Hi,
Just a small question I have.
I'm building another GC, but now with panasonic FC and a IC buffer a la Nuuk and maybe a bigger supply. I'm using now solid wiring and trying to build it clean.
I finished yesterday the 1st channel (without the buffer) and the chip got very hot despite my care with the wiring. Could be that I didn't put the input cap and the resulting offset close to 250mV if I remember right heated the chip? I don't know, but this has easy solution.
But my concern is that oscillation is caused. As they were very close, I used the same point for both signal and power ground, and there comes my question from: Is there any technical reason to not to tie together SG and PS?
Just a small question I have.
I'm building another GC, but now with panasonic FC and a IC buffer a la Nuuk and maybe a bigger supply. I'm using now solid wiring and trying to build it clean.
I finished yesterday the 1st channel (without the buffer) and the chip got very hot despite my care with the wiring. Could be that I didn't put the input cap and the resulting offset close to 250mV if I remember right heated the chip? I don't know, but this has easy solution.
But my concern is that oscillation is caused. As they were very close, I used the same point for both signal and power ground, and there comes my question from: Is there any technical reason to not to tie together SG and PS?
Hi Raka
250mV offset is way too much and i suspect it is not offset but oscillation. No scope? Offset will certainly not cause the chip to heat up, especially without a load. The main reason for separating SG and PG is to avoid heavy PS currents producing voltage drops in the SG line and thus modulating the input voltage. Generally combining the two grounds may not be good for the sound but would hardly lead to oscillation. Got ps decoupling close to the chip?
cheers
peter
250mV offset is way too much and i suspect it is not offset but oscillation. No scope? Offset will certainly not cause the chip to heat up, especially without a load. The main reason for separating SG and PG is to avoid heavy PS currents producing voltage drops in the SG line and thus modulating the input voltage. Generally combining the two grounds may not be good for the sound but would hardly lead to oscillation. Got ps decoupling close to the chip?
cheers
peter
No scope.
The caps are soldered to the chip pins, like I did with my first GC which had no temperature rise.
After I fitted the cap I got 0.0mV, but I did this after half an hour of listening music at medium level.
I separate both grounds, and then linked then with a single wire, right? Well, what if instead a wire I joined both? The only difference would be the impedance of the link, right? but what impedance has a 2cm wire?
The caps are soldered to the chip pins, like I did with my first GC which had no temperature rise.
After I fitted the cap I got 0.0mV, but I did this after half an hour of listening music at medium level.
I separate both grounds, and then linked then with a single wire, right? Well, what if instead a wire I joined both? The only difference would be the impedance of the link, right? but what impedance has a 2cm wire?
Raka said:
Tying the signal and power supply grounds together at the same point is good practice. But more important in terms of oscillation is what's going on with the other power supply leads, i.e. your positive and negative rails.
To avoid oscillation, you want your voltage lines and your power supply ground line(s) as close together as possible to keep the inductance of the power supply leads as low as possible. The greater the separation between them, the greater their inductance and the more likely you are to develop oscillation problems as the signal effectively gets fed back by way of the power supply lead inductance and the amp's compensation capacitor.
So the solution may be as simple as dressing up your supply leads. As analog_sa mentioned, try some decoupling caps near the supply pins and see if that takes care of the problem. If it does, that will at least tell you where the source of the problem is.
se
Nuuk said:Steve, what exactly does 'dressing up' involve? And I am talking leads here
Well, argyle knit stockings are all the rage this year and would look just adoooooorrable on most any pair of leads.
You didn't really think I'd let your "I am talking leads here" get in the way of a smart-*** comment from me did you?
No, I just mean keeping the power supply leads in close proximity to keep power supply inductance low whether by twisting, braiding, what have you. Don't know just how Raka's got the leads running or if it may be in a more cobbled-together-just-to-see-if-it-doesn't-blow-up-and-then-I'll-do-it-all-up-neater-later stage or what so just threw that one out.
se
Steve Eddy said:
Well, argyle knit stockings are all the rage this year and would look just adoooooorrable on most any pair of leads.
You didn't really think I'd let your "I am talking leads here" get in the way of a smart-*** comment from me did you?
No, I just mean keeping the power supply leads in close proximity to keep power supply inductance low whether by twisting, braiding, what have you. Don't know just how Raka's got the leads running or if it may be in a more cobbled-together-just-to-see-if-it-doesn't-blow-up-and-then-I'll-do-it-all-up-neater-later stage or what so just threw that one out.
se
I didn't understand the first two paragraphs, but suppose that are kind of ironic jokes
Then, if I simply twist a little bit the four power supply leads of the first channel I probably avoid this problem, right? I'll try asap.
Steve, I have a very good looking arrangement
, not a rat nest like the first one (that worked cool, btw)
Raka said:I didn't understand the first two paragraphs, but suppose that are kind of ironic jokes
Ironic? Naaaah. Smart-***.
Just having some fun with some word play. Nuuk asked about "dressing" the leads so I came back acting like some gay audio fashion consultant.
Don't know about probably. Just offered it as one possibility to consider.
Steve, I have a very good looking arrangement
Hehehe. If that's the case, then it may well be something else. But knowing what it isn't gets you closer to what it is.
2:00 AM here so I'm off to bed for now.
Good luck!
se
Nope, nuuk that was a bad joke
Seriously, you all know that I'm a newbie, but I'll try to explain (sorry for my language mistakes) the point I understood very well the dressing point:
You have to put together the wires, the closer the better. Then, you pour some drops of vinegar and olive oil together with two pinches of salt. You remove at high frequency and you're done.
Seriously, you all know that I'm a newbie, but I'll try to explain (sorry for my language mistakes) the point I understood very well the dressing point:
You have to put together the wires, the closer the better. Then, you pour some drops of vinegar and olive oil together with two pinches of salt. You remove at high frequency and you're done.
Raka said:
In point-to-point
sometimes it's difficult do get the 1000uf caps as close to the chips as they would like.On the breadborad modules that I usually make, the 1000uf caps are very close to the chips, but right at the pins I bypass with 0.1uf polyester caps.
You may try to do that, maby that's the problem.
It may seam to you that the caps are close enough, but at a few cm from the chip they are too far.
If you can't get them closer, bypass.
Biamp
Hi Carlos, and thanks for all your answers, are very helpful.
My 1000uF caps are not more than a cm far from the chip. Yesterday I "dress" the power supply wires and with 5mV offset the chip got cool at low spl, but at medium got warm, but I think that this is because of the hot here. Now is sounding nice.
Nuuk, what is the 270K resistor in your buffer for?
BTW, I knew I forgot something when I started the thread:
I'd like to know if I could just biamp my speakers with the active filter built like the LPF a la Rasmussen. My idea is to buffer the input with 637 and then divide the signal to two channels (one GC each). My speakers have a first order slope for both drivers that could be translated to the chip pins easily, has this any problem?
I know the drivers well and the crossover point and slope is no problem, it's already studied.
I'm on my way?
Hi Carlos, and thanks for all your answers, are very helpful.
My 1000uF caps are not more than a cm far from the chip. Yesterday I "dress" the power supply wires and with 5mV offset the chip got cool at low spl, but at medium got warm, but I think that this is because of the hot here. Now is sounding nice.
Nuuk, what is the 270K resistor in your buffer for?
BTW, I knew I forgot something when I started the thread:
I'd like to know if I could just biamp my speakers with the active filter built like the LPF a la Rasmussen. My idea is to buffer the input with 637 and then divide the signal to two channels (one GC each). My speakers have a first order slope for both drivers that could be translated to the chip pins easily, has this any problem?
I know the drivers well and the crossover point and slope is no problem, it's already studied.
I'm on my way?
Nuuk said:
Nuuk, you can remove the 270k resistor.
It's doing anything there.
You use a resistor in parallel with the output of a device (preamp, cd player, tuner, whatever) to reference the output to ground WHEN there's anything conected there.
In the case of your buffer, it's inside of your GC and always connected to the following circuit (the LM3875), so it's doing nothing and you can remove it.
The way you have it, your buffer will never be "floating", it will always be connected to the LM.
And, important too, the LM will never be without a source connected! (the buffer)
Aaaahhh,
That's why they (Elektor and such) put a 47K + 22K from each source to ground and take the input of the circuit from the 22K resistor. Casually I was last night thinking about this. So it's to avoid the amp to be without input, right?
I was thinking that the reason will be to parallel and cancel some impedance before it...
Nuuk,
What did you measure about the buffer?
That's why they (Elektor and such) put a 47K + 22K from each source to ground and take the input of the circuit from the 22K resistor. Casually I was last night thinking about this. So it's to avoid the amp to be without input, right?
I was thinking that the reason will be to parallel and cancel some impedance before it...
Nuuk,
What did you measure about the buffer?
Raka, with the PSU I am using now, I get around zero mV on the output and just a very brief rise when I power down the buffer.
Last week we had a power cut here, so both sections (buffer and LM3875) went off together and then powered up together about 5 seconds later. No damage was done although as a matter of good practise, I would always switch the buffer on first and off last.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.