Gainclone heaven ?

[PeterSt]

Mani,

I don't think you ever saw it, but in my basement I have a 25Kg (sine) power regenerator as well. It sits there ready for action, but I never use it anymore. Ok, but why the story ?

Already at line 3 or yours I saw the same coming up that I experienced here : power regeneration - nice, but it won't stop any coming in mains disturbances over PE (or maybe the Neutral). So, undoubtedly that is what happened to you, but with the idea that two (or more) devices not on the same PE (/Neutral) explicitly create the problem (of potential difference).
The story is a bit similar to what I said maybe a year back or so : connect your PC to the nice dedicated (audio) mains source as well, or otherwise it's looking for trouble only. So, seems counterproductive (PC's perceivedbly throwing out noise) but I don't think it is (at least not net).

Anyway, recognizable all over ...
Peter


PS: Here too, at using the battery only (mains cord not in) all is fine. Connected to the mains - the thing is on separate PE (pin), but any distortion source will come through anyway (this is about my fameous PWM pump example).

[Scroobius]

A familiar story indeed - I posted quite some time ago that I originally had my PC on the house PE and the amp/NOS on the dedicated hifi PE. The thinking was to keep to PC hash away from the dedicated hifi PE. But moving the PC onto the dedicated hifi PE resulted in a dramatic improvement.

But I think that was in the days when my amp ground was connected to PE. Nowadays the amp ground is connected to NOS ground which in turn is connected to the PC ground (and PE? probably) so I am not sure what would happen now. The hifi PE is therefore only connected to the metal cases of NOS and the amp for safety reasons. So now if the PC was plugged into the house PE what would happen? the hifi PE and house PE would not be connected together (as they were before). But the hifi ground would become connected to the "dirty" house PE via the PC.

Out of interest how is the PE connected in your GC amp Mani?

P

[manisandher]

Hi Paul, we all know that grounding and PE are a nightmare story. My understanding is that my amps are configured in exactly the same way as yours are and the NOS1 too, i.e. signal ground is not connected to PE. But as you say, who knows what happens in other components or the PC?

... moving the PC onto the dedicated hifi PE resulted in a dramatic improvement.

Generally I would characterize grounding issues as increasing the brightness, glare and edginess of the sound. It can sound more detailed at first, but with prolonged listening you realise that this is superficial detail that becomes fatiguing with time.

Does your 'dramatic improvement' match this?

Mani.

[Scroobius]

Mani - yes I would say that what I experienced is similar to what you have described. But given that your grounding arrangements with your electronic ground not connected to PE other than via the PC is it just a result of being connected to a "dirty" PE. Well earthing as Peter points out is so complex and is different in almost all installations that it is impossible to draw meaningful conclusions.

It is difficult to compare with my experience at that time because in my case I ended up (without realising at the time) connecting the PE's together. I will try again tonight and see what happens just out of interest. 

[Scroobius]

Peter's new Phase Alignment (PA) feature turned out to be a big challenge for me and I learned a few things about how to get the best out of a GainClone on the way. Thanks Peter for your help (big help)!!!.

First of all I could not use PA because my GC's do not have any blocking capacitors in the signal path. So the first challenge was to find blocking caps that did not affect sound quality too much. The second challenge was that I came to realise that there is a problem with GC's that I kind of knew about but did not realise how important it was and that is DC offset.

Blocking Caps
The GC's amplify DC (and oh boy do they!) if you stick 300mv in then a whopping 11volts appears at the output to the speakers (34db gain). Wow. Not many amplifiers do that they mostly ignore dc. So the only answer was blocking caps. Having tried a few caps I did not hold much hope of finding any any of good enough quality to put in the signal path. I tried some caps but they were not of good enough quality or I could not get the right values. To cut a long story short I found a quartet of Blackgate 22uF NX Types 6.3v (my GC is biamped). Some would say the best capacitor ever made. So I installed them and to my amazement I could hardly tell the difference with them installed versus no caps installed. Maybe a bit less distinct at the top (straight after install) but very very good indeed - and they can only get better with burn in. I can easily live with these BlackGates. I found this useful in getting some idea of the value of cap needed http://www.st-andrews.ac.uk/~jcgl/Scots_Guide/experiment/lowpass/lpf.html. Peter Daniel recommends no more than 100uF. My GC's have a 22k resistor between input and ground so the input impedance of the GC is 22k and that is the resistor value for the RC low pass filter. If you use one capacitor to feed two amps. (bi amp) then the resistor value is going to be 2 x 22k in parallel i.e. 11k.

GC DC Offset and Phase Alignment
So it turns out that DC offset appearing at the speakers affects phase alignment (it does elsewhere but that is another matter). The really bad thing is to have a difference in dc offset between speakers i.e. if there has to be dc offset then it should be the same for both speakers (and the offset should be -ve i.e. the speaker cone should be pushed back slightly). This is where it gets to be a challenge. There is an inherent dc offset derived within the GC's which exists with no input connected. But when you connect an input impedance it reduces (a lot) depending on that impedance. But then there is also an offset that is derived by amplifying the dc offset that exists on the output of the NOS1 (say 5-15mv). So if you have no dc blocking caps you can get a BIG dc offset at the output. In my case this was 380mv R and 120mv L (roughly) with no blocking caps and this was derived from the NOS1 output dc offset because of the gain of 34db applicable to dc. That is a HUGE offset problem and has a big impact on phase alignment between speakers. It is a problem I have had for a long time that I did not realise.

So today when I installed the blocking caps the dc offset went up because of effectively removing the NOS1 source impedance of 30ohms but went down because the GC could no longer see the NOS1 output dc offset. In the end I decided to rebuild my amps by selecting LM3875 chips from a batch of 12 to find those that matched. I ended up with -80mv L and -80mv R for bass units and -87mv L and -85mv for the tweets. But note that offset is potentially not an issue for the tweets because of the capacitor that is in the crossover for all HF units (no cap for bass units).

After final install that offset changed slightly because the surrounding components are slightly different in the final build compared with my test rig. But anyway I ended up with a few mv difference between drive units rather than the 150mv before. The difference in sound was immediately apparent the image between the speakers is clearer with fewer strange phase effects.

But then I switched on PA and WOW WOW well you will have read the reviews elsewhere.

The message is that if you build GC's then to get the very best out of them you really MUST match the offsets of the LM3875 (or whatever) chips to at the very least match the offsets on the outputs to each speaker. It is unlikely you will get very low offsets (but you may be lucky) but at least get them as close as possible left and right at the speakers. But most important is that you will have to match the LM3875's for your particular install (i.e. with or without blocking caps) in other words the same LM3875's will produce different offsets depending on whether blocking caps are installed and if none then depending on the output offset of your NOS1.

It took me a whole day to do all of this but I have to say it has been well worth it.

So you can tweak yourself completely crazy with this one. Best of luck!!!!

All the best

Paul

[stefanobilliani]

Nice Post Paul ,

From what I read it means that your NOS1 is working perfectly well about the new PA feature of XXHE ( plops , etc ) .

Stefano

[Scroobius]

Hi Stefan - my first extended listen with PA engaged was yesterday and I listened to many familiar albums - and that would be the way to describe the sound - engaging it was difficult to pull myself away and even my wife joined in making requests for her favourites.

Everything here seems to be working perfectly (I have to be careful saying that I don't want to invite trouble) since reloading my o/s. But I do get "plops" sometimes when an album starts playing and maybe that is as the dc offset is applied (I guess) if in unattended it happens once at the start of the album. Other than that everything is fine. I have not started playing with settings yet.

I will catch up on "plop posts" now ha ha.

Peter - the sound quality I get now is STUNNING it has to be better than I thought was possible when I set out to build this system - I mean WAY better. I still have to pinch myself that the sound I hear is coming from mere loudspeakers in a very odd shaped room. Thanks for your total dedication and hard work - it is certainly appreciated here.

All the best

Paul

I must play with setting tonight.

[Eric]

Hi Paul,
I also spend the Saturdaynight with XXHE/PA. Same experience here, even if I am currently using a temporary amp+preamp (blocking DC). I ordered a pair(+) of Gainclones, including caps. I expect these to be delivered by Bert soon.

Cheers
Eric

[Scroobius]

Hi Eric - Good to hear from you I hope things are good with you. Any chance that Bert could get the dc offset outputs to the speakers (at least the bass units) matched closely? Also if you can bi amp or more not sure I can remember how many drivers you have that is more than well worth doing.

Any idea what blocking caps Bert is using?

Anyway all the best with your GC's it will be interesting to hear how you get on.

All the best

Paul

[PeterSt]

I must play with setting tonight.

Paul, is that because you are not allowed to drink a fine Whisky in the morning ? haha

Thanks for all the kind words. But I estimate I learned as much from your little project. But with your great summarize of it from your earlier post in here, I sure hope others can benefit from it as well. And let's say it is a completely logical thing, but who thinks of it ? okay, we did now, but it still took a few dozen emails, right ?

And so we go on and on.

Let me add -although a relative few will read this- that it is the Xth time that a single person (like you, but also others) shout outloud that sound again improved by 100% (the 100% is on my account) or even more (which is also on my account ). Still it is so. I myself am amongst those persons and while people may claim something like "5 % improvement" (FWIW anyway of course) this is *really* 100% and more.

I think I can say that this is another "fork" of improvement, but why not. Point is (for those who can't understand sh*t of all this) : the further you are, the more easy it gets. The better SQ just becomes grabbable, and as I said about a year ago : it even becomes fairly easy to hear where the improvement should be (or where the remaining faults are). Phase Alignment is such a typicle improvement, although it sprung rom theory. But for those who are using it for a week or so by now ... go back and listen to what sheer distortion you were listening to all the time. Not that you could hear that before ...
All is about the reference.

Ok. So now I should check my own L/R offsets, right ? haha.

But maybe to emphasize what this (last larger post of Paul) is all about :
The fact that an amplifier blocks DC does not prohibit that it may have DC offset itself at the outputs. So, in the case of the GainClone chips this is the most easy to have, because the chips are not equal for it. This is a bit similar to the NOS1 which may exhibit a few mV difference between Left and Right just the same. However, while this few mV is neglectable, a few 10s of mV is not at all and this can happen at the output of amplifiers. Also non-GainClones;
Like Paul told, for the mid-high this won't be much of a problem, because the capacitors in the Crossovers (assumed they are there of course -> not with electronic crossovers !) will block the Offset from there. But not so with the woofers.

There is not so much you can do about it with any normal prebuild amplifier, unless you make a high pass at the speaker side that won't allow frequencies to pass under, say, 0.5 Hz. But this requires huge good quality capacitors (like 150,000 uF) and once you spent the money on them, it is still to see whether it works out for the better. But when you are trying out an amp you might want to buy, go out with your Voltmeters !

Super, Paul !
Peter

[BertD]

Any chance that Bert could get the dc offset outputs to the speakers (at least the bass units) matched closely?

There is NO DC-offset at the outputs. A servo regulation makes it 0.

Any idea what blocking caps Bert is using?

Duelunds 

Bert

[stefanobilliani]

Interesting to see that the much maligned capacitor is playing a role again , in a place where at least a couple of things are assumed not so correctly
For example . It is incorrect to say that any amplifier blocks dc , aswell as calling capacitors Blocking DC .
Capacitors do Couple for DC .
 
b- a gain clone chip which has I believe at least 60 db (90db?) open loop gain , and some 30 or 40 db of gain is going to waste by the feedback loop and is it comparable to the distorsion of a capacitor ... I mean in Music?

c- we are talking a bit of the PA feature of XXHE , that *for now* can be far more offensive than the vaste majority of capacitors on the planet .
Offensive for its inherent problems ( for now - for someone .... ) about plops at start , or at stop putting up volume over the +zone in db.

and more : AP is not about L and R phase somewhere I did read ( if right of course) and now suddenly and with reason it is ok to predict that the dc offset of the amps should have same values to every channels , for those concerned about the maybe not neglectable 300mV .
.

Maybe I am missing the point .
When and why is it safe to use PA in XXHE judging with the Voltmeter?

Is it really DC ,something that let your woofer excurs for a centimeter when music Stops ? What can capacitors really do about that?


Stefano

[Scroobius]

There is one thing I forgot to mention I had 3 out of 4 amps "blow" 2 were driving bass units 1 was driving a capacitor protected tweet. So maybe a humongous capacitor on the output of a GC (with none on the input) may not be viable for GC users.

I agree with Peter SQ with PA engaged is a very big step forward. I could not rate it % terms and I would find it difficult to describe apart from "clearer" but anyway I do not need to go back & turn off PA to hear the difference it is as clear as night and day here. Nice one Peter.

It is a bit too early for a whisky for me but maybe later ha ha.

P

[stefanobilliani]

There is one thing I forgot to mention I had 3 out of 4 amps "blow" 2 were driving bass units 1 was driving a capacitor protected tweet. So maybe a humongous capacitor on the output of a GC (with none on the input) may not be viable for GC users.

Well maybe that is telling us the GC is not good without the input being DC cap Coupled ....and maybe the enourmous gain of it looks toward wild instability if the input sees some of the source changes .... .
The output cap is another story .

stefano

[BertD]

Is it really DC ,something that let your woofer excurs for a centimeter when music Stops ? What can capacitors really do about that?

Yes, true DC which will misalign the cone in one direction and even burning the coils (depending on the output power and DC Voltage) if its on there too long (getting overheated while not moving!) but that is the least of the problem...

A capacitor in series with the DC/AC signal blocks DC and passes AC (this AC is making your diagramps vibrate and accepted as making music or "wanted" noise).

Edit: Best spot for this protecting capacitor is at the input. The value can then be a lot lower due to the higher impedance as seen by the capacitor. This position also prevents the amp to become instable when a source outputting DC (should not be allowed if you ask me) is connected to the amp.

Bert