I got the same value.
When I found the time I will put a power regulator based on gyrator on my 3886 (Rev-C). Putting a better Vreg on the 318 supply made a significant improvement in sound.
Did you mod the PBC to allow alternative regulator on your group buy?
Here's the PowerReg: (I've commented about it previously with little effect)
http://www.pinkfishmedia.net/forum/showthread.php?t=39306&highlight=powerreg
Cheers,
M
Andrew, thank you.AndrewT said:
I am afraid I wasn't very precise with my question. I am curious about the ripple current that the main caps in the supply see. Looks like you are reporting on the output voltage ripple. Do you know what the voltage ripple on your supplies is? This then should correspond to the PSRR, right?
I was motivated because I was pouring over cap datasheets last night reading the fine print on ripple current and various ambient temperature multipliers. Then I read Rod Elliott's article on power supplies (very enlightening!) and I realized that the relationships between toroid size and cap size and switch on time on the rectifier and all that jazz are not nearly as straightforward as I thought. Hence my attempt at a short cut: what's the measured ripple current in the caps?
peter
I have a DC on the mains problem. 260mV just about everywhere (even on the ground). I dont know why. The power company is coming out. I actually have one outlet in the house with 2.6V of DC.
Anyway, I have DC getting to the speaker. Usually 400mV at 200mA. I have a capacitor blocking it right now but it still buzzes at 60Hz.
This happens even with no source attached.
Both mains Line and Neutral now have a DC blocker circuit on them now. Just like This
http://sound.westhost.com/articles/xfmr-dc-f8.gif
Except that I used 1N4007 in triplet for each diode.
Then I added
http://sound.westhost.com/earth-f3.gif
and it didnt help any so I removed it.
Maybe the mains problem is not related to the DC to the speaker. Who knows but I figure it must be.
Peter suggested trying the DC voltage across R11, the 1R resistor. .49VDC there.
So with a source attached or completely removed this entire circuit is sending DC to the speaker.
I am using a IEC with EMI filter for mains then to a fuse then a switch.
After the switch we have both DC blockers then a terminal block with a CL60/33pf/CL60 circuit which is keeping my fuse safe to the 600VA toroid to a terminal block that sends out 25/0/25 to the amp.
The speaker ground and input ground both go to the board where ground was generated from the 0V combo of the 25V secondaries.
Earth ground goes straight to the chassis and NOWHERE else.
Uriah
Anyway, I have DC getting to the speaker. Usually 400mV at 200mA. I have a capacitor blocking it right now but it still buzzes at 60Hz.
This happens even with no source attached.
Both mains Line and Neutral now have a DC blocker circuit on them now. Just like This
http://sound.westhost.com/articles/xfmr-dc-f8.gif
Except that I used 1N4007 in triplet for each diode.
Then I added
http://sound.westhost.com/earth-f3.gif
and it didnt help any so I removed it.
Maybe the mains problem is not related to the DC to the speaker. Who knows but I figure it must be.
Peter suggested trying the DC voltage across R11, the 1R resistor. .49VDC there.
So with a source attached or completely removed this entire circuit is sending DC to the speaker.
I am using a IEC with EMI filter for mains then to a fuse then a switch.
After the switch we have both DC blockers then a terminal block with a CL60/33pf/CL60 circuit which is keeping my fuse safe to the 600VA toroid to a terminal block that sends out 25/0/25 to the amp.
The speaker ground and input ground both go to the board where ground was generated from the 0V combo of the 25V secondaries.
Earth ground goes straight to the chassis and NOWHERE else.
Uriah
I think I understood what you were on about. The question is precise.schro20 said:
I do not have a method for measuring smoothing cap ripple current.
I do have a method for estimating the measured voltage ripple on the smoothing caps.
I suspect there is a formula that relates voltage ripple to ripple current.
I gave you my results for a chipamp powered from a +-22mF supply.
I did a further measurement.
Using a 4 to 8 ohm speaker and increasing the output voltage ~ 2.4Vac (~1W), the ripple on the smoothing caps ranged between 25mV and 60mV, ~ 75mVpp to 200mVpp (rail to Power Ground, double these values for rail to rail).
This is just 22dB down on the output signal. If the PSRR has deteriorated to 30dB at high audio frequencies then the rail ripple will create distortion ~50dB below the signal. That alone substantiates the case for high smoothing capacitance.
I just stuck a 400VDC .22uf film cap between ground and the chassis. So from the chassis to the transformer end of the DC blockers I get zero VDC. From the Earth GND to the transformer end of the DC blockers I get 270mV. There is no DC at all on the output of the transformer.
Still 490mv on output to speaker.
Uriah
Still 490mv on output to speaker.
Uriah
you don't need to block both the Live and Neutral.udailey said:
I don't know what VA rating your transformer is but, 4700uF seems far too small for the blocking capacitance. This has a reactance to 60Hz of 1r13.
The diodes will pass @~1.2V. The maximum peak current that the blocker is able to pass is only 1.06Apk ~115watts on a 110/120Vac mains system.
The peak charging currents to your transformer are likely to be very much higher than this.
I know but I was reading DC on both with two different multimeters. DC on the ground to.
Sorry, each cap in mine is 10,000uf. Rod Elliott said to double the 4700uf if in the US.
Didnt sound right to me either. However I asked Brian and he said Earth GND has no place on the amp. If you check the amp schematics you will see this as well, I think. Maybe I misread them though but I see nowhere that allows for earth to go to signal or output GND and there is nowhere on the board for earth either.
Transformer is 600VA 24/0/24 my little pic says 25V out but thats wrong its 24.
Uriah
Attachments
Hi,
all exposed conductive parts must be connected to Safety Earth.
Your Safety Earth must be permanently connected direct to chassis. Bolt the third earth wire direct to chassis.
Can anyone disconnect any of the cables coming out of your amp and touch the exposed ens while the other end is still connected to the power amp? A child might do just that, worse they may put the free end in their mouth.
You MUST connect all your Audio Grounds to Safety Earth, you have no choice in the matter. If your supplier is telling you otherwise, then he is incompetent.
Even 10mF is too small. That's just 230W at the current peak.
If the 10mF were in parallel then the peak current quadruples to >8Apk. and other than during a fault condition the caps will never have >1.5V across them.
all exposed conductive parts must be connected to Safety Earth.
Your Safety Earth must be permanently connected direct to chassis. Bolt the third earth wire direct to chassis.
Can anyone disconnect any of the cables coming out of your amp and touch the exposed ens while the other end is still connected to the power amp? A child might do just that, worse they may put the free end in their mouth.
You MUST connect all your Audio Grounds to Safety Earth, you have no choice in the matter. If your supplier is telling you otherwise, then he is incompetent.
Even 10mF is too small. That's just 230W at the current peak.
If the 10mF were in parallel then the peak current quadruples to >8Apk. and other than during a fault condition the caps will never have >1.5V across them.
He, he, it is just that this is the Only DIY regulator that I know how to build
(apart the same without monolithic preregulator: zener diode and gyrator). The fact is that it sounds so much better than LM317.Dear Andrew, what do you think about adding a gyrator based PS to the 3886 from RevC? (like in the link I posted above)
Cheers,
M
Okay, let me clearify my position on the grounding for this amp...
The question was (in an email):
My answer was:
...which was inaccurate and incomplete. For diagnosing noise that is most likely a ground loop, it is a good first step or remove excessive grounding (under safe workbench conditions, then reapply proper grounding).
What I left out of my reply was to indeed connect the center-tap ground to the chassis, so that the primaries and secondaries are working at the same potential. For this amp/layout all you need is this one ground connection, as all grounds are tied to. Grounding the connectors will yield a ground loop. This is not always the case, and depends on the layout in question.
If in diagnosing noise you find the Mains Ground to be the trigger, then you need a "ground breaker" to keep current from flowing from/to it unless there is actual voltage present.
In an ideal world, you would also fuse each of your secondaries, as a fuse will reach faster than a circuit breaker, which typically must wait for a certain point in the wave to trip (unlike a fuse) and will also have substantially higher current trip point (20A).
Incidentally, fusing the secondaries is how most consumer equipment works when there is no mains ground connection.
The question was (in an email):
My answer was:
...which was inaccurate and incomplete. For diagnosing noise that is most likely a ground loop, it is a good first step or remove excessive grounding (under safe workbench conditions, then reapply proper grounding).
What I left out of my reply was to indeed connect the center-tap ground to the chassis, so that the primaries and secondaries are working at the same potential. For this amp/layout all you need is this one ground connection, as all grounds are tied to. Grounding the connectors will yield a ground loop. This is not always the case, and depends on the layout in question.
If in diagnosing noise you find the Mains Ground to be the trigger, then you need a "ground breaker" to keep current from flowing from/to it unless there is actual voltage present.
In an ideal world, you would also fuse each of your secondaries, as a fuse will reach faster than a circuit breaker, which typically must wait for a certain point in the wave to trip (unlike a fuse) and will also have substantially higher current trip point (20A).
Incidentally, fusing the secondaries is how most consumer equipment works when there is no mains ground connection.
So Brian,
Let me see if I understand.
I need to take the 0V from my transformer and create a little star ground with it and Earth GND. Then take a wire from that star to the GND on the board.
I DO NOT need to take any extra wires from input or output to this star or anything connected to that star since this is taken care of on the board.
Am I understanding this correctly? Right now I have Earth and 0V tied and a wire from there to GND on the board but I do have extra wires from signal and output to the star GND. There is nearly zero noise and almost zero DC but there is always room for improvement.
Also, if I fuse the secondaries what rating should the fuses be for each channel?
Thanks for your help,
Uriah
Let me see if I understand.
I need to take the 0V from my transformer and create a little star ground with it and Earth GND. Then take a wire from that star to the GND on the board.
I DO NOT need to take any extra wires from input or output to this star or anything connected to that star since this is taken care of on the board.
Am I understanding this correctly? Right now I have Earth and 0V tied and a wire from there to GND on the board but I do have extra wires from signal and output to the star GND. There is nearly zero noise and almost zero DC but there is always room for improvement.
Also, if I fuse the secondaries what rating should the fuses be for each channel?
Thanks for your help,
Uriah
F4A fuses feeding the chip amplifier rails are far too high.
A 4A fuse will allow upto 250W into 4r0 or upto 500W into 8r0.
F3.1A <=150 into 4r0 or <=300W into 8r0.
F2.5A <=100W into 4r0 or <=200W into 8r0.
F2A <=60W into 4r0 or <=120W into 8r0.
F1.6A <=40W into 4r0 or <=80W into 8r0.
It appears that 4ohm speakers require ~F2A and 8ohm speakers require F1.6A fuses in the supply rails fitted after the main smoothing caps if one designs to the power limit of a 3886.
A 4A fuse will allow upto 250W into 4r0 or upto 500W into 8r0.
F3.1A <=150 into 4r0 or <=300W into 8r0.
F2.5A <=100W into 4r0 or <=200W into 8r0.
F2A <=60W into 4r0 or <=120W into 8r0.
F1.6A <=40W into 4r0 or <=80W into 8r0.
It appears that 4ohm speakers require ~F2A and 8ohm speakers require F1.6A fuses in the supply rails fitted after the main smoothing caps if one designs to the power limit of a 3886.