[j-curve]

It's a wonder that the old ED-1 Stax equalizer disappeared completely and modern Stax amps don't have a "diffuse EQ" button to fill in the upper mids and tame the unbridled highs. Many audiophiles reject equalizers for various reasons, some of which relate not to the EQ circuit itself (or algorithm) but to the surrounding "glue" of op-amps, transistors, A/D and D/A converters etc. The only "passive" EQ I'm aware of is horribly expensive and is not actually 100% passive - it contains some amplifying glue to restore the signal to a usable level.

Here's a circuit which aims to equalize Stax headphones and has no active components. You would need two of these for stereo

:-

Some details and specifications:-

The circuit operates at line level.
Input impedance: 10k ohms. (Can be driven with low impedance sources too).
Output impedance: 10k ohms. (Easily drives a Stax amplifier with a 50k ohm input impedance).
Insertion loss: 13dB (from a 10k ohm source when driving a 50k ohm load).

The insertion loss means you have to turn up the volume a bit. Stax systems usually reach "loud" by 12-o-clock so this shouldn't be a problem.

Construction
The resistors and capacitors are shelf items.
The inductors L1, L2 are the trick. Anyone notice anything unusual about them?

[Hint: Look up the values of inductors you might typically find in a speaker crossover.] Needless to say, these suckers are not shelf items! It would be nice if they were air-cored but some ferrite may be necessary, especially for L1, where I estimate 250 metres of wire could be necessary otherwise. This is not something you want to try and wind by hand. If you can locate a reel of enamelled wire with both ends of the copper sticking out then you should be so lucky. The good news is you don't need fencing wire for these - even as fine as SWG 34 (or AWG 31 = 0.23mm diameter) should be OK, as the impedances are quite high and a few ohms of resistance won't upset things.

The inductors should be placed well apart from each other in a metal box. If you don't have an inductance meter (who does?), the tank circuits L1-C1 and L2-C2 can be tuned with an audio oscillator and a multimeter. The inductors should be in place in the box when you do this.
L1-C1 tunes to 3.4kHz.
L2-C2 tunes to 10.5kHz.
If they oscillate at a lower frequency, the inductance is too high and a few turns have to come off the inductor.


Obviously, I haven't built mine yet, and it could be a while. Is anyone willing to give it a go? Any comments, suggestions?



Acknowledgements: The frequency adjustments for this design were chosen on the basis of three sources, as follows:-
1. My "DIY Response Plots" measurements and listening tests on SRS-2020, SRS-3030 and SRS-4040 systems.
2. A Lambda response plot produced by Moller et al. from the article "Transfer Characteristics of Headphones Measured on Human Ears", Journal of the Audio Engineering Society, v43, #4, April 1995, p215.
3. Comments by Jorg Stumpp, a.k.a. "stu" on Headwize regarding equalisation of Stax headphones and the ED-1 equaliser.

Keywords: filter boost cut treble etch

 

[Jupiter]

Could you use these inductors? (Available from Digi-Key for $3.75 each.)

Instead of tuning the inductors/caps, could you instead measure the frequency response of the Stax headphones? (I don't have a scope.)

 

[j-curve]

Hi Jupiter,
Those Toko inductors are a great find! I searched a bit but didn't turn up anything with those kinds of values, and they are so compact too(!) This involves compromises of course - they are obviously using a very fine guage of wire so the DC resistance is near the top of the tolerable range. [More than 1 ohm per mH would be unacceptable, I think.] The other issues are hysteresis distortion and magnetic saturation. I presume the Toko inductors are built on ferrite cores. Probably the only way to find out is to give it a go, using the closest values and playing around with a few capacitors until you get a good match.

Quote:

Jupiter: Instead of tuning the inductors/caps, could you instead measure the frequency response of the Stax headphones?

Yes, the measurements have been done previously. The tuning process is only meant to be done once, ie. set-and-forget, in order to get the desired adjustments to the frequency response. This design doesn't let you tweak the response on-the-fly. It is supposed to bring a Stax headset (Lambda series in particular) closer to the diffuse field response.

Here's the theoretical response of the above circuit when driven from a 10k-ohm source and with a 50k-ohm load attached:-

The boost and cut applied here are more conservative than those of the original ED-1 design.

 

[Jupiter]

Linkwitz designed an equalizer for Etys, which you can find here. Might be helpful.

I found some other inductors, but these are pretty expensive. ($47 for 100mH)

 

[j-curve]

Adjustable Boost/Cut Mod
The above design is simple but offers no adjustability. The amount of boost and cut is limited by R1 and R2, so if these were replaced with 33k-ohm linear potentiometers the EQ would be quite flexible. This would give you almost 6dB boost at 3.4kHz and 8dB cut at 10.5kHz. You can also wind the pots back to zero for flat response (with no phase shift either). Somewhere around 12 o'clock should sound pretty good I think. The potentiometers have to be dual-gang for stereo.

Toko Inductors and Tuning
Jupiter, thanks again for the Toko inductor tip. I don't think this circuit will push them anywhere near magnetic saturation, and ferrite hysteresis is minimal so distortion should be very low. Moreover, they are tiny and magnetically shielded, meaning they can all live happily together in a much smaller box. The "J" type with 5% tolerance will be best.

Since they are not adjustable, the tuning will have to be done using pairs of smaller capacitors in parallel.
eg: Instead of C1 = 18nF
Use 15nF + 3.3nF in parallel to lower the frequency.
Use 15nF + 2.7nF in parallel to increase the frequency.
etc. etc.

Phase Shift
Has anyone looked at the phase graph above? The peak phase shift of this circuit is less than 25 degrees(!)

 

[Jupiter]

j-curve,

Digi-Key doesn't sell the 36 mH inductor in small quantities, just the 33 mH and 39 mH. These are the J (5%) version.

At the moment I'm trying different measurement equipment and software. I'm looking at what equipment will give me the flattest frequency response and what software/method will be the most accurate.

Do you have the actual frequency response of the Stax SRS-4040? (Not compared to the HD-580.)

 

[j-curve]

Quote:

Jupiter: Digi-Key doesn't sell the 36 mH inductor in small quantities, just the 33 mH and 39 mH.

Good point. They give you a great price on the 36mH but you have to buy 100 of them. So it seems the 33mH is the go. Quote:

Jupiter: Do you have the actual frequency response of the Stax SRS-4040?

No, for the simple reason that my binaural microphones aren't calibrated. Apparently some kind of plot can be found in the handbook which comes with the Stax, but without knowing exactly how it was measured it might not help at all. And if you did have an actual frequency response plot, what would you do with it? It has little meaning unless you plot other headphones on the same graph, or a target curve such as a diffuse field response.

 

[Jupiter]

j-curve,

I'll get the parts next time I order from Digi-Key.

Maybe it would be better to measure the frequency response of the equalizer instead of the headphones. That should be pretty easy with RightMark Audio Analyzer.

 

[j-curve]

Good to hear that someone else will be giving this a go too! For my attempt, I'm thinking along the lines of hand-wound inductors but using ferrite cores. I found out that the popular Pultech eq's use ferrite so I'm not even going to consider air cores at this point. Quote:

Jupiter: Maybe it would be better to measure the frequency response of the equalizer instead of the headphones.

Absolutely. You remove a few degrees of uncertainty that way.

 

[Jupiter]

I got the SR-404 frequency response from Stax.

 

[j-curve]

Excellent! Looks like the harmonic distortion plot there as well, about 60dB down!

Now to figure out what it means... I think we need to know where the diffuse field response sits on that same graph, measured on the same basis. Was the Stax measured on an acoustic coupler?

 

[Jupiter]

I have no idea how the measurements were done. I'll do some measurements on my own and I'll post here if I come up with something that could be useful. (I got little experience with doing measurements.)

 

[j-curve]

Jupiter,
Here's a snapshot from that graph you linked, with a couple of lines superimposed:-

RED - Moller's Diffuse Field EQ Target for a blocked ear canal measurement.
BLUE - Moller's Diffuse Field EQ Target for an open ear canal measurement taken at the entrance to the ear canal.

Note however that since Stax have used a B&K 4134 microphone and a B&K 4153 "man made ear", it may be risky drawing conclusions from this graph. I haven't been able to find any detailed info about the B&K 4153 yet. The positioning of the microphone is important too (eardrum vs. entrance to ear).

If we assume the graph has meaning, the Passive EQ seems to bring the Stax plot closer to the blue line, but the fit is not perfect at all and there is still too much energy above about 14kHz, provided the amplifier used is flat to 20kHz. This is in conflict with Moller's test of a Lambda headphone on a blocked ear canal, where the response was within limits above about 13kHz. In other words, I've probably only confused things by adding those lines which are not on a "like-for-like" basis with Stax's measurements.

The Passive EQ will correct the Lambda's deviations from diffuse field EQ as measured by Moller. (See the article I referenced at the bottom of the first post in this thread).


PS: Hope you don't mind my cut'n'paste job. Send me PM if you want the graph removed.


Acknowledgement: Moller et al., "Design Criteria for Headphones", Journal of the Audio Engineering Society, v43 #4, April 1995

 

[Jupiter]

I found the technical specs:
B&K 4134 microphone
B&K 4153 artificial ear

I don't have an artificial ear, so I guess I can forget about microphone measurements.

Have you talked to Jorg Stumpp/stu about the project? It would be interesting to see the frequency response of his equalizer (using his settings).

 

[j-curve]

Thanks for digging up that technical info, looks interesting.

No, haven't spoken with Jorg Stumpp but there are a couple of relevant threads on Headwize where he gave details about equalising Stax. Quote:

Jupiter: I don't have an artificial ear, so I guess I can forget about microphone measurements.

Only if you want to verify Stax's plot for some reason. For doing your own measurements, binaural microphones will be fine, especially if they fit neatly into the ears or can be fitted into an earplug to block the ear canal. They don't have to be calibrated but I sure wish mine were...