Meridian Unplugged - Powered by MrTechGuy.co.uk Forums Discussion Meridian - Systems +3dB "LevelPerfect" Tweak Performance Improvement Poll

It doesn't actually explain why no G18. As there is a bunch of other stuff 518 does and would allow "Gain" tweaking on the fly.

The real reason no G18 is I think a G18 would be too geeky for Meridian's average customer these days and would not sell enough to justify itself.

I had always wished Meridian would expose 518 like features in the processors (which is definitely doable). But again, too geeky and would just generate service calls for folks that don't know what they are doing or do more harm than good.

All this tweak has done is expose more range in the 518 like Gain feature of the Processor (a rather critical range, namely 0) and a better understanding of it.

The real reason no G18 is I think a G18 would be too geeky for Meridian's average customer these days and would not sell enough to justify itself.

Excellent point, and a very interesting thread

Regards I

George,

Have you ever tried running the G68 in analog with the digital outputs connected to your DSP speakers? I wonder where full scale is setup for analog?

The volume on the G68 or 568 would change the digital volume to the DSP's. No?

I have used digital gain on my 518 for some older CD's that are recorded low. It does work quite well but like I said changing digital gain downward on a 518 and changing the trims in the 568 do not yield the same sound results in an analog setup.

Ron

I would bet if you set a 568/G68/861 to 0dB Calibration (which is -3dB)
And hooked the G68 digital out to any DAC (including DSP speakers).

It would sound very similar to a
SqueezeBox hooked to a 518 set to -3 Gain hooked to that same DAC.

Assuming you set the Noise Shape the same as G68/568/861 (C is the rumor).

When you talk analog with the processor there is just too much we don't know. I'm sure it could be poked at from many sides to figure what it's probably doing in analog as well. But it will always be harder with analog to know what's going on. And likly does not have this artifact at all.

If you set a processor analog, then as you adjust volume the digital outs go up and down. And if you apply offsets, if that changes the digital output or not I'm not sure what it would mean.

Also if the volume and speaker trim output is after internal processing as I believe it is the volume is probably in a 24 bit format for the analog DAC's and the DSP speakers (DSP's set to full output of course) and in an analog setup once my volume control hits digital full scale the analog volume takes over. Trims set to -5 is not a problem for analog. I am sure below a certain digital volume analog kicks in again. The -4 or -5 setting I use is probably a sweet spot or something in the processing is adding some compression. probably in the volume control in the processor. This doesn't get used in the case of the DSP speaker setups. DSP speaker volume in in the speakers.

You guys seem to think you are changing input bit depth but I don't think you are changing it with the trims. You are changing input bit depth to the DSP speakers but that is after processing of Trifield and any other 48 bit processing! The processor is always seeing full scale input if the recording is full scale. If not that's when a 518 is useful.

Ron

Read all my posts.

G68 is doing exactly what 518 is doing. Except you were locked into -3dB plus what ever you subtract for calibration. Using the tweak you are no longer locked into -3dB.

I measured it. It's full scale in and -3dB out (when calibrated to 0dB).

Doesn't matter how many bits is used internally in G68 or DSP Speaker. Folks keep thinking that somehow solves the problem. It doesn't. You've pushed your data down and that's not always good. If anything you may want to push it up if something is recorded low (As you said). If something is recorded too loud (i.e. clipped) there is nothing you can really do. Attenuating won't help it, that most signifcant data is lost if it was recorded too high. If it was recorded too low you've lost LSB bits which isn't nearly as bad. But by raising it you're sliding it into the better performing part of DAC.

You can treat the DSP speaker like any other DAC. Except you can't TRIM them in analog. That's what I'd like to see implemented in DSP firmware. Just like MI did on his "thirdparty DSP" setup. Setting all front to +3dB and compromizing other channels is the closest as we can get (which isn't that bad).

Just want to clarify one minor point that has come up in a couple of posts. The audio samples are signed integers, not unsigned integers. From the LPCM Wiki:

"If the sample is 16-bit signed, the sample range is from -32768 to 32767, with a centerpoint of 0.[8] (For example, signed LPCM data is used on Audio CD, DVD Video, 16-bit LPCM in WAV, audio/L16, etc.) If the sample is 16-bit unsigned, the sample range is from 0 to 65535, with a centerpoint of 32768."

Wiki on Linear PCM

Also, it should be clear from the range value, that the signed integers make use of Two's Complement notation so that only a single value, all zeros, represents 0.

Wiki on Two's Complement

(Sample range would be -32767 to 32767 if One's Complement notation were being used and there would be two values that represent 0.)

Dan

Three things:

1) I'm pretty certain that the 518 and subsequent Meridian resolution enhancement "stretches" the signal to the new word size rather than adding zeros in the LSBs.

2) Meridian must know about this. And the 861v6 is likely to apply 3dB of headroom as well; we thing the headroom is there for a reason, right? I know from personal experience that around 3dB of gain with my 518 can open up the sound of material which is delivered beneath digital full scale. So it's not too much of a leap to imagine that -3dB could cause an opposite effect. On the other hand, a 3dB reduction in a 24-bit signal isn't really that much, so perhaps there is a bug in the system (which includes both software and user )?

3) It's not occurred to me before, but the Meridian comms system probably has weakness in *any* multi-channel systems containing DSP speakers. As things stand, each DSP speaker responds to a single comms master volume. This requires that digital attenuation is present - 518-style - for all DSP speakers/channels which have a non-zero calibration with respect to each other. What Meridian needs is a new comms system which achieves relative calibration of each loudspeaker by applying *volume* offsets at each DSP speaker; this would allow a full-scale signal to be sent to each speaker.

George,

Did you try multichannel input bass frequencies to make sure the -3db setting isn't there for a reason? I bet at some point you guys will run into a problem and clip. Meridian wouldn't throw away resolution to the DSP speakers if there was no reason for it. Maybe there is dither added in the DSP speakers or even gain in the DSP speakers before the DAC in the speakers. For music it probably works fine but with multichannel I bet you'll find scenes where some clipping will take place. I am surprised Meridian hasn't chimed in on this one. Then again they never did with my finding although they did say it was interesting and they would listen. They never commented after that. That was back in 2000-2001.

Ron

It could all be done with the current hardware. It's just a DSP firmware change to allow for 1) Channel Ident 2) Channel Calibration Offset and a Processor firmware change to send the commands to the DSPs.

Processor would then broadcast the "CO" command for each speaker when the system is comes out of standby and say when the volume is changed.

CO103.5 Channel 1 (Front Left) -3.5db attenuation
CO203.5 Channel 2 (Front Right) -3.5db attenuation
CO203.0 Channel 3 (Front Centre) -3.0db attenuation
CO202.0 Channel 4 (Front Sub) -2.0db attenuation
CO204.0 Channel 5 (Rear Left) -4.0db attenuation
CO204.5 Channel 6 (Rear Right) -4.5db attenuation
VN65 Volume Number at 65 (Master volume as displayed)

Simple.

Regards,
Carl

Ron,

I know your not reading all the posts and just catching up.

We highly suspect it's there for headroom for Center Channel boost in Presets. Which so happens to be limited to what. 3dB. If they didn't lower everthing by 3dB it would allow clipping (if someone boosted their Center Preset). But who uses +3dB on center preset?

If you boost Center Channel +3dB you'll be at 0dB FullScale. I have not measured center yet. But I did rears. Rears allows for +20dB boost and you can make them clip !! BUt I think they know rears are usually attenuated and allowed more positive range.

I'll bet an 808 doesn't attenuate 3dB when hooked direct to 8K's. I'll bet it's perfect full scale.

And the more you attenuate for Speaker Calibration the more bits your just throwing away just for the purposes of equalizing them. That's a huge cost. Only because they didn't design DSP speakers themselves to allow for an ANALOG trims. They can do analog trimming inside preamps like yours and they DO. That's why the range is allowed to go positive on analog speakers. They don't do analog Trimming with DSP Speakers and I absolutely know that and that in itself is bad. But they easily could.

Direct preset in G68 now sounds like Direct wired. Rears are always attenuated in Trifield. I'm not worried about clipping. If someone wants to do more measurements, go for it. I'm convinced this is better and I'm pretty convinced why it works (see 518 manual) and why they did it (equalization and option of center boost).

Exactly Carl. That's exactly what they need to do and I'm amazed they have not.

I'd even be happy to manually do it each speaker. But they could make it completely invisible as you show. With existing hardware.