Line level crossover

Design and use of the various types.
prazza1
Posts: 38
Joined: Tue Nov 29, 2005 12:17 pm
Location: France

Line level crossover

Post by prazza1 »

I have a project of line level crossover for a multiway horn speaker. After having played with triode, I'd like to try transistor (SS ! why not ?) but with valve philosophy.

This filter would be:

First input buffer NPN Emitter follower. The emitter load is a transformer with a gain of +3db to +8db (to be adjusted) or two secondary with different gains (why not?).
The secondary is (are) followed by the passive filters, one for each way: low, medium, ... high.
Each filter is followed by its own buffer NPN Emitter follower loaded again by an emitter transformer, with a gain of 1:1.
So no active gain, only by transformer.
And this for left and right, of course.

My questions are about ... emitter follower transformers.
This transformer would have an idle DC current between 30 to 100 mA (to be adjusted) through it but with a low voltage: 6 to 15 V (to be adjusted too).
I'd like a primary Rdc >= 200 ohms for an idle voltage of around 10V = 200 x 50 mA. So Vb of transistor would be around 11V.
Z Primary 15 to 50 ohms.
Z secondary depending of ratio (of course).
Nickel?

Maybe a second primary for current cancellation, with current sourcing through it.

My big problem is the RDC for idle voltage. Any other solution? Maybe the RDC can be lower, so the idle voltage lower too, but the transistor emitter can go under 0V with signal because loaded by an inductance (primary of transformer): right?
What about a transformer like this?
Any thought?

Thank you.
Attachments
Buffer.JPG
Buffer.JPG (62.37 KiB) Viewed 38092 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

I If I read you correctly, you will have a 50 ohm source impedance driving a primary of a step up transformer that has a DCR of around 200 ohms an is running between 30 and 100ma.

The step up ratios you suggest seem a little low to me. I suspect for that small amount of gain you will be introducing a fair bit of compromise.

am I understanding you this far?

dave
prazza1
Posts: 38
Joined: Tue Nov 29, 2005 12:17 pm
Location: France

Post by prazza1 »

Dave,

We think always things are easy!
So: if we have 100K before transistor, with a gain of 2000, we can expect a transistor Z output of around 100K/2000 = 50 ohms. It seams to be the impedance driving the primary of the step up transformer. This is the nice property of the emitter follower: very very low impedance.
The problem is polarisation. For me, if I have a DCR of 200 ohms, with a current between 30 to 50, I can have a V Emitter between 6 to 10 Volts, so quite range for a small signal whithout clippling in follower mode.
The step up ratios you suggest seem a little low to me. I suspect for that small amount of gain you will be introducing a fair bit of compromise
Why?
About the step up: I need small gain maybe +6 to +9 dBV. So what problem ? I have made a think a little like interstage transformer between two DHT, where 1:1 is desirable. Wrong ?

I hope my explanations are clear (and my english too :( ).
Thanks.

Pierre
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Hello Pierre,

Your english is perfect, my understanding of circuits with sane is what is lacking :-).

The need of from 3-9dB of gain may add a fair bit of compromise to the design since small ratio transformers (other than 1:1) are typically more difficult to get wide bandwidth from. A 1:1 in this situation would easily have 200Khz bandwidth.

It sounds like an interesting project, If you want to proceed I'll crunch some numbers to see where we get.

dave
prazza1
Posts: 38
Joined: Tue Nov 29, 2005 12:17 pm
Location: France

Post by prazza1 »

Dave,
It sounds like an interesting project
Thank you. To be honest, inspiration come from http://www.royaldevice.com/BLU%20EYES%20PRE.htm site.

For 1:1 ok. Let's go.

Suggestions:

What about idea of second primary for current canceling with current source.
Do you think it would be interesting to play with bandwidth for each individual output buffer (remember, buffer + OT + 4 to 5 filters in parallel + buffers with OT by way), each dedicated for different small band less or equal to 3 octaves?

Thank you,

Pierre
Romy The Cat
Posts: 22
Joined: Fri Feb 13, 2009 1:52 am

Re: Line level crossover

Post by Romy The Cat »

prazza1 wrote:I have a project of line level crossover for a multiway horn speaker. After having played with triode, I'd like to try transistor (SS ! why not ?) but with valve philosophy.
prazza1,

The level crossover “passive” crossover is certainly a way to go for multiway horn but your desire to go for SS after using triodes made me wonder about your sense of actions. Is it something that you want because it is doable as most of unfortunate DIY people do or because are there any well-defined demons for sound that you were not able to fulfill with triodes?

If the second then be advised that in some cases some compression drivers do have some inclinations to amplification type and for reactance to minimum current. It is all depends from a few design characteristics of the given driver but mostly depends of type magnet was used. Ceramic and partially electromagnets (depends of the core type) mangier do what to be driven by SS and tend to sound better with more current on voice coil. Alnico and both type of rare earth magnets feel better with low currents, tubes and lower power. All Alnico are not the same, there are 7-8 type and they are all different. Also, the effect ware from the specific frequency range you are looking at and the material of driver’s magnetic return path and the center poll-peace.

Anyhow, the point I am trying to make is that triode might not be beatable and will not force to going SS … if it drives a favorable driver for a specific region of spectra. So, contemplating tube vs. SS for your given channel do not lose a perspective what kind driver you use in your given channel. BTW, each type of cone, suspension and magnet in your driver will allow a different plate loading abuse, so always start from a driver’s needs, at least it is what I do.

Rgs, Romy The Cat
prazza1
Posts: 38
Joined: Tue Nov 29, 2005 12:17 pm
Location: France

Post by prazza1 »

Romy,

I hate you! If I'm here asking Dave, it's because, one day, I have found your internet site. And here it was the beginning of a lot of bad nights, without sleeping. Internet is a wonderful tool but it has something of pernicious.
I come from commercial devices. But two years ago, I have changed for a single driver with big back load: it was very different of all commercial speakers, something that these last don't have. And I was fallen into love. It was droved by a 300B. But there was mismatch between output transformer amp and driver (missing 16 ohms output). So I have changed for an SS amp (from royaldevice.com) for two reasons: first the price (sometimes we have to count money!) and second for philosophy. Minimum of transistors with interstage and output transformers, into an original scheme. So I have tried and I have liked it. Since, I have sold my 300B and I have made an happy guy :lol: .
But! But. I have read, then printed some ideas from your internet site. And read, and read again, about drivers, horn, music... I know you don't like single driver. And it is for this reason I want to try multiway, but by changing slowly (we all need time to do our passion). Limiting extension of the driver (~ 200 – 1K ?) by adding upper bass, midrange, then …. A lot of tries in prospect.
I wanted to ask you some advice but I hesitated cause you have your experience and your point of view. I'm sure very different then my very small experience compared to yours :oops: .
And, you have added your advice here. So, thank you, Romy. I take note. And maybe in the future I will come ask you other good advices. And congratulation for all your work. If I could hear it one day….

Hey Dave, sorry to take up this space.

Pierre.
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

prazza1 wrote:
For 1:1 ok. Let's go.
I didn't try to talk you out of the idea of a step up, I just wanted to point out that 6dB of gain through 20dB of attenuation later on is an odd concept to me. Given your low source impedance and limited bandwidth requirements, a step up might be the best choice of compromise.
What about idea of second primary for current canceling with current source.
I love the concept, but it seems over the top for this application.
Do you think it would be interesting to play with bandwidth for each individual output buffer (remember, buffer + OT + 4 to 5 filters in parallel + buffers with OT by way), each dedicated for different small band less or equal to 3 octaves?
wow... not an easy question to answer. 3 octaves is easy and adding a decade on either side would be my goal. As you know, multi-way systems become very complex quickly. If you have a good blueprint to where you want to go, I can help but please understand that this will be a process and not an instant solution.

dave
prazza1
Posts: 38
Joined: Tue Nov 29, 2005 12:17 pm
Location: France

Post by prazza1 »

Dave,
I just wanted to point out that 6dB of gain through 20dB of attenuation later on is an odd concept to me. Given your low source impedance and limited bandwidth requirements, a step up might be the best choice of compromise
Yes, I agree. At the beginning, we can try one output buffer with 1:1, for a low Z output filter and wide band. We will see later for the input with or whithout gain.
I love the concept, but it seems over the top for this application.
Yes, but can't we try.
If you have a good blueprint to where you want to go, I can help but please understand that this will be a process and not an instant solution.
Ok, we forget at this step of project.

Dave, please let me say how you see what will happen next?


Pierre
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

I would guess the next step would be to map it all out with a schematic including the filters. We could build it in spice to see the function if you want. I'm not sure what solid state models exist, but i'm sure we can figure it out.

dave
prazza1
Posts: 38
Joined: Tue Nov 29, 2005 12:17 pm
Location: France

Post by prazza1 »

Dave,

I will use MPSA13 (sound good, high gain so little output). Maybe you can fill spice with its data.

We can make pattern like NPN + OT with gain + filter + NPN + OT 1:1.


Pierre.
Attachments
MPSA13_D.pdf
(104.63 KiB) Downloaded 539 times
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Spice model below! Had to change *.lib to *.txt .
Attachments
MPSA14.txt
(1.53 KiB) Downloaded 649 times
Brgds
Lars
prazza1
Posts: 38
Joined: Tue Nov 29, 2005 12:17 pm
Location: France

Post by prazza1 »

Thank you Lars.

There is small difference between 13 and 14, apart hfe. It's just for simulation, but never heard the 14.

Rgs, Pierre
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Hello Prazza,

The difference is only hfe! MPSA14 has double as high as MPSA13.
Brgds
Lars
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

I'll look at this a bit later,

just wanted to mention that you should be able to upload .lib files now.

dave
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Hey Lars,

Thanks for the model, now how exactly do you use it :-) Did I mention all things sand confuse me just about as much as all things software. I placed LIB file in my SUB folder but cannot seem to access it. I'll keep trying (hint a simple buffer .asc as per pierre's original sketch would be a help) I'm not asking for a handout here, I have spent about 30 minutes trying to figure it out and given an other hour could probably figure it out, but that hour would be better spent looking at a 300hy 10:1 tapped choke design :-)

dave
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Using trial and error I make the sand work by doing like this:
Go into lib/cmp/standard.bjt and add the lines below.

.MODEL MPSA14 NPN(IS=1.34E-14 BF=340 NF=1
VAF=136.7 IKF=0.38 ISE=7.84E-14 NE=1.5 BR=0.657 NR=1 VAR=92 IKR=1.87 ISC=9.0E-13 NC=2.0 RB=86.610 RE=0.08 NK=0.9 RE=0.58 RC=0.25 EG=1.180 FC=0.5 CJE=1.19288E-11 VJE=1.12097 MJE=0.301248 CJC=1.25659E-11 VJC=0.70336
MJC=0.325457 XCJC=0.9 TF=1.27E-9 XTB=2.12 XTI=3)
Attachments
prazza.asc
(1.41 KiB) Downloaded 592 times
Brgds
Lars
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Thanks Lars.... sort of :-) (see pics below for the problem)

Pierre, is this what you were thinking?

LT spice is free and can be downloaded here
Attachments
prazza xover.asc
the file
(4.29 KiB) Downloaded 503 times
WTF is up with the LF behavior?
WTF is up with the LF behavior?
Picture 24.png (27.05 KiB) Viewed 37867 times
Input to the buffers looks perfect.
Input to the buffers looks perfect.
Picture 23.png (28.27 KiB) Viewed 37867 times
I went for a 1st order at 1K to start.
I went for a 1st order at 1K to start.
Picture 22.png (31.86 KiB) Viewed 37867 times
standard.bjt.zip
unzip and replace "standard.bjt" in Program files/LTC/SwCADIII/lib/cmp/

I had to zip it to get past the extension.
(2.99 KiB) Downloaded 879 times
Last edited by dave slagle on Thu Feb 19, 2009 6:57 pm, edited 2 times in total.
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Trial and error!
Brgds
Lars
prazza1
Posts: 38
Joined: Tue Nov 29, 2005 12:17 pm
Location: France

Post by prazza1 »

Dave,
Pierre, is this what you were thinking?
Yes, nice thank you.
WTF is up with the LF behavior?
Sorry but what does mean WTF? :oops:

About hf_in and lf_in, it's ok. So the input buffer + filter looks perfect like you say.
However, at the output what do you think about lf_out. What does explain the rollof under 100Hz with same buffer than input?

Rgs, Pierre
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

WTF means What the F***. I do not know why the Lowpas is behaving in that manner.

maybe lars has an idea??

dave
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

You forgot to take DC in consideration. Still after my rework I don´t feel so good about this. What about 1mA through the filter choke?

I would go for tubes or diamondbuffers. See my capfree tube-attachment.
Attachments
prazzaxoverlarstube.asc
(4.03 KiB) Downloaded 458 times
prazzaxoverlars.asc
(3.7 KiB) Downloaded 486 times
Brgds
Lars
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Thanks Lars!

The DC current shouldn't be a problem. I have done a similar design for an inductive only crossover. Mine is adjustable from 125 to 500hz and has 1dB steps for level matching.

dave
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Hey Dave,

What do you think of using chokes for HF instead of caps?
Attachments
prazzaxovernocaps.asc
(3.69 KiB) Downloaded 464 times
pnocaps.PNG
pnocaps.PNG (62.47 KiB) Viewed 37807 times
Brgds
Lars
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

I like the idea of using the series 10K to set the source Z and then a shunt choke. I have even considered using a series resistance and a shunt cap for the low pass :-)

dave
prazza1
Posts: 38
Joined: Tue Nov 29, 2005 12:17 pm
Location: France

Post by prazza1 »

If we consider a pass band (lo and high) for each way like said at the beginning of the multiway project, we have a capacitor for each, so no DC into choke and/or transformer, isn'it? So maybe is more easy with no DC and then no effect between 2 transitors for polarization?

Sorry I don't know how to include picture like you do, but I have tried a little to draw with Spice. Please look joined file.
I will try to help you but how to simulate frequency respons from 10 to 100k like you do?

Dave and Lars thanks you!

Rgs, Pierre[/img]
Attachments
prazza_passband.asc
(4.59 KiB) Downloaded 463 times
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Hi Pierre,
In my eyes caps should be avoided as much as possible. Because of that chokes and transformers are used. If it is OK with caps we can as well take away the transformers :wink: .
Brgds
Lars
prazza1
Posts: 38
Joined: Tue Nov 29, 2005 12:17 pm
Location: France

Post by prazza1 »

Yes I agree with you.

But be careful, you have to make output buffers polarization through chokes with DC and by the first buffer like you have drawn. But for Dave it's not a problem. So I like your idea.

I will learn a little more with Spice to help you :idea: .

Rgs, Pierre
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Hey Pierre and Dave,
Why don´t go hybrid to get rid of the input cap?
Attachments
prazzaxovernocapshybrid.asc
(3.97 KiB) Downloaded 457 times
Brgds
Lars
prazza1
Posts: 38
Joined: Tue Nov 29, 2005 12:17 pm
Location: France

Post by prazza1 »

Dave, Lars,

I have learn a little more with Spice.
I have made simulation with little modifications: 10k potentiometer at the input (simulated with 1k/9k), input transformer with gain, choke like buffer emitter load. For filters I have introduced capacitor. It seams difficult to make differently. It is hard to adjust in simulation mode so in the real world it would be ten more difficult. But you have certainly more experience then me, so maybe you have better solution :idea: .

Rgs, Pierre.
Attachments
Xover.JPG
Xover.JPG (26.61 KiB) Viewed 37654 times
Linelevel Xover 03.asc
(5.45 KiB) Downloaded 483 times
Post Reply