Adding more channels with Fireface UC

I hope somebody familiar with the RME Fireface UC can help me here.

I currently use the RME Fireface UC as my sound module, and have six active studio monitors running from the six balanced analogue outputs, together with an SVS PC2000 subwoofer fed from the headphone output. Previously I had 4 Behringer Truth B3031A's as four cycling mono channels as my main output, and 2 Siprit By Soundcraft 4P's as a stereo pair for the pedal division. I have moved my system to a larger room (about 7 by 5 metres with a 3 metre ceiling) and also ordered the Salisbury and Zwolle samplesets, and I am now experimenting with surround configurations.

Running my system as three stereo pairs (the 4P's being the rear / surround channels) I am getting a lot of problems with what I think is intermodulation distortion (mainly using the Velosovo sampleset, my favourite, but also with others). I had this previously but it was less obvious, presumably as I was routing the output through more channels. Although I have yet to find the optimum speaker placement to achieve good surround sound, my understanding is that the greatest realism is with as many stereo channels as possible for the front, and a surround stereo channel at the rear. I think, therefore, that the solution to the distortion would be to add more channels, and I am therefore considering buying 6 or 8 Behringer Truth B2031A's (these don't have as much bass extension as the 3031A's, but now that I have the sub this is probably less important).

According to the specifications, the Fireface UC has eighteen output channels altogether. Physically there are the 6 balanced outputs on the back and the two unbalanced ones on the front (for headphones), together with an SPDIF digital output and an ADAT optical output. Presumably I can route one of the digital outputs through another digital-to-analogue converter to access the additional channels, though I am very unclear about what additional hardware I need to achieve this. The manual is very confusing (I am a doctor, not a sound engineer) and after re-reading it I am beginning to wonder whether the digital outputs simply duplicate the analogue ones, and I don't know where the additional channels fit in at all.

Apparently it is possible to use up to three Fireface UC's together, which I suppose would solve the problem, but they are £800 each and I don't have enough USB ports (I could add a hub but I don't know how reliable that would be; the existing ports are already quite temperamental). I would also have to think of another way of getting everything mixed down for the subwoofer.

Can I add a third-party DAC, such as the Behringer ADA 8200 (8 channels, only £140)? Or the RME ADI 4 DD (8 channels, £600 and maybe more likely to be compatible). Or should I get a different sound module altogether (probably a lot more expensive with 16+ output channels but maybe simpler)? I don't want to spend so much on sound modules that I have nothing left for speakers, particularly as I have just bought two expensive samplesets.

The trouble with Hauptwerk is that it is difficult to know where to stop...

Hello Julian:

From your post it seems you have 2 questions or concerns: How to relieve distorted audio and how to add more channels to your RME interface.

The latter is the easy answer. You are correct, an ADAT such as the Behringer ADA 8200 connected by a fiber optic Toslink cable will give you 8 additional outputs. The S/PDIF coax port will give you 2 additional outputs but you will need to purchase an S/PDIF to analog converter. With most of these converters you'll end end up with 2 female RCA analog plugs (L & R). But they are usable for Hauptwerk!

For the distortion issue most audio interfaces have some software that comes with it for setup and studio work. I'm familiar with RME just never worked with it directly. See if there are volume sliders or gain controls that could be set too high. An audio signal over driven pre or post audio interface is the first place to look and an easy fix; just turn the input gain down.

Your post says you are running 4 mono channels. Is there a reason for not running stereo? There are two areas in Hauptwerk for this setting: 1) Audio outputs menu, set as Mono or Stereo 2)when loading the organ Load organ, adjusting rank audio/memory option. Coincidentally I've experimenting with these settings the last few days with my 16 channel setup. I discovered that the best Mono performance (on 3 of my classical sets) was when ranks were loaded as "Stereo". In other words, the stereo pipe rank is summed into a Mono channel or should I say an Audio Group made up of multiple mono channels. However when I set the rank as Mono into a Mono channel I noticed a considerable lost of fidelity and detail. Your setup and organ samples may vary but with mine, stereo samples played back in stereo was the clear winner in the fidelity contest.

Is the distortion present when using only headphones? Do you notice any of the CPU, Audio. or Poly LED meters going into the red when playing? Review your audio buffer size in Hauptwerk. I think the default is 1024 but this can be adjusted if need be.

Finally configure a different Alt_Config with a basic, single stereo pair of speakers. Using a separate Alt_Config will keep your original intact. Maybe load St. Anne Mosley too, the idea is to start at a basic setup and build from there to discover the cause of your distortion.

Oh yea, the Behringer 2031A monitor sans tweeter is identical to the discontinued 3031A. The amplification, rear controls, 8.75" woofer are identical between the two. The ribbon vs dome tweeter and frequency crossover are the differences.

Hope this helps,

Danny B.

Dear Danny,

Thank-you for your prompt reply. If I can add additional channels with the Behringer ADA 8200 then that is clearly the cheapest solution. I believe there have been sound quality concerns with the ADA 8000 that have been addressed in the current model, though if there is a significant difference in quality then maybe I should pay more for the RME equivalent.

I'm pretty sure that the ditortion that I can hear isn't harmonic distortion, which is what you get with an overdriven amplifier, or digital distortion from an overloaded CPU. It comes with specific combinations of pipes (e.g. an 8 foot flue in the low octave in combination with a mixture playing particular chords) and I think it is due to the harmonics interfering with each other. I once had a similar problem with my in-car HiFi when listening to organ music, which I solved by getting a subwoofer and fiddling with the crossover cut-off points of the front and rear speaker channels (trying to get in-car HiFi to sound reasonable is a whole different nightmare). I have never heard it with headphones, though I haven't used Hauptwerk with headphones since replacing my original Focusrite Saffire sound interface with the Fireface UC when the Saffire died. This set off a chain of problems since the Fireface drivers wouldn't run on my dedicated Hauptwerk MacPro workstation as the OS version was too old, and the MacPro hardware wouldn't support a more recent version of OSX. I ended up running Hauptwerk on an old Windows 7 laptop (a Dell Precision workstation with a 4-core processor and 32GB RAM), and was pleasantly surprised that it ran very well (I am getting polyphony of over 2,000 and I can now use all of my RAM and not just the two-thirds that OSX allows).

I had a different issue with the Spirit 4P's where high-pitched pipes sounded clear on their own but distorted in combination (this did sound more similar to harmonic distortion) which I solved by only using the 4P's for the pedal. This was the reason I was cycling everything else as mono samples through the Behringers - I was trying to limit the number of samples played simultaneously by any given speaker.

Anyway, Martin's recommendation is to use as many stereo channels as possible (to minimise intermodulation distortion), so now that I have a subwoofer (and I am blown away by how this sounds with a 32 foot pedal) adding more channels seems to be the next step. What I am not clear about is how many channels are required before it becomes a matter of diminishing returns. I think a DAC and four pairs of 2031A's is probably my limit.

What I am not clear about is how many channels are required before it becomes a matter of diminishing returns. I think a DAC and four pairs of 2031A's is probably my limit.

Diminishing returns is an excellent question and somewhat subjective. The simple answer is the size of your room (studio, living area, church, etc) and the organ or organ you use the most often. From my experience and my studio (27 X 18) 4 stereo pairs for the main pipe work and 1 stereo pair for surround sound worked very well. I built these up slowly adding an additional pair about once every 2.5 months so I was accustomed to how it sounded before the addition. I estimated a 20% improvement. Adding a 5th pair to my array of 4 stereo pairs in an Audio Group (with the Cyclic routing algorithm) did not yield a 20% improvement just a wider dispersion. So I took some time and experimented on how to best incorporate them and additional pairs added several months later. Again YMMV.

Esteemed Hauptwerk member Mark C. has done extensive work and experiments with rank routing in a multi-channel setup. It is all documented here in 2 threads with various designs and critiques following. Both are worth reading and might even resolve your inter-modulation problem:

Early experiments:
http://forum.hauptwerk.com/viewtopic.php?f=17&t=13679

Later revisions:
http://forum.hauptwerk.com/viewtopic.php?f=17&t=14769&p=109957

Danny B.

Hmm... I've followed the links and done a lot of thinking about what I have found there. I am sure there is a great deal more that I haven't found on the topic of how to use multiple channels.

My original question has been answered - it would seem that the relatively cheap Behringer DAC should work with my Fireface, and hopefully the quality will not be that much worse than the (nearly 4x more expensive) equivalent RME unit, and that I would probably be better off spending the difference on more speakers.

If so, then potentially I will end up with my existing 4 Behringer Truth B3031A's, and up to an additional 8 B2031A's. The 3031's (before they were discontinued) were a bit more expensive, if I remember rightly, and they have better treble reproduction due to a ribbon tweeter. It would not seem to be a good idea to put the 3031's and the 2031's in the same output group, so assuming that I configure them as stereo pairs, I have a minimum of two output groups (2 pairs of 3031's plus 4 pairs of 2031's) and a maximum of 6 output groups. I am proposing to use my Spirit by Soundcraft 4P's for the rear / surround channel.

The samplesets that I have been using thus far are mainly Velosovo and Freiberg, also Bosch-Schnitger and Metz. At the moment I mostly play Bach, though I would like to make more inroads into Franck and his French successors. I have just ordered Salisbury and Zwolle.

This has all been led by distortion, which I believe is intermodulation distortion, and I also believe that this is due to too many pipes being reproduced by too few speakers (intermodulation distortion occurs whenever two or more different tones are reproduced by a system which isn't perfectly linear in its response, which means all real-life amplifiers and speakers). What I have heard takes two forms. The first seems to occur with a full chorus to mixtures is used to play multiple notes, and for certain combinations of notes one of the principals in the bottom octave seems to modulate the mixture sound from another note higher up the keyboard. I believe the solution to this is for the 16 and 8 foot flues to use a different channel from the mixtures.

The other type of distortion is within the upperwork, where the mixtures sound clean when playing one note, but when more than one note is played they become fuzzy. I noticed this particularly with the 4P's and less so with the Behringers, and it was improved further when I configured them as four mono channels with the notes cycling.

Having said that, with full organ the mixtures still sound a bit screamy (and also when listening to CD's through my reasonably good in-car HiFi system). The problem may in part be with my 58-year-old ears. I damaged them one afternoon in my late teens by shooting without ear protection (I couldn't hear the top octave of the piano for a week afterwards and I have had a permanent 4 KHz notch in my audiogram), and I think they have been damaged further by chemotherapy (platinum-based cytotoxic drugs, particularly cisplatin, are well known to cause sensorineural hearing damage, which can be mitigated to some extent by how the chemotherapy is shceduled - I was an oncologist myself before developing cancer myself forced me to stop working). I remember as a teenager hearing Peter Hurford give a recital on the Dutch organ in Eton School Hall which had been relatively recently restored by Flentrop; the upperwork added a beautiful percussive bell-like effect which I struggle to hear at all nowadays).

Enough of this rambling...

So there seem to be several schools of though as to how to configure multiple channels. The obvious one (to me) is to use one (or maybe two) for each division, and locate them in different places in the room. From what I have read in the Forum, however, that doesn't give very good results, as what we are really aiming for is to have as few pipes as possible being reproduced by any given speaker, and to avoid particularly problematic combinations.

So there is the method recommended by Martin Dyde, to use Hauptwerk's cycling algorithm with as many stereo channels as possible. Somebody else has suggested that this should ideally be a prime number of channels (which would avoid octaves using the same channel). This would certainly give the most even distribution of pipes and therefore the fewest sounding through one speaker at the same time.

The other method seems to be to give each type of stop its own channel, based on its harmonic structure. So one for 16 and 8 foot flues, another for 4 and 2 foot flues, another for mutations (tierce, larigot, nazard etc.), another for mixtures, another for reeds. Possibly using the scheme the pedals can have their own channel, since it is unlikely that you will need to play more than two notes at once and they will all be within two-and-a-half octaves of each other. It will also avoid the possibility of low-pitched flues interacting with mixtures, for instance.

Clearly there is plenty of opportunity for experimentation, though I have limited time to spend with the organ and I would rather use it playing, particularly given how long it takes for a sample set to load every time it is reconfigured. I am thinking that a good starting point would be to use the 3031A's to give me two stereo channels and have the mixtures cycling between them, and have four separate channels with the 2031A's, namely pedal, 16 and 8 foot flues, 4, 2 and 1 foot flues, reeds and non-mixture mutations.

This has also got me thinking about acoustic interference, which of course takes place within a real organ, and is utilised deliberately (e.g. for celestes, and for the faux 32 foot stop that relies on resultant tones), or else occurs as a consequence of the physical behaviour of the instrument (e.g. the howling wolf that you get with certain intervals if you try to play in A flat with a meantone temperament). One of the joys of Hauptwerk is how well these phenomena are reproduced. I could ramble on about this, too, particularly with how J S Bach seems to have found ways to turn the shortcomings of unequal temperament into deliberate musical effects.