Several updates have been made to the design since the last entry so this brings everything up to date.

Valves

Yep - it's pretty obvious, it now has a valve, but as you will guess from my other postings, it's a low-voltage valve design. While this is not optimal (ie. it's low gain and relatively low input impedance relative to high voltage designs), it does suit the application.

Noise

As noted before there was a thermal noise problem. That's now greatly improved. The high impedance (~100Kohm) filter networks have been replaced with lower impedance (~2.5Kohm) filters while maintaining the same responses. The signal levels through the filters are also massively increased so that the tiny amount of noise now generated has much more signal with it. To run with the lower impedances and higher signal levels plenty of buffering is needed so I use NE5532 op-amps for that.

No surprise, the noise from the filters is for all practical purposes inaudible. The next biggest noise source, a VCA, has also had the low-impedance treatment and that too is now greatly improved.

The current biggest source of noise is in fact the input pre-amp. This is a standard TL07x based affair which is needed to give high input impedances which guitars like, but at the same time it throws up some problems with noise. One possible option would be to put a discrete JFET buffer on the front of a NE5532.

That said, I think it's worth recognising at this point that the noise is now respectable and completely in line with other kit out there.

Clean-up Filters

One of the things that was bothering me before was the roughness of the sound. The system was really prone to the sound turning to mush. That's been tracked down to background very-low frequency content which dies down very slowly compared to the note. The solution has been relatively simple - some careful filtering to remove these frequencies which are causing unwanted intermodulation.

Some designs use relatively high frequency filtering (eg. 100-200Hz fixed HPF) on the input, but that is not brilliant for an ultra-flexible guitar processor. I have chosen some careful filters which should not affect normal operation, and the rest happens as per the design with the normal controls.

What next?

One thing that I have discovered is that the servo balancing I am using may be interacting badly with the second clipping stage. The key thing here is that the clipping is not perfectly symmetrical (this is analogue) so during long stretches of clipping, there is an enforced offset introduced. This totally throws the servo balancing as there is nothing it can do to balance the system and it drifts off taking the circuit out of balance.

There are several ways I can approach this including to remove the servo balancing and trim up this one stage old-school. A possibly better solution I am looking at is to re-arrange how the servo input is derived so that it doesn't get thrown by the clipping occurring.

The other thing that is needed is a DC converter to allow the circuit to run from standard supplies while giving it the wide dynamic range it currently has.