EV RE320 Modifications
Let me begin by saying that, while the RE320 is a pretty nice mic in some ways, I don't like it. There are just two positives: EV's Variable D technology, which minimises proximity effect, and that it's considerably cheaper than an RE20. The very large negative, in my opinion, is how it sounds. EV's own datasheet shows the extent of the problem - the range above a couple of kHz is a mess, whether you use the "flat" or "kick" setting.
Here, I'll be discussing how I turned the mess in the mid/high range into something that actually sounds very nice.
First, though, I must give thanks to Jeff Geerling for his excellent guide on restoring an EV RE20 which helped me work out how to access the passive circuitry in the RE320.
So, let's get cracking.
Firstly, I took some measurements to see what I'd be dealing with. The graphs are below. These are the curves produced by a stock RE320 in the two different switch positions. The consistent approx. 8dB bump at 5kHz is the main source of the awful sound, but there are other interesting artifacts, too, such as the low-frequency rolloff of the "flat" setting.
NB - The test speaker I was using for this experiment falls off very quickly below 70Hz and above 14kHz (to a point where calibrating for it would be a problem), so the frequency response at the extremes won't be accurate. These curves were taken in a definitely-not-acoustically-treated room, so some artifacts of the room have got in. The 140Hz hump, for instance, can be safely ignored.
Measurements in-hand, I opened the mic up (thanks again, Jeff!) and found a rather complex bit of circuitry which was responsible for the passive filtering in the mic. The 5dB dip around 350Hz on the "kick" setting, differences in output around 3-4kHz, etc, are all down to this small circuit board.
Naturally, I removed it so I could measure the capsule on its own. That response curve was different again, but provided the basis of what would be an interesting investigation: how should I go about simulating a new passive circuit?
It then occurred to me that it's possible to simulate a passive crossover for a speaker, and dynamic microphones are just speakers in reverse. So, it should be possible to (ab)use a crossover simulator to get it to tell me what will happen with microphones.
That's the subject of another post.
Here, we're going to stick with the RE320. The raw capsule response was effectively what you'd get if you drew a new curve which follows the maximum of either the "kick" or "flat" setting. Intuitively, this makes sense: the passive filter can only ever reduce the output.
After getting the simulator up and running and giving believable results, I ordered some parts. The primary focus would be to make the 2-10kHz range as smooth and flat as possible, and a simple LCR notch filter would manage exactly that.
In this case, the notch filter would be used to effectively "short out" the capsule in the region of interest, thus reducing its output there. The resistor in the filter would set the amount of signal that's being cut in that region, and the inductor and capacitor would, between them, set the centre frequency and width of the notch. I settled on 13mH, 100ohm, and 82nF as a good starting point. Since 13mH inductors are a little difficult to find, I ordered a 10mH and 4x 1mH. The resistors were 100ohm 1/4W rated (I think I ended up with 50x of those, since they literally cost pennies), and I ordered 5x capacitors of 82nF each. The idea being that, by combining resistors in various ways, I could choose the depth of the notch and similarly, adding or removing the small inductors would allow the exact notch frequency to be fine-tuned. I ordered a few capacitors because they can typically vary by 5-10% and indeed they did arrive with a range from 77-85nF. Happily, one of them was 82nF exactly, so that one was chosen.
It's a testament to the simulation software that 82nF and 13mH did line up perfectly with the ~5.2kHz peak on the microphone, and some tweaking of resistor values got me the following:
Which is a huge improvement. The 3.3kHz dip followed by 5.2kHz mountain is replaced with a gentle rise from 2-9kHz which gives the sound a sense of "detail" without it being particularly coloured.
The resistor value came out to be 150ohm as the best choice. The initial 100ohm resistor cut too much at 5kHz, which resulted in a flat response from 2-6kHz, but then a bump around 9-10kHz. Upon listening, I decided that a smooth rise sounds much nicer than a bump, so used a larger resistor which would allow the capsule to put out more in the 5kHz region.
It would have been possible, I suppose, to implement another filter to reduce the 9-10kHz region, but I wanted to keep things simple for now.
The components I used were physically small, even with 4x inductors in series, so they were easy to place inside the mic, replacing the existing PCB. I used some electrical tape to hold everything together, which resulted in a snug fit.
I usually prefer to let the graphs do the talking, but subjectively the sound is much improved. In comparison with the stock "flat" stting, there's more low-frequency output and a much smoother mid-high region. I really did hate the RE320 before I did this modification - the excessively coloured mid-high region gave it a "forced" yet "cheap" sound which might only sound good on a select few sources which need a large 5kHz boost. Now, though, this modded RE320 is one of my first-choice dynamic mics. The smooth response means it sounds good on anything, which is, in my opinion, the hallmark of an excellent microphone.