- This topic is empty.
September 4, 2008 at 8:46 pm #105813ThreeFingersOfLoveParticipant
I have a spectrum analyzer and when the H-8000FW is in bypass, it shows an average level of -100dB and of course everything is ultra-quite. When it's not in bypass mode and the Analog 1, 2, 3, 4 output levels are 0.0 dB the spectum analyzer shows -80dB and I can hear a barely perceptible noise in the background. It's a bit annoying – the noise is definitely there. I need to set the output levels to -6dB to not be able to hear the noise (the spectrum analyzer shows aprox. -86dB) and at -25dB to get -97 dB at the spectrum analyzer). The same is true about the inputs, with -6dB the noise disappears (or at least I can't hear it). So, are the converters that noisy? 🙁 Is it better to leave this at -6dB and use a pre-amplifier either before or after the H-8000FW?
September 4, 2008 at 9:36 pm #117832nickroseModeratorEventide Staff
It depends a little on what you mean by dB.
You probably mean dBu or dBv (I say this because dB is a relative measure, rather than an absolute one, so is not helpful here). See wiki.com for more information on dB, dBm, dBu, dBv, dBV etc).
In this case , the full level output of the H8000 is about +26dBu, so if you are measuring -80dBu with no signal, this suggests a signal to noise level of about 80+26 or 106dB, which is reasonable, depending on the measuring conditions.
Hope this is not too technical. What it suggests is that whatever you have after the H8000 is too sensitive, which is why you are hearing the noise. You are probably also not driving the H8000 with enough signal (the top green led needs to light from time to time.
Once you are driving the H8000 hard enough (use LEVELS/inputs/Pre A/D Gain if necessary) then adjust the LEVELS/outputs/Post D/A Gain to avoid clipping whatever follows the H8000). When all this is done, you will have a dynamic range well over 100dB, and are unlikely to hear any noise.
Essentially, you could have the world's quietest unit (the H8000 is quite good), and if you feed it into a sensitive enough input, you will hear noise. If you then turn up the gain, it will sound worse.
September 5, 2008 at 6:17 pm #117834ThreeFingersOfLoveParticipant
thanks for the valuable info. However, I think I found out that the algorithm itself causes this low noise! Patch is #3910, I have examined it and I think that in this as well as in most patches that use feedback, there exists this noise. Other patches are very – very quiet upon loading.
I do however have a question about the optimal input level. Given the fact that the converters have a 26dBu headroom, Italo mentioned that 4 green LEDs are optimal, you say that the top green led needs to light from time to time, whereas the manual clearly mentions that the signal should be in the yellow area. I am a bit confused… so which is it?
Can you please explain what are the benefits from having a 26dBu headroom? Shouldn't the user be responsible for making sure that the signal never reaches 0dBu?
September 5, 2008 at 7:23 pm #128984nickroseModeratorEventide Staff
The signal should mainly be in the green area, but this implies that the yellow LEDs will light on peaks.
You will get the best signal to noise ration when you have the most signal (no surprise) but you don't want distortion caused by the input clipping, so the red LEDs should light rarely, if at all.
Any real-world signal will have an average or typical level, but will have occasional peaks at a higher level. So, while the user can make sure that his average level is (say) 0dB, he can't control the peaks (without using a limiter or compressor, which will reduce the dynamics). Having the extra 26dB headroom gives you lots of space for peaks without distortion.
September 5, 2008 at 7:50 pm #128987sean.eParticipant
I think the confusion is due to the type of input. Italo has mentioned that for digital inputs, the LEDs should be in the yellow, but not so for analog inputs.
September 19, 2008 at 9:29 pm #129023mbMember
Respectfully, leds go from green to yellow to red.
Generally, it is not a good idea to have inputs averaging as high as possible, that is, top of the green into the first yellow indicators.
Lets say you have 4 analog inputs running into DSP A and DSP B-in parallel.
If you also have them summed at the analog outputs, a normal routing, and even if you are running just the THRU presets on both A and B, (NO GAIN added to the DSPs) you will be already over the top at the outputs.
Real world presets in the 8000 have far too many situations where preset gain stuctures will produce outputs FAR above the level comming in.
As a clear example:
A 1 kHz sine wave enters the analog inputs at -15 dBFS on the 8000 meters.
Bring that signal accross the 4 inputs of DSP A and B, sum them to get all 4 outs from each processor and you are already clipped and thats comming in at minus 15!
Immagine an input level starting in the yellow!
Yes, I know we can adjust levels here and there. But idealy, the only place we should adjust levels would be at the pre A/D converter at inputs and the post D/A converter at outputs.
In practicle use with many, many, presets, if you come in at more than 3 of 4 lights from the bottom of the leds, many presets will produce outputs that are too hot.
- You must be logged in to reply to this topic.