Scanspeak 23w4557 crossover point

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Looking at the Scanspeak 23w4457T02.

Quick calculation seems to suggest that in a 1cuft (28l) sealed box it can provide an F3 of 32hz and an f10 of just below 20hz and QTC of 0.76 which is very appealing for the bass driver on a 3 way. Ok, it is not cheap but - you can't take it with you !

So, looking at the FR plot it seems quite well behaved up to 1khz when it start to break up.

I understand this driver is not very sensitive (understatement) but will be using an active crossover and separate power amps for each driver so this should not be much of a problem.

Looking at using this with a 5" or possibly 6" mid cone driver in a separate sealed enclosure (favourites are SB MR13P-8 or if budget allows the Scanspeak 12MU).

So the questions is - has anyone used this and can comment on performance and secondly what is the highest you would cross over to the mid, currently looking at around 300-400hz LR2 ideally, would that be OK ?

For completeness, my alternative is the 22w8857 in a 42l BR tuned to 28hz which seems to have a similar break up pattern and has been used upto 750hz with a 2" mid dome (Paul K Cantilena). This gives a similar F3 performance but sharper low roll off compared to the 23W but in a much bigger and ported box (I would prefer the sealed box if it can work).

Thanks in advance for replies, much appreciated.
It also depends on the crossover slope, a 4th order could be arround 700hz, while a first order would have to stay under 400 i think. But best is to sim it, even with only the curve from the spec sheet you can find this out using the graphic capture function of programs like VirtualCad. it will not be precise, but gives you an ID what is the best region to start.
Quick update;

"Mating a 23W that is designed to go wayyy down with a smallish 12cm is not recommended (I gave you the "very polite version"), 23W good for up to 300 Hz if EQing the cone break-up at 1.2kHz . IF you insist my friend suggests to go with a 300 Hz, LR 4th order, 24dB/oct as a starting point that should "work" +/- ...

Other than that you are on your own.... And I can't help (not that I don't want to, it' just over my head and not my domain....) I told you what he mated them with though.
Many thanks both for the replies, much appreciated. Please pass on my thanks also to your friend @coolnose and point taken on the smaller mids, will look into 6-6.5" ones, do you know which version of the 15W he used and with what tweeter.

I have done a couple of sims using Boxsim 2.0 with crossovers at around 350hz both LR2 and LR4 with the MR13P and will try and post results late this evening.
So here's a few sims, thoughts please.

1) WINISD comparison of 23W Closed in 28l Qtc = .76 9 (Blue) vs 22W BR in 42l tuned to 28hz (Green) - 23W Closed looks very good and needs no low roll of at 60W (power of my amp, too small but Ok for a start), 22W needs a 2nd order highpass at 25hz to keep it under xmax plus a 3inch port to keep velocity below 20m/s.

2) Boxsim 2.0 active sims for 23W closed LR2 and LR4 for comparison with SB MR13P-8 and SB 26ADC.

Baffle 95x30cm - 23W @ 53cm, 13P @ 73cm, 26ADC @ 86cm all on centreline i.e. small floor stander that may have a divider half way up to produce 28L bass cavity.

23W - 28L moderate stuffing, simmed with 2xFs upwards (hence smooth bass roll off), xover at 300hz

13P - 3lL moderate stuffing, xover set using optimiser - LR2 = 425/1900, LR4 = 330/2440hz

26ADC - 2700hz crossover
(All data from datasheets using Vituix cad trace)
Only other controls are delays set on 13P and 26ADC with optimiser, no smoothing using parametric filters i.e. 13P surround 1500hz wiggle is still there.

Graphs show individual frequency response/crossover, off axis spread 0-30 degs, phase.

All comments/observations welcome.


  • Winisd 22w8857 BR 23w4557 Closed.JPG
    Winisd 22w8857 BR 23w4557 Closed.JPG
    110.9 KB · Views: 226
  • 23W4557 Mr13P-8 26ADC LR4 simple Offaxis.JPG
    23W4557 Mr13P-8 26ADC LR4 simple Offaxis.JPG
    383.1 KB · Views: 64
  • 23W4557 Mr13P-8 26ADC LR4 simple Phase.JPG
    23W4557 Mr13P-8 26ADC LR4 simple Phase.JPG
    229.9 KB · Views: 55
  • 23W4557 Mr13P-8 26ADC LR4 simple FREQ.JPG
    23W4557 Mr13P-8 26ADC LR4 simple FREQ.JPG
    254 KB · Views: 58
  • 23W4557 Mr13P-8 26ADC LR2 simple Offaxis.JPG
    23W4557 Mr13P-8 26ADC LR2 simple Offaxis.JPG
    386.1 KB · Views: 181
  • 23W4557 Mr13P-8 26ADC LR2 simple Phase.JPG
    23W4557 Mr13P-8 26ADC LR2 simple Phase.JPG
    222 KB · Views: 215
  • 23W4557 Mr13P-8 26ADC LR2 simple FREQ.JPG
    23W4557 Mr13P-8 26ADC LR2 simple FREQ.JPG
    261.7 KB · Views: 190
  • 23W4557 Mr13P-8 26ADC Box.JPG
    23W4557 Mr13P-8 26ADC Box.JPG
    71.9 KB · Views: 195
Quick update -

Over the past few days I have been trying to learn VituixCad, find a better mid driver and learn the basics of what to optimise in a crossover.

I think I have got to grips with the passive and active model in VituixCad and can replicate results in Boxsim 2.0 which give some confidence that the two modelling systems are doing their stuff correctly.

The search for better mids turned up the Audio technology 15H Midrange driver (C-Quenze 15 H 52 12 06 SDKM) which, although expensive seems to be very easy to work with. In terms of modelling (taking into account I am using traced curves from OEM datasheets) the process I used was -

  • Produced initial crossovers that provided a fairly flat frequency plot,
  • Inverted Mid,
  • Moved crossover point until I had two healthy nulls (-30dB),
  • Then optimised the crossovers so that on the phase plot, around the frequency that the woofer/mid and mid/tweeter crossed over, the graphs were coincident.
Not sure if this is the correct process but seemed to work quite well.

I have added plots below for both passive and active, normal and inverted mid using approximately the same crossover points/ note to get the phase to work on the active I added delays to the drivers, passive version this is all done in the crososover.

Anyway, from the plots it looks like the 23W4457, C-Quence 15H Mid and SB26CDC play very nicely together with pretty simple LR2 (W/M) and LR4 (M/T) crossovers with LPads to get the sensitivities down to the woofer (admittedly low). Ok, these drivers come at a price but looks like they could make a pretty good system.

One final questions, does box shape matter (within reason) for a sealed woofer.

i.e. fro an internal volume of 35l using 20mm birchply (excluding driver, mid cavity and bracing for this illustration) will there be a difference between -

  • Stand mount 29cm baffle, 60cm high by 29cm deep,
  • Floor stander 29cm baffle, 100cm high and 18.5cm deep ?
So, I'd be grateful of any advice as to whether the process above is sensible and whether there is anything I can do with the plots to improve them.

Thanks in advance of further comments, much appreciated.


  • 23W4557 15Hmid 26CDC passive.JPG
    23W4557 15Hmid 26CDC passive.JPG
    403.9 KB · Views: 52
  • 23W4557 15Hmid 26CDC passive invert.JPG
    23W4557 15Hmid 26CDC passive invert.JPG
    422.2 KB · Views: 49
  • 23W4557 15Hmid 26CDC active.JPG
    23W4557 15Hmid 26CDC active.JPG
    406.7 KB · Views: 42
  • 23W4557 15Hmid 26CDC active invert.JPG
    23W4557 15Hmid 26CDC active invert.JPG
    435.6 KB · Views: 51
@eriksquires. Many thanks for the comment, and I agree with the breakup and tried to keep it at about half the frequency of the 3dB 30deg point (around 1.1khz).

However, my only worry is the big difference in sensitivity between the woofer (82dB) and the mid (91dB) which pulls the overall sensitivity down to around 80dB. I understand that low sensitivity speakers can sound dull at low levels, this may be misinformation, but most of my listening will be at low to mid levels (very rarely crank it up loud) so would not want to fall fowl of this.

If this is the case then I may be better reverting to the SS 22W8857 or SB WO24P bass reflex which would, push this up around 85-87 dB.

Anyone deny/confirm the sensitivity worries.
Last edited:
@draki many thanks for the comment.

Thanks, although the two sims look like passive, using the “active” crossover models it simulates a fully active system that can easily be input into minidsp. I have done this in boxsim and looks fine aswell.

I am in the process of building a dsp box using a minidsp nanodigi plus 4 Khadas Tone Boards and a usb to spdif converter so will have coax, toslink and usb digital inputs. I also have Rotel RMB1066 poweramps (6x60w) so my plan is to go active anyway. Just thought I’d model passive to see if I can replicate the active results and learn crossover design principals. Hopefully going active counteracts any issues with low sensitivity.
Given the differences in sensitivity and how low you want to take this speaker, may I suggest you go active/passive, with a DSP in front of it all? This would let you use a 2 channel plate amp, and a passive crossover for the tweeter/midrange.

The problem that vexes all audiophiles is that the lower you go, the more room modes you get into. I can tell you that the absolute best sounding systems at shows are usually active/passive with bass EQ.

After a lot of in room measurements for me, I'd never attempt low frequency without either a sub, or an EQ anymore. It's' just not worth it.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.