I have finished both of the board, and found a weird problem.
The first one got 80mV DC offset, after aging for an hour, settle down at about 28mVm which is okay but I thing it's not the best status?
The second one is pretty high when finished, I got 178~190 mV, then again aging for an hour, gradrally down to about 46~50mV.
Then I forgot to turn off amp's power when I went to dinner, after about an hour, measured "-4~-6mV", this is weird result.
I decided to wired up the speaker, everything seens fine.
But why the DC offset is going negative, is that normal?
The first one got 80mV DC offset, after aging for an hour, settle down at about 28mVm which is okay but I thing it's not the best status?
The second one is pretty high when finished, I got 178~190 mV, then again aging for an hour, gradrally down to about 46~50mV.
Then I forgot to turn off amp's power when I went to dinner, after about an hour, measured "-4~-6mV", this is weird result.
I decided to wired up the speaker, everything seens fine.
But why the DC offset is going negative, is that normal?
I've read the this post before souring the parts, so I'm also using SMD291 to solder the SMD parts.
And also, I've clean the board by 99.5% ethanol, so I think the flux is not the problem.
I measure the 2nd board today and get -32~-38mV, which is worse than yesterday(-5mV).
Any suggestions?
Last edited:
Unless you have faulty components or mounted something in the wrong place 99% of DC offset problems are due to insuffucient cleaning.
from https://labproinc.com/blogs/chemica...l-alcohol-for-cleaning-electronics-in-the-lab
Ethanol’s drawback is that it can sometimes leave oil traces on the surface it evaporates from. These traces of oil can subsequently cause malfunctions in highly sensitive electronics. This drawback is highly minimized when using the purest form of ethanol, but the most sensitive electronic components may still have problems when cleaned with ethanol as a result.
(...)
Isopropyl alcohol (IPA) is the more common choice for cleaning electronics because it evaporates more rapidly than ethanol and also because it does not leave any traces of oils upon evaporation.
Ethanol’s drawback is that it can sometimes leave oil traces on the surface it evaporates from. These traces of oil can subsequently cause malfunctions in highly sensitive electronics. This drawback is highly minimized when using the purest form of ethanol, but the most sensitive electronic components may still have problems when cleaned with ethanol as a result.
(...)
Isopropyl alcohol (IPA) is the more common choice for cleaning electronics because it evaporates more rapidly than ethanol and also because it does not leave any traces of oils upon evaporation.
Thanks Dario, will get some IPA from local store and see if it solve the problem.
But I search on web and found the offset lower than +/- 50mV is nowmal and acceptable, I think it should not affect the sound quality or should be no damage to the speaker.
According to the formula, P = square(V) / R = 0.0025 / 6 = 0.0417mW, should only produce fairly low heat to driver?
But I search on web and found the offset lower than +/- 50mV is nowmal and acceptable, I think it should not affect the sound quality or should be no damage to the speaker.
According to the formula, P = square(V) / R = 0.0025 / 6 = 0.0417mW, should only produce fairly low heat to driver?
Fair enough, still the My_Ref DC offset is usually under +- 15mV so I would anyway try to achieve it.
Good news, after washing both of the PCB with IPA, the highest DC offest is now only 7.0mV, nice!
May I ask a question about the layout?
Can someone explain the positioning of R11 to me? I read that it lead to 10 fold less distortion compared to its old position, but couldn't find an explanation why. Or was that experimentally determined?
And is the outer feedback loop twisted on purpose to eliminate its inductance and EMF coupling? Because that's very clever.
Can someone explain the positioning of R11 to me? I read that it lead to 10 fold less distortion compared to its old position, but couldn't find an explanation why. Or was that experimentally determined?
And is the outer feedback loop twisted on purpose to eliminate its inductance and EMF coupling? Because that's very clever.
Sure
It depends on the return current path and the area it delimits.
Bigger the area, bigger the RFI emission/capture, it behaves like an antenna.
Can you explain what you mean by 'twisted' in this context?
I made a drawing for you.
The feedback path is a figure eight (a loop that is twisted once, if you will). The electric fields induced in each partial loop of the figure eight are of opposite polarity. If both partial loops are of equal size, no net voltage/current is induced in a uniform varying magnetic field. Thus mutual (coupling) inductance is zero.
I'm not completely sure about self inductance, though. There are active devices in the loop, thus currents in each branch will be different (LM318 output to LM3886 input; LM3886 output to LM318 input). But doesn't only the net current which flows in a loop count? Hmm.
The loops will never be perfectly equal since the LM3886 package adds a vertical component and there are resistors and capacitors in parallel. But to me the effective area in the PCB plane seems smaller than with a minimized "classical" loop.
There is the possibility that the magnetic field is non uniform, then the most densely packed non twisted solution might be slightly better. ...Or possibly not, as the difference in total enclosed area between both layouts is small anyway.
The area enclosed by the return current from LM3886 output through R11 to its bypass capacitors? Or the current into the input pins of LM318 through its regulator and R11? Possibly both?
Thank you.
Thanks, now I see what you meant.
The twisting was pretty much dictated from the circuit topology/schematic of Penasa's design, the only attention I had while designing the PCB was to have a perpendicular crossing of traces and the use of the copper gap between resistor pads so to minimize coupling between the two.
On Tomchr suggestion, while revising the PCB design, I and JosephK where focused on the second one (LM318/regulator).
The first one can be addressed only with a complete redesign, moving the entire PS/regulator complex to the left of LM318 (eventually with rectification and main smoothing on a separate board) and rotating the LM318 90° clock wise.
It could happen in a future, but not planned, revision that eventually could carry also a second current pump, like in Penasa My_Evo:
Forgive this topic. It's pretty off topic, but it is a little bit related.
Background
My current project is totally unrelated, a 3 phase sine drive for a modern PMSM motor to drive my vintage turntable. But I have chosen to use 3 x LM2886 to drive the motor. I will be using a fairly minimum gain (better noise) of 11 for the LM3886s and, since I'm pushing stability at low gain, I have been reading up on improving stability of the LM3886. Naturally, that led me to Tom Christensen's Modulus-86. He seems to love using a parallel inductor/resistor in the output to filter out very high frequency. In this project, it doesn't make sense to try it, but I wondered if any of you experimenters have tried this kind of output filter on MyRef? I'm guessing that it isn't compatible with the current pump approach because the filter would add it's non-linear impedance to the pump circuit. Any thoughts?
Jac
Background
My current project is totally unrelated, a 3 phase sine drive for a modern PMSM motor to drive my vintage turntable. But I have chosen to use 3 x LM2886 to drive the motor. I will be using a fairly minimum gain (better noise) of 11 for the LM3886s and, since I'm pushing stability at low gain, I have been reading up on improving stability of the LM3886. Naturally, that led me to Tom Christensen's Modulus-86. He seems to love using a parallel inductor/resistor in the output to filter out very high frequency. In this project, it doesn't make sense to try it, but I wondered if any of you experimenters have tried this kind of output filter on MyRef? I'm guessing that it isn't compatible with the current pump approach because the filter would add it's non-linear impedance to the pump circuit. Any thoughts?
Jac
The answer is in the orignal My_Ref thread, from what I can remember it was both not compatible with the current pump and considered useless in the circuit by Penasa