They (the ATL boys/girls/items) say that the number of circles (crop and non crop) and number of servo's (dc un-drive entitie's (also DUDE's) they call them) should tell all 🙂
Is this all you can do, do decent bribe, then I just have to reveal all 🙂 So here it is, the answer was in the fact that a cartridge does not have a center-tab 🙂 So no real ground needed 🙂 Now we still need to see that it actually works, have fun testing.
Attachments
What you said is correct.
Cartridge has no center ap which means that input gnd is floating.
Proven fact is that I run it balance with single end cable i.e. no gnd.
The only thing is that the scheme servo corrects directly the input signal which seems to be invasive IMHO.
Any tought on this?
Would you like to repost it?
We may take it from there! I would still be interested to try that version to see how it compares to the discrete's.
Btw I just tried the SE version of the circuit here and sound is very, very good.
Fast, detailed, deep base very refined. 😎
No (and yes), IMO it makes no difference where you connect the servo, as long as it influences the output it is in the signal, as long as it influences the output with surten degree then that 'degree' is the measure, no mather where it is connected.
Sorry, but I see I made a mistake 🙂 I did only servo the 'right' output, and not the left. This can be easily corrected I think, but a third servo is needed. Both 75 Ohm resistors should be driven by separate servo's. For this the center point should be connected to ground, and the other side of those resistors should be servo managed.
Would depend on the matching of the transistors. The mismatch on the input BJT's is multiplied by 40dB (that's a lot). Problem is, this does not always show in simulations, as all devices of one type match 100% (in LTspice, in other simulators you can do some Monte Carlo analysis).
Hand matching, not talking about simulators.
The 100 percent identical 'conditions' would be the real problem, no idea how to achieve that.
Just a (pre-lunch) mental exercise.
Other mind jump was to take the differential of both outputs and servo it down to zero, plus individual side servo's for the common DC level, but where to inject those ?
Maybe not 100% but on 99% I am working on and is been pre lunch and pre dinner for weeks
ok I will post it a little later.
It is the SE of the circuit I have already posted here.
After giving this a bit more thought, I come with this for input differential nulling. The previous scheme will not work, the input differential was null due to the fact that the cartridge-model had a series resistance of 0 Ohm (a bikini model would have been just as effective). I upped the series resistance of the cartridge to 50 Ohm and added some offset injection (R58) for testing. This version will work.
For output offset nulling I still have no final scheme, those guys from ATL where not as smart as it seems 🙂 I do have some alternative ideas to try, but I also need more time, that I sparsely have.
For output offset nulling I still have no final scheme, those guys from ATL where not as smart as it seems 🙂 I do have some alternative ideas to try, but I also need more time, that I sparsely have.
Attachments
Looking at it, I really think this servo is intrusive.
Basically it corrects right at the input of a 60-65dB normalized gainstage.
Plus the use of the 1m right at the input seems excessive.
Isn't there any different way to correct it?
I am also thinking how much the differential offset will wonder if the stage is built with monolithic BJTs.
Also like I said the differential offset can be corrected between the two legs of the differential input but I was unable to find a suitable circuitry for that.
I am really curious to see what the other alternatives you have in mind are 🙂
- Status
- Not open for further replies.