You can use LTspice to model the effect using the Hans Polak model from the Quad ESL-63 Delay Line thread:
https://www.diyaudio.com/community/...ater-delay-line-inductors.338927/post-6710863
Rx was set to 1.5 ohm and 0.001ohm(ie effectively removing the filter)
You can see that removing the filter lifts 100hz - 1kHz by 1-1.5dB, and reduces the impedance by 1.5ohm.
Changing Cx from 220uF or 47uF would put a broad ~1dB dip in the response centered on 1.5kHz
More discussion in Post #199 and 200 here: https://www.diyaudio.com/community/...ater-delay-line-inductors.338927/post-6229336
thank you very much for the advices and the plots.
Under
https://keith-snook.info/loud-speaker-stuff/QUAD-ESL63/QUAD-ESL63.html
there are also to find an interesting statement concerning the parallel network in series to the signal pad from Rob Flain, IAG Group Service Manager,
MKP (stable & electrically neat) vs. electrolytic capacitor (flawed long-term stability / may cause distortions when at LF and high AC currents). If I have had some 220uF MKP's handy, and if they would have physically fitted into the available space, of course I would have chosen the 220uF value. So my swap was certainly a compromise, and I hope/imagine a good one. I am open for further objections.
@ Bolserst: Thank you for your simulations and the forum links! I am eager to make some comparative measurements as soon as I can. I will then post the results. It's internally scheduled for within a week or so.
Done that. After a lot of searches I found exactly what I needed. Soda can sized / dumb priced Mundorfs (audiophile tax) were ruled out right from start.
As you can see it's a really snug fit, adjusted by less than 1mm in available space, two 110µF MKP caps wired in parallel, and I'm happy with the visual result.
Data : 110µF/600V - ESR @10KHz =0.0048Ω (https://www.mouser.fr/datasheet/2/315/RDL0000C251-1772694.pdf)
Also replaced original wirewound resistor with a ceramic one installed on heatsink.
All items sourced these from Mouser.
Last edited:
There is some more resistors in series with the transformers... maybe you want "audiophile" variants of those also?
If your are mentionning R1 & R2 (1R62) look closely to the above photo, the cemented wirewound models have been replaced with "audiophile" ceramic ones.
With those resistors in series and the fact that the transformers itself has several ohms, it's funny that some people experience "improvements" in sound by connecting THICK wires to the speakers...?? The worst and really BAD example is when they change the leads from delay line to the panels to thicker ones!? Why that is bad? Since you have impedance transformer of approx 1/12 000, which means that 1,65 ohms on primary side is equal to 20 000 ohms on secondary side in the speaker. So you can actually put in resistors in each of those wires of 2000 Ohms and you would not hear it, nor measure the difference in output, NO WAY! That's the obvious part... the next part is that the capacitance between those wires and the surrounding is critical! So going with THINNER wires is actually a benefit, since stray capacitance is important to be kept low!
Lots of engineering hours weeks months years, was spent on optimizing those aspects. Why they are so thick as they are, is only for mechanical properties, and handling building! Otherwise they would be as thin as the delay coil wires (several kilometers long in each coil by the way) like a human hair! Of course this would be impractical.. So IF you want to change those wires go for thinner ones to keep the stray capacitance low, or just leave it 🙂
They are called ESL 63 since they started the work developing it in 1963... it came out on the market 18 years later. And Walker, Baxandall and other engineers involved did a huge leap in technology. So big in fact that nothing in the design has changed in more than 40 years... I will reveal some modifications that really IS an improvement. It is based on that some compromises that was done in the 70ties can be handled today...
Lots of engineering hours weeks months years, was spent on optimizing those aspects. Why they are so thick as they are, is only for mechanical properties, and handling building! Otherwise they would be as thin as the delay coil wires (several kilometers long in each coil by the way) like a human hair! Of course this would be impractical.. So IF you want to change those wires go for thinner ones to keep the stray capacitance low, or just leave it 🙂
They are called ESL 63 since they started the work developing it in 1963... it came out on the market 18 years later. And Walker, Baxandall and other engineers involved did a huge leap in technology. So big in fact that nothing in the design has changed in more than 40 years... I will reveal some modifications that really IS an improvement. It is based on that some compromises that was done in the 70ties can be handled today...
"I will reveal some modifications that really IS an improvement. It is based on that some compromises that was done in the 70ties can be handled today..."
Looking forward to that. Surprise us.
Looking forward to that. Surprise us.
Make sure you check the power rating on those resistors, TO-220 power resistor power specs are specified assuming they are attached to a heat sink. That 35 watt resistor is only a watt or so without a heat sink.
From ohmite's data sheet: "Ohmite’s TCH35 TO220 package resistor provides 35W of steady state power when properly used in today’s well defined heat sink applications."
Here's a table from Riedon:
Note the heatsink vs free air power ratings.
Sheldon
EDIT: I suspect you knew that because you put the replacement 1.5ohm on a heatsink. But I thought I'd mention it since the new resistors cut the power handling on the clamp board by about 5x.
Last edited:
There is some truth to that, but maybe not entirely. For example they had an undersized filter capacitor in the power supply for decades; it stayed there through three series of modern quads, and is super easy to have caught and fixed. in the USA, that thing is over it's rated voltage. The panel support structure should have been tweaked slightly to provide mechanical support for the stators instead of relying on glue and hope.
Then the company was sold to IAG and it appears that they don't have any serious electrical engineers there, as the changes they have made over the years are pretty slap and dash and iffy. For example until the 2812 series, the circuit board for all electronics under the speaker were just the three ESL63 board circuits slapped together on one pcb board. Not well laid out for the location of the wires, and not re-laid out for better flow on the board as a single unit. But I shouldn't harsh on the EE guys, the mechanical design of the 988 and 989 is pretty bad too. The panels are so floppy in the 989 that they can hit and damage the dust covers in shipping. The bases bottoms are held in place with very small thread-cutting plastic screws best used on a $6 child's push-car. In fact all the fasteners and fastening methods for most things in that speaker are pretty bad. More akin to a cheap toy than a speaker that cost many thousands of dollars.
Don't get me wrong, I love these speakers, but no changes doesn't mean perfect, it can also mean lazy.
Sheldon
Also between those resistors is a 1uF cap that you might want to change. Everything else on that board isn't really a contributor to the sound of that speaker (unless it's triggering, then it effects the sound like audiophile parts wish they could).
I'd also suggest that you not mess with the HV supply. Remember those ceramic caps are completely air-gapped (or should I say neon-gapped) out of the circuit when the neon bulb isn't firing.
Sheldon
Thank you Sheldon, although I was aware about this, it's a good thing you pointed at it 👍
I always carefully read datasheets when they're available and indeed, power handling of the Bourns PWR220T in free air is 1.5W, which actually is 6 times less watts than a 10W cemented model. Using a heatsink will allow dissipating 50W. Although I am not using a lot of electric power when listening, I will bolt those resistors on a large and thick copper plate, that should provide better power handling.
I am looking forward to. Now asking after four days of waiting for your welcomed post: By when do you plan to reveal these modifications?
I have been mentioning about those mods for at least ten fifteen years.. I have not invented all by myself but together with MT Audio. Lots of work and measurement has been done, and times just runs. I guess its time soon for a thread about this. Let me think over the Easter how to present this.
I have had the idea of several products that probably never will be realized, and the community has always been so generous so maybe this is a good start to share 🙂 A teaser... look on the ESL 63 and 988, 989 etc etc directivity pattern, and for you who have heard them, there is an issue with the narrow sweet spot.
ESL 63 directivity pattern. It can be done better, and the sweet spot can be much more seamless. And i mentioned another clue in a previous post.
More meaasurements from Troels Gravesen http://www.troelsgravesen.dk/QUAD-ESL63.htm They are good but there is room for improvement!
I have had the idea of several products that probably never will be realized, and the community has always been so generous so maybe this is a good start to share 🙂 A teaser... look on the ESL 63 and 988, 989 etc etc directivity pattern, and for you who have heard them, there is an issue with the narrow sweet spot.
More meaasurements from Troels Gravesen http://www.troelsgravesen.dk/QUAD-ESL63.htm They are good but there is room for improvement!
So then, "a week or so" lasted 18 days ... I could now make a comparative measurement, showing that Bolserst's simulations (Post 41) are indeed a very neat fit to reality.
The measurements I made were close range (50cm) measurements, not reflecting the dipole's characteristic (which would be of no importance for this specific issue, anyway). And of course, with values for C25 such as 47uF (red) vs 220uF (green). The black curve shows the difference between the measurements with 47uF and with 220uF. There is this shallow (-1dB) and very broadband dip between 350Hz and 4.5kHz basically as Bolserst predicted.
My conclusion would be that this approx. four- to fivefold smaller value of 47uF is of no practical importance for my everyday listening. The overall DSP target curve corrected response will be different anyway.
This might look weird and wrong at first glance.
But, ask yourself what is thicker, the copper or the isolation?
If the first, it is wrong? If the latter ............?
This is is done with 20 kV, 24 AWG UL3239 silicone wire. Right, or wrong? Whether it adds anything is another useful question.
Agree to 100%, if you look on the bass frequency response that is depended on the 360k resistors in series and that together form an RC network with the panels. Did they change the resistors? Nope... Did they get a bump in the frequency response? Yepp sure! I have done some work there.
Me too!
Which is exactly what most people want and is sold as more/better bass response. 😕
Not saying it is good, not at all.
Just saying it could commercially / marketing wise, probably be a good move in the current bass driven audio world.