Tuesday, December 29, 2015

Squires Natural Reference 1 - Compression and Distortion

Comparing the 95 db and 70 db signals.  What errors you can see are probably due to me being lazy and using the Logitech Squeezebox volume control.

Distortion elements

Please note that while the shape of the distortion in these charts is accurate the scale is NOT!   You cannot read the actual distortion values off these charts due to a limitation (or feature) in OmniMic.  The distortion figures on on a scale not shown.  What I will say here is that the actual distortion figures are ridiculously low all the way to 100 db. :)

Distortion at 70 db

Distortion at 80 db

Distortion at 85 db

Distortion at 90 db

Distortion at 95 db
Energy Storage, flat EQ

Wednesday, December 23, 2015

Squires Affordable Reference 1, SAR-1

DIY Speaker

The SAR-1 uses a Dayton Audio AMT3-4 ($160/pr) tweeter and  ScanSpeak 7", 4 Ohm Revelator 18W/4531-G00 woofer ($420/pr) with a target crossover of 2.8kHz.

The speaker to the right however is not the SAR-1 but the Squires Natural Reference 1 (SNR-1) a much more expensive variant on the theme which uses the Mundorf AMT25CM1.1-R ($960/pair). The cabinet design and execution is from Taylor Speakers.

Done right the SAR-1 should look fairly similar.  The Dayton AMT3-4 is almost the same height but about 6mm narrower.  Both tweeters suffer inadequate flange lips at the top and bottom though, so be prepared for a really fun time mounting.

I'm really pretty proud of the SNR-1 though I know it may still be too expensive for many hobbyists. For the SAR-1 the goal was to get as close as possible to the SNR-1 at under a thousand dollars, and at least on paper it looks possible.


The total cost of the SNR-1 seemed too much for a DIY project for most so instead I have redesigned the crossover for the much less expensive Dayton tweeter while keeping the magnificent Revelator woofer. If you build your own cabinets the SAR-1 could be done for around $800 + lumber depending on the cost of your crossover parts.  Add another $850 for custom cabinets.

I hope at least some enjoy making a pair of SAR-1s, and I know many will enjoy criticizing it. I call it the "Affordable Reference" because compared to the Squires Natural Reference-1 (SNR-1) this is a much less expensive design due to the tweeter and crossover costs.  SNR-1 is still proprietary but I've used the SNR-1 frequency response and bass crossover as a reference line to design the SAR-1 from.  In terms of frequency response and phase angle matching SAR gets pretty close to the SNR-1 with some limitations, of course.

I should also thank SpeakerDoctor at DiyAudio for his help in pushing me along to use AMT's, particularly the Mundorf tweeters in my design. 


I can only attest to the sound of the SNR-1 which is in a word, fantastic. I do not know how close an $80 tweeter will get you to a $430 tweeter, but I can tell you the 7" ScanSpeak driver in this enclosure is simply amazing.    The Dayton data for the tweeter is also very good, but here is where we can separate the quality of the tweeters.  Let's discuss the two tweeters without crossovers.


As you can see, the Dayton tweeter has a constant rise from 2kHz, with a a peak at around 10kHz followed by a rapid rolloff after 12kHz of almost 20db/octave.  Sadly we're just not going to get the last half octave out of this tweeter.

Compare that with the Mundorf tweeter in yellow which is completely flat after 2kHz. Also the Mundorf tweeter is a 6 ohm unit while the Dayton is 3.  Just no getting around the frequency response and impedance challenges. Another potential issue is the power handling of the Dayton is a little limited, but then again, it's very efficient.  This is why you see 24watts of resistors padding it down.


Remember kids, in-room bass is never ever smooth. Fortunately by placing the speakers assymetrically you can get lucky and get each speaker to fill in the weak points for the other.

Anyway, my point is, for a 2 way, in a medium San Francisco apartment, the SS 7" in the right cabinet has more than enough bass for anyone. If you want better bass than this, you don't need a bigger speaker, you need a bigger home.

The $320 Illuminator 7" driver will add $200 and is almost a drop-in replacement, but has higher distortion which you trade for higher output.  If you really have a larger living room, this is an option you should consider.  Note the cut-outs are NOT exactly the same though.  If you plan on experimenting play close attention to the driver spec's.


To the right you can see the SNR-1 again.

We use ScanSpeak's recommendations mostly, a 1 ft3 enclosure but with a 6" long, 2" port. This port change is based on the Klang Tong Nada kit design. It can easily be adapted to a sealed enclosure as well by using a 0.4 ft3, but as I learned from Lee Taylor that the larger volume is totally worth it. It's like having twins. You were only expecting one baby, but you got two. The volume is a bit of a surprise, but totally worth it. The volume and port also lend themselves to floor-standing versions.


The crossovery is third order high low pass.

The simulation assumes 1.44" of woofer delay at 3'.  Without having the Dayton I can't tell how correct this is, but my gut tells me it will be within 0.25". Close enough to compensate for in the woofer, see below.

The Tweeter section has an unusually high number of inductors due to a notch filter and a treble boost formed by R6/L4/C2.  This adds a little compensation for the step roll-off after 12kHz.  It can be safely omitted.


For the SNR-1 I used Clarity MR caps in the tweeter and Clarity ESA caps for the woofer. Clarity Caps have proven to be amazing. Better to my ears than Mundorf Supreme's and SIO's in that they lack a lot of character the Mundorf's seem to add. The one area where they seem to have character is in strings, which take on a nearly warm, soapy feeling, as if the bows had been lathered in wet soap. I am sorry, that's probably a terrible description.

Sadly, the tweeter caps will still be expensive, around $180 per speaker pair. If that's still too rich the lowest I'd go with your final caps are the Clarity ESAs, around $79 for tweeter caps.

You will be tempted to skimp here, thinking that the Dayton tweeters don't justify the cost.  Please don't.  If you can afford them at all, take the plunge.  The tweeters will show you the difference.  How do I know?  I'm in the camp that says that any decent tweeter will demonstrate the effects of caps changing.  I'm not alone either.  Troels Gravesen has written similarly.

Of course, you are welcome to use your own particular favorite caps, but these are the only one's I feel I have any experience with so I can't comment on substitutions.

However if you'd like my opinion on your favorite caps, please write your question on the back of a pair of Deuland 10uF VSF speaker capacitors and send them to......


I really like the Jantzen air coil inductors.  Bargain priced, spot on specification.  If you want to get fancy for the woofer, go with either a Goetz foil or Mundorf either in 14 gauge should work and it's what I use.


I use Mills resistors exclusively.  Nothing added, nothing gained, absolutely transparent, and as a bonus, often smaller than equivalents from others.  Also, not too expensive, but you'll pay $60 for a set so maybe the MResist from Mundorf will help you save a little more.


Now, before you get TOO excited, this design is virtual and was designed with the help of OmniMic + DATS and XSim from Bill Waslo. The woofer measurements are based on my measurements in my cabinets in a room. The tweeter data however comes from Dayton. It's likely you'll need to adjust something before yours is completely dialed in. I strongly recommend the use of these tools.  XSim has been amazingly accurate, so if you trust that and you trust Dayton's data, this crossover should be pretty close.  You may have to experiment with the resistors in the woofer section to get perfect alignment.

I suspect that if the Dayton data is correct the most you'll have to do is dial in the correct tweeter to woofer distance and compensate for this in the woofer circuits.

Virtual Measurements

 OK, enough chit-chat! Down to the dirty stuff. In the first chart, the frequency response, I plot the SNR-1 with the Mundorf Tweeter above the SAR-1. Note the actual sensitivity is identical, I've offset for clarity.  Remember, the squigglies below 400Hz or so are in-room artifacts.

Next the individual driver response. 

Tweeter Inversion Test

To test the accuracy of the phase matching we invert one driver and look for a deep, symmetrical null.  Here you have it, textbook! At 1/24th octave smothing it almost 16 db.  Not the best, but not bad either. 



And finally, the impedance plot.  After examining the lowest point, I decided to remove the impedance compensation network as it was vestigial. By not using it the low impedance point is now 3.2 Ohms at 1.4kHz with almost exactly 0 degrees.  A little low, but with such low phase angle hopefully most SS amps will deal well with this faux pas.

Saturday, December 12, 2015

Noes - XSin vs. Actual

A big thanks go out to Bill Waslow for his excellent and free XSim.  Together with OminMic and DATS I did in weeks what it would have taken me years if ever.

To make the long story short, here is a comparison between the simulated crossover, and XSim's expectation.  Blue is simulated.  Differences in phase are probably due to inexact placement of the microphone between early measurements and now. 

If we take that data, and overlay it on the simulation from LEAP

Saturday, December 5, 2015

I love XSim! 

Here is the test schematic

Sorry about the part numbers.  I can't reorder them.  On top is the 2nd order design, on the bottom is the first order. 

Here we have the impedance of each, seen from the amplifier's point of view:

The top, yellow curve is the 2nd order filter, the bottom green line is the first order. 

Next, let's take a look at the power dissipation in the series resistors in each. 

R1 and R6 are part of the first order , R3 and R5 part of the second order filter.  You can see in each case the power of the 2nd order filter lags (is lower) than the first order filter, except at the very end for the first resistors in the circuits.  

Here are the simulated frequency responses, but please bear in mind having a good response was not part of the experiment.   Tweaking the crossover becomes very arbitrary.  The goal of this experiment was just to examine how power and impedances changed based on 1st vs. 2nd order.  Also, the first resistor is not really needed, it's again there for the investigation.


Thursday, December 3, 2015

XSim - Simulated Noes Xrossover

A quick posting for the currently simulated crossover.  As soon as the real one is built we'll compare!  

And here is a weird pre-ringing issue with the tweeter.  minor, but weird.