Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Analog Line Level Preamplifiers , Passive Pre-amps, Crossovers, etc.

Few questions about a filter/crossover
Few questions about a filter/crossover
Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 16th October 2019, 12:43 AM   #1
Fergutor is offline Fergutor  Argentina
diyAudio Member
 
Join Date: Dec 2014
Default Few questions about a filter/crossover

Hi everyone.
Is this schematic (from a car amplifier but I want to implement this feature in a home preamp), the filter, a crossover? So the HPF signal is what's missing from the LPF one in the exact way as a proper crossover?
7520a-22-inputfiltro.jpg

Looking at this Rod Elliot's crossover it seems to be the same thing...is it? So if it is the same thing I guess I could use it as such running the LPF + HPF signals without problem and with the same performance as the Elliot's crossover right? Or if I wanted the LPF + FLAT signal too without problems?

Another question is that as I will use one or the other as base because both lack features, from the other and in general, could I simply add the feature the other lacks? For example the FLAT of the car amp in Elliot's crossover, or the 11.2K R2 resistor in Elliot's one that controls the Q value in the other one, or what I guess is a "differential balanced output" from the car amp in the Elliot's before the crossover, etc?
What if I add somewhere a baxandall tone control?, in the FLAT signal or before the crossover? Or a sort of parametric eq (some car amps has this just after what the schematic show but only boosting and not cutting).
Thanks
  Reply With Quote
Old 16th October 2019, 06:37 PM   #2
Fergutor is offline Fergutor  Argentina
diyAudio Member
 
Join Date: Dec 2014
Here's the schematic from the Rod Elliot's crossover (project 148):
p148-f1.gif

So is it the same thing? Would it work as a proper crossover if I use the LPF and HPF simultaneously (in the previous schematic)?
  Reply With Quote
Old 16th October 2019, 11:11 PM   #3
Mark Tillotson is offline Mark Tillotson
diyAudio Member
 
Join Date: May 2018
Location: Cambridge UK
These are (2nd order) state variable filters, a common way to get high pass, band pass and low pass simultaneously. Yes, you can use this as a cross-over.
  Reply With Quote
Old 16th October 2019, 11:17 PM   #4
Fergutor is offline Fergutor  Argentina
diyAudio Member
 
Join Date: Dec 2014
Quote:
Originally Posted by Mark Tillotson View Post
These are (2nd order) state variable filters, a common way to get high pass, band pass and low pass simultaneously. Yes, you can use this as a cross-over.
So the sum of both is a rather flat result right? And not an improper mess with a peak or a dip in the middle?
  Reply With Quote
Old 16th October 2019, 11:54 PM   #5
sgrossklass is offline sgrossklass  Germany
diyAudio Member
 
Join Date: Nov 2004
Location: Germany
P148 has an additional inverter on one of the outputs, you'd probably need that.

Be warned that the car audio circuit has its quirks. It seems the resistor values may have been originally chosen for a TL072 or similar and are not at all optimized for a low-noise bipolar opamp like the NJM2068. The flat amp stage right after input level control is screwing up output noise levels, I would change it to noninverting with resistor values at least a decade lower than now (e.g. Rf=4k7 / Rg=820R, and a couple hundred ohms in series + 47k bias on the noninverting input).

So maybe taking the ESP circuit and adding the two preceding stages would be an idea.

Having the balanced input is neat (definitely not a bad idea in car audio), but again 47k resistors are a tad on the high side - I'd suggest 22k or 10k.

If you already have a balanced receiver, you might as well add some actual balanced inputs as well - e.g. XLR. If you have 10k/10k for the RCA, I'd suggest adding XLR via 22k to 39k (i.e. circuit config = 22k...39k / 10k when using XLR, 10k / 10k on RCA). XLR pin 1 goes to chassis.

A Baxandall tone control would definitely go ahead of the actual state variable crossover, e.g. after flat amp.

Last edited by sgrossklass; 16th October 2019 at 11:59 PM.
  Reply With Quote
Old 17th October 2019, 02:16 AM   #6
Fergutor is offline Fergutor  Argentina
diyAudio Member
 
Join Date: Dec 2014
Quote:
Originally Posted by sgrossklass View Post
P148 has an additional inverter on one of the outputs, you'd probably need that.

Be warned that the car audio circuit has its quirks. It seems the resistor values may have been originally chosen for a TL072 or similar and are not at all optimized for a low-noise bipolar opamp like the NJM2068. The flat amp stage right after input level control is screwing up output noise levels, I would change it to noninverting with resistor values at least a decade lower than now (e.g. Rf=4k7 / Rg=820R, and a couple hundred ohms in series + 47k bias on the noninverting input).

So maybe taking the ESP circuit and adding the two preceding stages would be an idea.

Having the balanced input is neat (definitely not a bad idea in car audio), but again 47k resistors are a tad on the high side - I'd suggest 22k or 10k.

If you already have a balanced receiver, you might as well add some actual balanced inputs as well - e.g. XLR. If you have 10k/10k for the RCA, I'd suggest adding XLR via 22k to 39k (i.e. circuit config = 22k...39k / 10k when using XLR, 10k / 10k on RCA). XLR pin 1 goes to chassis.

A Baxandall tone control would definitely go ahead of the actual state variable crossover, e.g. after flat amp.
Thanks for all that!
As I knew there were variations in the schematics I have, even if all of them follow the same style, I started looking at them with what you said in mind and look at this one, it has changes in line with all your advices!:
reference_5350a-19-CH1.jpg

As far as I can see now it's only missing the AUX output wich I guess it can be added easily, right?

But what I'm not sure is where the signal is going in that input, as this other is going to the non-inverting, while the original was in the inverting and they look the same.
PX6002-21-inputfiltro.jpg

I guess is going to the inverting input as there's a capacitor to ground, but while in the other two the cap is a 1nF, in this is a 220uF!!! Why such difference??

Also that FL1 (with RJ45 GND) don't know what that is...

And that transistor at the end...What changes should I make on that part?
(I know I probably should just use the Elliot one with the preceding stages as you say, it's just that as the Elliots doesn't have the 3 outputs, and I need those...)

And about that input, I really don't understand it: is it the one called "differential balanced input"? But does it need a balanced source (I don't have)? My idea is to use it with a smartphone for example or an old CD player, bluetooth dongle, soundcar output (for Hi Res), etc.

Last edited by Fergutor; 17th October 2019 at 02:24 AM. Reason: Some text corrections
  Reply With Quote
Old 18th October 2019, 03:42 PM   #7
sgrossklass is offline sgrossklass  Germany
diyAudio Member
 
Join Date: Nov 2004
Location: Germany
Quote:
Originally Posted by Fergutor View Post
Thanks for all that!
As I knew there were variations in the schematics I have, even if all of them follow the same style, I started looking at them with what you said in mind and look at this one, it has changes in line with all your advices!:
Attachment 788244
Well, you definitely overdid it there. You do know that the LM833 is only an average load driver? Also, beware: TI sells two LM833 variations these days, one the original NatSemi part (called "LM833-N", LM833N in DIP), one their own take on the MC33078 (LM833P in DIP).

R101A is used to limit level adjustment range, why did you change that (unless that is precisely what you set out to do)?
R102A at 2k7 is too low in relation to the output impedance of a 20k pot (~5k max). Or did you want to "fake log" a linear pot?
R103A/R110A @ 1k8/100R is lower than needed - when the input voltage noise of your opamp is about the equivalent of 1 kOhm, you don't have to go as low as 100R externally. And why this much gain anyway, wasn't it 9 or 10 dB less before? That'll bring the noise output of this stage right back up!
R115A/R117A at 1k/1k: No, why? Think of the poor opamps that have to drive this. A 4558 won't like this too much, and the LM833 won't be entirely happy either. And what for anyway? The crossover and gain stage are going to be more noisy for sure. You could use 10k and it wouldn't matter much (noise output ~3 V, vs. gain stage with LM833 at 4k7/820R: >6 V).
R116A: I don't know why that is even there is the first place, it merely increases noise gain. Could be used to tame an opamp at less than minimum stable gain, but I don't see the point here.
Since you need all 3 outputs, you may have to duplicate the output buffer.

Seems like you could use an opamp noise calculcator. Here is the one I use.

You should also start doing some sort of level planning. Particularly, how much output level and how much output noise you are going to need. If you have both a flat amp and an (inverting) output amp, you can pretty much shift levels in the crossover around, so you can find a good compromise between noise and distortion performance. What are you planning to follow this up with? You'll need some level adjustment for your drivers, and you can have a lot of fun with output noise when using high-sensitivity drivers.

Quote:
Originally Posted by Fergutor View Post
As far as I can see now it's only missing the AUX output wich I guess it can be added easily, right?
Indeed.
Quote:
Originally Posted by Fergutor View Post
But what I'm not sure is where the signal is going in that input, as this other is going to the non-inverting, while the original was in the inverting and they look the same.
Attachment 788245
The first circuit had a noninverting input stage and inverting gain stage.
This one has an inverting input stage and noninverting gain stage.

Both are giving the same absolute phase. Since inverting phase on a balanced input is trivial (swap the inputs around), you can do so as needed.
Quote:
Originally Posted by Fergutor View Post
I guess is going to the inverting input as there's a capacitor to ground, but while in the other two the cap is a 1nF, in this is a 220uF!!! Why such difference??

Also that FL1 (with RJ45 GND) don't know what that is...
I have no idea what you are talking about here. Mind marking these components on the schematic?

Quote:
Originally Posted by Fergutor View Post
And that transistor at the end...What changes should I make on that part?
This looks like some sort of attenuator and wouldn't be doing very much with the values shown. Do you even need that functionality?

Quote:
Originally Posted by Fergutor View Post
(I know I probably should just use the Elliot one with the preceding stages as you say, it's just that as the Elliots doesn't have the 3 outputs, and I need those...)
I don't see what the problem is. Don't you effectively have all your 3 outputs available once you've added the flat amp stage?
Quote:
Originally Posted by Fergutor View Post
And about that input, I really don't understand it: is it the one called "differential balanced input"? But does it need a balanced source (I don't have)? My idea is to use it with a smartphone for example or an old CD player, bluetooth dongle, soundcar output (for Hi Res), etc.
Yes, it's a balanced input, and no, it doesn't need a balanced output. It won't deliver perfect CMRR with unbalanced source impedance (which may have been part of why they went with 47k resistors originally), but it'll be a whole lot better than a plain unbalanced input still, plus your sources would generally have low(ish) output impedance anyway. Remember this is a car audio device, and cars have all kinds of nasty ground currents running around, plus everything is fed from the same DC power source and grounded to the same chassis.

One or two of the buses I am riding around here clearly did not have that much thought put into their PA systems, as you'll have plainly audible whining and whistling whenever the system unmutes to make an automatic station announcement. That's what happens when you get a ground loop in a car.

Your home may not be as electrically hostile as a motor vehicle, but let me tell you, it still is really easy to get into trouble with ground loops. A lot of perople will build their DIY power amps as IEC Class I devices, i.e. audio ground ultimately connects to protective earth, often "hard" (directly). Well, guess what your PC will do, just the same. Connect both, and you'll get a first-rate ground loop. Then you'll have to bust out your audio isolation transformers - or have a balanced input.

There are various ways of further improving balanced input performance, BTW:
1. Add unity gain buffers up front in each leg, which allows you to have high input impedance regardless of resistor values in the bal/unbal stage (so you could easily go down to 4k7s or even 3k3s there for lower noise).
2. Most buffered inputs like that would just have 100k bias resistors to ground. A more clever version: 10k + 10k from each leg meeting, then going via 100k to ground. (Similar if you need higher input capacitance, you could have e.g. 1n + 1n and then 100p to ground. An additional 47-100p from each leg to ground right at the input is recommended.) This more than doubles common-mode input impedance (for better CMRR in the face of mismatched source impedance) while keeping differential input impedance reasonably low (for less noise and "antenna effect" with the input open). You can even go higher than 100k for the common resistor, as any voltage appearing over it from opamp input current is purely common-mode and hence suppressed by CMRR.
3. Turning an unbalanced input into a (decent enough) balanced one is quite trivial as long as you know what's behind it: You just need to replicate the output impedance on the "signal" side in the connection to the "ground" side. Lots of home studio equipment actually does it like this - if they have 100 ohms and 22 F in series with the signal output for "+out", they'll add another 100 ohms and 22 F to ground for "-out". Of course this means that the balanced receiver will see a signal that is partially common-mode, and its distortion performance may suffer if its common-mode linearity is not good, but for rejection of ground loop noise this is irrelevant.

Rod Eliott has several articles on balanced connections, these may be a good read.
  Reply With Quote
Old 18th October 2019, 11:32 PM   #8
Fergutor is offline Fergutor  Argentina
diyAudio Member
 
Join Date: Dec 2014
Quote:
Originally Posted by sgrossklass View Post
Well, you definitely overdid it there.
No, it's just another schematic I found that it seemed to follow what you were saying. So none of that is my work and no intentions there either, of course.

But are those that bad? I assumed that as they were car amplifiers, where sound quality isn't precisely appreciated or even needed, they wouldn't be very good...but that bad?!

I only know very very basic stuff. I do my projects mostly by copying from good sources and hoping they work. And up until now all but one of the many things I did worked. How good, not sure, but worked apparently as intended.

I'm going to do this: tomorrow I will make a block diagram of what I want, post it here and you tell me your thoughts (if ou want of course).

Quote:
Originally Posted by sgrossklass View Post
You do know that the LM833 is only an average load driver? Also, beware: TI sells two LM833 variations these days, one the original NatSemi part (called "LM833-N", LM833N in DIP), one their own take on the MC33078 (LM833P in DIP).
I was planning to replace them with NE5532 that are supposedly the same chip, although today I saw there was a difference in the bipolar input transistor...no idea.

Quote:
Originally Posted by sgrossklass View Post
I have no idea what you are talking about here. Mind marking these components on the schematic?
I can't mark those components now but the capacitors I'm talking about are the ones from signal to ground in the input (C101/C201), that capacitor is 1nF in the other schematics and 220uF on this one.
FL1/FR1 is that just under it.


Quote:
Originally Posted by sgrossklass View Post
I don't see what the problem is. Don't you effectively have all your 3 outputs available once you've added the flat amp stage?
The problem is all it lacks. But if it's easier to add to the Elliot's one what it doesn't have than rectify all the problems with the 5350a that is basically complete...Thing is, due to my insufficient kowledge, I'm not sure if I can do it, so I was "hoping" that one of these car amplifiers preamp stages would do...but as I said they seem to be even worse than I expected!

Quote:
Originally Posted by sgrossklass View Post
Your home may not be as electrically hostile as a motor vehicle, but let me tell you, it still is really easy to get into trouble with ground loops.
I "live" next to a radio antenna. I made this other thread asking precisely about this project, and one of the things in the thing I was thinking on copy there () were the common mode chokes because of the antenna problem. And I recently read that this car amp's balanced inputs could be useful as well...but I still don't know...

I will also read again more carefully all you written, I appreciate a lot all the info. Will post tomorrow the block diagram of my plan here (or on the other thread if you prefer).
  Reply With Quote
Old 19th October 2019, 03:38 AM   #9
sgrossklass is offline sgrossklass  Germany
diyAudio Member
 
Join Date: Nov 2004
Location: Germany
Quote:
Originally Posted by Fergutor View Post
I'm going to do this: tomorrow I will make a block diagram of what I want, post it here and you tell me your thoughts (if ou want of course).
Sounds good.
Quote:
Originally Posted by Fergutor View Post
I was planning to replace them with NE5532 that are supposedly the same chip, although today I saw there was a difference in the bipolar input transistor...no idea.
They are definitely not the same chip, they just belong to the same class - the first wave of low noise, low distortion bipolar audio opamps from the 1980s. LM833 and MC33078 were lower-cost alternatives to NE5532. Samuel Groner has measured all 3. TI's NE5532 can actually be worse in low/unity gain noninverting stages than the alternatives due to its comparatively bad common-mode linearity (as evidenced by the high frequency linearity test), but other manufacturers' ones seem to be better in this respect.
Quote:
Originally Posted by Fergutor View Post
I can't mark those components now but the capacitors I'm talking about are the ones from signal to ground in the input (C101/C201), that capacitor is 1nF in the other schematics and 220uF on this one.
FL1/FR1 is that just under it.
Ah. With the 220 cap, no idea what they were thinking there. This usually is just for shunting RF, it makes absolutely no sense to be using a cap that large, as it would eliminate pretty much any benefits of the balanced input stage. (Even so, having the 1n capacitor only on one leg throws off balance at higher frequencies.)

FLI/FRI appear to be just additional / alternative inputs, coming from somewhere via twisted pair cabling if the "RJ45" is any indication.
Quote:
Originally Posted by Fergutor View Post
The problem is all it lacks. But if it's easier to add to the Elliot's one what it doesn't have than rectify all the problems with the 5350a that is basically complete...Thing is, due to my insufficient kowledge, I'm not sure if I can do it, so I was "hoping" that one of these car amplifiers preamp stages would do...but as I said they seem to be even worse than I expected!
I mean, the basic topology isn't bad, I just don't agree with all component values.

BTW, when seeing the schematic that had NJM2058 marked on it, the penny finally dropped. Those are quad 4558s, and they are not substantially better at load driving than TL07x, and not all that low noise either. That explains the highish impedances.
Why then they went with NJM2068s when they needed duals, no idea. Maybe those weren't much more expensive than other alternatives in SIP-8. (It's one of the few interesting parts you can get in that oldschool package, alongside NJM4556A, NJM5532, NJM2114 and maybe the NJM4565 if you need a low-Ib part. Oh, the NJM2114 seems to have gone EOL recently. That was JRC's "super 5532" - never got all that popular it seems, even though it did make an appearance in higher-end soundcards from both Creative/EMU and Asus.)
Quote:
Originally Posted by Fergutor View Post
I "live" next to a radio antenna. I made this other thread asking precisely about this project, and one of the things in the thing I was thinking on copy there () were the common mode chokes because of the antenna problem. And I recently read that this car amp's balanced inputs could be useful as well...but I still don't know...
Balanced inputs tend to be a good idea regardless of whether you are living next to an AM flamethrower or not. Some extra attention is likely to be required in your scenario. A common-mode choke between the 47-100p to ground at the input and the T'd larger caps + bias Rs is likely to be a good idea, as well as a few hundred ohms in series in each leg...

BTW, when you talk about "copying", you are basically talking about coming up with your own layout, are you? As they say, the layout is the circuit, and there still is enough to go wrong there...
  Reply With Quote
Old 19th October 2019, 04:44 PM   #10
Mark Tillotson is offline Mark Tillotson
diyAudio Member
 
Join Date: May 2018
Location: Cambridge UK
Quote:
Originally Posted by Fergutor View Post
So the sum of both is a rather flat result right? And not an improper mess with a peak or a dip in the middle?

The sum of all three (suitably phased/scaled) is exactly the input, that's how the circuit works. Sum high and low and you get a notch filter.
  Reply With Quote

Reply


Few questions about a filter/crossoverHide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Questions on driver selection and crossover filter stappvargen Multi-Way 4 17th December 2007 11:00 PM
active sub filter questions paulspencer Subwoofers 19 22nd October 2005 02:24 PM
active filter from rod elliot and sub filter questions wuffer Parts 0 2nd October 2005 02:12 PM
Active Filter Questions ChevS-10 Car Audio 2 26th June 2005 07:37 PM


New To Site? Need Help?

All times are GMT. The time now is 02:11 AM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 14.29%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Copyright ©1999-2019 diyAudio
Wiki