Multi channel attenuator withPGA2311/PGA4311

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
-ewildgoose

Thank you for the tip!

I went to the link you recommended: www.goldpt.com but wasn't sure which one I would need. Perhaps the Mini-L?

How does it work if I wan't balanced in- and outputs?
Which value would be best? The 25K version?
I need 2*4 channel attenuation. Is it possible to link eight sections in a row?
Which resistors should I choose? Is there a great difference between them?
Roughly, how much money would such a solution total?

Lots of questions I know, but I didn't find the answers at the FAQ section.

I would be eternally grateful for any additional help here.

BR
Roland
 
Actually, samples for TI do come from Digikey, unless TI has suddenly opened up a new warehouse at the same address as Digikey's current one :D

There is most likely just a reimbursement program or they have a certain number of each respective IC that is set aside for sample purposes.

Anyways, more on the topic...I have been testing some PGA2310 that I have and they work quite well. The specs are nearly indentical to the PGA2311 (aside from the supply voltage). The microcontroller part isn't really that bad...if you've never done digital logic before, the learning curve for something this simple isn't all too steep. You can incorporate an optical encoder into the mix as well for a nice little setup. If you haven't already seen this page, I got a lot of good information here:

http://www.mhennessy.f9.co.uk/preamp/analogue_research.htm
 
-ble0t

Thank you for the link! I went there and found lot's of interesting stuff, although some of the images didn't work.

I particularly looked for a balanced 2*4 multi-channel attenuation, but couldn't find anything like that.
Perhaps I didn't look in the right place, or those pages were out of order somehow.

My problem is that I need very detailed and step by step instructions, because unfortunately I know absolutely nothing about IC's, programming and micro controllers.

BR
Roland
 
I guess I'm slightly confused...you now do not want to use the PGA2311 and instead use passive attenuation (i.e. use some stepped attenuators)? Good luck finding an 8 channel version of a stepped attenuator...I wouldn't even want to think about doing all the soldering/wiring for one of those :eek:

Anyways, I wouldn't be too frightened about the microcontroller aspect of things. In fact, there is an even easier site for using the PGA chips. Here is the link for that:

http://home.tu-clausthal.de/~tpa/PGA/index.html

He uses the optical encoder I mentioned and he even has a full schematic for it, which is quite easy to follow.

Hope this helps a bit...the best way to learn is to take the plunge and fix it as you go :D
 
I've learned that the PGA4311 would be most suitable for my project but haven't seen anyone use that one the way I would have to, yet. But I will of course look into the German link.

Yeah, I know, I would probably have to buy a few soldering irons, to last me through all the required soldering, if it's going to be a stepped attenuator. :D

And thanks once again for your help.

BR
Roland
 
paradigm said:
[BMy problem is that I need very detailed and step by step instructions, because unfortunately I know absolutely nothing about IC's, programming and micro controllers.[/B]
It's not that hard, really. I programmed the code for my first microcontroller project yesterday in like 1 hour. Of course the things the µC has to do are quite simple (checking 3 pins and if 1 of them is high turn on a relay, if they go low keep the relay turned on for 1 more minute), but then again, if it takes me only 1 hour to learn it (and emulate to see if it's working as it should), it shouldn't take that long (certainly not more than 5 hours) to write some code to control 2 PGA4310's.
 
-ghemink

Somehow, I still would consider the DEQX to be a potential option and even more so if I could find a used one.

I've been thinking about the DEQX after having tested to almost exclude the low-mid JBL2110A by raising the upper hig-bass(JBL2225) to almost meeting the mid JBL2445J/Selenium CD-horn at 600Hz (48dB Butterworth).

I think it would be possible to take out the 2110A without loosing any sonic quality, which btw surprised me. Didn't think the JBL2225 could match the JBL2110A that well.

If I did this, it would mean I could use one DEQX.

But now I'm coming to what puzzles me.
In a passive x-over, the transient/impulse response drops with higher x-over caracteristics, but havent been able to hear any significant sonic degradation even with 4th order filters.

Do basic passive x-crossover caracteristics, not apply when it comes to digital active x-overs?

BR
Roland
 
Roland,

I will drop you an e-mail in a few days.

In a passive x-over, the transient/impulse response drops with higher x-over caracteristics, but havent been able to hear any significant sonic degradation even with 4th order filters.
Do basic passive x-crossover caracteristics, not apply when it comes to digital active x-overs?

That is probably because you implemented the filters correctly. Too often people just take a second-order crossover for example without accounting for the driver's complex impedance. When it sounds bad they blame it on phase shift and what not while the real reason is the 10 dB hump... ;)

Digital crossovers follow the same set of rules as analog crossovers if they are implemented as linear filters. Which, as far as I know is true for all digital crossovers on the market intended for audio.

For other applications you can do all sorts of magic tricks (suck as a perfect brick-wall filter) in the digital domain. Image enhancement is probably one of the best examples.

/Magnus
 
Yes, phase shift is probably audible with fourth order analogue active crossovers, but if you have done them well, then at the listening position the phase distortion probably mostly cancels. Especially if you have corrected for the time delay?

In the analogue passive world, people can't shift drivers forward and backwards in time quite the same way, so stuff like inverting the phase of the tweeter and having a two step baffle and other stuff like that is done to get things "nearly" in phase.

You can get phase perfect crossovers at lower frequencies in the digital domain though. You should be able to get very steep then (I'm playing with something which looks like 100dB/octave at 120Hz and doesn't seem to have any noticable pre-echo in the impulse response. You could go higher and whilst there is some preecho in the impulse response you almost certainly won't be able to hear it.

At 2Khz though, you need to be more careful about pre-echo and less careful about phase (I think).

The DEQX thing will do more advanced crossovers than your Behringer units, but I would not have thought that you *need* that at this stage. However, they will sell them directly to you second hand... Drop them a line and ask about it.

The room correction seems to be a red herring at this stage - I can't see that there is much support for EQ generally (at least much better than your Behringer units).

As for the passive pre-amp, I tried the following:


+ve ---Rs---+--- +ve
............|
............Rv
............|
-ve ---Rs---+--- -ve


Where Rs is a shunt resistor of order 5K, and the Rv is a variable resistor of order 10K.

Flatten that out and you can see it's three resistors in series between the two signal wires and it's a potential divider so that voltage is split across the three resistors proportionally to resistance.

Now I tried this with a cheap log pot from Maplins, I bought quite a few stereo ones and found that you could disassemble them and reassemble them as a large multiway device. However, I found that because it was a *log* pot, then although it would go up linearly if used as designed, it increased in volume too quickly when wired as above, ie only using the variable resistor part (third leg left unconnected).

I haven't tried using a linear pot, and perhaps foolishly went and bought an expensive DACT 8 way device and the remote control for this from Bent audio....

So my new design will probably be to simply wire it up as per a normal variable attentuator with the constant 10K load between +ve and -ve, and the variable bit is wired to the amp. This way the amp will for example see the actual signal -ve supply on the -ve input, but the +ve signal will be attenuated. This means that in some cases you will have the +ve input to the amp being significantly negative, and at a lower potential than the ground....

However, as I understand it, this is not a problem for most amplifiers, they only care about the voltage differential on their differential input and don't care whether the input is actually symmetric or has an effective DC offset when you consider both channels together. I do need to think that through some more though before anyone wires anything up based on this idea and smokes their amp... (Comments welcome)

Anyway, some ideas there to try. Otherwise look at building some of these PGA boards. There is a german company which supplies kits for these that have the remote control and multiple stereo boards. You pretty much just need to wire them up and off you go...
 
Hi Magnus!

Great to hear from you! I'll be waiting for your e-mail. :)

-Devil_H@ck

I think I'm capable of compreheding many different things, but I strongly feel that programming and putting together the schematics of a balanced multi-channel preamp, unfortunately isn't one of them. ;)

I can do the mounting and soldering, no problems, but understanding what goes where and why in this area, is way out of my league.

BR
Roland
 
I've learned that the PGA4311 would be most suitable for my project but haven't seen anyone use that one the way I would have to, yet.

The "german link" is my page, so i may drop in here.

The PGA4311 is a SMD as i konw, and at least for me its easier to handel 2 DIPs than one SMD.

Using more channels in daisychain would be minor changes to the code provided on my page. In case you have questions or are interested in cooperation, feel free to email me.

you now do not want to use the PGA2311 and instead use passive attenuation (i.e. use some stepped attenuators)? Good luck finding an 8 channel version of a stepped attenuator..

In this case a relay volume control would be a solution, also to find on my www.
 
paradigm said:
-ghemink

Somehow, I still would consider the DEQX to be a potential option and even more so if I could find a used one.

I've been thinking about the DEQX after having tested to almost exclude the low-mid JBL2110A by raising the upper hig-bass(JBL2225) to almost meeting the mid JBL2445J/Selenium CD-horn at 600Hz (48dB Butterworth).

I think it would be possible to take out the 2110A without loosing any sonic quality, which btw surprised me. Didn't think the JBL2225 could match the JBL2110A that well.

If I did this, it would mean I could use one DEQX.

But now I'm coming to what puzzles me.
In a passive x-over, the transient/impulse response drops with higher x-over caracteristics, but havent been able to hear any significant sonic degradation even with 4th order filters.

Do basic passive x-crossover caracteristics, not apply when it comes to digital active x-overs?

BR
Roland


Hi Roland,

This is the good part of the DEQX unit. It corrects both the frequency and phase response of your drivers, it corrects the delays between drivers and it uses phase linear crossovers. So absolutely no phase shift. So you get perfect impulse response. Since I want to avoid issues with pre-ringing as much as possible, I most of the time use only 48dB/Oct filters, this is the lowest slope that is currently allowed. You can get slopes upto 300dB/Oct if you want, but I don`t want to use those steep slopes, and don`T really see why we need them.

As far as I understand most other digital crossovers, and I think also your behringer all use non phase linear crossovers with a phase response that is similar to their analog equivalent.

Best regards

Gertjan
 
Hallo till!

At this point I would be open to any suggestions on which way to go and I would love to cooperate with you on this, but I'm not sure if there is all that much I can add to the design.

However I do pretty much know what it is, that many others and I want for christmas. ;)

With the fast rising numbers of class-D modules out there I expect the problem I have, will be shared by many others in the near future.
I think that it would simply be great if it were possible to keep the Audio signal digital and balanced, for as long as possible through the sound-system and with these affordable and easy to build class-D amps, I think it's inevitable, that the number of bi-amped, tri-amped, quadra-amped and even higher numbers, will go up, thus increasing the demand for balanced multi-channel preamp/attenuation devices/kits.

I have btw visited your site several times and I have it bookmarked in my DIY folder, but I never saw a balanced 6 - 8 channel solution there, so I had to keep on looking. But I would of course be very happy if I could be of any help in this.

Sehr Froh sogar!!!

Auf wiederhöhren!
Roland


-ghemink

Yes, 300dB slope seems very steep. I don't know if that would make it possible for me to go even lower on the 2445JSelenium horn (which is an exponential horn), than the recommended 800Hz 12dB/oct, as I already can go as low as 500Hz with 48dB/Butterworth, without any coloration or other problems.
The Selenium 14-50 horn works down to 500Hz although there is a drop, down 10dB, the last oct. But that isn't that much of a problem with the 2445J though.

About the UltraDrive, it does compensate for phase/time alignment, with a Microphone (ECM8000).

BR
Roland
 
Sorry if this is a side-track on your selected path but I will be going the route below most likely as posted in a different thread (and your HF thread):

I quote myself:

"Go to:

http://www.siliconchip.com.au/

Search for volume control an select the 6-channel volume control project. Add more boards for as many channels you need. Seems I'm going to try this with 8 balanced channels...

Should you want to purchase a kit for it then http://www.altronics.com.au/ sells a kit at AUD 200 complete with case and stuf it seems."

I will not use the display as it seems right now. The .HEX can be downloaded for free from the Silicon Chip site. I will most likely try to modify it to run LM1972 instead of LM1973 but I will initially try to run LM1973 and was considering using one LM1973 per channel giving me 2 chip channel for a balanced and using the last channel in each chip for an unbalanced connection. This would with 10 chips give me 4-5 stereo balanced volume controls and a separate 4-5 stereo unbalanced controls for maximum of 8-10 stereo controls. Should be more than enough. All with remote control. Since I am not used to PIC programming I was also considering using perhaps one controller/receiver per channel plus a main one which with some simple OR logic between the controllers and the volume control board would allow me to set volume individually on each channel.

BTW: One should obviously not put too much trust in a single person's verdict of you don't know him but Thorsten Loesch (nick: Kuei Yang Wang) seems to have tested most/many of the potentiometer/VC chips out there and seems to prefer the sound of the LMs. They also have the distinct advantage of having no intregrated OP so that you may choose whatever you like e.g. OPA627 for midrange and OPA134 for others. Quite attractive IMHO.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.