Jim Brown answers this question, “What should be on product data sheets?” He gives some very insightful information. Thank you Jim!
The full version of Jim’s article includes sections on Intercom and Infrared systems.
It’s time to talk about documentation – what manufacturers need to tell us about their products, and what we need to tell our customers about the systems we sell them. This time around, we’ll focus on what manufacturers need to tell us. Back in 1996 I wrote, “Over the past ten years, we have observed with dismay an ongoing “dumbing down” of product data sheets, with the result that many important specifications are not available to the designer or prospective purchaser. There has also developed an increasingly neglectful approach to service manuals.
It is as if the laissez faire practices of the computer industry have infiltrated the professional audio world. But unlike the computer industry, where equipment is often obsolete in 2-4 year cycles, professional systems are generally intended for a life cycle of 20-25 years. In our world, systems often outlast the vendors of their components.” In the intervening years, it hasn’t gotten any better.
A product data sheet should tell us everything we need to know to make a buying decision, and to integrate the product into the system. Ah, you say, integration information is in the manual, not the data sheet. But designers rarely have a manual – we’re working from the data sheet! When a system is being designed, it’s all on paper (or bytes) – we haven’t bought the boxes yet, so we don’t have manuals, and we can’t open the product up to check it out (nor do we have time to do so). What should be on a product data sheet? Here’s my list from 1996, with comments added today.
For small signal electronics (including digital products, intercoms, infrared systems, wireless mic systems, and the inputs of power amplifiers):
- Input and output configuration – balanced or unbalanced? If a transformer is used, who manufactured it? Does it have a Faraday shield?
- Input and output circuit types – for example, are the outputs symmetrical, or “impedance balanced” with asymmetrical drive (good) or “servo balanced” (bad)? Is the input a good differential receiver (good), or a Whitlock “Ingenius” circuit (great!)?
- Input and output impedances – that is, the actual output impedance of the line driver, and the input impedance of the input stage (including the low pass filter)
- Input and output circuit balance – what are the tolerances on the balance of input and output R and C?
- Common mode input impedance vs. frequency – this specification, and the preceding one, can tell us a lot about how well the circuit can reject noise [Bill Whitlock has observed that this and the previous item could be covered by a realistic CMRR specification per IEC60268-3]
- Rated output load impedance in the form of equivalent parallel R and C
- Input and output clip levels and “reference levels” – we can’t even begin to design a system without knowing these levels, yet very few power amplifier data sheets provide this information!
- Minimum and maximum gain
- THD and IM Distortion 1 dB below clip and 20 dB below clip. Most systems run well below clip most of the time – how do they sound when they are closer to the bottom of their A/D and D/A converters than the top?
- Output noise, in volts or dB referenced to clip, with gain and bandwidth referenced
- Input noise, in volts or dB referenced to clip, with gain and bandwidth referenced
- Any configuration and/or switching options, and how they affect the above parameters
- Input and output connectorization – what connectors are used, and does their wiring comply with AES14 (XL-connector wiring), AES26 (polarity), and AES48 (pin 1 problems)?
- A plot of the amplitude response showing +/- 0.5 dB limits and the low pass filter response to protect against RF interference.
- Are the outputs also protected by low pass filters? Many headphone amplifiers and power amplifiers have serious susceptibility to RF coupled onto their output wiring.
- AC input power (watts or amps) – we must provide electrical load information to the project EE, and heat loads to the project ME
- Mains voltage/frequency options (some of us are designing projects outside North America)
- UL/CSA listing (and comparable safety agency compliance for other countries)
- CE or FCC Class B compliance for digital products (RF interference to wireless mics)
For digital equipment:
- A/D and D/A linearity (Richard Cabot has written a lot about this)
- All parameters listed for small signal devices
- Latency from each input to each output, with no processing and with all possible processing conditions
For power amplifiers:
- All parameters listed for small signal devices, but with output parameters referenced to all rated conditions of output power and load impedance
- AC input power at idle (in the “on” condition).
- AC input power for moderately compressed audio near clip.
- Audio output power for uncompressed pink noise just below clip for the rated load conditions
- The K-factor with no signal (in the on condition), with uncompressed pink noise 20 dB below clip, and compressed pink noise at clip for all rated load conditions
- Start-up behavior (i.e., does the amplifier do anything to control or randomize the timing of inrush current at power-up, and what is the nature of the inrush current, worst case?) If the unit has an automatic turn-on circuit, how quickly does it turn on, and how long does it stay on?
- Re-start behavior – If the unit was on when power dropped, will it remember that for the duration of a short (less than a half hour) power failure and turn on automatically when power is restored? This can be important in a system where the amplifiers themselves are the sequencing system – we don’t want the sound operator in a facility to cause a panic when he/she does a mad dash through the audience to restore power after a failure! Can the memory be defeated (so that it can be used with an external sequencing system)?
- Details of all signal processing integral to the amplifier (compression, limiting, “smart” turn-on, etc.). Can it be disabled without opening the box (and without modifying the circuitry)?
For any product controllable via electronic interface:
- All parameters listed for small signal and/or power amplifiers, as applicable
- Listing of any industry standards with which it complies
- Full details of any non-standard protocols
- I/O connectorization
- Small signal parameters as defined above for all inputs and outputs, including patch points.
- A complete signal flow diagram that I can read without a magnifying glass. What sends are pre/post EQ, pre/post fade? Is phantom power present on line inputs? Can it be turned off “per channel?” What is the monitoring/solo logic? if there is LCR panning, how is it implemented? What options are there to configure these features differently?
- A high resolution FFT showing the spectral content of the electrical noise – is there synchronous noise (for example, a clock and its harmonics) that will be present at multiple inputs and add coherently to degrade signal-to-noise ratio by 3 dB per doubling of the number of inputs?
- The A-weighted and C-weighted acoustic noise level produced by its power supply – many otherwise fine mix consoles add 20 dB (or more!) to the low ambient noise levels of a well designed performance space!
- Good data on the equalization provided.
For signal processing equipment:
- All parameters listed for other small signal equipment
- The nature of any dynamics processing – what controls are available to set thresholds and transfer characteristics, and the transfer characteristic itself. Can the processing be configured to have a “soft knee,” or is it only a hard transition from no processing to gain reduction (or expansion)? What range of time constants and compression/expansion ratios are available?
- Specifications for all linear processing it provides – equalization, crossovers, etc.
- On-axis sensitivity referenced to 2.83 volts (swept), or to the voltage equivalent to one watt for a resistive load equal to the nominal impedance (for example, 2 volts for a 4 ohm loudspeaker)
- Impedance versus frequency (swept), and the AES impedance
- On-axis amplitude response to a resolution of at least 1/10 octave
- Off-axis amplitude response in 5 degree increments with a resolution of at least 1/3 octave. This data should be presented as a family of curves normalized to the on-axis response in increments of 10 degrees within the nominal coverage angle of the loudspeaker (5 degrees for narrow-coverage devices).
- Modeling data should also be provided in electronic form as complex response data (that is, magnitude and phase, not normalized) for direct use with EASE 4 and other modeling software.
- If the loudspeaker is a multi-way device, the data should also be provided as a .dll file (that is, separate data for each driver, including crossover response and spatial relationships between drivers), so that arrays can be designed and modeled with greater accuracy. AES papers by Dave Gunness are the definitive references for these techniques.
- Peak and average power ratings versus frequency
- Input and output connectorization
- Crossover parameters – filter type, frequencies, number of poles each way, which (if any) components are passive (internal to the loudspeaker) and which are active
- Any switch or control options (crossover level, for example)
- Accurate physical dimensions, including the reference point for all electroacoustic data
- If the loudspeaker is active (that is, one or more integral amplifiers, crossovers, signal processing), all parameters listed for amplifiers and signal processing equipment
- Physical principal on which mic operates (i.e., dynamic, condenser, electret condenser, ribbon, etc.)
- Generic polar pattern – omni, cardioid, super/hyper-cardioid, bi-directional
- On axis sensitivity referenced to 1 Pa
- Minimum load impedance, specified as parallel R and C
- Output impedance, including a plot of Z vs. frequency for dynamic mics
- Swept on-axis response at 1 m, and if intended for closeup use, also at 1-2 inches. This data should be at a resolution of at least 1/10 octave (load conditions should be representative of real world use – 1 K in parallel with at least 2 nF)
- Off-axis amplitude response in 1/3 octave, 5 degree increments, normalized to on-axis (in other words, data suitable for use in EASE and other modeling software)
- Maximum SPL for rated distortion with rated load
- Any switch options, including the effect on electrical and electro-acoustic parameters
- For lavaliers, a schematic diagram with pinouts (to allow design for use with wireless mics)
- For all phantom powered components:
- Minimum current and voltage at which the component meets all of its specifications
- Current drawn from 48v/6.8K phantom circuit
- Voltage to which the mic pulls the phantom powered mic line
For all battery operated components:
- Current drain and tolerances on battery voltage
- Battery types utilized
- Any physical restraints on specific brands of battery
For wireless systems:
- All parameters listed for battery operated components
- All parameters listed for small signal electronics applied to inputs and outputs
- All parameters listed for microphones applied to acoustic performance
- All parameters relevant to the powering and connection of a lavalier microphone
- Microphone element used
- Operating frequencies on which the unit can use without modification
- Spurious voltage levels of local oscillator and its harmonics at the antenna connector with a 50-75 ohm source driving the receiver input
- Form of modulation and RF bandwidth
- Intermediate Frequency and bandwidth, low or high side injection
- Noise reduction system used – system, ratio, and pre-emphasis
- Input voltage versus frequency at which intermod or desense occurs
- Acceptable frequency spacing between transmitters in the same system
- RF overload parameters – third order intercept, blocking dynamic range
- Measures included to reduce intermod and spurious responses. Does the transmitter include a circulator? What selectivity is between the receiver input connector and the first electronic stage?
For all products using digital technology, including CD and DVD players, DSP units, power amplifiers, and switching power supplies:
- Does it conform to FCC Part 15 Class B?
Some things I don’t care to ever see again:
- Polar graphs (too hard to interpret, too easy to gloss over problems)
- Frequency response numbers with no dB reference or limits
- “Typical” parameters
- “Architects and engineers specifications”
Now, I realize that at this point, the marketing guys are rolling their eyes in disbelief – not only can we not put all this info on a data sheet, but why in the world would anyone care about all of this obscure information? But the hard reality is that a good system designer needs to have nearly every single piece of information listed to make an intelligent purchasing decision, and/or to integrate the product into a system! These are marketing questions! Consider these real life design issues.
Circuit configuration – are all the inputs and outputs balanced, or will transformers be required at some interconnecting points?
Circuit capabilities – can the output stage drive the capacitance of long cables between the mix location and the amplifier racks?
Noise immunity – are the inputs and outputs well filtered against RF, and sufficiently well balanced to minimize the coupling of hum and buzz from strong magnetic fields?
Does the equipment put out enough RF trash to degrade the performance of my wireless mic systems? To degrade the reception of AM, FM, and TV when needed? To interfere with neighbors? Russ O’Toole recently encountered a digital power amplifier that was “running hot.” When he asked me about it, I first suspected oscillation, and suggested that he hang a scope across the output. He found several volts of unfiltered digital trash in the low MHz range, which, when connected to exposed loudspeaker wiring, caused enough interference that he couldn’t listen to strong local broadcast stations!
Gain/level matching — are resistive pads (attenuators) needed at equipment inputs because the input stage clips well below the output level of the equipment that drives it? Many power amplifiers have this design “feature,” thanks to the disastrous pair of IEC standards (IEC 61938 and IEC 60268-3) that not only call for 0.5 volt input sensitivity, but fail to require specification (or even measurement) of input clip levels on product data sheets! All of this to make it easier for manufacturers to sell the same product to both consumers and professionals, but make it more difficult for us to use it. The EMC Working Group (SC-05-05) of the AES Standards committee has initiated a project to address this mess (AES-X152).
One of the most common errors in the setup of audio systems is the improper gain matching that results when power amplifiers are run wide open (that is, at full gain, so that 1 volt produces full output). These errors can easily degrade signal to noise ratio and susceptibility to hum, buzz, and RF by 20 dB, and the solution is nearly free – if the information is on the data sheet!
If I want to use a compressor/limiter to protect the input of my DSP unit (or power amplifiers), will the input stage of the compressor/limiter clip before the output stage that drives it? For years, the otherwise excellent dbx compressor/limiters had this fundamental design flaw! What were they thinking?
Many sound reinforcement systems are at the edge of what latency can be tolerated by virtue of the acoustic time of flight between loudspeakers. Latency of digital snakes, digital routers, digital mix consoles, digital power amplifiers, and digital signal processing systems can easily push a system over the edge. Latency to the audience is usually, but not always, far less of an issue. Manufacturers demonstrate their ignorance of our world when they fail to publish this specification, or are careless about defining or minimizing it.
Every system designer needs a reasonable estimate of the electrical power required, the k-factor, and the heat produced by our systems, but few of us have a clue about what these requirements are for the systems we design. Why? When is the last time you hung a ammeter on the feeders to an operating sound system? I routinely do so when I inspect a system, and I rarely see currents that exceed quiescent (that is, on and idling) currents by a factor of more than about 2:1. When was the last time you saw manufacturer’s data for the current drawn by a power amplifier carrying actual program material near clip? I never have! Is that number “soft?” Of course it is, but it’s a lot closer to 2x quiescent than the 10-20x quiescent values listed on data sheets for sine waves and heavily compressed pink noise!
Neil Muncy has observed that the electrical power system for nearly every sound is over-specified by a factor of at least 5:1. Neil notes that the sound system isolation transformer’s magnetizing current is often greater than the full load of the sound system itself! Why? We simply don’t understand the load requirements of our sound systems, because we’ve never measured them and the manufacturers give only us “bench” test data (sine waves, highly compressed pink noise) that don’t relate to the way we use them.
Why is this a problem? Isn’t it a lot better to be conservative, to play it safe? Of course not – good engineering is putting money where it is needed, and not where it isn’t! The money it takes to over-engineer the electrical system – bigger copper, that over-sized transformer, bigger conduit, and other parts of the system to support the exaggerated current requirements, more labor to install it – as well as the HVAC system to over-cool the amplifier racks, ultimately reduces the money we have to spend on things that will actually make the sound system work better – steel conduit, a better mix console, a better selection of mics, more signal processing toys, surround/EFX loudspeakers that had to be cut from the budget, etc. jb