Hissing + DC offset of Bridge-Parallel GainClone

Status
Not open for further replies.
Jump to Latest
#22
Does this ssssss occur only when music is present? If yes, I think you gave an oscillation somewhere and by this you get distortion products which are audiable.

Using these extremely fast LM IC's demands good pcb and wiring. Your feedback for instance. Those caps, are a death sin if you are using current feedback amps. Feedback is totally different in mechanism.
 
#23
peranders said:
Using these extremely fast LM IC's demands good pcb and wiring. Your feedback for instance. Those caps, are a death sin if you are using current feedback amps. Feedback is totally different in mechanism.
Click to expand...

LM6172 is voltage feedback. The cap in that position is recommended to guarentee stability. 2pF is recommended but higher values work OK aswell.

As op amps go this is a good one

mike
 
#24
OK, I'm half right. The IC _is_ a current feedback with buffered inverting input which makes it looks like a voltage feedabck.

OK, but 110 kohms seems very high for a high speed device.

PRINTED CIRCUIT BOARDS AND HIGH SPEED OP AMPS
There are many things to consider when designing PC boards for high speed op amps. Without proper caution, it is very easy to have excessive ringing, oscillation and other degraded AC performance in high speed circuits.

Quote from the datasheet.

This IC needs very careful layout. I doubt that you can connect it "hard-wired".

Why don't you rewind and use a normal audio opamp first? Try to isolate the problem.
 
#26
peranders said:

Text from the datasheet. This is something to consider. 110 kohms isn't a very pure resistor at the frequencies the LM6171 works.

May I suggest 100 ohms / 1 kohms or max 220/2k2
Click to expand...

Good point, also both the PSU decoupling caps and the feedback cap should be very close to the chip.

The chip probably won't work properly if mounted in an IC socket. I got hissing just like you when I tried it.

I got success with LM6182 by soldering a decoupling cap right accross the legs of the chip. I used a single stacked film 1uf low inductance straight accross the + and - supply legs

mike
 
#27
mikelm said:
I got success with LM6182 by soldering a decoupling cap right accross the legs of the chip. I used a single stacked film 1uf low inductance straight accross the + and - supply legs
Click to expand...
Notice that similar devices rather often need SMD caps and a good pcb in order to get clean supply voltage. Short leads are essential in order to get high resonance frequency.

An advice to you Mike and everybody else: read carefully the datasheet and especially the application hints. Try also to get application notes if there are any. Even though you won't understand much, some information is possible to pick up.
 
#28
peranders said:
An advice to you Mike and everybody else: read carefully the datasheet and especially the application hints. Try also to get application notes if there are any. Even though you won't understand much, some information is possible to pick up.
Click to expand...

Yes I always try to get the datasheets.

I have LM6172 & 6182 datasheets

But what I am trying to say is this method mentioned above actually does work for LM6182 and I have just remembered that I used it successfully for LM6172 aswell.

1uF stacked film low inductance cap.

The idea came from thorsten.
 
#29
DC Offset and other probs

overmind said:
Each LM3875 opamp is hardwired and the components are placed closely. I have tried placing the by-passing cap as close to the pins as possible. The hiss is still there. I am using Wima 0.22uF 160V MKP10 caps for by passing.
Click to expand...

God bless everyone who hardwire's the LM3875, LM3886 etc. It's just not a good idea. I am now firmly committed to PCB's with wide trace leads (50 mils) for the inputs and SMT resistors.

Secondly, if you are suffering from oscillation, replacing the 0.1 ohm resistor with a choke -- about 10 turns of #18 wire on a 1 watt carbon resistor should give you enough uH.

The method you have of nulling the DC offset -- this is somewhat akin to what folks do in order to use dual supply op-amps as single-supply -- and the resulting problem is that you pick up any power supply noise unless you decouple the potentiometer wiper.

LM6172 ? With these high speed opamps you need wide traces (50 mils) and a ground plane -- prevent any capacitative coupling, all the supply pins have to be decoupled as closely as possible to the supply pins. This wouldn't be my first choice in this application. High speed opamps shouldn't be hardwired, wire-wrapped etc. Any noise you are generating from the voltage regulator is going to merrily propagate its way through the system.
 
#30
Re: DC Offset and other probs

jackinnj said:
LM6172 ? With these high speed opamps you need wide traces (50 mils) and a ground plane -- prevent any capacitative coupling

This is an interesting opinion - but it is simply not true.

all the supply pins have to be decoupled as closely as possible to the supply pins.

that's why it's a good idea to solder the cap directly onto the pins

High speed opamps shouldn't be hardwired, wire-wrapped etc. Any noise you are generating from the voltage regulator is going to merrily propagate its way through the system.

what regulators ? chokes & caps seems to work very nicely for me.

I took the o/p chip from a friends pioneer CD player and popped 6172 it's place. with the decoupling cap on the pins - it work like a dream first time no problems. He said it sounded like a new CD player.
Click to expand...
 
#32
mikelm said:
sounds like oscillation - this could explain the hiss and the offset

are the chips seperately decoupled ?

.1uf - 1uf across the supply pins of each chip is normal
Click to expand...
Note that the LM6171 takes it power from each supply voltage and returns it to ground, therefore I think it's wise to have decuopling from each supply pin down to ground AND the ground must be good.

I doubt though that it's possible at all to hardwire the LM6171! Has anyone out there succeeded in this? I think pcb with groundplane is a must in order to get max performance.

Dr Meier has made an power amp of 44 opamps but his pcb is rather good.

BTW: What has happened to http://www.meier-audio.com. Has his site been hijacked? Please don't sinbin me but this URL used to be Meier-Audio's!
http://home.t-online.de/home/meier-audio/ seems to be working...
 
#33
peranders said:

Note that the LM6171 takes it power from each supply voltage and returns it to ground, therefore I think it's wise to have decuopling from each supply pin down to ground AND the ground must be good.

THIS IS A THEORY - inpractice 1uF stacked film cap works very well

I doubt though that it's possible at all to hardwire the LM6171! Has anyone out there succeeded in this? I think pcb with groundplane is a must in order to get max performance.

I have an LM6182 as the o/p chip of my cd723 hard wired, no pcb at all, glued upside down with it's little legs sticking up in the air. with one 1uF staked film cap for decoupling - it has been working very well for about one year and sounds good

THIS IS NOT A THEORY

however when tried it in an IC socket it hissed just like described above.


...🙂......mike
Click to expand...
 
#34
Re: Re: DC Offset and other probs

mikelm said:
jackinnj said:
LM6172 ? With these high speed opamps you need wide traces (50 mils) and a ground plane -- prevent any capacitative coupling

This is an interesting opinion - but it is simply not true.
Click to expand...

It's Texas Instruments, Linear Tech and Analog Devices who all recommend the above. What may be true in your case may not provide the type of reliability and signal immunity that is called for in a commercial design.

The technique you are using is called "dead bug" -- with the IC's feet wriggling in the air. We use this in VHF, UHF ham applications.
Click to expand...
 
#35
Hi All,

I have tested the LM6172 input buffer individually. It worked fine. I had also used it in my CD player, hardwired + IC socket, I have observed that this opamp needs a decoupling cap between the +/- Voltage supply for stable operation.

The hiss seems to be orginated from Paralleling/Bridging of the LM3875 opamp module. As I have bypassed the LM6172 input buffer from the LM3875 opamps.

Two things I will try,
Determine what is causing the problem? Paralleling or Bridging. The LM6172 will be by passed for this test.

Add some inductive components at the output of the LM3875 opamp module before they are paralleled or bridged.

I will report back.
 
#36
Tested the amp again without the LM6172 input buffer.

Bridging - almost no problem. There is a very slight noise, no problem with that.

Paralleling - hiss. I think this is oscillation. The hiss is there with or without the music.

Just with Bridging, the amp sounded quite good. Cleaner and more control compare with normal GainKlone.


If there is no oscillation, the DC offset voltage at the output is not significant, matching of components pays off.


Now the problem: how do I parallel the LM3875s?

I will try adding a 10ohm resistor + about 10 turns of wire around the resistor. The 10 ohm + turns of wire will be connected in parallel. The combination will be connected in series after the 0.5 ohm output resistor of each LM3875.

Will report after I try it.

Thanks to all that helps.

Regards,
SW Yeo
 
#38
http://www.diyaudio.com/forums/showthread.php?postid=230603#post230603


.....how hard can it be...i have build this thing - a pcb with 4 non inverting lm3886 in parallel, and servoes.
for one bridged monoblok i use one of those pcb's per leg...4 against 4.....
and i also have a "input/driver pcb with opa627 - lpf - drv134 powered by alw/jung regulators....very nice IMHO.

no pictures yet, i don't have a digital camera.....

edit:
btw this thing is fully regulated - with lt1083, i got them cheap

edit2:
a pair of lt1083 per lm3886.......!!!!!
 
#40
overmind said:


I will try adding a 10ohm resistor + about 10 turns of wire around the resistor. The 10 ohm + turns of wire will be connected in parallel. The combination will be connected in series after the 0.5 ohm output resistor of each LM3875.
Click to expand...

Again tested without the LM6172 input buffer.

Just added the 10ohm resistor + 12 turns of wire (in parallel) after the 0.5 output resistor. The LM3875s are parallel after the resistor+coil.

THE HISS is GONE. :bawling:
DC off set is almost zero.

Much better sounding than the single LM3875 Gainklone.

Will adding the LM6172 and report back

Thanks everyone.
 
Status
Not open for further replies.
Top Bottom