Parallel effects loops have become common in recent years, particularly on higher-end guitar amps. In this article, I will attempt to explain how and why they are different from a normal serial effects loop, as well as their not-so-obvious problems and how to deal with them.
The Serial Loop
It used to be that all effects loops simply inserted the effects inline, or in serial, between an amplifier's pre-amp and power amp sections. If no effects are connected, the signal passes straight through from pre-amp to power amp; however, inserting a plug into the effects return jack interrupts this signal, so that it passes through the effects loop instead. This is illustrated (in its most basic form) here:
Stop the Suck
Unfortunately, many vintage pedals and rack processors suffered from the dreaded "tone suck" — a rather non-specific term, like so many used by musicians, which encompasses a wide range of issues. This became more widespread with the advent of digital effect processors; because, although many love the sound of digital effects, some cringe at the thought of digitizing the dry guitar signal.
But it wasn't just digital effects. There are a whole host of issues that may corrupt the signal, and the more effect units one chains together in an effects loop (or in any signal chain), the worse these problems may get.
The Parallel Loop
To keep these dirty little effect boxes as far away as possible from the amplifier's pristine signal, some amp designers started incorporating parallel effects loops. The signal is split after the pre-amp: one path remains "dry" through the amp, and the other, parallel signal is sent through the effects loop — its return, the "wet" signal, summed with the dry signal in the amp. The idea here is to keep at least part of the signal pristine by not sending it through extraneous circuitry:
Because a parallel loop mixes the effects with the dry signal, it only works well with side-chain effects: these are typically delay-based effects such as delay/echo, chorus and reverb. Conveniently, these are the type of effects one would place after the pre-amp (as opposed to before), which is where the effects loop is located in the signal chain.
So, which effects work where? Here's a table that may help. Keep in mind that this is a general guide, and rules are made to be broken. Some effects can work in different places and will sound different depending on where they are placed:
| Before the Pre-amp
||In a Serial Loop
||In a Parallel Loop
*It may be difficult to achieve a strong effect when placing these in a parallel loop.
Due to an increased awareness of tone suck, more pedal-based effects use true-bypassing, an analog dry signal path, and/or higher-quality circuitry. In addition, there are now switching systems that remove entire effect chains from one's signal. These advancements have mitigated the need for parallel effect loops.
If you're lucky, you have a loop that can be configured for either serial or parallel operation, which gives you the most flexibility in how you want to incorporate your external effects. If this is the case, don't be afraid to experiment both configurations.
Don't Get (Too) Wet
Parallel loops aren't intended for heavily effect-laden tones, which are common in some music genres. This is because the dry signal is always present: the wet signal is simply mixed in alongside it.
Ideally, you'd want the dry signal to attenuate as the wet signal increases, so that the sum of the two is the same level — regardless of how where you set the mix. This allows you to increase the wet signal to 100% — where no dry signal remains — for some ultra-immersive effect tones.
Blow Your Stack
One of the best things about stacking side-chain effects is the multiplicity of effect layers created. When stacking a delay and reverb, for example, you don't just get two layers of sound (delay & reverb) — you actually get four:
- dry signal
- delayed signal
- reverberated signal
- reverberated delayed signal
And when stacking three effects, you get eight layers of sound! Each side-chain effect is doubles the number of sound layers from the signal before it. Here's an illustration of effect layering for a chorus > delay > reverb signal chain:
So, how does this work in a parallel loop? With just one effect, it's easy: simply turn this effect's Mix control to 100%. Or, if it has a Kill Dry switch, you should use that instead.
However, with multiple effects, we run into problems. Let's say we try turning all the effects to 100% wet signal — here's an illustration using our previous example:
Oops. Well, we've certainly cut out the dry signal, but we've also lost all of our effect layers except one! This is certainly not what we want. Maybe we should try turning only one of the effects to 100% wet?
This is better, but we're still missing three effect layers (not counting the dry signal). By now, I think you get the point — there's no way to get there from here.
Actually, there is a way: subtract the dry signal at the end of the effect chain. This sounds easy enough — but depending on what these effects are doing to the dry signal, it might be nearly impossible. Is your dry signal being digitized, and therefore delayed due to converter latency? Does the dry signal's gain change with the effect's Mix control? If the answer to either of these questions is "yes," then, while technologically possible, you're not going to find an adaptive dry signal canceler at your favorite music shop.
Making It Work
Obviously, the easy fix is to just use a serial loop. But if you really want to use a parallel loop, what are your options? By my estimation, you have two:
- use a single effect processor with Kill Dry, or
- use multiple effects configured as a "parallel-serial" chain.
By far, the best option is to use a single effect or effect processor with a Kill Dry function, which ensures that the dry signal is completely removed. If you have an effect that digitizes the dry signal, then you definitely want to configure it for 100% wet or Kill Dry. Any dry signal passed through will be delayed slightly due to converter latency. Adding this to the dry signal in the amp will create a comb filter: a repeated notching of the frequency response (and looks like a comb) that sounds like a stuck flanger. Not good.
The other option is set up your effect chain as what I call a parallel-serial chain. You'll get some dry signal through the loop, but you won't lose any of your effect layers. If set up properly, most of your dry signal will come through the amp, not the loop. Here's how to dial in a parallel-serial effects chain to minimize the amount of dry signal through the loop:
- Start by setting your loop's return level fairly high (you want to really hear the effects) and all your effects to 50% wet. This may or may not be 50% on the Mix knob — check the effect's User Guide if you're not sure where this is.
- For the effect that you want to be the most prominent, leave it's Mix knob at 50%.
- For effects that you want lower in the mix, lower their Mix knobs. Here you want to listen for relative effect level. This is the level of each effect relative to one another, not to the dry signal.
- Once your relative effect levels are fairly close, reduce the loop return level to balance the overall effect level with the dry signal.
- At this point, you can tweak your effects' Mix levels, but remember to leave the most prominent effect at 50% Mix.
For a parallel-serial chain to work well, you must not have any effects that digitize the dry signal. This bit of dry signal coming through the loop will create a comb filter with the dry signal in the amp. In this case, you should set this effect to 100% wet (or Kill Dry). If you are chaining it with other effects, you'll lose effect layers; but this is better than the alternative.
If you hadn't noticed, I'm not a big fan of parallel loops; because I think they are usually more trouble than they're worth.
However, there is one case in which I think they're ideal: you have a single effect processor that digitizes or otherwise sucks tone from the dry signal. If you can set this processor's mix to 100%, then a parallel loop is the way to go.
Otherwise, I'd prefer to use high-quality effects in a serial loop. But that's just me.