M - What?
(first published in Line Up magazine)
Perhaps there was a time when all station outputs were continuously monitored by an engineer watching a PPM and listening but stereo put an end to it! It had to, because most radio and TV stations were broadcasting not only in stereo but also in mono. Even more complex issues arise when multi-channel surround formats are considered but this article considers only the stereo broadcast hardware now in general use. If surround material is sent over this, it has to accept the confines of the mono and stereo transmission system.
It often helps to consider first the extreme cases of wide and narrow images. Double mono is a stereo signal, but with a left channel totally identical to the right. There is no difference between left and right so subtracting them leaves nothing. Such a signal is excellently mono compatible as it is mono but image width is a bit lacking!
A fairly wide stereo signal is created if spot microphones are panned fully left or fully right, but this is not the widest possible. Double mono with a polarity inversion between left and right i.e. "out of phase" mono, is even wider. When we only had left-right stereo, the concept of images wider than the space between the speakers could seem a bit odd but nowadays this is easily appreciated with a Dolby Pro Logic decoder. Such a signal gets taken way outside the L-R plane and deposited behind you!
Broadcasters may sometimes have the luxury of mixing an independent mono signal but usually the mono level is controlled by manipulating the left and right signals. It is worth remembering that FM radio is an application of M-S signals so common that people sometimes forget that the main FM signal is a mono (M) signal. Stereo reception is only possible because a separate difference (S) signal is transmitted on a 38 kHz sub-carrier and FM stereo radios include an M-S to L-R conversion matrix. Accurately metering the level of the mono signal that will eventually be derived for transmissions purposes is therefore essential, even if this signal does not exist in a discrete "routable" form in the audio control room.
This immediately imposes a problem for the radio broadcaster who must take material that originates in stereo and reduce it to mono. Should he just add left and right or do something more subtle? Whatever is done in the systems that combine left and right for transmission must also be used in the metering systems if they are to show anything useful. TV also has the same challenge as it must derive mono sums to feed the FM mono service from the left-right signals being digitally or otherwise encoded.
In the UK, BBC Radio were the first broadcasters to address the issue of monitoring mono and stereo at the same time and came up with the solution using two twin needle PPMs. The "stereo" meter shows left and right signals on nautically inspired red and green needles and the other meter had a conventional white needle to show the mono level and a yellow (later, orange) needle to show the difference signal. The aim is to control signals to peak to the preferred levels on the left/right outputs and at the same time achieve a similar degree of modulation on the mono outputs to what would have been achieved had this been under separate manual control. Sadly this can never be done perfectly as the size of the mono sum depends on what is being summed.
For example consider the super narrow, double mono signal. Whilst not the most obvious stereo material, a centrally panned mono microphone for announcers/DJs/commentators is absent from few broadcasts. The green line represents around 8 mS of audio peaking to 40% or around 8 dB below peak. We'll avoid getting bogged down in instantaneous peak levels versus what a PPM will read if we accept that these examples are all comparative and the dB figures are voltage levels not powers.
If this is a right channel signal and we add an identical left to it giving mono by simple addition we get a signal (white) now hitting 80% or around 2 dB below peak. This is 6 dB higher than our left and right signals. Using a summing network that gives (L + R) -6 dB could be nice with identical levels on left, right and mono outputs - but only for double mono signals.
The extreme case of out of phase mono, (R=-L) obviously has a L + R of nothing, regardless of attenuation! Real world stereo is somewhere between these extremes and the diagram to the right gives an idea what can happen. The red and green lines are left and right and the white line is the mono sum which is now something like 3 dB greater than the left and right signals.
Very wide "spacey" images might have values less than 3 dB above the left and right signals but BBC research for then new stereo radio services concluded that 3 dB was a typical figure for the stereo they expected to broadcast.
This resulted in the definition of mono summing networks as being (L + R) 3 dB, later known as "M-3" or just "M3" systems. Stereo line-up tone was therefore sent at -3 dBu on each leg. Simple summing gave a +-3 dBu mono, but the 3 dB attenuation returned it to 0 dBu
Stereo television brought a new challenge because pioneering work, again from the BBC, determined that over-wide stereo images were undesirable. Whilst low level atmospheres could be as wide as sounded suitable, it was found that the main programme content should not be too far from centre. This has the effect of placing that part of the mix that usually gives the highest levels in something that was close to double mono. Summing matrices of (L + R) 3 dB meant stereo television ran the risk of having mono levels up to 3 dB higher than might be desired. Added to this, many television sources were and still are true mono so there was a real risk of over-modulated mono transmissions! A solution might have been to hold back L / R levels a little more, but this does not fit with the ever present desire of stations to ensure they sound no less loud than their competitors!
Back in the early 1990s, UK commercial television partially adopted a compromise with the 3 dB dip replaced by 4.5 dB Double mono signals of 0 dBu gave a simple sum to +6 dBu which the "dip" held back to +1.5 dBu Whilst not a bad scheme for programme audio it gave line-up headaches! The 3 dB tone required on left and right for M3 systems could be set fairly accurately on a UK PPM reading 3¼. Meters with VU style scaling and a 0 mark at ++4 dBu, as fitted to many VTRs, read -7.
M4.5 systems implied the left and right line up tone should be 1.5 dBu, a PPM position of 3 and 5/8ths which is impossible to set with any degree of certainty! At this point, engineering minds took the technically correct view that even this, still allowed over-modulation and the only safe design for mono combiners was to have a gain of (L + R) -6 dB. Line up tone should then be 0 dBu on L and R which means derived mono is also 0 dBu
This is the declared norm now for many places, including ITV network and satellite companies and also for the delivery of digital material to the BBC. However, matters are not totally clear-cut as the BBC's analogue delivery specification still specifies M3, a standard that still exists in many TV systems! In the radio world it is a reasonable prospect that most programmes will be truly stereo so here, M3 continues, generally without question.
For the converted...
M3 and its sister, S3 also occur in M-S to L-R conversion matrices with good reason. Conversion matrices are symmetrical; put M-S in and out comes L-R. Put L-R into the same matrix and you get M-S. This makes for simple conversion as many times as it might be useful, but now consider the amplitude. If we apply 0 dBu to each leg of an M0 matrix i.e. no attenuation, the summed output will be +6 dBu and the subtracted output will be nothing. If these signals are applied to another identical matrix the summed output will be +6 dBu and the subtracted output will also be +6 dBu The signal that began as 0 dBu on both legs is now +6 dBu on both legs! Using an M3 and S3 gain structure allows matrices to be cascaded as often as required with no overall level changes which is clearly a useful feature!
Whilst there may be a good argument to retain M3 for M-S conversion matrices, M6 is becoming established so it might seem nice to have this everywhere. Sadly this is likely to take some time to achieve and is easily understood with a moment's thought. There are many places stereo to mono conversion or a related process takes place in a typical broadcast audio system but should all of these be modified?
Sound desk stereo channels often have selectors at the input to produce double mono through both left and right paths with this double mono being either the left or the right input or perhaps the sum of both. In this case it is fair to assume that the source is either true stereo with some width or that only the left or right input is being used. This suggests the combiner should be M0. In any event, this is a desk source that is hopefully going to be subject to careful control!
Mixer mono inputs have pan controls and these often have a central dip of around 3 dB This may seem odd in an M6 environment, but consider for a moment what is being done. The purpose of the pan pot is to allow the source to be moved across the sound stage in a smooth way. Ideally, the attenuation between end and centre should follow a square law so that at the halfway point the level is square root 2 of what is was at the end. Such a potentiometer is not simple to produce and mixer makers commonly use a linear potentiometer with a "pull up" resistor.
With carefully buffered designs, this resistor is typically 35% of the value of the potentiometer so a channel with 10K pots might have 3K6 resistors. To test the virtue of the 3 dB dip, increase the value of the pull up resistor or remove it all together to get a -6 dB. pan dip. This destroys the approximately constant power concept of the 3 dB pan pot which was why the design had survived for so many years. 3 dB therefore continues to be an appropriate figure for most systems.
We can also question whether pan pots should have a centre dip, or an "end rise". Stereo mixer inputs often have balance controls and it seems reasonable that with them set central, signal levels pass without gain or loss. In an M6 system where the norm is for the pan pot to be in use, it can be argued that, like the balance control, the centre point should have unity gain so left and right outputs are at unity and so is the M6 mono. Panning left or right gives a lift of 3 dB to that side and full cut to the other with the M6 mono being 3 dB This also means pan pots can be switched in without the 3 dB drop that happens with many mixers!
Monitoring & metering
Now consider the mono function on the sound desk's speaker control panel. Here the purpose is to allow the effect of mono audio to be judged but the appropriate gain will depend whether it is heard on one or two speakers. Many commercial mixing consoles send mono to both speakers so the use of at least M3, if not M6 is essential or there will be an apparent increase in sound level. Which value is right depends whether we assume the programme is close to double mono or is wider stereo and the final decision should be to work from the same assumptions as the rest of the broadcast path. If it is M6, each of the two speakers should receive an M6 signal.
More specialised broadcast mixers give an option to route the mono signal to only one speaker. In this case an M6 mix will almost certainly sound quieter than when listening in stereo and it will be necessary to adjust the speaker level control.
Meter panels are more straightforward. If the desk derives its own mono output then the meter should display this signal with the mono combiner complying with the M3 or M6 system design. Where the desk has mono metering driven from the stereo outputs (as with UK style "L-R/M-S" twin needle pairs), the meter must be calibrated to M3 or M6 depending what is done in mono combiners later in the chain in MCR or at wherever the mono feed for the transmitter is produced.
Line up tone sources should follow the levels shown in the table below but remember that not all two channel paths are stereo. News VTRs may have voice on one track and background effects on the other, so are treated as two mono tracks.
Meter standard Mono M3 M6 dBu 0 -3 0 UK PPM 4 - 3¼ 5 VU (0VU=+4dBu) -4 -7 -4 Digital (0dBFS=+18dBu) -18 -21 -18 Digital (0dBFS=+20dBu) -20 -23 -20
All this is intended to show that going from M3 to M6 is not just a matter of adjusting meters. In large systems, many hundreds of signals may need re-calibration and it is little wonder that few organisations have achieved absolute consistency! Of course the doubter might say that mismatches typically cause errors of 3 dB and automated processing towards the back end of most transmission paths will make greater adjustments than this. However, such a view ignores the possibilities of cascaded errors in complex signal chains and cannot be a proper basis for a well engineered system.
All material is copyright PHM © 2004.
P H M
(P H Music) : PO Box 383
Bury : BL8 4WX : GB
tel/fax: +44 (0)1204 887161