Leak Sandwich (original) rebuild and TH parameters

While I'm waiting for my MA drivers to arrive to put together with a KEF B139 I thought I'd get on with a refurbishment project.

I picked up a pair of Leak Sandwich 15 Ohm speakers recently and have finally resolved to rebuild them into larger cabinets - with a separate section for the mid/tweeter and importing a foreign tweeter. Barlow, who designed the drive units always wanted to add a tweeter considering that asking his smaller sandwich unit to handle 900-16,000 Hz was a tad of a tall order... and then the cabinet is too small for the 13" bass drivers.

I got around to figuring out on my learning curve how to measure the TS parameters using REW and a soundcard - as a precursor to a new sealed cabinet. I tested each driver 3 times and got pretty consistent results for each - but there are major differences between the drivers leading to huge differences in cabinet size.

I'm placing them below - partly because they may be helpful to someone else but also I'd be interested in why such large differences in VAS of 166 and 274 litres? I'll add the measurements of the mid-tweeter (if anyone is interested) in a separate post to save space.

For comparison Leak tested the free-air resonance at 19Hz, in cabinet (I think at 58Hz) with a cabinet size of approx 65/70 litres.

Speaker A

TS Parameters file
Room EQ V5.30.5
Dated: 03-Feb-2024 12:20:35
From measurement L Feb 3

Zmin 11.76 ohm
fmin 153 Hz
f3 878 Hz
Le(f3) 1.513 mH
Motional impedance parameters
R0 25.23 ohm
CMES 423.4 uF
L0 49.122 mH
beta 0.3437
omega0 291.5
Blocked impedance parameters
RDC 10.60 ohm
dR 0.23 ohm
Re 10.83 ohm
Leb 216.7 uH
Le 3.845 mH
Rss 100000.0 ohm
Ke 0.1665 S-H
Thiele-Small parameters
fs 30.6 Hz
Qms 1.301
Qes 0.882
Qts 0.526
Fts 58.2
Mms 87.59 g
Cms 0.309 mm/N
Rms 12.942 kg/s
Vas 166.10 litres
Bl 14.383 Tm
Eta 0.55 %
Lp (1W/1m) 89.62 dB
Dd 28.00 cm
Sd 615.8 cm^2

Speaker B
TS Parameters file
Room EQ V5.30.5
Dated: 03-Feb-2024 14:08:12
From measurement L Feb 3

Zmin 11.59 ohm
fmin 150 Hz
f3 861 Hz
Le(f3) 1.521 mH
Motional impedance parameters
R0 29.83 ohm
CMES 388.5 uF
L0 88.961 mH
beta 0.1273
omega0 248.0
Blocked impedance parameters
RDC 10.70 ohm
dR 0.18 ohm
Re 10.88 ohm
Leb 194.6 uH
Le 3.377 mH
Rss 99372.6 ohm
Ke 0.1847 S-H
Thiele-Small parameters
fs 25.9 Hz
Qms 1.546
Qes 0.688
Qts 0.476
Fts 54.4
Mms 73.85 g
Cms 0.511 mm/N
Rms 7.778 kg/s
Vas 274.67 litres
Bl 13.786 Tm
Eta 0.71 %
Lp (1W/1m) 90.72 dB
Dd 28.00 cm
Sd 615.8 cm^2
Last edited:
  • Like
Reactions: 1 user
For reference purposes here are the parameters for the Sandwich mid/tweeter (which is NOT) the paper cone model used on the very first Sandwiches.

TS Parameters file
Room EQ V5.30.5
Dated: 03-Feb-2024 16:53:16
From measurement L Feb 3

Zmin 9.95 ohm
fmin 1315 Hz
f3 4875 Hz
Le(f3) 0.238 mH
Motional impedance parameters
R0 7.82 ohm
CMES 117.5 uF
L0 0.797 mH
beta 0.1153
omega0 4705.0
Blocked impedance parameters
RDC 8.80 ohm
dR 0.14 ohm
Re 8.94 ohm
Leb 63.2 uH
Le 0.395 mH
Rss 43.8 ohm
Ke 0.0749 S-H
Thiele-Small parameters
fs 500.6 Hz
Qms 2.213
Qes 3.303
Qts 1.325
Fts 377.8
Mms 0.15 g
Cms 0.662 mm/N
Rms 0.217 kg/s
Vas 1.64 litres
Bl 1.140 Tm
Eta 6.34 %
Lp (1W/1m) 100.24 dB
Dd 7.30 cm
Sd 41.8 cm^2
This is the three way

For those who want to see the sandwich cone

Duraluminium - Polystyrol - Duraluminium on a straight cone.

I was thinking drivers, but graphs are good too. More context wrt what you ar edoing
Context is:

The original Sandwich is a very compromised design. It seems to be acknowledged that the box is far too small for the woofer and the mid-tweeter is being asked to cover far too big a range - 900Hz to 16,000Hz. Don Barlow (the sandwich designer) wanted to incorporate a tweeter as well but either Leak didn't have one or couldn't/didn't want to source a 'foreign' one. The result is that the bass is ok-ish but the lower mid-range is horrible. There is a paper on this by Stephen Spicer where he revised to XO to produce substantial improvement. See Leak Crossover

I was thinking therefore of

a) building much larger new boxes in the same manner as Leak - 9mm ply damped with 9mm bitumen. Keep the unique woofer bracing and add a horizontal brace to the wood block to the sides.

b) i) Revise the XO ala Spicer to XO at 750 rather than 900 to avoid any cone breakup or ii) use the BRY5 Class D amp (which has onboard DSP) for the woofer and bujild a new XO for mid and new tweeter

c) Add a tweeter - model tbd.

d) on box size - as the current design (similar in size to the Leak 2060) was intended as a floor stander but really needs stands - increasing the box size (or at least the height) turns it into a floor-stander. However using the data I posted earlier each woofer is so different that recommended box sizes are wildly different! What I came up with was something around twice the capacity - say around 140 litres.

Here is the woofer - 15Ohm original. The TH parameters posted above above are for these original Sandwich drivers - not the versions from the Sandwich 600.


...and fore and aft of the original Sandwich tweeter/mid - 15 ohm. note the unusual mounting of the unit on the baffle.



...and the impedance graph from REW
  • Like
Reactions: 1 user
I was lucky to have worked with Don Barlow in Jurassic Times. He was a true polymath but the Sandwich cone wasn't one of his best ideas ... though he was obsessed with it till his dying day.

He did inspire Raymond Cooke, then R&D Manager at Wharfedale, to try something similar. The Old Man, Gilbert Briggs, didn't think much of his efforts so Raymond left and founded KEF. The Old Man was heartbroken as he was grooming Raymond to take over Wharfedale. And that's the inside story of the B139's (and KEF's) genesis.

The Rank Organisation bought Wharfedale and later LEAK which is how I got to work with Don. The speaker in #7 is from the Rank days. The Treble unit is actually from the Wharfedale Dovedale 3, which was surprisingly good when I put it into a DBLT decades later ... better than the LEAK Sandwiches of the same period.

But the 12" Sandwich unit (about the same vintage as an early B139) might be a better candidate for a large TL.
Last edited:
  • Like
Reactions: 2 users
@kgrlee thankyou for such an interesting story - it's always good to hear of the people behind what they create. I always wondered why Cooke left Wharfedale. I have a lot of respect for those people back then (and I'm including the sound engineers at say Decca and the Beeb) - they were pioneers, often gifted amateurs, for whom there were no boundaries to the imagination.

I have the original 13" bass drivers - as well as a Sandwich 600 that has the 12" versions - although one of the 12" is slightly later having replaced the original - it has the brown fabric surround and black fabric dustcap - although the magnet seems the same. I rather like the 600 - it's got rough edges though - nasty resonances in the lower mid-range area and I'm assuming that a XO rebuild will help? Further the output from the newer bass unit is much higher than the other (the original) - and putting a tone generator through them (via the XO) presents a different sound. Again I'm assuming that this is down to the Elcaps being no longer of use - and the carbon resistors look to have over-heated - so I'll replace with higher power versions.

I'm proposing to just restore the 600s - hoping that the two different woofers 'even out'.
But the 12" Sandwich unit (about the same vintage as an early B139) might be a better candidate for a large TL.
I've measured the 13" woofers from the original (not the later 12" ones yet - I ought to do that today) and the parameters suggest only using in a sealed box.

So my project (as of today) is to a) put them in a much larger box b) retain the sandwich mid and XO and keep the current Hz of 900 c) introduce a foreign tweeter (maybe the contemporaneous T15) and d) redesign the 'upper' part of the XO at say 3,000 Hz.

As someone who obviously knows these units well I'd appreciate your thoughts. As I say it's an experiment to create what Leak could have done in 1961 to improve the original Sandwich. I'm not expecting the last word in sound reproduction - it's a tinkerers project - call it experiential industrial archeology!
Today a design like this is called FAST system. Advantage is it can have phase behavior of a fullrange driver. Adding a tweeter destroys this concept.

There is this modification from Stephen Spicer with transformer - I would try this one - it's tested.

Or you modernize the xo with new parts where necessary and use EQ. A smartfone app will do if you can measure and adjust.

Like this you can also keep the box volume of the original and eq the bass to taste
Today a design like this is called FAST system
I'm not sure that the original Sandwich would qualify as a FAST system? After all the midrange/tweeter is hardly a FR.

Adding a tweeter destroys this concept.
...or turns it into what Don Barlow reputedly wanted to create in 1960 but couldn't.
There is this modification from Stephen Spicer with transformer - I would try this one - it's tested.
I have that paper from 20 odd years ago. I read it with interest

I'll quote from his paper (my emphases in bold):

"Listening to the Sandwich MkII speaker, my main complaints are:
  • too much output between 1kHz and 3kHz
  • too little output above 6kHz
  • a mild bass lumpyness, and not enough extension
  • a lack of efficiency"

The problems in the treble seem to be the result of choosing a two-way design with
a low crossover frequency of 900 Hz. This cross-over frequency was chosen to
avoid driving the woofer into cone-break-up, which first occurs at about 1.5kHz. So
the tweeter is required to cover 900 Hz to 16kHz a demanding job. Of course, a 3-
way design would have eliminated this problem Barlows paper describing this
speaker recognised that a 1 sandwich diaphragm is required for extended treble
response, so one concludes that it was not pursued on the basis of cost.
(Later mod-
els produced in the 1970s were of a 3-way design).
The excess output from 1 to 3kHz may be
the result of juggling the tweeter output accepting excess output between 1kHz
and 3kHz in order to prop-up the sagging response at 10kHz or so while still living
within a +/-4dB envelope overall."

He goes on to say:

"A new crossover was developed that re-equalises the treble unit by:
  • reducing treble output below 3kHz (easily achieved)
  • increasing treble output above 6kHz (not so easy)
For increasing the treble output, a 1:2 transformer is used, so that the voltage from
the amplifier is stepped up by a factor of 2 (6dB). This reduces the system imped-
ance at high frequencies to 4 Ohms, instead of the nominal 15 Ohms. At the same
time, lower-HF output is easily reduced with series resistors and capacitors. The
revised crossover design and measured response is shown in the figures.
Alterations to reduce the bass lumpyness have been investigated, including the ef-
fect of reflex loading which does not appear to assist in this case as it increases the
system Q. To date, room placement appears to be the best tool to tackle this, or
alternatively a much larger sealed cabinet."

So... seeing as a) he's done this very interesting modification already and b) I can't seem to find suitable transformers I thought I'd i) put them in a bigger sealed box (my TH measurements suggest a box at least twice the size) and ii) introduce a 'foreign' tweeter. I'd tried to get in contact with Stephen Spicer via the Melbourne Audio Club but I've not had a reply from them.

...and finally he says:

"The result is useable, but probably not as a primary loudspeaker
perhaps a nice pair of speakers for the TV or workshop?"

It's a pity you're in Germany and not SW England, Dragan! I suspect we'd have much to share beers over.
Last edited:
Sounds like a good idea, the beer and listening to good loudspeakers!

I have experience with use of measurement assisted dsp. Together with a current feedback amp (cfa) the bass lumpiness and the like can be cured. Maybe with the use of cfa amps alone.

I have the Leak 600 and did these changes to it:

- after all the years such a speaker must be opened and checked:
Inside the damping bitumen alike pads all were fallen off - reglued it to the inside walls
  • all connections of the xo were welded as these unwelded connections were heavily corroded - big difference in sound
  • all electrolytics in the xo got 0.3uF MKP in parallel in order not to touch original parts.
  • the middle tone driver got correctly damped, the damping was badly distributed - big sound difference! No damping on the internal sides of the middle tone enclosure but all stuffing behind the magnet not preventing first internal reflections, very audible indeed.

- biggest fault of all: unexpectedly the black textile domes were deformed by being once pushed maybe by children's hands. Nothing you would have to change usually. - - - But after taking a needle and bringing the dome to original shape again - the domes in my boxes were sticky treated on its backside(!!!) and the textile dome sticked (!) to the pole plate completely changing the original sound. Absolutely a mechanical nogo.

- the mylar dome of the tweeter of this three way box got a thin foil aluminum tweak making it sound more precise, watch here where I did this tweak on another box:


- in the end linearization with Behringer deq2496 after all the other mods, but at that time I had no cfa amps nor I knew about them anything at all

Last mod with eq is not as important as the previous ones. My boxes had no bass lumpiness.
  • Like
Reactions: 1 user
After one bass was scrubbing I had them changed their textile surrounds to a well damping rubber surround and changing the yellow spider for a more mellow one, bringing fs more down and get the scrubbing fixed.

This mod was done by professionals specialized in doing this.
Last edited:
  • Like
Reactions: 1 user
We know that the famous ever being sticky damping treatment collects dust over the time if applied to the front of a textile dome.

But whoever had the clever idea putting it on the backside did not take into account the kind of massive sound deterioration due to sticking on the pole plate after being pushed.

This was in my Hifi experience one of the most unexpected things I learned.

Later than I treated the middle tone domes with thin alu foil, too. Like the mylar tweeters. Also this sounded better in my ears. It's an aluminized sandwich cone so why not the textile dome, too?
  • Like
Reactions: 1 user