Rothwell Love Squeeze/NE-1 pedal build

Following my previous post, I have since built the Love Squeeze / NE-1 combo pedal for my bass. It’s pretty good, I love the tone control portion. The compressor I don’t really use much as it makes slap bass lines sound flat and I mostly just play slap. There was a major electronic issue with the compressor which I will discuss later. For now, check out the build pictures in the slideshow below.

Right off the bat, it sputters when used with my jazz bass especially with notes going from low E to A. Tried the compressor with my telecaster and there were no such issues. I was able to reduce this sputter by lowering my bass pickups, the sputter just stops after it goes lower than 10%. Very strange! Obviously the JFET is doing some weird oscillating gating thing when the circuit is fed low frequency signals…

Listen to the sputtering above, it has a purr like quality, my cat seemed to love it 😅

I can see that the compression effect is kicking in abruptly then cutting out before kicking in again, every few cycles. This behaviour can be simulated in SPICE as well, if I had bothered simulating the circuit response to the range of the bass guitar, I might have picked up this issue before starting the build. I was too busy making it simulate AC/DC bass riffs though…

Compare V_IN and V_COMP in the graph above.
V_n001 is the negative biased output from the sidechain follower, n001 node is highlighted in purple.

Turns out some people online also encounter this issue with the love squeeze compressor design, there seems to be 3 explanations commonly brought up to explain this sputter. Along with their respective fixes:

1) Undesirable JFET characteristics → replace JFET
2) Lack of headroom running at 9V → run at 18V
3) Sidechain follower high pass issue → adjust high pass frequency

My JFET is soldered in and I didn’t feel #1 was a satisfying explanation, after all the issue shows up in SPICE. As for solution #2, I was not about to run the pedal at 18V as it is a combination pedal and I don’t want to waste a 1044 chip in order to pump my incoming 9V to 18V. Explanation #3 is convincing as the issue only appears at low frequencies. Looking to the sidechain we see the following:

The C4-R1 high pass filter attenuates frequencies below 106Hz, I was thinking that if it was blocking all frequencies below 106Hz, then what I should observe is that low frequency notes have no compression and the effect suddenly appears above 106Hz. However, the attenuated low frequency signals are obviously still enough to be able to cause the sidechain to output a negative bias to the JFET gate. In fact, it causes the sidechain to spike every few cycles, this sudden high compression followed by little to no compression on the next peak, then high compression again, gives that sputtering unpleasant sound.

The solution idea now will be to allow all frequencies into the sidechain unimpeded by the C4-R1 highpass.
The question however is why does the sidechain output cut in and out sharply and intermittently?

I will explain what was done in real life first. The first thing I planned to do was to bring the high pass frequency down from 106Hz to something usable for the bass guitar. The plan was for C4-R1(1.5n,1Meg) to be modified to (120n,200K), with that the new corner frequency would be ~6.63Hz. Well clear of any bass notes, the lowest being 41Hz.

Meanwhile in reality, I didn’t have a 120n capacitor, only a 68n. Desoldering was a pain because my newly purchased solder wick was of shoddy quality, so instead of replacing R1, I soldered another 1Meg in parallel, halving R1 to 500K. This made sense because the capacitance I wanted was halved so I might as well almost double the resistance to make up for it.

Amazingly, this behaviour could be predicted in SPICE as well.

Now look at C7-R10 inside of the negative peak detector:
•R10’s purpose is to bleed or discharge the capacitor C7.
•While C7’s purpose is stabilising output of the negative peak detector.
•While the negative peak detector’s purpose is to supply the JFET gate with a negative bias proportional to the magnitude of signal seen by the sidechain.
•The JFET in turn has its resistance increased and hence decreases gain of the main opamp.

During every cycle, C7 is charged, then discharges during the second half of the cycle over a time constant of R10*C7. Which is in this case 100ms. The abrupt change in V_n001 due to the charge/discharge cycle should be the reason behind the buckling shape of the output waveform and hence, the idea now is to lengthen the discharge time of capacitor C7. This will minimise the ripple across the capacitor. There should be no issues with response of the compressor because I am only increasing the discharge time (decay / release) slightly and since I’m using a bass guitar, I won’t be reaching guitar shred god levels of BPMs. I’m no Les Claypool either.

Yet, I return to that question, why does the compression cut in and out intermittently when the high pass is set wrongly? I set up the SPICE model again this time probing at different net points.

I reversed all the mods above and added a new trace probed at the net labelled “detector”, to observe the waveform going into the negative peak detector section. Input volume at 0.8V and 0.15V, as expected, the sputtering goes away at lower volumes. The question remains, the input is a perfect sine wave, why does the sidechain output intermittent hard spikes.

Tracing further back the source of the signal, I look at the signal entering and leaving the sidechain non-inverting opamp. Here it gets interesting, looking at the top right graph, it seems that the signal did not have such a pronounced spike entering the opamp. It only develops that sharp and intermittent spike after amplification.

Granted, the signal coming in is by no means sinusoidal, something funky has already happened to it earlier on. Also the starting points of those spikes are already visible in the incoming signal, only amplified disproportionately by the opamp.

For the sidechain opamp, gain = 1+R7/(R8+Z5). The capacitor C5 in the path of inverting input to ground causes a weird filter effect. Such that gain is varied with frequency, for a DC signal reactance is infinite so we get unity gain. So for low frequencies, we have lower gain and higher gain at high frequencies. This creates something like a high pass with corner frequency ~72Hz, but low end is not attenuated to ground only boosted less. I’m not sure if this high pass was the intended purpose or to control DC gain of the opamp but I supposed it is the former. The DC gain is not important at this stage, any DC component gets removed by the downstream coupling capacitors anyway.

Once again this is an issue that only arises when using a bass guitar due to the low end extending past corner frequency. Anyway, it seems the small “pilot” spikes that are in the incoming signal appears to the opamp as a higher frequency compared to the rest of the 40Hz signal and hence is amplified with disproportionately more gain. Resulting in the sharp increase in compression.

By increasing R8 fivefold, the corner frequency is reduced to ~14.4Hz and the resulting output is much more well-behaved!

This frequency dependent opamp gain explains the strength of those spikes. The intermittency is explained due to the intermittent occurence of high frequency changes in signal voltage. There still remains the question of why this incoming signal is non-sinusoidal and having these intermittent weird high frequency artifacts.

I look at another trace, “OPAMP1_IN”, no issues here. The deformation must be coming somewhere between the sidechain input and the main opamp’s feedback circuit. Again I find another pesky 72Hz high pass, R6-C3 this time it attenuates frequencies below 72Hz going into the JFET. Honestly no clue how it causes deformation of the wave that badly, but gut feel is to squash that high pass way down by increasing R6 by a thousand fold for starters. Aaand, it clears right up, but now we see that the sidechain is receiving a phase delayed version as compared to the starting signal. It was not apparent before due to the messy wave, this phase delay is likely coming from the C4-R1 high pass of 106Hz that we identified as the first problem. It is lagging around 1pi.

To be sure, I looked at the case where only C4-R1 high pass is edited down to 6.63Hz again. Bleed resistor, R10 is also at the modded value of 20K. Opamp out is actually still lagging, but by a lesser amount now, maybe around 1/2 pi. Makes sense… Maybe the buckling observed previously was more of an artifact of this phase delay and less due to the release time, although lengthening that would help to mask the effects.

Now in addition to the mods done to the actual pedal, we also decrease both 72Hz high passes down to 7.2Hz by increasing both resistors R6 and R8 by one magnitude. The waveform clears up even more and no phase delay at all. I might have to do these mods to the actual pedal as well. I’ve simulated the difference to try to decide if it will be worth it. Notes are E A D G E E.

There is a buzziness in the simulated version that I also can faintly hear on my actual pedal. With these new mods, the buzz disappears all together, at least in the simulated WAV file….

I will create a new post to update on the real differences if I do decide to mod the compressor again.