Active crossover design : buffer and summation

[ppan]

Hello world!

For my mono Bluetooth loudspeaker project, I decided to DIY again an active crossover (my first home made was around 1992!). A Linkwitz-Riley of course!

1. My first question is about buffer.

From Mr Linkwitz himself, here is the way to go:

Here is the Elliott Sound Products point of view::

What is the best design?
Especially if I need to add a pot volume before it.

NB: the source is a LQSC BT module. Amplifier: TDA7293 or LM3875 stereo (one channel per speaker). Independent power supply for crossover, amplifier and BT module.


2. For a different project, the second question is about source summation (subwoofer case) and crosstalk issue.

First example, I saw many time, summation before filter section:

I assume RA & RB are involved in the cutoff frequency formula.
Is it correct?

Second example, summation after a stereo design with 2x10K resistors:

Is it not a better way to avoid crosstalk?

Have a nice week.

Pierre

 

[Boden]

So it is a mono project. But what loudspeakers do you intend to use, so at least we could guesstimate what the intended x/o frequency is. We then could help with the RC values.

And what crosstalk are you afraid of?

 

[tvrgeek]

I run a pair of XKitz which are very clean, but with a single supply and artificial ground, HUGE thump on power on and off. Lesson is to build a solid tracking dual supply.

I might suggest reading the books from Doug Self to understand the little details in execution that matter. I always model in LTSpice.

I suggest independent regulators but no real reason for independent raw supplies, especially for low current stages. Correct placement and selection of caps for the power supply makes independent regulators not needed. Physical layout of the bypass caps matters.

 

[stv]

Is it not a better way to avoid crosstalk

if you build a mono speaker why do you care about crosstalk?
after all summing stereo to mono is maximizing crosstalk to get a sigle signal.

I would first sum the stereo signal to mono (with buffer) insert the (mono) volume pot and then feed the signal to the (mono) crossover.

by the way: the schematics you posted seem to be crossovers for stereo + mono-subwoofer

 

[ppan]

So it is a mono project.

Hi Boden!
In the first case yes. In the second, it's a stereo one with subwoofer.

But what loudspeakers do you intend to use, so at least we could guesstimate what the intended x/o frequency is. We then could help with the RC values.

I make loudspeaker from my first age (so around 40 years) and I don't need help about speakers and cutoff frequency.
My question is about buffer design (see example 1 & 2).

And what crosstalk are you afraid of?

Is crosstalk not a fact when you connect two signals?

(...) HUGE thump on power on and off. Lesson is to build a solid tracking dual supply.

Hi tvrgeek!
When the amplifier do not provide it, I always place a relay cutoff module between speaker and amplifiers: soft start and (at least) immediate shutdown. Except when it's a music or PA sound system.
Also, my trick is to choose bigger caps for the crossover dual power supply (at least 2x470µF) which provide delay: the relay cut the output signal before noise.

if you build a mono speaker why do you care about crosstalk?

Hi stv!
The second question and design is not a mono project

the schematics you posted seem to be crossovers for stereo + mono-subwoofer

That's it! What do you mean "I would first sum the stereo signal to mono (with buffer)"?

 

[stv]

The second question and design is not a mono project

I see! i was confused by

For my mono Bluetooth loudspeaker project

That's it! What do you mean "I would first sum the stereo signal to mono (with buffer)"?

I suggest to buffer the signal before feeding it to the volume pot.

if i undestand your question correctly you would like to know whether it is better for the mono-woofer to

1. first sum the signals and then low-pass it or
2. first low pass both signals and then sum them

to avoid any possible crosstalk you could either include buffers before summing the signal for solution 1. or just go with solution 2.
whatever is cheaper!

edit: eventual crosstalk via summing resistors depends mainly from the source impedance of your signals. so, if you sum the signal e.g. coming directly from a volume pot (without buffer) there may and will be some crosstalk. if the summing network is connected to a pre-amp (source impedance very low) and the summing resistors are high compared to source impedance, there will be no/little crosstalk.

 

[ppan]

I suggest to buffer the signal before feeding it to the volume pot.

OK.

To give you the point, I found this post (it's not LR circuit).

We see 3 stages:
- two 21dB gain buffers (which is Mr Linkwitz buffer option),
- master volume
- summation buffer before lowpass section.

When KMTech choose the second buffer (Elliot Sound voltage follower option) and my solution 2 for summation:

 

[stv]

We see 3 stages:
- two 21dB gain buffers (which is Mr Linkwitz buffer option),
- master volume
- summation buffer before lowpass section.

volume pot and summation without buffer in between will give you some crossfeed. the volume pot circuit is not shown, so maybe there is a buffer after the pot. no problem then.

When KMTech choose the second buffer (Elliot Sound voltage follower option) and my solution 2 for summation:

no crossfeed here!

 

[ppan]

Yes. I will choose the second design for my next sub projects. Not the cheapest, not the smaller, but it's not an issue

What do you think about the two buffer design?

 

[stv]

your summing buffer looks fine to me, i suppose you chose the inverting op-amp on purpose.
if you don't need it to be inverting you may even be ok without the buffer, simply connecting the summing resistors (maybe a bit less resistance, rod elliott chose 5,6 kOhm in his project 73) to the power amp.

 

[ppan]

Here is my stripboard of the 2.1 KMTech LR4 crossover:

And the LochMaster 3.0 file.
I omitted the six 100nF caps directly soldered on the pins 1-5 (or 5-9).

A mono version for BT amplified speaker. 3 pots onboard:

And the LochMaster file.

All IC = NEC C4570HA, SIP version, very flexible (symmetric design) and good performance.

The final built:

Tweeter: Davis Acoustics
Woofer: 2x Focal 7K515 (one filtered with Audax LA 4mH inductance) Fp=40Hz
Amplifier: TDA7293 stereo module
Power supply: SMPS240QR and DIY 2x12V for crossover
Bluetooth module powered by the SMPS 12V aux
Genuine TDA7293 here (tested)

Some other (old) projects here.