- Slides: 75
Setting Up DSP Processors • Gordon Moore CTS • • Gordon. [email protected] com 1 -800 -821 -1121
DSP § Digital Signal Processors § Can be any device that modifies a digital signal (video, audio – anything) § In this class – we will be talking about Audio § NOT a “How-to” for specific manufacturers § Use their training
§ Apologies to any Manufacturers whose screens are not featured – no slight intended. § The CONCEPTS are the core of this course - no brands endorsed or rejected.
What to expect § Signal pathway organization § Setting the array of “modules” or functions available
Why? § Most DSP processors are barely used to their best capability. § Units returned for service usually have just the simplest functions enabled while many enhancements are not touched.
Most commonly unused or forgotten § § § Compressors Limiters Input filtering NOM bus Delays
Most commonly set up functions § § Input Gain (But often badly set) Routing (inputs to outputs) Equalization for outputs Controls interfaces
DSP organization § Flexible Architecture § Fixed Architecture § Dedicated function (one capability only) § Multi-function with fixed pathway Hybrid Architecture – some fixed, some flexibility in routing
Flexible Architecture All functions can be configured in a “drag-n-drop” environment. Audio functions can be placed almost anywhere along the signal chain in any order. Characterized by a drag and drop GUI (Graphical User Interface) and/or “fuel gauge”.
Advantages Complete flexibility – you can do some amazing things within a single box and develop very complex signal paths. Excellent choice for systems where complexity and/or multiple applications may come into play. Airports Large scale paging systems Complex communications systems Very little you cannot accomplish.
Disadvantages § Flexible Architecture may be more costly on a per channel basis § Requires more DSP power – memory register stacks must be allocated for any eventuality – code space cannot be optimized.
Fixed Architecture § Dedicated function § Does one type of function § Compression/limiting § Or § Equalization § Or § Signal Routing
Advantage Fixed architecture is simple to set up and operate – may not even require a computer Less cost for the box – may not be as cost effective as a combined DSP capability Sets up much like analog counterpart Excellent choice for existing system upgrade
Disadvantage § Very limited in scope of function § May not be very scalable
Fixed Architecture Multi-function Has multiple functions in a fixed pathway Generally fairly cost effective Limited in terms of routing and or set up choices. Advantage – predictable known good pathway No gauge Disadvantage – few if any routing choices
Hybrid Architecture § Combination of routing choices plus fixed multiple function signal pathway § Advantages – § Allows flexibility as far a signal routing goes – what inputs show up at what outputs § Optimizes DSP processing power – memory stacks and registers can be more tightly packed § Disadvantages § May not fulfill all needs in a system
Enough Boring stuff – Let’s set up some functions § § Input Gain Filters – Input and/or Output § § § Equalization Feedback suppression Crossovers Noise reduction Dynamics § Compressors § Limiters § Gates § Routing
Dynamics § Those functions affecting gain structure and levels
INPUT GAIN § § § Most important setting – GET THIS RIGHT! Always set up – but not necessarily well set up Microphones § Handheld Vocals = 35 d. B minimum § Handheld Presentation = 45 d. B § Gooseneck desk = 45 d. B § Boundary mic = 55 d. B § Any further away = 60 db+ § Ceiling – as hot as you can get it
INPUT GAIN § Multimedia § Unbalanced? Consumer = +10 § Balanced? Professional = 0 to -8
Signal to noise ratio Poor S/N Ratio Line Level Mic Level Noise Floor Mic Preamp Mixer Signal processors Amps
Signal to noise ratio Line Level Mic Level Noise Floor Good S/N Ratio Mic Preamp Mixer Signal processors Amps
Signal to noise ratio Line Level Mic Level Noise Floor Mic Preamp Mixer Signal processors Amps
THRESHOLDS § The level at which the desired function becomes active § Generally speaking a lower threshold level means it will activate earlier. § Recommended starting threshold for most line level (post preamp) functions = 0 d. Bu
AGC and/or Levellers § § Automatic Gain Control RAISES gain if signal too low Compresses if signal too high BE VERY CAREFUL with these § Can run a room into feedback if used on amplified inputs § Primary application – to capture weak signals for recording or transmission § Start with threshold set at 0 d. B – keep gain centered at line level
Ambient Level Control Uses a reference microphone to measure room noise level that automatically adjusts system for noisier environment Reference microphone may be dedicated microphone – only purpose is reference, or may be designated microphone – used in system but designated to be the reference signal Read the manual
COMPRESSORS § § Control dynamics - loudest to softest Useful for keeping level under control Meek versus motivational speaker Rarely set up
Compressor settings - Ratio § Ratio – The amount of actual level increase above threshold that will yield ONE decibel in actual gain change after the compressor. § Example – 3: 1 ratio § For every 3 DB the gain increases above threshold, the final level will change only one d. B § SO, if level jumps 9 DB, the final level will jump only 3 d. B § FM is broadcast at typical 10: 1 ratio
Compressor settings – Threshold, Attack, Release § Threshold - Level at which compressor begins to engage and affect level. § Attack – time in milliseconds the compressor begins to make changes after level exceeds threshold § Release – time in milliseconds the compressor lets go after level settles below threshold. § Makeup or post compressor gain § Compensation in level to make up for compressor reduction in signal.
Suggested setting for compressors § Speech systems (conference rooms, boardrooms, etc) Ratio = 3: 1 Attack = 10 -20 ms Release = 200 -500 ms Threshold = 0 If initial input gain was set to result in 0 d. B level, then it would take a 60 d. B increase at the mic to hit the +20 d. B limit of input (clipping) MUSIC or Multimedia – try increasing ratio to 6: 1 BE CAREFUL – too much = bad
LIMITERS § Basically, a compressor with an infinite ratio § Absolute ceiling to maximum level § Protects downstream gear by preventing severe clipping and overdriving amps and speakers § Many amplifiers have built in limiters to protect themselves. § ALWAYS set limiter threshold above threshold of compressor § Otherwise, compressor will never engage § Good for spikes like dropped microphones, cymbals, plosives (P, D, T)
Suggested Limiter settings § Threshold – 15 db higher than compressor = +15 d. B if 0 d. B for compressor § Attack – faster than compressor = 2 ms or quicker § Release – 200 ms or less
Some dynamics filters can be frequency specific § Compressor with low pass filter good for controlling proximity effect while allowing high frequencies to pass unaffected.
EXPANDERS § Increases gain if signal very low such as weak talker § BE VERY CAREFUL HERE – Expanders, in an amplified environment, can push system into ear bleeding feedback. § Primarily intended for recording and or transmission.
GATES § Gates activate a channel, allowing it to pass, once the level passes above threshold. § Found in some automixers § Useful for noise control (noisy multimedia source, for example) § Originated in music
Attack , Release, Threshold § Attack – try relatively fast settings, 1 ms to begin § Release – start at 50 ms § Threshold – depends on place in chain § If after input gain – 0 d. B is a good starting place § Lower if not getting a reliable start
Spectral § Affects the frequency response of the system
EQUALIZATION (Inputs) § Equalization is one of the most commonly used functions § Input EQ is generally for tonality control – adjusting the tonal content so each input sounds similar.
EQUALIZATION (Outputs) § Generally used for speaker compensation § Adjusting for “quirks” or characteristics in the loudspeaker response. § You cannot EQ a “room”
Equalization patterns § § Pass – Low, High or Band Shelving – Low or High Parametric - Notch Graphic
“Q” § No, not James Bond’s gadget guy § The ratio of filter width to depth at 3 d. B roll off points
Low Q – Wide band width
High Q – Narrow Band width
Filter Slope or “Order” § Rate of attenuation on filter - Shown in terms of d. B/octave § Octave – doubling of frequency § § § First order = 6 d. B per octave Second Order = 12 d. B per Octave Third Order = 18 d. B/Octave Fourth Order – 24 d. B/Octave Each order equals another 6 db of roll off. § That means 4 times factor in power level § If 6 db down from 100 watts = 25 watts.
First Order – 6 d. B per octave (High pass)
Same filter – Second order
Eighth Order- 48 d. B/octave
Parametric § § Fully configurable Boost or Cut adjustable Center Frequency selectable Q selectable
Shelving – Boost or cut, High or Low
Filters - Graphic EQ
CROSSOVERS § Used for bi-amplified and tri-amplified systems § Low frequency content sent to bass amplifiers § Mid and high range sent to appropriate amplifiers § § § Separate amplifiers involved Large scale concert systems High order - 4 th to 8 th order filters Bass – 250 Hz or lower For tri-amplified – start at 4 K for high pass
FEEDBACK SUPPRESSION § Should be set up LAST after Equalization § Smooth response FIRST, then take care of Feedback nodes § A time domain issue coupled with Frequency domain § Notch Filters – very tight § But too many can badly affect content
Set up tip – One mic at a time § 1. Before final EQ – run feedback “eliminator” first § 2. Make note of first three feedback freqs § 3. Construct three very tight notch filters at INPUT on those frequencies § 4. Reset the feedback filters § 5. engage feedback filters again after equalization and system is at operational levels.
NOISE REDUCTION § Popular new algorithms that “sample” the noise floor § Noise floor – acoustical and electronic noise that is NOT wanted in system § Air Conditioning § Fan noise § Laptops near boundary mics § computers § Projectors § Electronic noise – noisy sound cards in computers etc.
Setting Noise Reduction Filters § § Canceller Depth – depends on the amount of noise Quiet conference room with little to no noise may not need this. Computer and projector fan noise – try starting at 9 d. B. Heavy room noise – large attendance training room or bad air conditioning rumble, try 12 d. B § Remember, these filters remove spectral content – none are perfect and they will affect your room response. § DON”T GET CARRIED AWAY!
Signal Generators § § White noise – equal energy per frequency Pink Noise – equal energy per octave Tones Primary use – test and measurement § Use pink noise to set up your levels in the room § If your gain structure is correct – this will allow amplifiers settings to be accurately set up § Use test tones for gain staging and/or speaker alignment § Secondary Use § Noise masking – covering conversation or background ambient noise § Alarms
ROUTING § Matrix – § Rarely overlooked § Determines which inputs go to which outputs § (technically speaking “gozindas to gozoudas”) § Some traps here § Watch for § Feedback loops § NOM bus assignment – VERY IMPORTANT
Selecting NOM bus and action § NOM= Number of Open Microphones § Determines interactions of microphones in automixing § Failing to select correct NOM interaction can affect echo cancellation, and gain before feedback § Choices may include § Chairman Over-ride § Auto mix (or Normal) § Background (or Ducking)
Feedback loops § Common error in setup – requires careful documentation and double checking – especially when dealing with mix minus conferencing systems. Sending an input BACK to itself.
Delays § § Often neglected Primarily used for time alignment § Loudspeaker stacks in large venues § Input alignment § Loudspeaker alignment § Secondary use for spatial referencing (Haas effect) § Localization of sound based on first heard § Set up tip – § Try setting for 10 ms or 10 ft or 3. 3 m FARTHER than distance would indicate. § Example – Speaker is 68 ft from stage – instead of 60. 5 ms, try 70. 5.
Some architectural Do’s and Don’ts § In Flexible Architecture – General rules – always exceptions § Do NOT put compressors before Equalizers/Filters § The EQ can take out energy that would falsely trigger compressor – so put compressor AFTER filtering § Do NOT put limiters before compressors § Do NOT put limiter thresholds lower than other devices – especially compressors § Be careful about thresholds for automixers – (gated designs) § Too low and room noise will open mics § Too high and you may lose the first characters § Shared gain mixers will not have thresholds
Hybrid Architecture Do’s and Don’t § Don’t create a feedback loop – be careful where you send inputs § Don’t go overboard on settings § “Too much of a good thing is wonderful” does NOT apply here.
Troubleshooting tips § Ringing or severe echo in audio § Internal feedback loop § Signal generators or incoming conferencing too loud § Poor input gain structure – HOT IT UP! § HISSsssss or noisssse in system § Bad gain structure – inputs too low – Amplifiers too high § University reasoning – wrong solution to the right problem. § Amps set to full output § Inputs set low § Correct solution – lock up the amplifier controls.
Questions? § Fill in your reviews please!