Helix - I/O latency measurements

[Orvillain]

So....

We have a click source as audio. This is being routed out of my Presonus Quantum, into my Reamp box, and back into the 1st input on the Quantum. We can see that the input and output are exactly matched in terms of timing. Those with a keen eye will see the first sample of the looped back recording, is slightly malformed compared to the input signal.

So.... next level..... putting the Helix in the chain. So reamp box, into the Helix guitar input. Then out of the Helix main output, into the 1st input on the Quantum. The Helix patch is completely empty. Just a straight link from input to output. The input gate is off, and I have only the guitar input selected - not the multi input.

This 3rd take is with the Helix in the chain, and without any processing going on. The offset is 86 samples, or for my sample-rate of 48khz, 1.792ms. That means if you plug in to your Helix and then plug it into your guitar amp or audio interface (in the analog realm) that the analog to digital conversion of the Helix adds 1ms of latency.

But what about if we start utilizing the effects loops?

So.... I've added a jack cable to each effect loop, sending the send of the loop back into the return, to chain the FX loop and make sure if I insert it into my Helix patch, that it will be part of the measurement.

So... for Loop 1....

The total offset is 174 samples. Or about 3.625ms.

For Loop 1+2....

The total offset is 261 samples. Or 5.438ms.

For Loop 1+2+3....

The total offset is 349 samples. Or 7.271ms.


For Loop 1+2+3+4.....

The total offset is 437 samples. Or 9.104ms.


And if you bypass all of the loops and don't enable trails, what happens to the timing then?

Quite good. We're back to the original baseline value.


But .... this isn't particularly "real world" because there aren't any effects in the loops. What happens if we put some in? Let's say we put the Meris LVX into Loop 1. What is the measurement like then? I've added the Meris LVX into loop 1, and I've enabled a preset to ensure it is processing, but I have set the mix value to 0% meaning it will be completely dry.

Pay close attention here:

So the latency is higher at 295 samples, which is about 6.146ms. By adding in another digital pedal into loop 1, we've added in a whole new set of analog to digital conversions, which has resulted in more latency than merely using two of the Helix's loops without any effects in them. Furthermore, the Meris pedal has inverted the phase of the signal - you can see the polarity has been switched, so when the original waveform went up, the new output waveform goes down... and vice versa.

What happens if we use the Strymon BigSky instead?

The BigSky seems to introduce no additional latency. What about the Strymon Timeline?

Same again. At this point I am suspecting the LVX of weirdness..... I will investigate. I will load a simple blank preset, ensure that buffered bypass is enabled, ensure that there are no insert effects going on in the preset, and check again.

EDIT: This is due to analog dry-thru in the Strymon pedals, and in certain situations the LVX will not have analog dry-thru.

 

[Orvillain]

Yes. The LVX does weird shit to the signal. It really shouldn't be flipping the polarity of your input signal. I'm using the same cables I've been using for everything else, so it isn't a cable causing this. It is definitely the Meris processing the audio. I'd call this a bug to be honest.


But as you can see, some pedals introduce latency, and some don't. It depends on how they process their signals. I have to be honest, I'm surprised at the Strymon results. I was not expecting them to not add any additional latency. I'm a little skeptical of myself, but I did check multiple times.

 

[Orvillain]

Here is the Eventide Ultratap in loop 1. You can see... the total offset is different again - 326 samples, which is 6.792ms.

 

[Digital Igloo]

The Strymons have dry through (as does DL4 MkII), which is why they appear to not add latency—only the wet/effected signal is A/D/A converted. I assume the LVX has dry through as well; perhaps it's not enabled?

 

[Orvillain]

Finally.... 4 digital pedals. All 4 Helix FX loops in use. Eventide Ultratap, LVX, BigSky, and Timeline in each of them.

712samples, or rather 14.833ms, of latency introduced.

 

[Orvillain]

The Strymons have dry through (as does DL4 MkII), which is why they appear to not add latency—only the wet/effected signal is A/D/A converted. I assume the LVX has dry through as well; perhaps it's not enabled?

Yeah I figured this might be the case.

From reading on TGP with the LVX, the dry-thru is only active when spillover is disabled. I have it enabled. So in such a situation where you use the Helix to bring a pedal in and out... do the spillover on the Helix, not the LVX itself. I guess that would bring the latency down somewhat.

 

[Orvillain]

Soooooo... the real take home for all this is:

Be mindful about how many of the loops you use, which pedals go into them, and the length of your cable runs and all the other time domain stuff we need to consider.

If I'm introducing 14ms of latency just because I'm vain enough to want my specific pedals in the loops... well.... time to grow some balls and just use the built in DSP on Helix itself!!

Or if I'm using specific pedals that don't have analog dry thru, I need to expect that it is going to add some amount of latency; and that latency is in addition to the inherent latency of using one of the loops on Helix.

This is just physics really. I don't think it can be helped. I think the figures for Helix are pretty impressive.

 

[Digital Igloo]

Mentioned this in the FAQ thread, but for any time-based or parallel type pedals (delays, reverbs, many mods), you can avoid any additional latency through the FX Loops by setting their Mix parameters to 100% and controlling the Mix from the FX Loop blocks—that way, the dry signal is not delayed in the slightest. For serial pedals, however (distortions, drives, fuzzes, compressors, EQs, etc.), you will accrue latency through any digital FX Loop. In this case, you can always run them before Helix and if they're MIDI controllable, bypass them via Command Center Instant commands.

 

[jellodog]

@Orvillain If spillover is enabled on the LVX, then I believe that means that analog dry thru is not available. Disable spillover?

 

[Orvillain]

Mentioned this in the FAQ thread, but for any time-based or parallel type pedals (delays, reverbs, many mods), you can avoid any additional latency through the FX Loops by setting their Mix parameters to 100% and controlling the Mix from the FX Loop blocks—that way, the dry signal is not delayed in the slightest. For serial pedals, however (distortions, drives, fuzzes, compressors, etc.), you will accrue latency through any digital FX Loop. In this case, you can always run them before Helix and if they're MIDI controllable, bypass them via Command Center Instant commands.

Is it fair to say then.... slap your analog effects into the loops, and keep your digital effects out of them altogether??? Is that a good rule of thumb?

 

[DLC86]

View attachment 16794

Yes. The LVX does weird shit to the signal. It really shouldn't be flipping the polarity of your input signal. I'm using the same cables I've been using for everything else, so it isn't a cable causing this. It is definitely the Meris processing the audio. I'd call this a bug to be honest.


But as you can see, some pedals introduce latency, and some don't. It depends on how they process their signals. I have to be honest, I'm surprised at the Strymon results. I was not expecting them to not add any additional latency. I'm a little skeptical of myself, but I did check multiple times.

Strymon pedals have an analog dry path iirc (the timeline does for sure at least)

EDIT: oops, it seems this was already addressed

 

[Orvillain]

Nah, disabling Spillover on the LVX doesn't change the latency.

 

[Digital Igloo]

Is it fair to say then.... slap your analog effects into the loops, and keep your digital effects out of them altogether??? Is that a good rule of thumb?

It depends on the type of effect. Time-based effects accrue no latency through loops if you control the mix from the multieffect. Serial effects will accrue latency through loops, so you can run them pre-multieffect or just deal with another ms or two. Also note that if you always run specific effects in the same order, they can share loops, and therefore, share latency.

If the LVX does indeed have analog dry through, its throughput latency should be zero. Something's fishy there.

 

[Orvillain]

If the LVX does indeed have analog dry through, its throughput latency should be zero. Something's fishy there.

Yeah, I can't figure it out. I've tried disabling spillover and trails, and with a basic digital patch with the mix at 0%, I get the same result each time.

 

[Orvillain]

Here is the Axe FX III. 60samples of latency, which is 1.25ms, quicker than the Helix by a mere 26samples.

If I do something like this in the Axe FX III preset:

115 samples, 2.396ms. Again, marginally quicker than just the looped-back Loop 1 on Helix.

 

[Orvillain]

Aye, LVX comparision, with nothing else in the chain:

I cannot quite explain that to be honest.

Yellow is with relay bypass disabled. And as you'd expect; no latency.

 

[the-trooper]

I cannot quite explain that to be honest

Leo Gibson has entered the chat.

 

[Digital Igloo]

According to the LVX manual, "Analog Mix is used unless Spillover is enabled or a processing element is put into the PRE + DRY or POST MIX locations." Might that be it?

 

[Sascha Franck]

The offset is 86 samples, or for my sample-rate of 48khz, 1ms.

Fwiw, it's 1.8ms (well, 1.791666...).

 

[Orvillain]

According to the LVX manual, "Analog Mix is used unless Spillover is enabled or a processing element is put into the PRE + DRY or POST MIX locations." Might that be it?

Ahhh that'll be it. The looper is in the post mix position by default. I did wonder about that, but couldn't be arsed to test it.