- This topic has 3 replies, 2 voices, and was last updated 2 weeks, 3 days ago by tbskoglundEventide Staff.
January 9, 2023 at 2:26 pm #168378homerParticipant
I was wondering if somebody from Eventide could give me an idea about the to be expected latency with my setup, which is a H9k with the MADI expansion card connected to a DAW equipped with an RME MADI FX PCIe card via optical cable.
First, as some of you may know Cubase has a function to measure latency for what is called an external fx. I specified one such external fx using 2 madi channels routing the signal roundtrip to the H9k via optical Madi.Within the Audio connection section there is a function to measure roundtrip latency to compensate for it. Dumb enough I seem to be unable to get a proper measurement and end up with „0ms“ regardless. Does anyone have an idea why this does not seem to work or how to get it done?
Secondly, I was entertaining the idea that I might be able to use the H9k as kind of a simple ADC for recording instruments like bass or synths. While on paper that should be a no brainer, I was honestly thrown off by the effective latency in my setup. With my monitors having an analog and digital input section (spdif) I was wondering how to best feed them: (A) stay in the digital realm by passing the signal via Spdif or (B) pass an analog output to the monitors ADC and accept some minimal added latency.
While both options should not be too difficult to accomplish, the reality is that fhe MADI Fx card per se does not offer an SPDIF out and even the Analog out is only working with the respective breakout cable which I don’t own.
My pragmatic solution therefore has been to route the signal back to the H9k via MADI and use its Spdif output to feed the DAC section in the monitors directly, which sure enough added another layer of latency. Basically I have put the idea to rest for now but at least in my mind the H9k could be filling some specific gaps when used as ADC or DAC.
Anyway, while I do understand what the RME ASIO driver latency is I would like to understand what the H9k conversions and the Madi expansion routing are adding to it. Could somebody from Eventide help me to identify the values for x,y and z in the calculation below? That would probably also allow me to understand which value the external fx compensation would need to look like if Cubase is able to measure it properly.
- A to D conversion = x ms
- routing signal „Thru“ to the respective Madi channel on the extension board = y ms
- sending the signal to RME Madi Fx and it processing the input = ASIO latency
- routing signal back to the H9K Madi expansion board = y ms
- Internal routing of Madi signal „thru“ to Spdif to active monitors = z ms
January 10, 2023 at 1:18 pm #168404tbskoglundModeratorEventide Staff
All H9000 latency information can be found here: https://cdn.eventideaudio.com/manuals/h9000/2.0/content/setup/DAWLatencyCompensation.html
I am not sure what the issue you are having with the Cubase External FX latency measurement. You should first configure your FX Chain on the H9000, then bypass all of the algorithms, and then trigger the measurement from Cubase to measure the latency of the FX Chain.
There shouldn’t be a noticeable amount of latency when you use a Direct I/O connection to patch the analog inputs to the digital or MADI outputs for tracking.
I would suggest using the RME software to monitor your inputs before you get to the DAW when tracking and that should give you minimal latency.
January 12, 2023 at 5:11 pm #168438homerParticipant
I need to be more precise: I did setup a direct routing (Analog IN 1 > Spdif 1 OUT H9K > Monitor SPDIF IN) and that in fact yields a minimum latency. The real issue is that the in parallel to recording I need to replay the output of Cubase and this works only via MADI to the H9K Spdif out, meaning I do have a parallel setup. My questions was therefore what the overall latency would be, roundabout, considering the ASIO one is known.
Regarding the DAW latency compensation for the External Fx in Cubase: the link to section in the manual is not very helpful. Basically it is saying that H9K latency is dependent on the number of algo blocks between IN and OUT on the FX chain – makes sense. Anyway, it is suggested to keep one Algo block in always in order to stabilize latency and that a dedicated DAW „mode“ is planned, which should take care of managing the dynamics of different number of Algos patched in (reminds me of a lot of promises around the H9K that never materialized). My problem is that when Cubase pings the H9K via Madi nothing seems to come back for it to measure it properly. I recall this was working on my old H8000 with Spdif flawlessly and still I dont have any clue if that is a Cubase issue or something wrong with the MADI implementation and integration on the expansion board on your end. Thx
January 13, 2023 at 5:05 pm #168459tbskoglundModeratorEventide Staff
Sorry, I’m not sure why Cubase is not measuring the latency correctly for you. I don’t have a copy of Cubase to try this with at the moment, but I just tested a similar feature in Logic with a Dante card and it worked correctly. I bypassed the algorithms in the FX Chain, pinged the external FX plug-in, and got a latency measurement. I then removed an algorithm and pinged again, and I got a smaller latency measurement.
Do you have another piece of hardware you can use as an insert in Cubase to try and measure the latency to see if it works there? If you can get audio to pass through the MADI card, then it should work when you ping it for latency.
Is the analog output on the PCIe card a TRS 1/4 stereo jack? If that is all you are missing to connect the card directly to your monitors, I would suggest purchasing a cable or adapter for that rather than routing the audio back to the H9000 and then to your monitors via SPDIF.
Once a Direct I/O routing has been made, that output may not have another input source routed to it. So I’m not sure what you mean by having parallel paths where both the analog inputs and outputs of Cubase via MADI are routed to the H9000’s SPDIF outputs simultaneously.
- You must be logged in to reply to this topic.