Philips Motional Feedback loudspeaker tales

[Studio263]

Its been a bit quieter recently so I’ve had the chance to do some detail work on Philips MFB (Motional Feedback) loudspeakers, RH 532s in particular. The RH 332 was the original model, released in 1974. It has a small (15 litre) cabinet of which 9 litres is used in the acoustic section. The rest contains the amplifiers and signal processing equipment for the Motional Feedback system.

In simple terms, an active crossover (low level) is used to split the signal into 35 > 500Hz, which is handled by the MFB woofer, and 500Hz > 20KHz, which is handled by the mid range driver and tweeter. These last two drivers are driven through a conventional crossover, which is very simple in design as all the bass notes have already been filtered out by the electronics. The bass amplifier is rated at 40W, the midrange / treble one is 20W.

There are numerous advantages to active operation. No cable losses, no intermodulation between high and low frequencies, optimised driver matching and damping due to the short interconnections, no need for elaborate protection systems since none of the amplifier output connections are exposed, no chance of ruining the tweeter if the bass amplifier is driven into clipping since they are not interconnected and last but not least, the design of the crossover can encompass characteristics which are not practical with standard passive components, since there is no need to worry about efficiency or the availability of large value power carrying capacitors and inductors.

Add to this Motional Feedback. The Philips MFB process works using a piezoelectric accelerometer in the centre of the woofer cone. The signal from this is processed to yield one representative of position, which is compared to the incoming programme signal. As in any negative feedback system, the two are then compared and an error signal generated. This is used to correct the position of the woofer cone at any given instant. The effect is to linearise all the woofer’s characteristics over the bandwidth that the system operates. This includes the damping of resonances, the correction of non-linearities and the limits of the magnetic interface, the flattening of frequency response and the resolving of phase errors. MFB pushes the pursuit of damping factor to its natural conclusion and demonstrates the importance of this factor in the process of accurate reproduction. There are other methods of applying MFB, including resistive pads, sensing coils and capacitive proximity sensors. None are as effective as the Philips method however, with is implemented in surprisingly simple circuits when one considers what it actually does.

I have to say, this series of loudspeakers gives the most realistic sounding and convincing portrayal of recorded music that I have yet heard. In a normal design, if one wants bass which defines the beginning and end of each sound precisely one must use small drivers and well damped cabinets. If one wants bass which is rich and fulsome one requires large drivers and ports, transmission lines etc. It is not normally possible to have both, no matter how expensively constructed the loudspeakers are. MFB solves this problem and so reveals details in recordings which others miss, such as the variation in the ‘texture’ of different bass sounds in a recording. With conventional loudspeakers one is often listing to little more than momentarily excited cabinet resonances. RH 532s are good and the follow-up model, the RH 544, offered an improvement with its excellent dome midrange unit. The RH 545 studio model is arguably the finest loudspeaker available, and used three separate amplifiers with electronic crossovers for all. I repaired some last year, their performance was simply staggering. At the other end of the scale, the tiny RH 541 wasn’t really a ‘proper’ MFB design – it used a single wide band amplifier and a passive crossover to split the drive between an MFB woofer and a tweeter. The passband of the MFB system meant that in theory its corrections never interfered with the tweeter, but it still doesn’t sound as special as the bigger ones.

Component quality is the RH 532s is good, most of the parts are made by Philips themselves so there’s no need to go mad with capacitors etc. I did notice however that C506 in the MFB loop is operated with only 0.6V of bias, not enough to keep the oxide layer in good condition. This turned out to be true, so I fitted tantalum capacitors in this position instead. C559 was also tired in both units, this sets the time constant for the auto power on / function. You really need the full service manual to work on these, there were 14 versions made with many component value changes between each one. The indispensable MFB Freaks website can provide everything that is needed, although it is all in Dutch.

One unusual service adjustment is the adjustment of the MFB sensitivity. This defines by how much the MFB process reduces the gain of the bass amplifier, which when left ‘open loop’ is about 33 when everything is working properly. I say open loop, but this amplifier is stabilised by three independent feedback loops – a DC one to stabilise the midpoint of the output stage, an AC one to flatten the amplifier’s basic response and of course the MFB system on top of that. The MFB gain is set at 125Hz with an input of 10mV. Originally the target gain was 12.5, but this was later increased to 14 and then to 15.5. The preset control gives a range of 4.5 (maximum MFB) to 20 (minimum MFB). Too much makes the loudspeakers sound clipped and dry, not enough makes them sound like ordinary loudspeakers. Having tried all the settings the gain of 14 seems to be the best compromise, although it should be noted that a different woofer was fitted from series PL9 onwards. This brought with it the 15.5 gain setting, which may not be a good idea with the earlier ones.

Having done this, the results had to be validated. I don’t have access to an anechoic chamber so I resorted instead to outdoor testing, which is nearly as good in some respects. The loudspeaker was mounted about a metre off the ground (over the grass to avoid reflections) and the microphone placed just over a metre away. The type used was a Sony ECM-909 with a TC-D5 ProII tape recorder as the pre-amplifier. Test signals came from a Philips PM 5129 function generator and were recorded on a Philips digital storage oscilloscope. Working around the problems of wind and other outdoor noises, the MFB level in each unit was adjusted and its effect noted over the 35 > 500 Hz operating range of the system. A pleasingly flat response was noted, the microphone being the main limiting factor. I do have access to a high speed (50k / samples per sec) precision laser distance meter, so I may further attempt to validate the system with that.

[Studio263]

The next problem is what to connect the loudspeakers too. There are three input sensitivities, 1V into 3k ohms and two settings to match into the loudspeaker outputs of conventional amplifiers, one for 20W and below and one for around 40W. Both of these see a resistive load of 100 ohms. It seems wrong to connect two power amplifiers effectively in series, even though into 100 ohms most are working as a simple voltage follower producing little in the way of noise or distortion. On the other hand, 3k ohms is a punishing low impedance for a typical pre-amplifier to drive, and imposes a premature bass roll off for those with small output coupling capacitors. For example, the Aurex pre-amplifier I tried seemed bass light, unsurprising as the power amplifier it is intended for use with has an input impedance of around 50 k ohms.

You can see why they did it though; long cables operated at high impedances are prone to picking up noise and interference, so it all makes sense in a way. When these were a current model Philips offered either the RH 832 quadraphonic music centre or the RH 551 stereo pre-amplifier to go with them. Both are lovely, but not exactly easy to find here. Looking at the diagrams for both the output circuits are identical, built onto little plug-in modules. Since the same outputs also drive the headphone socket this gives an idea of the nature of the signal. 220uF coupling capacitors are used, 100 times bigger than what the Aurex pre-amplifier has.

The answer was to make up a replica of the Philips circuit onto a little piece of Veroboard and fit it into something else. Looking around, the most suitable unit I had was a Braun Regie 510 receiver of about the same era. This has a “reserve” line level input DIN socket (to which I connect a CD player) with two pins free. These could be used as the MFB output and avoids the need to drill any extra holes. The Philips circuit uses 2.1V zener diodes to set the bias for its output transistors which I couldn’t find in the junk box, so I used green LEDs instead – they drop about 2V. The resulting circuit unit was easy enough to hide away in the Braun chassis, leaving just the power supplies to find. Both were obtained from the 55V tuner / preamplifier supply, the 45V line coming from across C441 and the 12V (which draws about 60mA) from a resistive dropper taken from the regulator transistor’s input (across C804). To make up for the extra voltage drop across R802 a 390 ohm 2 watt resistor was added in parallel with it. The feed to the Philips amplifier circuit came from C804 via two 180 ohm 2 watt resistors in series, decoupled by a 2200uF capacitor. The input was connected to the slider of the volume control, in parallel with the input to the built-in power amplifier. I did worry that the loading of two circuits would cause a drop in sensitivity or a change in frequency response but I couldn’t hear or measure one, so as things stand both outputs work in parallel.

The overall effect is very pleasing I have to say. Using a Philips 14 bit CD player as a source makes an excellent system, one that would be hard to really improve upon I think. CD and MFB are natural partners, record players struggle to produce really consistent bass at low frequencies due to things like arm resonances but CDs suffer no such limitations. The two eras crossed over, but only just. The last MFB loudspeakers (F9638), marketed alongside the CD104 and CD150, were a real disappointment and a sad end to what had been a truly great achievement in loudspeaker design and technology.

[unitaudio]

I love the neatness of the conversion to the receiver and the fact that all sockets and functions work as they originally did. I've never heard Philips MFB speakers but I've always been intrigued by the design philosophy of them.

Regards,
Paul

[jamesperrett]

I've just fixed an N4506 reel to reel for someone who intends to use it with some MFB speakers (the N4506 has a preamp section specifically for this purpose). Hopefully I can get to hear the combination in action at some point.

[Diabolical Artificer]

Great writeup and good work. I have an article somewhere about these, ETI maybe, I'd written them off as a gimick. I've heard it said on here a few times that Philip's were pretty farout as regards design and unconventional ideas.

Andy.

[TonyDuell]

I bought a pair of 22RH541 Motional Feedback speakers some time ago, I must get round to testing them, restoring them if necessary and trying them out. I always thought it was an interesting idea and should work well.

The 22RH541 is a simple 2 way unit with a passive crossover. Feedback is taken from the bass driver and the bandwidth of the feedback path is deliberately limited so the signal only has an effect on the bass. I would have prefered some slightly larger 3-way unts with active crossovers, but you take what you can get.

I've heard that there were some much larger floor-standing Motional Feedback speaker units made. But they are very rare and probably not as impressive as the small ones. It's not too hard to get good bass from a larger cabinet. Getting it from a bookshelf unit is, though.

[rontech]

The reference to outdoor testing triggered a memory I had. It was an article by Mr Briggs founder of Wharfedale Loudspeakers. In his early days he used that method until the neighbourhood complaints encouraged him to re locate his production facilities.

[Edward Huggins]

l saw the massive ammount of R&D invstment Philips put into these. The Company was riding on a high in the 1970s and there were basically no constraints on the Engineers invloved. These MFBs were built to achieve a desired specification and not really to a price. I doubt if the R&D budget was ever recouped from their sales. I heard them on demonstration in Eindhoven and their sound was breathtaking.

[mark_in_manc]

I half-remember a tale about Celestion's (Foxhall Rd?) works, with a pit in the car-park into which a driver in a baffle could be mounted - 2pi radiation with no reflections, but only on a dry day

[Trevor]

i have a small pair in the garage picked up in a car boot
I have used them in the past very nice speakers for the size and i agree with the comments
i did design a pair of similar speakers the used an rf carrier and detector this worked well and it was offered to a rather famous speaker manufacturer but it never went any where
I also remember a small motional feedback speaker that never got any where
Trev