Redesign 544: „Rev/One 2018“

Discussies over ontwikkelingen van MfbLabs
Plaats reactie
Peter B.
MfbLabs Gebruiker
Berichten: 68
Lid geworden op: ma 22 sep, 2014 15:38 pm

Re: Redesign 544: „Rev/One 2018“

Bericht door Peter B. »

Hello everyone again!

It turned out that the original RH544 transformer could not be used for the RevOne 544 2018 project. The mounting location is not sufficient for the installation of a standard toroidal transformer. It was necessary to develop a special, unusual model. A friendly company has agreed to take the job. This company is still evolving but its ambitions are to produce high quality transformers. They use extremely special transformer sheet metal from Krupp / Germany. They manufacture the toroidal cores themselves and perform a very precise heat treatment. Due to its technology, Novatech is able to achieve such very low no-load currents.

A few weeks ago I received the first four ordered toroidal transformers. They must be assembled with the help of a rectangular panel of insulating material. In the coming weeks I will order these panels for assembly. I have seen similar images in the Piratelogic, which were also used for the transformer installation. The price is not fixed yet but will not be high.

Here I present a data sheet of the transformers, as well as some pictures with them.

Regards,
Peter
RINGKERNTRANSFORMATOR.pdf
[img]
DSCN7693.JPG
[/img]
[img]
DSCN7692.JPG
[/img]
DSCN7691.JPG
DSCN7688.JPG
Je hebt niet voldoende permissies om de bijlagen van dit bericht te bekijken.
Hans
MfbLabs Gebruiker
Berichten: 1581
Lid geworden op: do 19 jun, 2014 14:25 pm
Locatie: Ach(t)

Re: Redesign 544: „Rev/One 2018“

Bericht door Hans »

Peter B. schreef:It turned out that the original RH544 transformer could not be used for the RevOne 544 2018 project. The mounting location is not sufficient for the installation of a standard toroidal transformer.
With a few different design choices that would not be a problem, see the 544Mk2 documentation on www.mfblabs.nl/544 where a standard toroid transformer is used.
Groeten,
Hans
Hans
MfbLabs Gebruiker
Berichten: 1581
Lid geworden op: do 19 jun, 2014 14:25 pm
Locatie: Ach(t)

Re: Redesign 544: „Rev/One 2018“

Bericht door Hans »

Peter B. schreef:....Subsequently, at the location of the inverting operational amplifier, a level shifter has come.
What is the purpose of the level shifter?
You still need a capacitor or a DC servo to block the remaining and variable DC component. Like a capacitor the DC servo behaves as a first order high-pass filter, the only advantage of a DC servo is that you can use a smaller capacitor for the same corner frequency of the LPF.
Peter B. schreef:R 6 verursacht eine Stufe –Verzehrung in dem Signal (bei +/-2,00V vom Generator und im Knoten R 5 - R 6 )
Why do you need R6? The disadvantage of R6 is that it injects noise from the +12V supply into the signal path. If you like to have a little DC shift at the output of U1, you could set the +input of U1 at a 1.7V lower DC level. But even if the +input is at V+/2, the opamp will not clip, the FET will clip first.

In the simulation that gives the cross-over like distortion, you start the AC signal at the top of the sinewave, this gives a DC offset at the start of the simulation. The input signal must start at a zero crossing.
Peter B. schreef:In the simulations, instead of the complex piezo circuit diagram, I only use one capacitor as a replacement. It may be that I had seen somewhere in their publications how you had specified the piezo capacitance of the Philips sensor plate by about 1.5nF.
The piezo in all 532, 541, 544 and 545 woofers has a capacity of 1.5nF, I think +-10%
Peter B. schreef:My request to you is whether, with the mention of her name, I could also use circuit no. 3 in "RevOne 544"?
No problem.
Groeten,
Hans
Peter B.
MfbLabs Gebruiker
Berichten: 68
Lid geworden op: ma 22 sep, 2014 15:38 pm

Re: Redesign 544: „Rev/One 2018“

Bericht door Peter B. »

Hallo Hans,
Sorry for my late reply. The translation from German to English is still a problem for me. And an insatiable urge for exactness transforms me into a thinker. This hampers me and is very annoying.
(DE: Entschuldigung für meiner verspäteten Antwort. Die Übersetzung aus Deutsch in Englisch ist immer noch ein Problem für mich. Und ein unstillbaren Drang nach Exaktheit verwandelt mich in einem Bedenkenträger :roll:. Das behindert mich und ist sehr ärgerlich.)

1.
With a few different design choices that would not be a problem, see the 544Mk2 documentation on http://www.mfblabs.nl/544 where a standard toroid transformer is used
The new transformer has two secondary windings and an integrated shield.
I have already seen your 544MK2 documentation after it has been released. I had not written here yet. Now I take the opportunity to stress that I find your work very professional and thoughtful. I like using individual modules myself so that different versions can be implemented quickly as needed.

2.
What is the purpose of the level shifter?
In the MFB processor part https://mfblabs.nl/forum/download/file.php?id=3803, I would like to have no isolation capacitors. Thus, the phase shift should be lower in the frequency range of the bass channel. In the last version of the partial diagram here I introduced a servo circuit. And I do not want this servo "fighting" with a big DC offset. For the other blocks of the processor, the offset is very small.

3.
Why do you need R6? The disadvantage of R6 is that it injects noise from the +12V supply into the signal path.
R6 was left that way by a simulation with NE5534. Then I put a TL072 and so R6 stayed there unchanged. If I decide to use NE5532, then I will also use R6 as recommended here:
Met een extra weerstand van 47k tussen 12V en IC300A.2 neemt ook bij een NE5532 de uitstuurbaarheid van het opampcircuit toe tot een piezo spanning van iets boven 1,5Veff.
De J.C. waarde wordt, behalve bij het opamp zonder de extra 47k, bepaald door de instelling (Vgs) van de FET en de spreiding daarop.
(https://mfblabs.nl/forum/viewtopic.php? ... 8&start=20)
Because of a better signal-to-noise ratio I would take an active smoothing filter. So the desired voltage + 12V is gained from + 15V.

4.
If you like to have a little DC shift at the output of U1, you could set the +input of U1 at a 1.7V lower DC level. But even if the +input is at V+/2, the opamp will not clip, the FET will clip first.
I did simulations with different voltages on + input U1. Finally, the circuit was tested with + 4.3V. The voltages I chose were in the range of + 4.3V ... + 6.0V. The voltage settings are derived from a deer of several consecutive resistors.
SIMULATIONEN für RevOne 544 Teil2.pdf
5.
In the simulation that gives the cross-over like distortion, you start the AC signal at the top of the sinewave, this gives a DC offset at the start of the simulation. The input signal must start at a zero crossing.
It turns out that I have a problem with the simulator. Thank You for your comment. I was wrong in the interpretation.

6.
The piezo in all 532, 541, 544 and 545 woofers has a capacity of 1.5nF, I think +-10%
- Thank You, Hans, for your confirmation.
****
I still think on the question if it is better to return to the rounds piezo elements for the MFB sensors? I found a piezo producer here in Bulgaria. He gave me a few pieces with dimensions: 10.3 x 10.3 x 0.35 mm. The question is, how the piezo plate is attached to the circuit board? Maybe two small segments of silicone should be poured, as was the case with Philips at the time. Here are upcoming tests with 8 "and 10" speakers. They are a special production. This makes the placement of the sensors easier. The scheme itself is different from the previous ones.

Regards,
Peter
Je hebt niet voldoende permissies om de bijlagen van dit bericht te bekijken.
Gebruikersavatar
Sandrowski
MfbLabs Gebruiker
Berichten: 1119
Lid geworden op: ma 27 apr, 2015 13:03 pm
Locatie: 't Darp

Re: Redesign 544: „Rev/One 2018“

Bericht door Sandrowski »

Peter B. schreef:I still think on the question if it is better to return to the rounds piezo elements for the MFB sensors? I found a piezo producer here in Bulgaria. He gave me a few pieces with dimensions: 10.3 x 10.3 x 0.35 mm. The question is, how the piezo plate is attached to the circuit board? Maybe two small segments of silicone should be poured, as was the case with Philips at the time.
The round piëzo's are extremely sensitive to base strain pickup and we haven't found any practical mounting method to keep this base strain away from them. This sensitivity causes an error in the measured response. With a few modifications to the round piezo's, Rene seems to have found something workable though. (see here and here).

Still, if you ask me, at the moment the Murata piëzo's are the best bet. The large ones are now obsolete unfortunately, so you will have to use the smaller ones (PKGS-00XP1-R) in parallel or 4 in series-parallel. These sensors are relatively easy to solder and seem to have less sensitivity to base strain (due to their smaller size?)

I don't think it's practical to try cloning the original Philips sensors.
Groet,
Sander
Gebruikersavatar
Rene_N
MfbLabs Gebruiker
Berichten: 1683
Lid geworden op: di 11 nov, 2014 10:01 am
Locatie: Breukelen

Re: Redesign 544: „Rev/One 2018“

Bericht door Rene_N »

Maybe you should work together with Salvador:

https://mfblabs.nl/forum/viewtopic.php? ... 6062#p6062

https://mfblabs.nl/forum/download/file. ... &mode=view

The best way to avoid any base strain pickup is mounting the sensor on a "mushroom". (Not sure how to translate the "Pilz" mounting :? )


Regards

Rene
Gebruikersavatar
Salvador
MfbLabs Gebruiker
Berichten: 937
Lid geworden op: vr 26 jun, 2015 20:35 pm
Locatie: Hilversum
Contacteer:

Re: Redesign 544: „Rev/One 2018“

Bericht door Salvador »

Rene_N schreef:Maybe you should work together with Salvador
It would be very interesting to work together in a project with Peter.

However, duplicate the Philips sensor is not possible for our QTY's. In China they made for me one layer (with on both sides conductive piëzo damped material) piëzo discs and rectangular piëzo's (exactly the shape of the Philips Piëzo). This all costed about 380 USD and resulted into nothing more than an expensive experience (and I did know the fact that it couldn't work well). The original Philips piëzo is a bymorph, which means it exists of two layers and have due to that more sensitivity. The rectangular piëzos I let made in China were single layered and the sensitivity was far away of the original Philips piëzo.
Also I had bad experiences with the company, because of lack of understand what I exactly wanted and they simply say that they can made it too fast without real research. One time I payed and after I payed, they said that the capacitance will be lower than I specified and the factory already made them... real bastards. Company name was CNIRHurricane Tech.(Shenzhen) Co., Ltd. And sometimes I become E-mails if I'm interested to do business again :roll: :lol: They can make a bymorph piëzo from a QTY of 10.000 or more pieces.

See also this and this topic.
Prototype MFB opnemer 28 mm ©Salvador Con.jpg
DSC_7169.JPG
Bymorph piezo 8x8mm total thickness 0%2c60mm%2c 1500pF 30percent.jpg
For me, I don't have any ideas to work again with custom piëzos or piëzo plate discs, the Murata piëzo is in a closed package and they have due to that not the extra distortion of ''air waves'' from outside influences.


One of my ideas is to put as Sander already said earlier, two H1's and one L1, power supply and automatic on/off circuit on one PCB, like the 541MKII PCB. For cooling the LM3886T's, I think a thick piece of Aluminium attached to the LM3886T's and the door would be enough, the construction is open.
Je hebt niet voldoende permissies om de bijlagen van dit bericht te bekijken.
De beste luidspreker, is de luidspreker die voor jou het beste klinkt.
Peter B.
MfbLabs Gebruiker
Berichten: 68
Lid geworden op: ma 22 sep, 2014 15:38 pm

Re: Redesign 544: „Rev/One 2018“

Bericht door Peter B. »

Hello everybody!

I have to apologize first because of my absence here and missing answers to the questions asked.
To make progress with RevOne 2018, I decided to stay in the province until the end of the development work and finish everything there in peace. For four weeks I have been living and working on the project in the small mountain town of Bansko.

One week ago, I received the two versions of the LP4 board. This constructive solution allows the use of normal electrolytic capacitors without being specially selected and without having height problems. The second version of the LP-4 is for standard electrolytic capacitors. Here I present the circuit diagram, the layout view from the CAD program and a few more pictures of the new boards. One of them (LP4a) was provisionally mounted with the electrolytic capacitors from Vishay.
s. The 4 pictures
[img]
LP4%20in%20544.JPG
[/img]
[img]
LP4%20und%20LP4a.JPG
[/img]
[img]
LP4a-Teilmontage%20%20der%20LP.JPG
[/img]
[img]
LP4-BS%20und%20LS.JPG
[/img]
This new circuit board has also changed the block diagram. The new version 2.0 can be seen here: BSBSchematic Prints
[img]
Blockschaltbild%20RevOne%20244%20V2.0.jpg
[/img]
BSBSchematic Prints.pdf
***************************
I would also like to address to you three: Salvador, Rene and Sandrovsky, and answering you in detail and individually. But now I still have no time and some other experiments are still ongoing. That's why I'll be brief:

1. I can tell you the following about the lap piezo discs:
* Installation only in the form of a mushroom.
* The material for the PCB should be made of aluminium only. This is similar to the printed circuit boards for LED lighting.
* The "Dust" cap (made of aluminium) above the loudspeaker coil should also serve as a kind of electrical shielding.

2. Special MFB speakers will have a longer coil. It goes over the attachment with the cone. The sensor electronics are mounted on a narrow, flexible PCB. It is glued as a ring around the coil. A second ring is made of thin copper foil. This ring is mounted over the flexible circuit board and serves as an electrostatic shield. On the top and in the conical membrane of the speaker, the dust cap is attached and glued.

3. I performed simulations for a new circuit with multiple parallel JFET transistors. The sensors are again from Murata - 4 pcs PKGS-00XP1-R. Please see this here: LSK489 http://www.micross.com

4. Multiple Models Piezo elements are manufactured in a factory near Sofia. When I met the people there, they were very busy. Now, with the new Salvador's information about Bymorph-Piëzo, I want to try again to correspond with these people about the sensor issue.

Greetings to all!
Peter
Je hebt niet voldoende permissies om de bijlagen van dit bericht te bekijken.
Gebruikersavatar
Sandrowski
MfbLabs Gebruiker
Berichten: 1119
Lid geworden op: ma 27 apr, 2015 13:03 pm
Locatie: 't Darp

Re: Redesign 544: „Rev/One 2018“

Bericht door Sandrowski »

That is quite nice! I like the mechanical design. Like you say, this way it allows one to use a wider range of capacitors. Also, plenty of space on the main circuit board now.

As for the piezo-based sensors (non Murata that is), I'm still confused as what you're trying to achieve. Aluminium PCB's will be rather expensive, and I would be worried that an aluminium dust cap might cause breakup problems. Also, the 'mushroom mounting method' might help to minimize base strain problems, but it will cause large resonances that will cause problems.

The LSK489 seems to be a nice JFET, i'll remember that one!
Groet,
Sander
Peter B.
MfbLabs Gebruiker
Berichten: 68
Lid geworden op: ma 22 sep, 2014 15:38 pm

Re: Redesign 544: „Rev/One 2018“

Bericht door Peter B. »

Зима 2018-3.jpg
Je hebt niet voldoende permissies om de bijlagen van dit bericht te bekijken.
Plaats reactie