. The opamp is a TLE-2074, the low-noise TL074.

I would like to limit the bandwidth to the buffers. The problem is that the board is done, and very cramped. I'm basically stuck to add stuff on the bottom side of the board. I have modeled sticking a 1uF cap from the feedback loop to ground on each of the buffers. That apparently creates a LPF that has less than 0.5dB reduction @10kHz, and starts dropping steeply at 20kHz. This fix would be easy enough to implement.

My questions are: Why does this work, and are there any drawbacks of using this method to limit bandwidth?

Thanks!

Mike

[PRR]

> I have modeled sticking a 1uF cap from the feedback loop to ground

The "feedback loop" is just a wire from OUT. Therefore you have put 1uFd directly on the chip's output.

I had too much coffee. My hand shakes and soldering is difficult. If I strap a 10 pound weight to my hand, the shaking is much smaller.

But it becomes a LOT more work to move. If instead of holding an iron I was waving a drumstick, it would be a poor answer. At first my hand would be too slow. My "internal feedback", sensing a problem and over-compensating, would probably over-shoot.

How fast can the chip output move with a 1uFd "weight" on it? Max output current is perhaps 20mA. 20mA/1uFd is 20,000 volts per second. A half-cycle of 10KHz is 0.000,05 seconds. The output can swing just about 1 volt peak-peak at 10KHz, and at that rate the output will be a triangle-wave.

For reference: 20,000 V/S is 0.02 volts per microsecond. Hi-fi guys sneer at 1V/uS, and even my ears may hear improvement with 5+V/uS chips on complex musical sounds. Your plan is a hundred times slower than "good sound". Maybe only a dozen times slower if your levels are lower than hi-fi. Maybe not, if you are pushing to distortion with +/-9V supply. Of course if you are distorting the lows-mids, ya may as well hammer the highs as well.

> the board is done, and very cramped.

Learn from my mistakes. Trust me. A too-cramped project WILL give you problems.

However you are where you are. Put a cap on the chip output, hear what happens. Do not trust the sim unless you know your TRAN from your AC and you know your opamp model intimately (many do not do-the-right-thing in "abnormal" operation). Try 0.1uFd to 1uFd.

If you can find/make the space: bypass your power rails all to heck. When the opamp stains agains the slugging cap, it will pull huge chunks of current from the supply rails, which will tend to sneak into other stages and encourage oscillation or just wheezy operation.

[.Mike]

Thanks Paul. I may have overstated how tight it is. It's tight compared to a lot of the designs I see around here, but not as crammed in as builds I've done in the past. All resistors are 0.4" or 0.3", everything has good spacing, all components are on a 0.125" grid, so they have even more space than perf, and traces are given plenty of space. Truth be told, I think that designing boards is my favorite part of this hobby.

I know that cramming a lot into a small space is a challenge, and I intended this to be difficult. That's why I did it. If I wanted to build something easy that didn't offer any challenge, I could whip together three boosters and a tonestack in separate enclosures using someone else's layouts in a weekend. There's just no fun in that for me. I want to learn, not produce. This won't be my last crammed-in build, and knowing a challenge lies ahead makes me want to cram in even more!

The nice thing is that I made this as separate boards-- a power supply board, a booster board, and a tone/volume board. I can always redesign or rework any of the boards in the future.

I think that after shielding the input wire and adding a cap to filter out high frequencies on the one booster that was giving me trouble, the problem is not with the layout. It's with the actual design itself. Gain on top of gain with a treble boost added on the end is going to pickup radio signals. It is a stupid mistake that I imagine shows my lack of experience. I breadboarded each circuit separately. It is something I hadn't even considered, but I will in the future.

The effect is usable. I only run into a problem at the most extreme settings. I doubt there is a practical use at those settings, but if there is room for improvement, I want to do it. Considering I don't get any interference until I add about 30 dB from an SHO into the (attempted) 20+ dB of the minibooster into the (attempted) 20+ dB of treble boost from the tonestack, I would say that is pretty freaking good.

I have already done an overkill job on bypassing the power supply. My house wiring is old and noisy. I think you even told me my house wiring was unsafe and not up to code, since my fuse boxes are in my kitchen cabinet.

I run my power supply at ~9.6V so I can use higher resistors to lower the corner frequencies in the power supply filters. The whole effect has power filtering, as well as each booster, each rail for the opamp, and the charge pump. It's overkill. I wanted to see how it would work.

I'll experiment some tomorrow. Thanks again!

Mike