New speaker design by NOOB

[thor.zmt]

This is not what I see in the schemo at 1mn 10s in the video, neither in the frequency responses. The 1st tweeter has both a high-pass and a low-pass.

Yes, this is the updated crossover the guy in the video, who according to you has no clue what he is doing proposes.

What I posted was the original stock crossover.

So, in my book it makes a true 3-way system.

Only if you fit the crossover designed by an "idiot (who) could not see a 3-way x-over if it bit his ass".

Now, I don't know these speakers, never seen them, nor heard them, nor measured them, never retro-engineered them, so, if this schemo and these responses are not authentic, what's the point of publishing them in a video?

You would know if you ACTUALLY HAD WATCHED THE VIDEO!

Thor

 

[thor.zmt]

I did my own bench testing on parallel caps back in 70s when I discovered phase errors in a RIAA phono preamp in the top octave caused by aluminum electrolytic capacitor's ESL. This top octave phase shift was measurable but not audible (tens of degrees at 20kHz). That capacitor was in the ground leg of a non-inverting RIAA preamp stage, so something like 360 ohms in series with 22uF.

That's a very poor way of implementing the DC blocker, especially in an active RIAA equalised phono stage.

Minimal rearrangement places this capacitor inside the feedback loop, where it should be, together with the output coupling one.

Seeing that your resistor setting the 3180uS time constant will likely around 360kOhm (with 360R in the lower feedback arm) you might even find just a 2.2uF film cap will do fine, or alternatively the 22uF electrolytic cap will do fine.

From my bench testing I empirically determined that a parallel film cap needed to be 10% of the total combined value to effectively behave like one large film cap. This is contrary to the popular wisdom that suggests much smaller value film capacitors in parallel with large electrolytic caps could make audible improvements.

Not sure about "popular wisdom", the way I heard this wisdom was to use the largest possible value filmcap to bypass electrolytic capacitors.

Coincidentally I also determined that a 22 uF tantalum capacitor also delivered better top octave phase response than the aluminum electrolytic cap.

Did you measure LF distortion? Tants ate really poor there. Sonically, subjective it depends.

Naim hifi gear famously uses (used?) Tants and sounded rather poor with them. Replace them and the typical naim "Hifi" turns to "natural".

What I mean with "Hifi sound" are leading edge emphasis, excessively "bright" sound. It might have worked semi ok sonically with overly dull 70's rock recordings (and using dull sounding mm pick-ups like Shure V15) it never worked with classical and sounded horrible with 80's recordings and DMM cuts using a good mc pick-up.

I make no claims that the tantalum sounded better than aluminum in this application, I only know it delivered better measured phase response at 20kHz. FWIW the tantalum cap was several times the size of the aluminum, but it was higher voltage cap.

That may have been the key factor. Common tants are relatively low voltage, high(ish) voltage types are made differently.

My sundry future phono preamp designs avoided that topology for that and other reasons.

Good choice, DC blockers belong inside the feedback loop.

Thor

 

[JohnRoberts]

Yes, this is the updated crossover the guy in the video, who according to you has no clue what he is doing proposes.

What I posted was the original stock crossover.



Only if you fit the crossover designed by an "idiot (who) could not see a 3-way x-over if it bit his ass".



You would know if you ACTUALLY HAD WATCHED THE VIDEO!

Thor

still making friends?

That's a very poor way of implementing the DC blocker, especially in an active RIAA equalised phono stage.

Minimal rearrangement places this capacitor inside the feedback loop, where it should be, together with the output coupling one.

Seeing that your resistor setting the 3180uS time constant will likely around 360kOhm (with 360R in the lower feedback arm) you might even find just a 2.2uF film cap will do fine, or alternatively the 22uF electrolytic cap will do fine.



Not sure about "popular wisdom", the way I heard this wisdom was to use the largest possible value filmcap to bypass electrolytic capacitors.



Did you measure LF distortion? Tants ate really poor there. Sonically, subjective it depends.

Naim hifi gear famously uses (used?) Tants and sounded rather poor with them. Replace them and the typical naim "Hifi" turns to "natural".

What I mean with "Hifi sound" are leading edge emphasis, excessively "bright" sound. It might have worked semi ok sonically with overly dull 70's rock recordings (and using dull sounding mm pick-ups like Shure V15) it never worked with classical and sounded horrible with 80's recordings and DMM cuts using a good mc pick-up.



That may have been the key factor. Common tants are relatively low voltage, high(ish) voltage types are made differently.



Good choice, DC blockers belong insidf the feedback loop.

Thor

and even more....?

This topic is nominally about speaker design.

JR

 

[thor.zmt]

still making friends?

Well, Abbey was talking utter loblox as we say on planet anagramia.

and even more....?

You mean I should not criticise a design method that is poor for being poor and recommend remedial actions?

This topic is nominally about speaker design.

Yes, it somehow fell into the "parallel capacitors suck/are the greatest thing since sliced bread" (delete as appropriate) hole.

The truth is simple.

In many cases there are possible improvements, but we need to be clear what problem we try to solve with paralleling capacitors and keep an eye out for potential gotchas.

Randomly paralleling very different value capacitors is not useful, neither is insisting that it should never be used.

Analyse the whole resulting circuit including ALL parasitics relevant to the operation frequencies plus some safety margins...

Thor

 

[abbey road d enfer]

Yes, this is the updated crossover the guy in the video, who according to you has no clue what he is doing proposes.

That was not clear at all.

What I posted was the original stock crossover.

OK, I'll have a look at it.

You would know if you ACTUALLY HAD WATCHED THE VIDEO!

This video is so boring I just wanted to get the gist of it. This guy is unable to captivate an auditory. He profers some things that are just embarassing.
There are too many of these pseudo science videos to waste time for them.
EDIT: OK, I admit I based my comment on uncomplete data.
My comment is still valid for the first video, though. Some guys should not expose themselves like they do.

 

[thor.zmt]

This video is so boring I just wanted to get the gist of it. This guy is unable to captivate an auditory. He profers some things that are just embarassing.

Yes, he is not great at making videos.

There are too many of these pseudo science videos to waste time for them.

What makes designing a proper crossover that turns a 2.5 Way speaker into a 3 Way one, with expected improve response smoothness "pseudo science"?

The fact that this guy is not a good public speaker or good at making videos?

Thor

 

[abbey road d enfer]

What makes designing a proper crossover that turns a 2.5 Way speaker into a 3 Way one, with expected improve response smoothness "pseudo science"?

The fact that this guy is not a good public speaker or good at making videos?

Thor

See my previous EDITED post.

 

[ccaudle]

when you have two capacitors in parallel one with high esr low inductance and another with low esr high inductance the resultant phase angle at the junction would be different to that of a single capacitor

This is not super complicated. Grab your favorite free analog simulator and check it out. You can find the relevant parameters in the capacitor datasheets.

it’s an interesting topic as there are many divided schools of thought

As I said before, be suspicious of any advice with no numbers. Sometimes the divided schools of thought are differences between those with experience designing electronics, using simulation, calculation, and test equipment, and people who do not. No need to put much stock in circuit design advice from the latter.

 

[thor.zmt]

As said before, be suspicious of any advice with no numbers.

Let's have some numbers then.

Wire inductance is appx 25nH per inch, or 300nH per foot or 1uH per meter.

In crossovers for speakers resistance is a few ohms, except for resonaces.

Capacitors are in uF up to several 100uF.

So if somehow accumulate 1m of wiring with a 100 uF capacitor in a woofer crossover (with 0.2 Ohm series resistance) we are looking at a resonance frequency of ~16kHz send a Q golf 0.5.

On the woofer crossover.

The case I make is extremely unrealistic, it illustrates there all real resonances will fall well outside the audio range, for realistic passive crossover circuits in speakers.

More realistic, let's parallel 5pcs 22uF capacitors to get 110uF, each around 3" long and 4" total wiring max per capacitor (depends a bit on physical layout). The lowest possible resonance frequency in this circuit is over 80kHz and we are still at the woofer lowpass or midrange highpass.

We will have five discrete resonances around there, above 80kHz and below 1MHz that will overall cause a broad not very high impedance peak.

Can we put this bed now?

Thor

 

[Adam Frandsen]

Let's have some numbers then.

Wire inductance is appx 25nH per inch, or 300nH per foot or 1uH per meter.

In crossovers for speakers resistance is a few ohms, except for resonaces.

Capacitors are in uF up to several 100uF.

So if somehow accumulate 1m of wiring with a 100 uF capacitor in a woofer crossover (with 0.2 Ohm series resistance) we are looking at a resonance frequency of ~16kHz send a Q golf 0.5.

On the woofer crossover.

The case I make is extremely unrealistic, it illustrates there all real resonances will fall well outside the audio range, for realistic passive crossover circuits in speakers.

More realistic, let's parallel 5pcs 22uF capacitors to get 110uF, each around 3" long and 4" total wiring max per capacitor (depends a bit on physical layout). The lowest possible resonance frequency in this circuit is over 80kHz and we are still at the woofer lowpass or midrange highpass.

We will have five discrete resonances around there, above 80kHz and below 1MHz that will overall cause a broad not very high impedance peak.

Can we put this bed now?

Thor

For someone less knowledgeable in dummy terms: If I need to reach a capacitance of 187.5uf, is it better to recalculate the crossover to hit a more standard value, put a 6.8uf in parallel with a 180uf or use three identical 62uf in parallel?

 

[abbey road d enfer]

For someone less knowledgeable in dummy terms: If I need to reach a capacitance of 187.5uf, is it better to recalculate the crossover to hit a more standard value, put a 6.8uf in parallel with a 180uf or use three identical 62uf in parallel?

With two identical or close value in parallels, the resulting combination is extremely close to a capacitor of double value of identical technology.
In the case of two largely different values, like 180 in parallels with 6.8 or 8.2, there is a good chance that the technologies are different. In that case the capacitor with the highest value will dominate, except maybe for ESR/ESL.
Mixing electrolytic and film caps requires some attention to detail. In any case, some analysis must be performed. It can be via calculation, or simulation, or very often guided by the designe's "instinct" (which is actually the result of years of experience and study).
Remember that most capacitors have a tolerance of at least 1% (often much more - 5 or 10% is not uncommon), so the 0.5 in 187.5 will likely be ignored.
And be aware that such a capacitor (or combination thereof) is bound to be quite expensive (a 100uF costs about $50).

Now, for an example of utter BS about x-over capacitors

 

[Adam Frandsen]

With two identical or close value in parallels, the resulting combination is extremely close to a capacitor of double value of identical technology.
In the case of two largely different values, like 180 in parallels with 6.8 or 8.2, there is a good chance that the technologies are different. In that case the capacitor with the highest value will dominate, except maybe for ESR/ESL.
Mixing electrolytic and film caps requires some attention to detail. In any case, some analysis must be performed. It can be via calculation, or simulation, or very often guided by the designe's "instinct" (which is actually the result of years of experience and study).
Remember that most capacitors have a tolerance of at least 1% (often much more - 5 or 10% is not uncommon), so the 0.5 in 187.5 will likely be ignored.
And be aware that such a capacitor (or combination thereof) is bound to be quite expensive (a 100uF costs about $50).

Now, for an example of utter BS about x-over capacitors

Okay, and yes the price is super high, why is that? I looked at V Caps and they are crazy expensive, saw a mundorf ecap 180uf at about 5 euro, but it is only 23 vac

 

[thor.zmt]

For someone less knowledgeable in dummy terms: If I need to reach a capacitance of 187.5uf, is it better to recalculate the crossover to hit a more standard value, put a 6.8uf in parallel with a 180uf or use three identical 62uf in parallel?

Parallel capacitors. Anything that works. Be aware that rolled up capacitors are quite microphonic. I prefer to use metal encased oil filled MKP "Motor Run" capacitors for crossovers. For extra points get NOS PCP Oil filled "toxic waste" ones at maximum discount.

Also, make sure your calculations are corrects. Drivers are not resistors. Especially dome tweeters have strong resonances that can interfere with crossovers. Cone drivers in case of midranges again will have a strong fundamental resonance usually not far from the crossover frequency and all drivers are more or less inductive.

On top, drivers have their own acoustic roll off, unless they remain flat in response at least two octaves either side of the crossover frequency.

As drivers are usually not coincident, there will phasing issues at frequencies with a wavelength longer than ~ 1/4 of the spacing between drivers.

As drivers often end up near reflective barriers to sound (floor, ceiling, walls to sides and behind this usually needs to be taken into account.

Crossover components are not ideal either. Especially the resistance of inductors needs to be accounted for.

All of this means a simple textbook calculated crossovers never work. Never, ever. They cannot, because they start from incorrect assumptions.

With a Pro-Audio system we even in the 80's had active crossovers with steep slopes (8th order LR) and 3rd octave band EQ's to correct problems. Even so, it turned out that random "mix'n'match" systems rarely worked well and systems properly designed with a system controller proved superior.

So again, before you worry what components to buy at the MINIMUM use makers datasheets and a speaker simulator like vituixcad to see if things stand a remote chance of working well.

Once you got your chosen drivers, mechanical arrangement and crossover to work IN THEORY, buy the drivers, build the box and MEASURE your drivers in the box.

Then and ONLY then can you make a first cut crossover to test and measure as well as listen in real life.

There you may measure or perceive problems, because your simulation did not simulate enough off axis cases etc et al, or you optimised one set of parameters that do not matter a lot while compromising others that actually do.

A solid sim is not a reliable good turnout, but without one you are doing what may be described as: "looking for a small black cat in a large dark room"...

Thor

 

[thor.zmt]

saw a mundorf ecap 180uf at about 5 euro, but it is only 23 vac

It is a bipolar electrolytic capacitor with a large tolerance, high ESR and limited linearity.

It is best to stick with high grade industrial film capacitors.

As said above, oil filled Motor Run are best for large values.

Plastic encased WIMA MKP/FKP for the Rest.

FKP is the best choice, but they are large, expensive and very limited values.

MKP / Oil Filled MKP are a very good choice.

But worry about capacitor choices AFTER (long after) you valideated your design in a simulator software and then build and measured actual drivers in the actual article.

Thor

 

[Tubetec]

Ive used 250/400v motor run caps in tube circuits on the HT line , there pretty good value for money compared to others kinds and give excellent performance .
I had wondered about their use in crossovers to replace electrolytics , it might save money compared to expensive 'audio' grade foil caps components and give very similar performance ,

CBB61 ceiling fan type caps also look good , there available in a wide variety of single ,two and three capacitors blocks ,with many different value combinations , use the sections in parralel to get the values you need.

 

[abbey road d enfer]

Okay, and yes the price is super high, why is that? I looked at V Caps and they are crazy expensive, saw a mundorf ecap 180uf at about 5 euro, but it is only 23 vac

I looked at these V Caps, they're expensive because they use expensive materials and production methods. It's not a guarantee of very good performance. For example they have paper-in-oil capacitors, which are often requested by vintage aficionados, because that's what was use in reverred vintage products, but actually, they're not terribly good objectively.
As Thor mentioned, you can't compare an electrolytic against a film.
Film capacitors above 50uF are expensive and bulky; that's why they're often replaced with non-polarized electrolytics.
In a speaker I designed, I needed a RLC circuit using a190uF cap. I had it custom-made by SCR; even ordered in 100's they cost about €15 each.
I couldn't use an electrolytic, because the two factors (Rser and Rloss) that result in power dissipation (heat) are too high in 'lytics (system operating at about 50+Vrms - 400W/8 ohms) .

 

[MicUlli]

If you really want to build a "good" speaker you should go active. Using Hypex stuff is cheaper than passive components and you would be able to tune your design by software..

 

[Adam Frandsen]

If you really want to build a "good" speaker you should go active. Using Hypex stuff is cheaper than passive components and you would be able to tune your design by software..

I know this is one way to get good results, and honestly I thought about it - was gonna make a poweramp of the eigentakt eval kit to pair with the speakers. But I simply like the idea of an all analogue system with no switching and no digital stages except for the DAC. I don't mind getting a UMIK-2 etc. , but I don't want a DSP in my system.

 

[abbey road d enfer]

I know this is one way to get good results, and honestly I thought about it - was gonna make a poweramp of the eigentakt eval kit to pair with the speakers. But I simply like the idea of an all analogue system with no switching and no digital stages except for the DAC. I don't mind getting a UMIK-2 etc. , but I don't want a DSP in my system.

I understand you want to impose that "no digital" constraint on you, but I really think you should consider active crossovers, particularly for the woofer and low-mid, because that's where you need high value capacitors and inductors, and where their parasitics create the most problems.
I've designed several systems with a simple 2-way electronic x-over, where the mid and high used a passive x-over. You get most of the advantages of active x-over and the relatively low cost of passive.
Indeed, non-digital x-overs do not offere the same degree of control of the parameters as DSP solutions, but these hybrid solutions have given excellent results in the past.
Several custom studio monitors used this principle.

 

[Adam Frandsen]

I understand you want to impose that "no digital" constraint on you, but I really think you should consider active crossovers, particularly for the woofer and low-mid, because that's where you need high value capacitors and inductors, and where their parasitics create the most problems.
I've designed several systems with a simple 2-way electronic x-over, where the mid and high used a passive x-over. You get most of the advantages of active x-over and the relatively low cost of passive.
Indeed, non-digital x-overs do not offere the same degree of control of the parameters as DSP solutions, but these hybrid solutions have given excellent results in the past.
Several custom studio monitors used this principle.

I think of it as autotune, you throw whatever at it, fine tune and it sounds ok - but it loses authenticity.