A Buffer ? Why ?

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

In my quest on learning audio electronic basics, here's another question coming to my mind :

What's the purpose of an input buffer ?
The bridgeclone (2 paralled chip) by example is using a buffer and the PA100 circuit in National's AN1192 doesn't use any.

Those two schematics looks very similar... Any advantages in using a buffer ? Is it dangerous when omitting it ?

Thank you in advance for your help !
 
I don't know for the PA100, but most circuits in application notes are built with the non inverting configuration

the gainclone is inverting, and has 2 disadvantages due to this:
low input impedance
the volume pot is inside the feedback loop, and feedback changes when the pot is turned


using a buffer solves these problems
 
Merci :)

Thanks Bricolo,

PA100, as you guessed, isn't inverted.

What about DRV134 ? Is it any good ? I saw several people with OPA627 but ending with little problems...

I was planning a separate power supply for the -15/+15V feeding the DRV134 and another winding feeding my input selector PCB.

Any ideas or comments about those buffers ?

Thanks !
 
Original PA100 schematic from AN1192
 

Attachments

  • pa100.jpg
    pa100.jpg
    21 KB · Views: 1,492
Konnichiwa,

Elkaid said:

What's the purpose of an input buffer ?

To isolate what is behind the buffer from that in front of the buffer and visa versa.

Elkaid said:

The bridgeclone (2 paralled chip) by example is using a buffer and

Actually, according to MY suggestions it does not, it would use a transformer. The Circuit in the original Bridgeclone thread uses a buffer and an inverter to both isolate the input from the low impedance load of the Inverter and the two other Amp's, plus it accomplishes the "phasesplit".

Using a Transformer with some stepdown is a better chocie IMHO....

Elkaid said:

the PA100 circuit in National's AN1192 doesn't use any.

It is a simple parallel circuit, not a bridged circuit, plus it is non-inverting.

Elkaid said:

Those two schematics looks very similar...

Do they? I would say they are about as different as possible while still using the same chip.

Elkaid said:

Any advantages in using a buffer ?

Depends upon the application.

Elkaid said:

Is it dangerous when omitting it ?

Usually not, the worst thing is that the source is loaded too much and starts to distort and sopund bad....

Sayonara
 
Salut,

Using an input transformer is appealing indeed. Is there any suggestion for input transformer cheaper than Jensen ? I'm not sure if the sound difference worth 100$/tx.

I would like to try "standard" input transformer (something like up to 40$) and then if everything's working fine, upgrade to higher quality components.

If anyone has input transformer suggestions, please don't hesitate to share your ideas.

On a side note, just to make sure that I understand a bit the role of a DRV134 buffer by example, would the following works ?

Thank you a lot
 

Attachments

  • buffered parallel gainclone small.jpg
    buffered parallel gainclone small.jpg
    20.5 KB · Views: 1,416
Elkaid said:
Salut,

Using an input transformer is appealing indeed. Is there any suggestion for input transformer cheaper than Jensen ? I'm not sure if the sound difference worth 100$/tx.

I would like to try "standard" input transformer (something like up to 40$) and then if everything's working fine, upgrade to higher quality components.

If anyone has input transformer suggestions, please don't hesitate to share your ideas.

Sowter seems to have some relatively inexpensive input transformers.

And check out this thread started by tiroth.

http://www.diyaudio.com/forums/showthread.php?s=&postid=232854#post232854

He's actually using a Talema power transformer as a line level input transformer with apparently decent results.

se
 
If you want a buffered amp, that is inverted and as well as paralleled, than build such an amp, regardless of if you've seen a schematic of it.

It's just that the DRV wouldn't be a good buffer to use. I use an OPA627 for gainclone buffering services. You could also easily use a BUF chip or even an Elantec EL200x buffer if you can get ahold of any.

The DRV is a good balanced line driver, that's why you may have seen it in bridge/parallel circuits.

You do only want a parallel amp and not a bridge/parallel amp, right?
 
go back to your room, and play with your tube(s).....

There are tubes in all rooms... you have to be more specific

If you consider the inverted topology as comprising of an I/V converter fed from an input resistor which merely converts the input voltage to current in the vitual ground point, you'll see that neither the input resistor, nor the preceding circuitry are part of a feedback loop - they just set the gain.
 
Bricolo said:
Could you stop telling us that we are always wrong, without telling why?
Prove us you know something about electronics, just one time

Ok here it goes,

If you look at the open loop and phase graphs on the datasheets of the chips, it follows that the chips are going towards instability at noise gains lower than 10. At a noise gain of 1 most of these chips are likely to oscillate. So for a stable amp keep it in the gain range above 10.

Now for the inverting GC: The gain is roughly defined by R_fb/R_in. So if you make R_in to large the noise gain drops below 10. Look what happens if there is no low source impedance (when the input is not connected to anything). Well then the noise gain becomes 1 and your GC is likely going to oscillate. Who was telling that R_in is not part of the feedback loop?

Note that an input potentiometer is in series with the input resistance. So depending on the position of the wiper the noise gain of the chip varies. For a stable inverting amp a low impedance buffer (after the input poti) is mandatory to keep noise gain high enough and the chip out of dangerous instability regeons

Cheers ;)
 
oui, en effet !

Thank you guys

You do only want a parallel amp and not a bridge/parallel amp, right?

You're right :nod:
A buffered, inverted paralleled gainclone !

And since I'm only beginning to understand exactly which part of the circuit does what... I'm not very good at designing my own ones...

I'll give it a try during lunch time. Be warned : It may be the most useless schematic you'll ever see in your life.

Thanks and see ya
 
Parallel Inverted Gainclone

Konnichiwa,

Well, is it practical to make a paralleled inverted Gainclone? I suppose so. I can see no reason whatsoever why it should not in a technical sense would "work".

What I would question however is the SENSE of doing so. Paralelling devices (no matter of what type) always causes problems as while these devices SHOULD IN THEORY share the load equally, they very rarely if ever do.

The result is INVARIABLY a drasitic decrease in sound quality. Just build the same discrete amplifier circuit with one pair of output transistors or with a huge bank of them.

Okay, a paralleled Gainclone will be more load tollerant, but if you have such incompetently designed speakers that their impedance falls much below 3 Ohm I would question the speakers and would then also consider if it is possible to break up the parallel connections of drivers and rather give each driver it's own amplifier.

Often speakers have a so-called 2 1/2 Way configuration which places 2 more or less identical drivers in parallel, one working only at LF the other up to crossover frequency. If such a speaker uses fairly low impedance drivers it may indeed drop below 3 Ohm. In this it is usually possible to convert the speaker to bi/tri wiring so that each driver (plus related X-Over parts) get their own binding posts which then allow each driver to be given a non-paralleled "gainclone".

Just for an aside, 5A peak (limit of LM3875) into a 3 Ohm Load allow 40W to applied to the Speaker and around 30W into a nominal 4 Ohm Load. This should be achievable using a 18V+18V secondary transformer. In most cases I would expect this level of power to be fine for "normal use" and 4 Ohm speakers with an average (87db/2.83V/m) sensitivity.

Sayonara
 
Often speakers have a so-called 2 1/2 Way configuration which places 2 more or less identical drivers in parallel, one working only at LF the other up to crossover frequency. If such a speaker uses fairly low impedance drivers it may indeed drop below 3 Ohm. In this it is usually possible to convert the speaker to bi/tri wiring so that each driver (plus related X-Over parts) get their own binding posts which then allow each driver to be given a non-paralleled "gainclone".


a very good advice.....:drink:

in my case i need a lot of watts (atc sl100p) so i need both the bridging and parallelling, but your speakers most likely only need some good stable and lowpowered amps - the biamping thing is the best thing for you, soundwise that is....

elkaid, tell us about your speaker....
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.