Bastl Instruments Tromsø is a neat little module, having three simple functions combined in a 5hp space: VCO/LFO, Comparator, and Sample & Hold. I have four of them and use them all the time as building blocks for control structures.

A little room for improvement

Tromsø’s comparator circuit is just a bit too simple. The schematic above details the circuit. The input voltage (which is normally taken from output of the triangle wave oscillator) is compared with the voltage set by the threshold pot and threshold input. The result is a pulse of variable width that is buffered and sent to the comparator output (and normally connected to the S&H gate input). But there’s no hysteresis. For a technical explanation of hysteresis, try this link to Analog Dialog. I had not had any issues with the circuit, until…

I was testing my newly acquired Tilt module from After Later Audio, checking its ADSR function and gating it with the comparator out of Tromsø. Here’s what I saw on Mordax Data.

The green trace on top shows the pulse wave from Tromsø. The red trace shows the output of Tilt. What’s wrong is that the envelope looks like batman’s ears. When the gate falls the release stage should immediately commence, but instead there’s a re-trigger before the release. This didn’t happen with gates from other modules. The correct behavior is seen below.

I scoped the comparator output and found this.

Look just to the right of the vertical center line. See the faint trace, falling from +5 volts to -5 volts, over the time of two milliseconds (1 ms per time division). There is a distinct glitch around zero, the set threshold, which unfortunately isn’t clear in this photo. I suspected that this glitch might be causing the re-trigger of Tilt.

Circuit modification

On the schematic above, see the hand-drawn one meg resistor added between pins 1 and 3 of IC5A. It adds hysteresis, meaning that as the voltage on the output begins to change, a small amount of positive feedback causes it to change faster.

There is a second, similar comparator on the trigger input to the S&H, needed because an external trigger can be supplied. I made the same modification to that comparator.

After making these mods, the scope trace looks a little better.

Here you need to look at the left side of the screen to see the transition, a light trace. Now it’s taking only about 1 ms, so it’s improved. And indeed, after this mod, Tilt no longer re-triggered on the falling edge.

Transition slew rate

We see that the 10 volt excursion take 1 millisecond, which in volts/microsecond is 0.01 volts per microsecond, which is a pretty slow rate of change. In fact it’s terrible for triggering the LF398 S&H chip, whose data sheet says the triggering voltage should be at least 1V/uS. Whoops. Well, it seems to work.

Making the modification

Adding the two 1M resistors was easy. Both go on the same chip, one between pins 1&3, and the other between pins 8&10.

Knob change

For fun, I tried using Davies style clone knobs. The original little wood knobs have a tiny circular depression for the indicator. It’s much easier to see the knob position with the Davies. And the Davies are a bit larger and easier to handle, I think.

I went with cream colored knobs, instead of the green, for the comparator threshold.