[sdiy] Ultrarough draft of my hot-rodding of the Music Easel ring mod

[Grant Richter]

> Questions:
>
> * Any other cool mods I should add?

Yes, with all due respect professor, here are a few fun ideas:

1. Use an LM837 or other modern low noise chip. They have the same  
4.5nV/Hz rating as the 5532s but would reduce current draw by like 50  
ma.

2. Bring out IC1a pin 1 to a front panel jack through a 1K. This is a  
highly distorted signal, or a free "fuzzed" version of the MODIN signal.

3. Along the same lines you could AC couple the bottom of R22 to a  
panel jack, and the top of R19 same way which will give you the  
positive and negative half-wave rectified distorted signal.

4. Make the SYMTR trimmer a combination of trimmer and front panel  
control. The you can have a fixed trimmed carrier null, but also be  
able to unbalance the carrier in a controlled way like Don does in  
the bottom section of the 285.

5. IC5B is doing nothing exciting (no offense) so why not make it a  
floating microphone preamp normalized to signal in. You need some  
kind of preamp on a ring mod for microphone or instrument pickups.

6. Add and XR2206 or ICL8038 for an on board crude sine wave carrier  
source. With an onboard carrier and onboard preamp you have a useful  
stand alone module. A ringmod is pretty much useless without a  
carrier source and preamp. (IMHO)

>
> * For my inverting buffers, I use 68K on the input and feedback,  
> since that's what Buchla used to buffer the signal input. Based on  
> the various bits of theory discussed on this list, to compensate  
> for input currents, I should put 68K || 68K from the + terminal to  
> ground. I've done so. But, I notice Buchla's choice of + to ground  
> is much lower - namely 4.7K (although that is shared between  
> several op amps). What's going on there?

Those resistors are used to compensate for the error introduce by the  
bias current of the inputs. If the bias current is 5 microamps like a  
5532 they matter, but if the bias current is in the nanoamps there is  
hardly any error. And for a JFET inputs, you can skip them altogether  
since the picoamp bias introduces a 2 microvolt error.

>
> * If I want a noninverting buffer, I'm guessing it's bad form to  
> just plug the input from the jack straight into the - terminal of  
> an op amp in case someone plugs something bad in. So I put in a 1K  
> please-don't-blow-up-the-op amp resistor. I also put 100K from the  
> - to the + terminal, and + terminal to ground, because it seemed  
> like modules should have some noninfinite input impedance because,  
> well, they just all seem to do. Is any of this sensible?

Perfect, bam, just right, the 1K keeps out the static lightning  
bolts, and the 100k keeps them from floating with no input. Voltage  
drop is only 1% which is nothing in dB terms.

The only times I have seen a resistor in the feedback loop of a unity  
gain follower is for a high frequency unit where you are compensating  
a rogue pole created by parasitic capacitance at the input pin. The  
LM318 is an example, without the 10K and 4.7pf in parallel the thing  
oscillates like the dickens.

I have never seen them used for op-amps with a GBP less than 10 MHz.  
Audio op-amps have GBP of around 3 MHz and aren't bothered by a stray  
5 pf.