H90 insert loop latency

[matfagan]

Hello problem solvers,

I wonder if anyone else has noticed this issue?

Any time I make an insert loop active I get latency on that program. It’s only subtle but I can feel when it’s there, the note attack is late. It goes away when I deactivate the loop. I’m only trying to insert an overdrive pedal, nothing crazy. Is there anything I can do to eliminate the latency?

Any help much appreciated 🙏

[matfagan]

Anyone?

This is turning out to be a deal breaker, I’m running my overdrive before the h90 as the insert loops are not useable. How is this unit not able to send the signal through it’s loops fast enough?

I’ve tried both inserts and both bypass methods, and I’m still in lag town. I can’t find any info on this issue, am I the only one experiencing this?

[Hermetech]

There will always be double the inbuilt latency when using inserts due to the double A/D/A/D/A trip, unfortunately, but it shouldn’t be noticeable if it’s low enough. It would be good to know what the H90’s latency is in samples or milliseconds for each A/D/A process.

I haven’t tried the inserts yet, but if there is noticeable latency as you say, I will also be very disappointed. I hope you get it sorted or some kind of feedback from Eventide.

Will be watching this thread!

Just to double check, in the Insert Routing menu you can set the latency to 0, it hasn’t been set higher by mistake by a previous owner or something?

[Infection]

If I remember correctly, the latency is about 2.5 ms for each conversion, so 5 ms without inserts from input to output, and 10 ms with serial insert.

It is to much latency for me to use inserts for anything other than parallell  processing like reverb and delay.

[apalazzolo]

Infection wrote:

If I remember correctly, the latency is about 2.5 ms for each conversion, so 5 ms without inserts from input to output, and 10 ms with serial insert. It is to much latency for me to use inserts for anything other than parallell processing like reverb and delay.

 

IIRC the insert latency is only 1.5ms, not 5ms. I doubt anyone can hear or feel that. 1.5ms of latency is not a design flaw. It is the current state of technology.

I’ve used a UAFX amp sim as a series stereo insert for over a year.  That’s about 2.5ms of UAFX latency.  With both sides of the H90 and the insert, I’m pushing 9ms.  Now, I am using headphones, I’m no speed demon, and I won’t add any other digital pedals, but I’m fine with 9ms.  I played an all analog rig (standing more than 9 feet away from my amp) all over New England for years.  I never had a problem with that either.

I believe the OP is experiencing something noticeable.  But, I suspect an errant setting (as suggested above) or some other problem.  I assume he’s inserting an analog drive which would add zero latency, so that’s not it either.

[Infection]

H90 latency

According to Eventide, it’s 5ms in dsp bypass mode.

[apalazzolo]

I stand corrected.

[sled]

I agree with you, I find the insert loops to be perfectly usable. In fairness, I am usually only using one, but in that insert I’ve got a MIDI loop switcher with all of my analog drives and my preamp (the switcher loops are also analog, so no extra latency there). After the H90 I’ve got a stereo cab sim, then it goes to my interface. The resulting 8-9-ish ms of latency is not enough to be significant as far as I can tell.

As you alluded to, it takes sound about 1 ms to travel 1 foot… so if I’m using my DI rig with 10 ms or less of latency, it’s going to feel at least as responsive as playing with an amp that’s 10 feet away. I doubt you’d find many people who feel like they suddenly can’t play with an amp at that distance.

If you’re playing with a DI rig using an H90 and both loops, AND your speakers are really far away, now we might be getting into “problematic latency” territory. But if you’re using headphones, or monitors at your desk or something, or in-ears on stage, there’s really not any reason why the latency you experience should be much different from most other real-world scenarios.

I don’t want to invalidate anyone’s experience or anything, maybe some people are more sensitive than others… just thought I’d chime in to anyone who feels disheartened at seeing latency numbers that seem “big.” The H90’s inserts are definitely usable with analog gear. I’m having no notable issues doing so.

[Infection]

People are different. Some notice latency at around 4-5 ms, others at 20+ ms, and I talked to a guy who was okay with 50 (he played ambient stuff with tons of effects and swells).

I think the psychoacoustic aspect also plays a role. If you see a speaker or sound source at a distance, your brain will expect the sound to be delayed and it will feel natural. Not so much If the sound is artificially delayed with a speaker close up or headphones.

 

[Hermetech]

I play with a real tube amp, as it happens about ten feet away. Can’t abide daily practise with headphones, I like to feel the air move in the room, it’s all part of the interactive experience of playing the electric guitar for me. So let’s say 10ms latency as sound travels through the air. I did some tests last year with guitar DI into PC with amp sim software to stereo monitors (also the same distance away), with differing buffer sizes, and I start to notice the added round trip latency at around 5-6ms (so 15-16ms total, acoustic and A/D/A latency), and it starts to negatively affect my playing, particularly single note lines. Adding an extra possible 12ms of latency to the 10ms acoustic latency by using the H90’s two inserts in series, and you’re talking a whopping 22ms total. It’s not unusable, but it’s not ideal either, for me at least.

[Hermetech]

@Infection Any idea what it is in True Bypass (Relay)? That’s what I always use. I just looked in the latest manual but the info is not there. Obvs zero when Bypassed, but is it the same 5ms when the effect is engaged?

I start to notice anything above about 5-6ms with electric guitar.

[tbskoglund]

The latency of the H90’s main path is dependent on the preset routing; Series: ~4.5ms and Parallel: ~3.8ms.

Each Insert will add an additional ~3.8ms of latency, plus any additional latency compensation specified in the Insert settings.

Program and Preset level parameters such as mix, bypass, kill dry, and bypass mode do not contribute to the total latency.

 

[apalazzolo]

Is there a latency difference in the main signal path between these three scenarios?:

1. Slots A and B are in series and both are running presets.

2. Slots A and B are in series and one slot runs a preset while the other is a non-bypassed “INIT Program”.

3.  Slots A and B are in series and one slot runs a preset while the other is a bypassed “INIT Program”.

I’m trying to understand if can reduce latency by not using one slot and, if so, do I actually need to bypass an “unused” slot to do that.

Thanks.

[tbskoglund]

Do you mean using a “Thru” preset in slot B? All of those scenarios would be the same latency.

 

[apalazzolo]

Yes, I meant a “Thru”.  Thank you.

[aleksisundell]

What about latency when using dual routing mode? Is it similar to normal routing, as in with both algorithms on a single path in series, that path will have a latency of 4,5 ms and in parallel 3,8 ms?

What about when you move one algorithm to path 1 and the other to path 2? What will be the latency for the individual paths?

I have the H90 in/out 1 in a true bypass loop switcher, so that I can remove the first path entirely from the signal path when I don’t use it. Can I reduce the overall latency of a path by moving one algorithm to the unused path?

How much of the latency is caused by A/D/A conversions and how much is actual processing latency? I assume from the series vs parallel numbers that the algos run on separate cores and passing the signal between cores adds around 0,7 ms?

[tbskoglund]

Yes, the latency in Dual mode is the same as in Insert Mode. 1 Algorithm on the path will be the same as 2 Algorithms in parallel: ~3.8ms.

[dannysomerville]

I think I understand this but want to make sure. I plan to use dual mode in relay bypass and will use one path at the front of my pedalboard for pitch shift, compressors, etc. I suspect this would get used sparingly. The second path will be after my drives for reverb, delay, modulation and will be on most of the time.

If I have one path relay bypassed and the other path with 2 algo in series it will be 4.5 total

If I have one path relay bypassed and the other path with only 1 algo it will be 3.8 total

if I have one path with 1 algo and the second path with 1 algo it will be 7.6 total (3.8 x 2)

is the above correct?

[dannysomerville]

I also see now that outputs 3/4 cannot be relay bypassed so I suppose there’s no way to get zero latency bypass without a loop switcher even in relay mode.

[tbskoglund]

Yes, your latency calculations are correct and relay bypass is only available for I/O 1-2.

[siggy76]

Does the insert loop latency remain also when the loop is bypassed?

[Hermetech]

Thanks for clearing that up. So a total of 11.4 ms to 12.1 ms latency if both Inserts are used? I’ll see how I get on with he distortion pedals in the Insert when my cables arrive, might have to keep them before the H90 if it’s noticeable.

[tbskoglund]

If everything is routed in series, yes.

[matfagan]

Thanks for the replies everyone, looks like this is just a flaw of the h90 we have to live with. The insert loops are simply not usable. Perhaps the technology will improve in time but for now I’m going to be running my overdrive before the h90. If I need to use modulation effects pre overdrive I’ll have to swap some cables around… not ideal.

It’s disappointing because one of the major selling points for me was the convenience of the routing. The h90 can still co-exist with my analogue gear, it’s just not as seamless as advertised.

[Herr Mosa]

Remember – it’s not “just a flaw of the h90”, it’s intrinsic to digital signal processing.

Why blame Eventide?

[matfagan]

Perhaps I should clarify, I’m not ‘blaming’ Eventide. They’ve done an amazing job with the h90, and I understand that latency is something that comes with ADA conversion.

As Infection noted earlier, ‘people are different’. I’m not sitting in my bedroom making ambient soundscapes, I play in all kinds of performance situations.

I posted here looking for a solution, or advice on how to minimise the issue. It needs to be said, the h90 has retired around a dozen pedals I’ve been swapping in and out for years depending on the gig, and now it’s all there in one unit. Incredible stuff, however…

If the ability to move pedals around in the signal chain is important to someone considering the h90, they should be aware it’s not flawless in its application.

[Author]

Posts