Some of you may have been following my reported progress
with trying to build a pair of ER Audio ESL-IIIB kits
which were purchased by a customer of mine from
http://www.eraudio.com.au/index.html

I have had the job of "assembling" these kits to a satisfactory level of
operation.

I got started back in about February this year and could not proceed as
quickly as
planned because of difficulties encountered resulting from the poor
design of the kits.

Before people think I have no prowess at building kits or speakers or
amplifiers, just take a look at
my website at http://www.turneraudio.com.au and compare the amount of
information and depth
of understanding compared with the all too brief techno info at the ER
Audio site.
At the ER site there is no proper electronic schematic for the ESL3B as
it should be, no impedance
measurements, and no measured performance data.
These should be stated with ESL, because some amps will NOT be able to
drive the ER speakers,
whereas the speakers at my site are all easily drivable, and everyone
knows about SEAS
drivers being OK, so nothing I have made is any worse than the best
speakers made with SEAS drivers such as Sonus Faber and
Vienna Acoustic. I gurantee a flat full range response, ER don't
gurantee anything.

I have now overcome most of the ERA problems; I was not warned by
anything at the ER Audio site.

Here is the history so far of what happened, and it includes process
details and
some solutions worth remembering.

The kit was supplied with spacers already glued to the 1,200mm x 550mm
frames for the 3 panel
speaker.
8 of the 12 machine screws riveted to the 12 stator plates were too
long, and I had to cut them down with a Dremel tool
to 6mm. Failure to notice if the screws are too long will result in a
completely stuffed panel,
because the stators frames won't press down flat properly onto the
frames they are glued into
with polyurethane glue.

The only real mistake I made was with the first attempt to stretch a
membrane of the 3.5uM gossamer thin
mylar membrane. It is just so delicate! It was easy to tear using tape
and a spring balance
for sequential tensioning.
And sure enough it tore, and I had to begin again, and the second
attempt went perfectly until I glued the
first frame with its stators in place onto the surface of the stretched
out membrane using super glue.

Unfortunately, a pin hole of invisible size allowed super glue to seep
through to the bench underneath,
and the membrane was ruined as I lifted the panel off the bench.
Its easy to sand it off the frame and clean up to do it again.

Could practice make perfect? I was about to learn....

Use of superglue as recommended in the ERA 33 pages of kit instructions
was abandoned!!!

I sanded down the surface of the MDF sheet I was using as the bench top
with very fine 600 grade
wet and dry sandpaper to remove whatever might have caused a damn
pinhole.
The bench sits on top of a heavy oak dining table I built 31 years ago
from thick oak planks.
( I don't have dinner parties anymore,
and the accoutrements of a gentleman have more productive uses these
days )

The third membrane went fine when I stretched it, and I used the
polyurethane glue
to glue to the frame after painting on the glue thinly to the plastic
frame.

I learnt that the easiest way to weigh down the frame on top of the
membrane already
stretched out on the perfectly flat solid bench is with about 24 house
bricks.

After lowering the frame with wet glue onto the membrane, cover the
frame with a damp
towel. The moisture in the towel will ensure the polyurethane glue cures
well in 24 hours.
Place the bricks on top of the towel gently without allowing the frame
to slide around.
When all bricks are placed, gently go to each prick and press down hard
with
about 20Kg in turn, and this will help remove air bubbles in the glue.
The towel prevents brick particles falling into the speaker panel.

You need to be very clean about everything with the construction; I used
my loungeroom
because of the good lights, and clean environment.

Next day I very carefully cut around the edge of the membrane and I had
a perfect looking
membrane when I lifted the bricks and towel away and lifted the frame.

The conductive strip of copper foil was placed on the second frame and
there were no problems.
This self adhesive strip 4.5m wide by about 0.2mm thick is actually the
foil
they use around bits of glass used in leadlighting windows, and is
easily available.
DON'T drill the 4.5mm dia holes for the 12 x 4mm dia plastic bolts too
close to spacer edges
or else the membrane will not glue to the spacer properly and will begin
a tear at the hole.

If you don't understand this, you will have torn membranes, and need to
relocate
bolt holes, after filling up wrong ones, and this is a PITA.....

I applied the high resistance coating to the membrane. All seemed to go
OK
but I used twice more than recomended, and Rob Mackinlay from ER later
said I shouldn't use too much.

After waiting a week for the coating to cure, I completed the basic
assembly with supplied
plastic bolts and plastic surrounds and I had a basically completed
speaker, but without
the timber surround or box for electronics underneath.

The kit does not include any surround timber frame or box, and one is
left to make all that oneself,
its rather like being given a speaker kit with drivers but without a
box.

I can only imagine the terrible attempts being made by amateur
audiophiles who have no
carpentry or joinery making skills or tools.
You won't want to see their speakers pictured on the Web.

Then came time to hook up the input transformer and build the not
supplied board for a suitable
cross over for the speakers.
I designed and built the un-supplied crossover and required board.

The electronic parts supplied were a regulated LV dc power supply which
can be
adjusted for its dc level. This then feeds a supplied EHT supply which
converts the sub 10Vdc to
up to about -5.5kV for application to the membrane.
The EHT supply is connected with -EHT voltage taken to the copper foil
strip around the frames
and the ground terminal of the EHT supply is taken to the CT of the SUT
secondary.

The SUT is supplied, but has half the core size and half the primary
turns really needed for
saturation proof operation.

I have advised ER Audio with full details of an alternative design for
their transformer,
and i look forward to them addopting my recomendations.

The supplied SUT has far too thin P to S insulation, resulting in 390pF
of shunt capacitance
appearing across the secondary, which transforms to massive 3.16uF at
the primary.
The treble panel has 100pF capacitance which transforms to 0.81uF,
and the total input capacitance seen by the amplifier = 4uF,
( about twice what Quad ESL57 managed ).

The bass panel capacitance of 800pF transforms to 6.5uF at the primary,
but the series resistance
prevents this much C ever being experienced as a load on the amp, so
bass panel C is quite entirely benign.

I tried a very basic crossover network about which I will have more to
say,
and hooked up the speaker and tranny to an amplifier capable of 21Vrms
max into a
high Z load, and up to 70 watts into 3 ohms, 14.5Vrms.
Output Z was 0.2 ohms, BW 7Hz to 30kHz, 2 ohms connected at low levels,
and THD 0.2% at max po into 8 ohms, 50 watts, 18Vrms, and mainly class
A with a
quad of mosfets.

I turned on the EHT and gradually increased the EHT until I heard all
kinds of spluttery spittery sounds
and saw numerous blue discharges occuring all over the panels. I backed
off immediately
until the noises stopped and conducted sound tests using pink noise and
test gear as described in my
speaker building pages at
http://www.turneraudio.com.au/loudspeakers-diy.html

Then I began testing the responses at 1M, 2M ,4M distances at a 1.2M off
floor level, on axis, with the
speaker taped temporarily to a stout dining chair I also built 31 years
ago.
Bottom of the speaker is 450mm above the floor, speaker is vertical.

I got a truly appalling response.

The sensitivity seemed appallingly low.
For the same quite low level of pink noise, I needed maybe 5 times the
voltage I'd need for my own Sublimes
shown at http://www.turneraudio.com.au/loudspeakers-new.html

The power handling was atrocious, and LF capablity suffered from arcing
and distortions.
My customer visted me in the middle of testing and was not impressed at
all.
We talked about quitting, but I ain't no bloody quitter.

I obtained the loan of a high impedance HV probe to measure up to 20kV
safely,
and found that any EHT increase above 2.2kV cause horrid discharge noise
and premature clipping.
ESL sensitivity is about proportional to applied EHT voltage, and I
wanted to be able to apply
5,000V if possible, like ESL57 and ESL63.

I then thought that the arcing seen must be because the poor insulation
qualities of
the black powder coating on the suplied stators.
So I stripped off membrane No 3, and coated all the stators
with two generous coats of anti corona paint, Isonel 642,
and put on a new membrane No 4 but with two very thin conductive coats
on the membrane,
this time using the right total amount and to get an even coating.

I also placed in the pattern of 10 anti resonance silicone pads but that
made no difference
to following results.

I re-tested after the rebuild and curing times, and although there was
not such a lot of arcing and EHT leakage, the response and problems were
quite
unacceptable. I turned off the EHT supply, and considered using more
membrane tension
of 2Kg / 60mm of side length, not the low 0.8Kg recommended.
ESL63 use 2.5Kg for the same material.
Lots of crackling noises continued for minutes after turn off of the EHT
and after the
voltage had rapidly fallen to 0V.
It was the membrane partially releasing from its stuck position.

When I unbolted the two halves of the speaker the next day after the
dismal tests,
I found that about 70% of the area of both bass membranes
had become glued to the stators on the side away from that where the
membrane is coated.
I was able to gently prise the membrane off the starors with a bit of
bent copper wire
poked through slots in the stators.
Anti corona paint is slightly sticky with mylar.

No wonder I had such horrible test results. But it had taken two
attempts before I became aware of the
problem and I guess that EVERY SILLY MAN WHO TRIES TO USE ABOVE ABOUT
3KV
will have what I call membrane "stiction" problems.
There are posts I've seen in these silly little private forums where
they all bull**** to each other
of guys cranking up the EHT and I believe they don't realize
their problems. With stuck membranes, the speakers still work at mid &
treble, but
poorly, and definitely not as intended, and with queer bass.

Back to my drawing board I went, and at this point I didn't think it
would be wise to consult
ER Audio for advice, and I posted on these groups to see if anyone else
had had problems.

One Collin Topps of the UK answered my initial posts to groups in
private emails
which began cordially, but ended rudely. Collin is the UK ERA sales rep.
He said repeatedly nobody has had any problems building ER Audio panels.
He refused to accept that I was having serious bothers, or offer any
advice that was relevant and useful.
Then he ended up telling me to sod off, and I thought that this
technically dumb salesman
in the UK acting for ER Audio as their sales agent has
some basic things about public relations he has yet to learn.

He'd sure never get a job selling anything I made.

I continued to post on the panels and learn more about ESL construction
from other ESL amateur productions around the world.

Since then, I decided that the membrane must not be allowed to
travel too close to a stator, ideally not more than half the actual
distance
between membrane and stator surface, so if this is 4mm, as it is in
ESL57 bass panels,
then 2mm is enough maximum movement distance. Quad achive this by using
2mm thick PVC plastic sheeting for bass stators, with conductive stator
material being conductiove paint on the outside of the perforated
plastic sheeting.
So the membrane CANNOT move more than 2mm and cannot be less than 2mm
away from the stator
surface.
The of force of electrostatic attraction increases proportionally to 1/d
squared,
so if d is only a few thousanths of an inch,
F becomes huge.
No wonder my ERA membranes were sticking like cling wrap around a
pumpkin at the grocer.

The bass panel membrane to stator distance in ERA panels is 2.4mm.
I thought that if I applied about 0.8mm of non conductive material to
the inside surface of the stator,
It would keep the membrane getting too close to a stator, and the extra
tension
would always overcome the force caused by the EHT without signal.
I'd have 1.6mm for membrane movement, not as much as in a Quad speaker,
but enough.

So I applied four coats of silicone paint made from roof and gutter
silicone sealant
bought for $5 per tube at a plumber's store.
To coat all the stators for one speaker I used 1.3 tubes of silicone,
and about
600ml of white spirit for thinning.
The thinning is done by using a small cheap brush with dabbing action in
a clean tuna fish can
until the paint has thick honey consistency and no visible lumps.
The silicone is applied by dabbing action with the brush laid flat on
the perforated steel sheeting,
not letting silicone close off perforation holes.
It self levels fairly well, and partially wraps around the sharp metal
edges of the holes in the stators,
so probably enhances air flow caused by sound waves.

24 hours is fine between coats of silicone. It sticks well to the anti
corona paint I already have on the stators.

The silicone becomes rubbery, and smooth surfaced, even though some
uneveness
of this coating thickness occurs, but basically, I got the right amount
of
goop applied to all stator surfaces.

I finally stretched membrane No 5 last week and have re-assembled the
speaker
for a new test.

After the frame was first glued to the stretched and much tighter
membrane,
I still found the membrane tended to stick to the stator if pressed over
to it gently by hand.
So I then bought a small can of Johnson's Baby Powder and squirted in a
lot of powder through perforations,
and with an air blower, blew it all around everywhere.
Then i sucked off the excess woth a gentle vacuum cleaner.
On the opposite second frame and stators, i just dusted the powder on
and
removed excess with vacuum.
The panel was then re-assembled.

Did my ingenuity prevent the problems I was having?

After connection of the speaker to the amp and without signal, I raised
the EHT and
at 3.2kV while I watched the bass membranes carefully I saw them whip
over to a stator and remain hard against it.
Hmm, same old problem.
However, when i reduced EHT, the membrane let go without all the
crackling
sounds they'd made previously after turn off of the EHT supply.
The powdered layer of rubber was doing the right kind of trick in
preventing
stiction.

I was able to leave the EHT on all night at 3kV, but just occasionally,
some
small click noise would occur, so i have settled on having 2.7kV as the
maximum EHT voltage
which could be safely applied to these panels before problems would
occur.

I began to seriously test the speaker response and do serious
comparisons again
with my own blameless Sublimes.
They sit side by side, and comparisons can be fairly made, same room,
200Cubic metres,
well out from a wall, well damped etc.
The same amps, test signal, test gear and mic is used to eliminate any
chance
of making an unfair comparison.

Finally, I got the speakers to measure +/- 2dB between LF pole = 35Hz
and HF pole = 22kHz,
and to give as flat a response as I could get without adding
a ridiculous number of crossover compensation networks.

This response was measured at 3M with mic 800mm above the floor, and
about exactly
how they will be listened to in my customer's room, which is not quite
as good as my own,
but measurement of his existing speaker response has been little
different to what i get
in my rooms. If I get speakers to sound well here, they always travel
well.

My impressions with the sound with music is good, but the 50 watt class
A amps I have been using
for trials run completely out of headroom at only modest levels equal to
about
1/3 watt average into my Sublimes which are 5.6 ohms average, and have
88dB/W/M watt sensitivity.

After numerous calculations, I figured I needed 5 times the applied
voltage for the same SPL
with ERA panels compared to my own speakers.
At least a 100 watt amp would be needed for quite modest levels of
sound,
not because 100 watts would ever be generated, but merely
to get enough voltage.
Collin Tops said he was using an SET amp with 16 watts only,
and maybe he likes quietspeakers.

But I prefer loudspeakers.

And for these ERA quietspeakers, there is not the slightest bit of
speaker protection circuitry included
as it is in Quad ESL63. What is the safe maximum voltage?
not as much as for Quads, that's for sure.
Anyway, I did clip the 50 watt amp a few times when i tried to go loud
with some busy Salsa music from Buena Vista Social Club,
and nothing bad happened, so the 6 coats of goop I have on the stators
seemes to resist arcing.

A 16 watt SET amp with Rout 2ohms would be a horrible thing to drive
such ESL.
There would be a serious HF roll off, but perhaps Collin is an old giza
with stuffed hearing,
and so may not mind if the HF pole = 6kHz.
And without my networks and crossovers, the response could be worse,
peaked up high between
500 and 6kHz, ie, without bass or treble.

He will have to conduct all the same well conducted tests i have done
before he'd ever know WTF he's actually built.
He should not be offended when I say I don't trust what salesmen tell
me.

Nobody from ERA has had the courage to discuss their speakers in the
public forums.

I have just measured the Z, and with the details of networks below got
the following Z figures:-

10Hz, approx 80ohms,
20Hz, 41,
50Hz, 14,
60Hz, 11,
80Hz, 10,
110Hz, 11,
200Hz, 18,
300Hz, 20,
500Hz, 19,
1kHz, 15,
2kHz, 11.5,
10kHz, 5,
20kHz, 3,
40kHz, 2.

If you plot the above points on a graphed sheet of log paper, and
join the dots with gentle curves and you will have the Z.

The average Z between 50Hz and 2kHz = 14 ohms.

Therefore a tube amp set up for a match to 16 ohms will work,
but it'd need to be able to make 30Vrms, which is 64 watts into 14 ohms.
The amount of signal above 2kHz which feeds the low Z average of 5ohms
between 2khz and 20kHz
is small, and as long as the Rout 0.5 ohms there will not be a serious
loss of HF detail.
One of my 8585 tube amplifiers with a quad of KT90 in PP per channel
is powering 3 stacked pairs of ESL57 no problems, even though the Z =
0.6 ohms at 18kHz.

But 16 watts from SET would be hopless, unless you had the speakers so
close they are like
giant ESL headphones.


DO NOT use the recommended circuit resistances so vaguely shown in the
ER instructions.

I found that the best way to hook up the ESL-3B speakers is as follows:-

BEFORE the the input to the primary of the step up transformer, ( SUT
):-
Make a series L + R network of 4mH plus 27 ohms.
Connect this across the P winding.

Connect one end of the P winding to the 0V terminal for the black lead
from the amplifier.
To the other live primary end, connect a series C & R network of 150uF
plus 1.5ohms across to
the speaker terminal to which the red wire from the amp connects.

These two input networks will have several effects.
The 27 ohms plus 4mH makes the input loaded with 27 ohms at low F,
thus damping the low Z of the series 150uF and primary shunt inductance
of 32 mH.
The 150uF will prevent LF entering the speaker and causing the
rather poor quality SUT from saturating all too easily with high
transient LF voltages.
The -3dB cut off is at 50Hz, and the attenuation is second order, barely
enough.
To get better headroom for these speakers, use a second input cap in
series of
100uF, giving -3db at about 100Hz, then use a sub. More signal will be
then applicable
above 100Hz.

The 1.5 ohms adds to the 0.5 ohms of SUT winding resistance which is
important to damp the series resonance
between the awfully high amount of shunt capacitance in the SUT and the
leakage inductance at around 20kHz.
Therefore input resistance is a minimum of 2ohms by 50kHz, so not many
amps should blow up.


AFTER the secondary of the SUT, place the following networks
from each end of sec to the bass panel terminals:-

You will need the following parts,

16 x 100k resistors, metal film, 1%, 2 watt rated.

2 x 0.0033uF capacitors rated at 4kV.
Find good quality polypropylene caps.

To each bass stator, connect 100k to a pair of common points for the
rear and front pairs of bass panels,
Then series 4 x 100k to make 400k from each common point to the
ends of the SUT secondary.

To each treble stator, connect a parallel pair of 100k to make 50k to a
pair of common points
for front and rear stators.
From the common points, connect the 0.0033uF to the sec ends of the SUT.

The 0.0033uF plus 50k form a CR HPF.

MAKE SURE the treble panel with its crossover is REVERSE PHASE CONNECTED
TO THE BASS PANELS.

The -3dB points in the signals in the electronic circuit thus created is
160Hz for bass,
with signal rolling off at 6dB/octave above 160Hz, and 700Hz for the
treble panels, also
rolling off at 6dB/octave.
Keeping pass signals out of the treble panel is a good idea.

The speaker phase shift will be maximally quite large, but maybe under
60degrees,
but I found the treble acoustic signal produced by the bass panels will
affect
the treble panel badly if not phase reversed ( like many other dynamic
speakers ),
and despite the attenuation by the R & C filter LPF effect of bass
series R and the bass panel capacitance
which totals 800pF.
The ONLY way I could flatten the response and remove what was a wide 6dB
dip in the resonse at 5kHz
with reverse phase connections.
This what I measured, and the acoustic response has little corelation to
the signal applied.

I will perhaps draw all the relevant curves of bass and treble signals
applied and
and include all this info at a website page.
I hope to include a schematic of the equivalant LCR model of the the
panels and SUT.
I hope there would be a net positive result amoung builders of ESL.

This should all have a very positive effect on understanding
of what is really involved to make a pair of ESL speakers.
Hopefully, all of what i say will fertilize the minds who created the ER
Audio ESL speakers.

They can scream long and hard at my critique, and i don't give a ****,
their
marketing is quite deceptive, and they say
it'll be easy, and take only so long, but no, its not true,
and they are bull****ting, and the total time I will end up spending on
R&D,
and patching up the mess of the thing they supplied plus do all the
woodwork
will amount to 10 times the time they say.

My customer here was lulled into the project, along with me, under what
now seem like
false statements about the product.

Had I known of all the difficulties and parts quality and non
inclusions, i would never ever
have agreed to my building of one of these kits after my customer bought
it.
He isn't technically minded,
and could NEVER EVER have successfully built a kit from ERA.

If I allowed quite low trade rates of $50 per hour to build these
speakers and fix problems and do R&D,
I would find I would be entitled to claim a payment of maybe $5,000 for
the labour
of building these speakers, plus the cost of supply of items not
included, about $400.
being a gentleman I won't ask more than a fraction of the figure from my
customer,
but frankly, the sooner I see these speakers leaving my workshop, the
better,
and I don't ever want to build another pair that I did not fully design
myself!!!!!

The original kit price was $2,000, and so the whole deal of up $7,400
value
is quite poor value when a fully restored/rebuilt pair of Quad ESL57
could be had for
about $4,000 from one John Hall in Melbourne.

I have a Quad ESL57 which is unrestored and it even outperforms the ERA
on most counts.

How anyone could say the ERA speaker is better product than a Quad ESL57
or ESL63, as some have suggested
is quite beyond me; maybe these people had old speakers with lots of
faults, and replaced them with ERA
because they thought they'd be better.

The other complaint I have about the ESL3B is that its ONE panel, not
two bass with one treble, like ESL2B.
This means that if a membrane tears or ****s up in just one of the 3 sub
panels, the WHOLE
3 subpanels have to be fixed, rather than just fix one of the 3 sub
panels.

There is BS spiel about why the ESL3B is a better than others, but ERA
are pushing the flat ESL3B because
it appears to me the cost of production is less for the less complex
product.
but the repair bills for YOU, the buyer, will always be higher.

All the comparisons one reads make silly one liner statements without
the pages of detailed back ups.

So don't make me waste more time reading your replies which are not
backed up with facts.

Patrick Turner.