Saiga

Dedicated to those large boxes at one end of the room
Post Reply
User avatar
Scottmoose
Needs to get out more
Posts: 1802
Joined: Sat Sep 08, 2007 11:03 am
Contact:

#1 Saiga

Post by Scottmoose »

It's been some time since I posted a complete multiway design online, and those that I did were on a long-dead forum, so I thought I'd dig something out of the collection to put here. To state from the outset, I doubt anyone here will particularly care for the design itself, and for reasons explained below, I have some reservations about whether it could be reliably built now. However, I thought it might be interesting as an example of a design process. Other people design in different ways, and so do I -the operative word there is 'a', since large parts should vary with objectives in any case, whoever the designer is. This is just how I approached this particular speaker.

Caveats noted, some background:

About 15 months back, I got a request from someone I know through a US forum for some assistance designing a speaker for his small studio. What he was after was either a standmount or floorstander that could be used as an evaluation tool in a [relatively] small room attached that also doubled as a lounge / coffee-production space. Basically a hybrid requirement between a studio monitor and a hi-fi speaker. The criteria laid down were:

-Modest cost, of about $500 - $700 all-in
-Not too large (tall was OK, width & depth limited)
-Neutral balance as they would be evaluating vocals & acoustic instruments (mainly bluegrass)
-Listening distances of about 2m - 4m (similar to many homes, which is why I picked this one out)
-To make use of a pair of tweeters they already had (see below)

The initial suggestion of an 8in midbass + dome tweeter was rejected due to width. Rather than dropping down a single size, the next suggestion was to shift to an MTM design with 5in drivers. < 45Hz output was not really needed, two 5in - 5 1/2in drivers have more radiating area than single 6 1/2in unit, and the MTM configuration, although slightly out of fashion at the moment, does have some advantages in terms of lobing etc. The final layout chosen was partially inspired by Lynn Olson's Ariel, using the same basic height, width and driver positions, but with a uniform 1in radius on the front and top baffle edges. The speaker didn't have a name, so for the sake of giving it a handle, I've called it Saiga, after the endangered antelope.
Attachments
Saiga.jpg
'"That'll do," comes the cry of the perfectionist down the ages.' (James May The Reassembler)
Website www.wodendesign.com
Community sites www.frugal-horn.com & www.frugal-phile.com
User avatar
Scottmoose
Needs to get out more
Posts: 1802
Joined: Sat Sep 08, 2007 11:03 am
Contact:

#2 Driver selection

Post by Scottmoose »

As noted, the tweeter for this project was fixed, largely because they happened to have a pair of quality dome tweeters NIB that they wanted to make use of. Fair enough, especially as it's a good one.

The Dayton Audio RS28a-4 is a custom 1 1/8in unit built for Dayton (Partsexpress's house brand) by Usher. As I understand it, it's essentially Usher's top model, which is not on general sale, with the beryllium dome replaced by a black anodised aluminium type. The motor design is excellent, with full symmetric drive (lots of copper plating + caps) and a large amount of linear travel. It's well-made, especially for the money, has a nice grill, aluminium faceplate and looks like something taken from the drawer marked 'evil'. Response is essentially flat out to about 15KHz where it has a roughly 5dB dip. A drop in output around this region is standard for aluminium dome tweeters due to the mass, unless they have a Helmholtz resonator over the front. The RS28a does have a small phase shield attached to the inside of the grill, but not sufficient to compensate for that. Personally I think that's a wise tradeoff, even if it does have a small effect on HF 'air'; YMMV as always. What is notable though is that its breakup mode is at 30KHz and of a very high Q -exceptional for an aluminimum dome, especially one slightly larger than the norm. Distortion performance is excellent; higher levels are very low, 2nd & 3rd order harmonic gradually rising < 2KHz. While I don't think it can be pushed as hard as some say it can because of this, it's still a bit of a tank and can take more of a thrashing than Dakota Johnson in a rubbish film with the word 'Grey' in the title. ;) Good value too -if you happen to live in the US. If you don't, then sadly, it's less attractive as once import etc. is factored in, the price has risen significantly and a lot of the value has gone. On a similar note, I mentioned a caveat above: the RS28a is it. It has a reputation for some consistency issues. The original units which first shipped about 10 years ago had ferrofluid; at some point that was removed without telling anybody. There have also been some other apparent changes since then. Enough that I have some doubts about whether the current units will actually work as intended here. Pity.

The midbass driver is the Seas ER15RLY. It was a toss up between this and the CA15RLY, which is largely the same, but with a coated cone. Low-mass CCAW voice coil, usual Seas build quality, nice extended and fairly linear response, reasonable distortion figures. Relatively shallow profile cone, ensuring reasonable off-axis dispersion. Quite flexible & suits a variety of different cabinet alignments. These 5in Seas units take off like a stabbed rat; the wide BW, low-mass moving components and decent sensitivity give them very good transient response. They can't get as low as a Revelator, but they're about 1/3 the price, so value for money is excellent.

I don't always bang on about some of this, since there are a lot of other factors and bald figures aren't anything like enough -as Toole points out, excessive reliance can be placed on HD measures (and to a point, FR etc. also). However, this particular speaker is designed with more reference to studio requirements than Hi-Fi per se, and those factors necessarily have to take a higher place than I might use in some other designs, intended for other purposes.
Attachments
ER15RLY.jpg
ER15RLY.jpg (112.25 KiB) Viewed 6706 times
RS28a-4.jpg
Last edited by Scottmoose on Sat Feb 11, 2017 10:53 pm, edited 3 times in total.
'"That'll do," comes the cry of the perfectionist down the ages.' (James May The Reassembler)
Website www.wodendesign.com
Community sites www.frugal-horn.com & www.frugal-phile.com
User avatar
Scottmoose
Needs to get out more
Posts: 1802
Joined: Sat Sep 08, 2007 11:03 am
Contact:

#3 Cabinet design

Post by Scottmoose »

As noted above, the general enclosure design & its proportions reflect the requirements laid down. Lynn Olson's Ariel inspired the height, width and driver positions on the baffle, although the cabinet itself is otherwise different in type and configuration. Since I knew they didn't have a 1 1/4in router bit available, a uniform 1in round was assumed for the front baffle vertical and horizontal edges, and the horizontal side edges of the top baffle. Round-overs and chamfers tend to provoke large amounts of debate; for what it's worth, I tend to find them useful for narrow cabinets, and measurements and simulations both indicate that a > 3/4in makes a noticeable difference. Audible? YMMV. I think it is.

I'm a rotten artist, so the drawing is basic (apologies). The enclosure itself isn't in bald terms very exciting and actually very simple indeed. It's just a vented box floorstander with some attention paid to the detailing. I considered a quarter-wave, but a conventional vented box is usually easier for people in studios to work with, as they're familiar with them & there's no potential surprises involved.

The top is tripled: it's a high pressure area and a good location for mass and structural rigidity, helping load the other panels. The front baffle is doubled. I would have preferred it to be tripled to ensure the most solid platform for the driver output, but we're dealing with 5in drivers here, and with a 1 1/2in (sorry, I tend to think in Imperial rather than metric) baffle, the driver cut-outs already need a fair bit of relief to avoid obvious restrictions / reflections etc. and ensure decent airflow. 2 1/4in would have been a pain. Optional tweeter backing plates are indicated: two rectangular pieces of material laminated together, the first with a cut-out for the driver, the second solid other than a hole for the internal wire. The top & bottom edges can be chamfered if desired. Small difference, but not difficult, so can be useful. All internal faces are lined with 1in acoustic fiberglass (or similar). The bottom of the driver chamber is doubled to improve rigidity, and a small compartment is provided immediately below to house the crossover. Below that, the void can be mass-loaded with xyz material. I divided it into two sections to provide some additional structural stiffening.

The acoustic alignment itself is the lightly damped type I tend to aim for with vented boxes unless circumstances demand something different; it tends toward 'practicality'. Rather than a single vent, I've used two; there's reasonable flexibility in terms of positioning; I like a high and low position. Slightly higher losses, but exciting a room at two different points can have some advantages in averaging behaviour out, despite the long wavelengths involved.

Assumed build material was 3/4in MDF or HDF. I'm not crazy about it as a material, but it's consistent, and the stiffening & multiple thicknesses tend to reduce some of its issues. The original pair used HDF and no brace was felt to be necessary in the upper (driver) chamber. With MDF, I might consider a window-brace or similar, since it needs all the help it can get. Lining the box with bitumen, or another heavy material can be a double-edged sword. It lowers panel Fs and Q, but without significant mechanical coupling, it does not increase stiffness per se, so it can become more trouble than it's worth if the box already possesses decent rigidity.
Attachments
Driver deflection.gif
cabinet alignment.gif
MTM ER15RLY+Dayton RS28a.PNG
Last edited by Scottmoose on Sun Feb 12, 2017 8:18 pm, edited 5 times in total.
'"That'll do," comes the cry of the perfectionist down the ages.' (James May The Reassembler)
Website www.wodendesign.com
Community sites www.frugal-horn.com & www.frugal-phile.com
User avatar
Scottmoose
Needs to get out more
Posts: 1802
Joined: Sat Sep 08, 2007 11:03 am
Contact:

#4 Driver responses

Post by Scottmoose »

Measured driver responses taken on the speaker baffle, on-axis with the tweeter. Distance is 2m (approximate minimum listening distance, and also the minimum distance a multiway of this size really needs for reasonable summing). Levels are normalised to regular 1m values to facilitate comparison. I can't recall the gating window, as it was some time ago & I wasn't the one taking the measurements. Box response is spliced in from 200Hz downward; 1/24 octave smoothing applied to both.

You can see the baffle-step coming in at approximately 570Hz; nothing startling and given the intended use and room positioning relatively near a front wall, not necessarily a major concern. The ER15RLY has some minor resonances coming in at 5.7KHz, 7.4KHz and 9.2KHz (roughly). These are not particularly severe and for the intended use here are unlikely to need any direct attention in the filter or any mechanical modifications to the drivers. Most significantly, there is wide overlap between the midbass and tweeters, and since the latter has considerable travel and good distortion performance at the low end, this gives significant flexibility in selecting a crossover frequency and slope[s]. LF output is in line with the intended lightly damped alignment & should in practice give useable output to ~40Hz, covering most acoustic instruments.

Overall, the driver responses give a flexible starting point and easy drivers to work with, which is all to the good. The close spacing between the midbass drivers and the slightly offset tweeter (the speakers are mirror-imaged, with the tweeter intended to be on the inner side) avoid obvious phase related notches in the response, providing you remain on the horizontal axis of the tweeter. The MTM configuration is slightly more compromised than a TM (or MT) in the vertical axis, but this is the trade-off inherent to the type, and some of the effects can be countered in the filter.
Attachments
Driver responses.gif
Last edited by Scottmoose on Sun Feb 12, 2017 4:10 pm, edited 2 times in total.
'"That'll do," comes the cry of the perfectionist down the ages.' (James May The Reassembler)
Website www.wodendesign.com
Community sites www.frugal-horn.com & www.frugal-phile.com
User avatar
Scottmoose
Needs to get out more
Posts: 1802
Joined: Sat Sep 08, 2007 11:03 am
Contact:

#5 Crossover design

Post by Scottmoose »

The crossover for this design is an asymmetric 4th order Bessel at 1.7KHz. Like the enclosure, the filter is actually rather simple, with just seven components needed to hit the target slopes. The circuit diagram, and the system frequency and impedance responses are attached below.

The crossover itself uses a 3rd order electrical high pass plus series padding resistor on the tweeter, and a damped 2nd order electrical on the midbass drivers (which are wired in parallel). There is some leeway for component selection; I specified quality MKP or film & foil types for the capacitors, exactly which being left to the builder. Performance will be reduced however with an MKT or electrolytic type, so these are better avoided in this case. Resistor power-ratings are given, otherwise there is leeway there also. Where there is no leeway are the inductors. The DCR of these is factored into the design, hence the reason the type (air core) and winding gauge is specified. Both the filter and cabinet alignment will be thrown out if these are altered.

The frequency response is broadly neutral. The 5dB dip inherent to the RS28a can be seen at the top end; not too much to worry about as it's otherwise very clean. There is a very shallow 1.5dB depression 1KHz - 3KHz after Harwood, enhancing subjective image depth perspective and ensuring there is no obvious 'strain' through the midband. There is also a similar slight depression centred on roughly 475Hz, giving a touch of acoustic scaling as employed by the BBC (and in different terms by W.E. several decades before) to the response. Although this is normally employed only on mini-monitors, it can within reason be used to make larger speakers sound subjectively more extended than is technically the case. This last was not deliberate on my part; it was simply fortunate circumstance through a combination of the nominal baffle-step frequency, the midbass driver characteristics and their interaction with the large primary inductor L2.

As far as impedance goes, the speaker is a fairly easy load for most amplifiers, with an impedance minimum of about 4ohms and a maximum of roughly 22ohms @ 1.2KHz. The resistor R2 in the low pass circuit, in addition to tweaking the phase response serves an additional purpose of preventing a direct 25uF capacitive shunt to ground, which some amplifiers may dislike. Not likely in this case, but it's no bad thing to avoid when possible.
Attachments
System impedance.gif
System FR.gif
Saiga XO ER15RY RS28a.PNG
Last edited by Scottmoose on Sun Feb 12, 2017 8:16 pm, edited 3 times in total.
'"That'll do," comes the cry of the perfectionist down the ages.' (James May The Reassembler)
Website www.wodendesign.com
Community sites www.frugal-horn.com & www.frugal-phile.com
User avatar
Scottmoose
Needs to get out more
Posts: 1802
Joined: Sat Sep 08, 2007 11:03 am
Contact:

#6 Re: Saiga

Post by Scottmoose »

The electrical order of a crossover is rarely the same as the acoustic order. Here we have a 3rd order electrical on the tweeter and a 2nd order electrical on the midbass drivers combining to give a 4th order acoustic XO. The tweeter tracks the target 4th order Bessel high pass slope, while the midbass drivers follow a slightly shallower low-pass to compensate for the offset between the voice-coils of the midbass units, which are acoustically slightly more than an inch behind that of the tweeter due to the driver sizes and construction.

A 4th order Bessel variation was selected as it gives good power-handling for the tweeter (superior to lower or higher order filters). 1.7KHz was a 'natural' frequency for both units, which could be easily achieved without requiring a large number of components. It also tends to be easier with MTM designs to get good phase-tracking with a lower XO frequency than a higher, providing the tweeter can handle the power. Originally, D'Appolito favoured 3rd order (acoustic) crossovers for MTM designs; he later shifted to 4th order, and the general consensus has been since then that even order filters are preferable for speakers of this configuration, with 3rd order tending to sound slightly aggressive. The reason is open to question; phase has been suggested. Others suggest that it's due to 4th order having a dip in the power-response around the XO frequency (3rd order Butterworth has flat amplitude and power responses at the crossover). On balance I suspect the latter is more likely to be the case, since if a 3rd order [quasi] Linkwitz is created with the individual filters down -6dB at the XO frequency, it seems to work OK for MTM designs too.

The three plots attached show the FR with (in grey) the deep symmetrical reverse-null at the XO frequency when the tweeter polarity is inverted, confirming this is functioning as an even order Bessel / Linkwitz. The power response (sum of the on and off axis outputs) is also attached. The dip around the XO frequency can be seen; the decline in the HF is also typical, if slightly over-stated due to the relatively large dome of the RS28a. As a rule, the power response should ideally track the frequency response shape with a decline in the HF, and where relevant, a dip around the XO frequency. The final plot shows the acoustic phase, again on axis with the tweeter, with the ideal 180 degree wrap at the XO frequency and tracking over a very wide bandwidth. I can't do much better than this in phase terms for a speaker of this type.

As noted, this speaker was designed for a specific purpose, and to a large extent personal preferences are irrelevant since the people who asked me (as a favour, this was not a commercial commission) to design it knew what they wanted in performance terms and got it. From a technical perspective, it's quite decent. Whether it would make a good home loudspeaker is a different question. Assuming the tweeters haven't been changed, and if a monitor type was desired (i.e. very neutral, without much character), then it should do quite nicely. But that isn't what everybody wants -I don't run a speaker like this either although if I had sufficient space I suppose I should keep something along these lines around as a baseline. YMMV as ever. It's here really as an example of one type of design & some of the things that went into it. Some parts of it might be of interest to a few people, so I decided to stick it up. It might do as an opening to talk about filter design & some of the issues around it.
Attachments
System reverse null.gif
System power response.gif
System FR + phase.gif
'"That'll do," comes the cry of the perfectionist down the ages.' (James May The Reassembler)
Website www.wodendesign.com
Community sites www.frugal-horn.com & www.frugal-phile.com
User avatar
Toppsy
Shed dweller
Posts: 2400
Joined: Mon Jul 23, 2007 12:10 pm
Location: red rose country

#7 Re: Saiga

Post by Toppsy »

Personally I cannot see how these are considered a complicated design? It is a simple box and the filled spaces below the drivers will add mass and stability and the doubling of the baffle will add stiffness both conditions I believe Richard Dunn is championing when recommending his mods (though with steel plate in the case of improving older poorly designed and built speakers). No bad thing.

Scott has commented these are not to his preference in personal speakers and they also are not to my liking as I am not a fan of studio monitor speakers modern or vintage. They tend to be too clinical a sound to my ears and lack excitement and do not engage the listener for a domestic environment, which is perhaps what a professional studio will perhaps be wanting in a speaker?

However, with different drivers, minimal internal damping and a simpler XO filter design I could imagine they would be rather good.

I have one recommended modification to the cabinet design. That would be rather than have the separate XO space as shown on the attached plan re-site this space to the rear baffle and have a removable panel so to gain easy access to the XO board. This could be placed to the rear of the upper filled area. As shown it would be very difficult to gain access to the XO.
User avatar
Scottmoose
Needs to get out more
Posts: 1802
Joined: Sat Sep 08, 2007 11:03 am
Contact:

#8 Re: Saiga

Post by Scottmoose »

Aye, good suggestion. Were I to do one like this again, I'd certainly adopt your suggestion for the XO compartment Colin. :) Out of choice, I'd also prefer to use an ML quarter-wave alignment rather than the BR, but that wasn't entirely in my hands in this particular case. Shifting to the QW box would mean no mass-loaded cavities at the bottom of the cabinet, but it would allow greater LF extension, and allow the rear panel to be doubled, further increasing structural rigidity, which you can't get to any useful extent by simply adding mass via bitumen or whatever.

I actually like the drivers used in this box well enough. The RS28a in theory is good -in practice I'm not completely convinced about consistency, and outside the US the value is gone what with shipping, import etc. The ones I've heard have been very solid, but I know some buyers have found inconsistent assembly & of course there have been some changes over the years, so in general I gravitate toward Seas, Scan or SB units.
'"That'll do," comes the cry of the perfectionist down the ages.' (James May The Reassembler)
Website www.wodendesign.com
Community sites www.frugal-horn.com & www.frugal-phile.com
User avatar
Scottmoose
Needs to get out more
Posts: 1802
Joined: Sat Sep 08, 2007 11:03 am
Contact:

#9 Re: Saiga

Post by Scottmoose »

Re studios & their expectations, it's a good point. It's not all that different from home use inasmuch as you get different monitors for different roles. These were largely meant for evaluating early mixes etc., so broadly speaking they wanted neutral, i.e. largely flat FR response and good phase characteristics. It's strange -many believe early recordings were far better than current ones. No engineer I, but to the best of my knowledge many were. They didn't have to contend with the modern market, loudness wars etc. But there also seems to be a widespread belief that this was done by engineers using speakers that had no attention whatsoever paid to their technical performance characteristics. Perhaps some were, but in many cases that's far from the case. I completely agree with Richard's comment elsewhere about the quality of some of the Altec Lansing etc. studio monitors. But the reason the products that came from W.E. / Bell Labs, Altec, RCA, Jensen, Stromberg Carlson & others were so good was the sheer level of engineering that went into them. Some top engineers at the big US telecoms companies were earning 6-figure salaries in contemporary money; they had massive resources and they invented, designed, built & measured just about everything possible (digital aside). Some of the equipment might look simple, but that's usually deceptive. The W.E. / Bell (later Altec) 755A wideband driver is a good example. Still impressive, nice relatively neutral response, especially for an 80 year old wideband design, out to about 11KHz: state of the art at the time given the carrier mediums. Altec 604 Duplex (coax) in its many variations. Still in many ways and various forms a very advanced unit. The cabinet designs were usually efficient sealed or vented box alignments damped out where appropriate using the old click-test to avoid audible ringing. The horns were often exceptional, and they / latterly JBL introduced a lot of new technologies, e.g. the Mantaray horn types etc. Not exactly hardware that can be said to have been casually slapped together or implemented without thought. They knew what they were doing & why; they would manipulate behaviour to achieve a desired end where appropriate. Not everything they made was up to much, but a goodly proportion of their top end pro (and consumer) audio gear was excellent.
'"That'll do," comes the cry of the perfectionist down the ages.' (James May The Reassembler)
Website www.wodendesign.com
Community sites www.frugal-horn.com & www.frugal-phile.com
User avatar
Scottmoose
Needs to get out more
Posts: 1802
Joined: Sat Sep 08, 2007 11:03 am
Contact:

#10 Re: Saiga

Post by Scottmoose »

Found this while sorting some files out & thought I'd add it to the thread for the sake of completeness. This version wasn't built, I simply put it together as a variation at the time. Nick & I discussed the general idea some time ago, & I've played about with it on & off since then. Nothing new; JBL have been doing this in the crossovers of their top models since the '80s & it goes back a lot further in wider electronics of course. Quite popular in some DIY circles, mostly in the US. It does seem to make a difference -better? Probably. Depends exactly how it's implemented and with what. This one assumes the straightforward Solen MKP caps and 9v that JBL normally use, although that can be upped to 18v or higher if desired, or different caps employed. The actual type of high value resistor and its value aren't especially critical; a metal film & roughly 1megaohm - 2.2megaohms are typical. I like the results you can get; it's not very practical for commercial speakers but easy to DIY.
Attachments
Saiga XO -full.PNG
'"That'll do," comes the cry of the perfectionist down the ages.' (James May The Reassembler)
Website www.wodendesign.com
Community sites www.frugal-horn.com & www.frugal-phile.com
Post Reply