DIY planar magnetic headphones
Overview
This project, dubbed Monstein, repurposes the planar magnetic (PM) elements from a pair of Monsoon MH-500 computer speakers to serve as full-range headphone drivers. Here I've mounted them on a spare Sennheiser HD-580 headphone frame.
Monsoon hifi, home theater, and computer speakers were manufactured in the 1990’s and early 2000’s, and are distinguished by their planar magnetic driver. Unlike the NXT “flat panel” drivers of the time these involve a lightweight membrane which is driven over its entire surface area, resulting in uniform pistonic motion with low distortion and near ideal acoustic properties. Even in their range of computer speakers, Monsoon’s engineering and construction are surprisingly good. Moreover, without the original subwoofer the MH-500 model can be found on ebay for $20-$40/pair.
There are several reasons why the PM element in the Monsoon MH-500 speaker has a very restricted frequency range and is normally used as a tweeter:
- The PM element is only about 3” x 5”. Since this is an open-backed dipole driver, acoustic cancellation from the unenclosed rear radiation limits the low end of the response to about 2kHz
- The diaphragms cannot sustain the excursions needed for real bass reproduction in a loudspeaker application
- A crossover filters out frequencies below 1.5 kHz to prevent damage
- The magnet elements are felt-lined and actually rest directly on the diaphragm, restricting its motion and raising the resonant frequency
Fortunately with headphones the diaphragm excursion requirements are greatly reduced. Furthermore the close proximity of the diaphragm to your ear alleviates the problem of acoustic cancellation: the rear radiation has to travel a greater relative distance to diffract around the driver and reach your ear, and is thus much lower in level. By eliminating the crossover and modifying the driver slightly we can obtain full-range headphone reproduction.
Disassembly
Start by prying off the Monsoon logo in the center of the metal grille. Remove the screw hidden under the logo. Now the grille can be pulled away. Set it aside for later use. From this point on, be very careful with the PM driver. The diaphragm is fragile and the magnets are very powerful and will pull nearby metal screws, small tools, etc into them.
Unscrew the 2-½ inch “woofer” and remove it from the cabinet. Remove the little piece of padding inside the enclosure. Use wire cutters to free the two wires coming from the PM element, preserving as much length as possible. (Note that some units have a passive crossover nested inside the enclosure, whereas other units use a multi-conductor DIN cable to power the woofer and tweeter independently, presumably with an electronic crossover filtering the respective signals.)
Unscrew the PM driver and use a flat-blade screwdriver to carefully pry the hot-glued portion loose so you can remove it. Note that the wires go through two holes in the enclosure and may be held in place by hot glue. Be very careful to set the PM driver in a clean location away from any tools or screws or anything metal that will be attracted by the magnets.
Measurements
To judge the effects of modifications, I set up a measurement system using pink noise as a source and a FFT (Fast Fourier Transform) audio plugin to render a frequency magnitude graph over the 20 Hz to 20 kHz audio range. Measurements used an electret instrumentation mic with the tip placed about ¼” from the diaphragm. Some of the graphs show a peak at 60 Hz due to electrical hum and very low frequency bumps from my family’s footfalls, and some of the unevenness in the top two octaves can be attributed to mic placement. I do not have access to a professional artificial head measurement system with spatial averaging. Nonetheless, the results are useful for comparing drivers and the effects of modifications. I recommend some such measurement system if you undertake this project.
Modifying the PM element
As seen in Graph 1, the stock PM element (white) has a cutoff frequency around 800 Hz due to the felt-backed magnets contacting the diaphragm. After modification (green) the low frequency cutoff is extended to the 100 Hz to 200 Hz range. (This measurement is from a later Monsoon sample -- as will be discussed below, the post-modification resonant frequency varies significantly among samples.)
Graph 1. Monsoon stock (white), modified (green)
The first step is to invert the felt-backed magnets so the diaphragm can move uniformly. We will also add shims to increase the distance between the magnets and the diaphragm to avoid contact and to enable low-distortion bass reproduction. This may reduce the efficiency of the drivers, but that is not a major concern for headphone applications.
CAUTION. Modifying the PM driver requires patience and very steady hands. If you are not extremely careful you will destroy the diaphragm and will soon be back on ebay looking for another sample.
Before removing the magnets, label them with a magic marker so you can restore them to their original polarity: use an arrow to indicate the top. Also set aside clean locations where you can put the magnets. They are so powerful that they will pinch your skin if you are holding one when another loose magnet suddenly jumps toward it.
Loosen the hot glue around the metal clip with a flat blade screwdriver, being very careful not to get the screwdriver near the diaphragm. Carefully lift the catches of the metal clip one at a time until it can be removed. At this point the magnets should still be held in place with hot glue. They must be removed one at a time with steady hands, because once they are free they will pull toward each other and may destroy the diaphragm.
Set the PM element on a clean surface and, holding one magnet down with one finger, pry its hot glue off with a screwdriver until it is free. Still holding it firmly, give a pull up and away from the diaphragm and set it aside. Repeat with each magnet, prying off hot glue as needed and holding them in place until you’re ready to yank them away.
We will flip the magnets over when re-installing them, so that the non-felt side is toward the diaphragm and no longer contacts it. This reverses the polarity as well, so you should reinstall them in the opposite direction. Therefore at this point you should mark the felt sides with a magic marker, with the arrows pointing in the reverse direction from what you had marked on the other side. (If you forgot to mark the magnets, lay all four side by side in the orientation that allows them to stick to each other, and then mark the tops so you can reinstall them in the same orientation. Make sure the orientation is the same for both the Left and Right speakers.)
Cut ⅛” strips of electricians tape and apply two layers near both ends of the non-felt sides of each magnet, where it rests on the PM element frame. The tape will act as a spacer to increase the distance between the magnet and the diaphragm.Trim the tape carefully.
Now reinstall one magnet, bracing your hand on the frame of the PM element. First insert the end under the L-bracket and then gently set the other end onto the side where you removed the metal clip, and push it into the cutout. Holding it in place, put a drop of hot glue on this end and wait for it to cool.
Next do the other magnet on the same side. This will be a little trickier because of the interacting magnetic fields, so brace your hand carefully. Once again slide the end under the L-bracket and snap the other end down and secure with a drop of hot glue, waiting for it to cool.
Flip the element over and follow the same process with the two magnets on this side. A firm, steady grip is required as the number of interacting magnetic fields increases.
Finally, slide the metal bracket on making sure the clips snap into place.
Mounting
Determine how you’ll connect the PM element leads to a headphone earcup/cable. Since I used a spare Sennheiser HD-580 headphone (with damaged drivers) as a frame, I removed the original Sennheiser driver and cut off the clips which held it to the earcup. Then I sawed the cable jack portion off from the original driver and hot-glued it back into place on the earcup. This arrangement let me use the original detachable cables.
The next step was to solder the PM element leads to the earcups, being careful to use consistent polarity on the left and right channels.
Mark and drill matching holes in the PM element frame and in the headphone frame you’re going to mount them on, again taking great care not to damage the fragile diaphragm. You can then use nuts and bolts to attach them. I was surprised to find that the driver’s measurements changed significantly as I tightened the mounting bolts due to subtle deformation of the PM frame. I had to add standoffs to prevent this. I also used an X-Acto knife to remove the thin fabric layer on the frame, only partially finished in this photo:
Due to the large magnets, these headphones are heavy. Fortunately the abandoned Sennheiser HD-580’s have a well-padded headband and large, soft, circumnaural ear pads.
As a finishing touch you can cut the original Monsoon grille down to make a nice grille, conveniently held in place by the powerful magnets. The completed headphones (bottom of page) look a little like a prop from a 1950’s science fiction movie.
Driver Matching
One issue that was apparent from listening tests was a lack of stereo focus and imprecise image placement. This suggested poor driver matching. Graph 2 shows the vastly different post-modification frequency response of the L (white) and R (green) drivers in my original attempt.
Graph 2. Monstein v1 Left (white), Right (green)
My first thought was to try to lower the resonant frequency of the Right driver to match that of the Left. I tried removing the magnets and stretching the diaphragm gently with my finger, but the results were unpredictable and unsatisfactory. Heating the diaphragm with a hair dryer raises the resonant frequency but also in unpredictable ways.
Resonances in a clamped panel are complex and difficult to control, and this may explain why so few manufacturers attempt planar magnetic (and electrostatic) designs!
I ended up buying two more pairs of MH-500's on ebay to select the two most closely-matched (post-modification) elements. This added $60 to the overall expense, and gave me lots of unwanted practice disassembling and modifying the drivers! Graph 3 shows the range of frequency responses among the 6 post-modification PM elements.
Graph 3. Comparison of 6 modified PM drivers
I selected the two with the lowest resonant frequency and played around with heating and stretching the diaphragms to get a better match. The final frequency response is shown in Graph 4 after mounting on the earcups, with the mic ¼” from the rear of the diaphragm as before. There remains a difference in the height of the resonant peak, but the mid and upper frequency range is well matched.
Graph 4. Rear of mounted drivers, Left (white), Right (green)
Resonance Control
At this point the modified Monsoon elements sounded very natural and open with band-limited material such as female voice and acoustic guitar. However on full-range material a large upper bass resonance could be heard, coloring male voice and bass guitar. This is easily seen in the graph traces above.
These elements are not designed for full-range use, and lack damping or a specialized suspension to control low-frequency resonance. My first thought was to modify them further mechanically. This patent application looked promising, in which the authors suggest applying adhesive tape or a coating of viscous material around the periphery of a planar magnetic speaker diaphragm. I experimented by applying felt and foam strips to the PM diaphragm, but was not able to produce a useful improvement. Instead I tried graphic equalization with mixed success...
Then a few months later I purchased a pair of Monsoon MM-700 elements on eBay for another project. Since this model uses only the planar magnetic elements and subwoofer (without per-channel woofers) the elements are capable of full-range response without modification. (As satellite speakers however they have a poor response due to rear-wave cancellation.) Furthermore close-miked measurements indicated a well-damped bass cutoff. Upon disassembly I noticed a layer of felt sandwiched behind (but not contacting with) the diaphragm. Upon removal a large bass resonant peak appeared. Apparently at low frequencies the closely-spaced felt layer acts as acoustic damping on the diaphragm.
Applying the same idea to my modified MH-500 elements, I hot-glued a layer of felt to the rear of the drivers a few millimeters from the diaphragms. Measurements showed this to be very effective, and I settled on two such layers. Graphs 5 and 6 show the Left and Right channel response curves, with Red indicating the final result with two layers of felt.
Graph 5. Left measurement in earcups (white), single layer of felt (yellow), two layers of felt (red)
Graph 5. Right measurement in earcups (white), single layer of felt (yellow), two layers of felt (red)
Amplification
WARNING. These elements have an impedance of about 6 ohm, similar to that of conventional speakers. Most headphones have an impedance in the 15 to 60 ohm range, with some professional models in the 200-600 ohm range. This shouldn’t generally be a problem, but I had a strange experience with my studio monitoring setup: plugging the headphones into the headphone jack only partially muted the main speakers. I had to instead plug a separate headphone amp into the headphone jack to fully mute the speakers, and then plug my headphones into the headphone amp. I did not experience this problem plugging the headphones into my MacBook: the laptop speakers muted automatically. Unlike high-impedance professional headphones you can probably obtain sufficient volume from mobile devices like an iPhone. Nonetheless I'd recommend a standalone amplifier capable of delivering a little extra current.
Otherwise these headphones are not sensitive to amplifier selection. Planar magnetic drivers have a nearly constant impedance, unlike dynamic drivers whose impedance varies with frequency. Therefore the amplifier's output impedance will have no effect on the sound.
Final Listening
After the resonance control modification the response sounds generally smooth and well balanced. Highs are extended with no trace of sibilance and the headphones sound transparent and natural, particularly on well-recorded acoustic material. Bass is not as extended as my reference Sennheiser HD580/HD600 headphones and image placement is not as specific as I would like due to the driver mismatch around 200-400 Hz. But what is most immediately noticeable is an out-of-your-head sense of openness. (Just be aware that due to the open backs anyone sitting nearby will also notice.) The sound of the Monsteins is like sitting in the middle of the performers instead of sitting in the audience. It's quite addictive!