LM4780 wihout Ci

[Dxvideo]

Hi all,

Last night I was playing with my all past amplifiers. To have some comparisons with my new toy (JFET Buffered LM3875 amp)..

While I was testing my LM4780 amp (the reference design of National) had also a DC offset measurement on the outputs and saw -1,5mV on each channel which is perfect for a chipamp. However it was an old amplifier and there were Ci capacitors (100u electroytics) on NFB line...
So I have decided to remove that caps to hear my LM4780 without them.
The result was significally different! I think it became a real hifi amp after that cap operation.. Especially just for the bass improvement was really worth to make that operation.
You must remove you Ci capacitors and see the results..
Anyway,
But there was a problem as you can guess. The DC offset was increased to ~100mV for the right and ~105mV for the left channel.
Do you think its an acceptable offset for that amplifier?
Do you know a way to reduce it (expect DC servo ?

Thx in advance for the replies..

[artquake]

expect DC servo

i think you meant "except" DC servo ... so yes you can, when you lower the input impedance then DC-Offset will fall too, not to zero but it will be lower ... and i think also that you know this allready

[Dxvideo]

Hmm..
You must be right. I have a 47K as input impedance. I think 20K is fairly enogh. I wont go lower because in this case my CDP cannot drive it..
Thanks..

By the way, DC servo matter getting more complex situation. Today I read a lot things about it and I think I will give up that idea..

[Dxvideo]

I was almost forgeting;
Before this amp a 100K pot is ok? Or should I select a lower one?
PS: No buffer or preamp before it..

[AndrewT]

Hi Dx,
are you using fig5 of 4780 datasheet?

what other changes have you made apart from 22uF to 100uF for Ci?

[artquake]

Do you know a way to reduce it

Feeding the difference (Amount of negative or positive offset) from dc rails to eliminate offset to zero with a trimmer ... I did that on a prototype LM3886 for nulling offset ... the amp is still working and my father is still using it ... and i changed the trimmer to a fixed resistor after running it two months ...

I think you should really explain more to the operation and usage of your LM4780 ... are both internal 3886's for seperate speakers or are you bridging them or paralleling them ? Even thinking that the answers you expect should also fit into a BPA usage ... ? I don't know ... i only can imagine ... suppose ... but that does not help to answer the question nor it is bringing you nearer into a decision ... at least in my thoughts ... the answers depend to the boundary conditions you comunicate ... and thats the point the way of a right answer begins ...
In fact i am a newbie into electronics, as i am learning a lot in reading here and the links to other places ... it's not easy to ask the right questions that i know ... but maybee it's only in my thoughts that you define more precise your requirement's ? Due to my knowledge maybee ...
Anyway ...

I wont go lower because in this case my CDP cannot drive it..

What makes you so sure that it won't ? Any more informations on this ? And if you mod the output driver of your cdp ? Would that be a solution ... think bigger ... you have to make allways some compromises even if it's a small thing ... it's on you ... or are you only using your CD-Player for this amp ? I don't think so ... maybe in future you'll connect a squeezebox or even other things directly to it ...

Wasn't it you - i do not remeber or i am to lazy to search now - that used a PGA2310 ? ... why not trying it with that ? ... many problems would be solved ...

And: When

By the way, DC servo matter getting more complex situation

What do you mean ? I think most of the things are written about this issue in the threads are correct, but theoretically. I'm besides a musician and know how instruments sound (i know thats very subjective) but the only cause i added a servo was because i wanted that my chips run cool and do not fight each other and don't produce stress ergo heat ... for what else ... for what else ? For better sounding ? The need of hearing the difference ? ... If yes something else was wrong ... even the recordnig studio's have some offset ... could they hear that ? ... stop stop stop enough writing ... not my day today ...

That's all ...

artQuake

[Dxvideo]

Originally posted by AndrewT
Hi Dx,
are you using fig5 of 4780 datasheet?

what other changes have you made apart from 22uF to 100uF for Ci?

No. It was like figure 1... But a load resistor and a DC blockage capacitor (4u7-MKT type) was added. And as mentioned Ci is disabled.

By the way, last evening I replaced the input load resistor with a 20K and DC blockage with 2u2 MKP capacitor also onboard PSU caps are replaced with Panasonic FM type 1500u s...
The final DC offset is 45 - 55mV (left and right).
I am feeding it with my regulated PSU (which is made for LM3875 with Pedja buffer!) at +/-30v. And with a ~1,5°C/W heatsink and after 4 hour non-loudy listening the 4780s surface was about 65°C..
The sound was really good.

[Dxvideo]

Originally posted by artquake

Feeding the difference (Amount of negative or positive offset) from dc rails to eliminate offset to zero with a trimmer ... I did that on a prototype LM3886 for nulling offset ... the amp is still working and my father is still using it ... and i changed the trimmer to a fixed resistor after running it two months ...

I think you should really explain more to the operation and usage of your LM4780 ... are both internal 3886's for seperate speakers or are you bridging them or paralleling them ? Even thinking that the answers you expect should also fit into a BPA usage ... ? I don't know ... i only can imagine ... suppose ... but that does not help to answer the question nor it is bringing you nearer into a decision ... at least in my thoughts ... the answers depend to the boundary conditions you comunicate ... and thats the point the way of a right answer begins ...
In fact i am a newbie into electronics, as i am learning a lot in reading here and the links to other places ... it's not easy to ask the right questions that i know ... but maybee it's only in my thoughts that you define more precise your requirement's ? Due to my knowledge maybee ...
Anyway ...


What makes you so sure that it won't ? Any more informations on this ? And if you mod the output driver of your cdp ? Would that be a solution ... think bigger ... you have to make allways some compromises even if it's a small thing ... it's on you ... or are you only using your CD-Player for this amp ? I don't think so ... maybe in future you'll connect a squeezebox or even other things directly to it ...

Wasn't it you - i do not remeber or i am to lazy to search now - that used a PGA2310 ? ... why not trying it with that ? ... many problems would be solved ...

And: When

What do you mean ? I think most of the things are written about this issue in the threads are correct, but theoretically. I'm besides a musician and know how instruments sound (i know thats very subjective) but the only cause i added a servo was because i wanted that my chips run cool and do not fight each other and don't produce stress ergo heat ... for what else ... for what else ? For better sounding ? The need of hearing the difference ? ... If yes something else was wrong ... even the recordnig studio's have some offset ... could they hear that ? ... stop stop stop enough writing ... not my day today ...

That's all ...

artQuake

To trim the supply is a crazy idea! But it must be right...
Last evening I could reduce the offset until ~50mV range . I think its enough for me. (with your way
But I will try the trimming way today. I have an adjustable regulated power supply an its very easy to me... I will inform you about the results.

And my circuit is (as mentioned) is the basic configuration;

But it was an old amplifier and laying somwhere in my room. Last night I rerun it just for a test.
Two years ago it was using with my computer and with +/-24v SMPS s. However by the time, I was bored from its sound and replaced it with a MOSFET amplifier (Nico Ras design). And last night I rediscovered its sound.
Anyway,
I mentioned, my CDP cannot drive it if I go lower than 20K. Because I have tested that situation with a LM4702+MOSFET amplifier. It was another reference application (AN1645) and the input load resistor was 6K8. At first listening impressions with my CDP, I was not satisfied from its mid and treble details and shared that experiences with the others. With a DIYers recommendation I've built a LM4562 preamp for it and its sound has improved too much.
Anyway, I ve tested that two or three times. Everytime I see, if I use a low value resistor on inputz (like 10K pot or 6K8 input load) then my CDP looses some details..

[AndrewT]

Ci is not present in Fig1.

To match the 4u7F @ the input, Ci1 & Ci2 should be >=330uF, not 100uF.
Have you tried this?
Now try 330uFelectro//2u2Fpropylene for both Ci1 & Ci2.
Ci1=CF1 in your diagram.

Look at the Krell Klone input bias adjustment.

It uses an adjustable resistance ladder across a pair of signal diodes fed from decoupled rails off the main supply rails.

[Dxvideo]

Yes, theyre mistaken named. In that figure theyre CF1 and CF2..
However I've already BYPASSED them. So I have no caps on NFB line..