Thank-you for your comments, Xelb.
Do you think that the layout of the board is the cause of the oscillation? Is it component selection? Could it be the 100k pot rather than using a 10 k? Could it be the crappy speakers with uncertain impedance?
(I do not have a symmetrical board because I was trying to get the big capacitors and the feedback resistors as close to the correct LM3886 pins as physically possible. Carlos made a big deal out of having the supply capacitors and their snubbers really close, and others have made similar comments about the feedback resistors. )
I presume you have a jumper between the Pins 1 and 5, because there appears to be no supply V+ supply to pin 1. I do like the overall layout style of your board. I will have a little think to see if I can make it all fit 4 similar boards along a 30 cm heat sink.
If you think it is the board as the main problem, I guess I will have to change it. 🙁
Your input is appreciated, and I will cogitate and formulate new plans,
Regards,
George.
Do you think that the layout of the board is the cause of the oscillation? Is it component selection? Could it be the 100k pot rather than using a 10 k? Could it be the crappy speakers with uncertain impedance?
(I do not have a symmetrical board because I was trying to get the big capacitors and the feedback resistors as close to the correct LM3886 pins as physically possible. Carlos made a big deal out of having the supply capacitors and their snubbers really close, and others have made similar comments about the feedback resistors. )
I presume you have a jumper between the Pins 1 and 5, because there appears to be no supply V+ supply to pin 1. I do like the overall layout style of your board. I will have a little think to see if I can make it all fit 4 similar boards along a 30 cm heat sink.
If you think it is the board as the main problem, I guess I will have to change it. 🙁
Your input is appreciated, and I will cogitate and formulate new plans,
Regards,
George.
Another quick though
When I look closely at the LM3886 package, there appers to be little copper ends appearing through the plastic at various points. Are these structural, or do they have some internal connection?
If they have some internal connection, then perhaps it is the way I am mounting the package onto the heatsink.
I doubt it, but always worth asking.
Regards,
George.
When I look closely at the LM3886 package, there appers to be little copper ends appearing through the plastic at various points. Are these structural, or do they have some internal connection?
If they have some internal connection, then perhaps it is the way I am mounting the package onto the heatsink.
I doubt it, but always worth asking.
Regards,
George.
Oscillation and DC offset
Hello Carlos.
I appreciate hearing from you.
No, the chip is NOT isolated from the heat-sink, but I have used the LM3886TF version, which I thought was supposed to be insulated electrically. I use white paste from an old kit to thermally join the chip to the sink. Also, at present the heatsink is just sitting on a piece of chip board and nothing else is connected to it ... (at least I think nothing else is connected to it!)
Pics coming in a bit, if I can get my son's digital camera!
Regards,
George.
Hello Carlos.
I appreciate hearing from you.
No, the chip is NOT isolated from the heat-sink, but I have used the LM3886TF version, which I thought was supposed to be insulated electrically. I use white paste from an old kit to thermally join the chip to the sink. Also, at present the heatsink is just sitting on a piece of chip board and nothing else is connected to it ... (at least I think nothing else is connected to it!)
Pics coming in a bit, if I can get my son's digital camera!
Regards,
George.
Oscillations continued
Thank-you for your patience everyone.
Carlos, as requested, I have put some pictures of my effort on my homepage. (Rather than clogging up this board with pictures of failures, I have put them on my website, which I have just modified for the purpose. That is why it has taken so long. I have rarely done this before either!)
Pictures of my oscillating LM3886 Amp
I hope this sheds some light.
Regards,
George.
Thank-you for your patience everyone.
Carlos, as requested, I have put some pictures of my effort on my homepage. (Rather than clogging up this board with pictures of failures, I have put them on my website, which I have just modified for the purpose. That is why it has taken so long. I have rarely done this before either!)
Pictures of my oscillating LM3886 Amp
I hope this sheds some light.
Regards,
George.
Hi George,
your pics don't show clearly the presence of 300 pF cap between pins 9 and 10 of the LM3886 chip. Check that, and if you don't have it, add one. If you have it, try another, bigger one (but don't go over 1 nF). That should stop the oscillations. Also, try to make those wires shorter. You don't want antennae (especialy in input signal part of the circuit - potentiometer and input wires should be short as possible and twisted together).
Cheers
your pics don't show clearly the presence of 300 pF cap between pins 9 and 10 of the LM3886 chip. Check that, and if you don't have it, add one. If you have it, try another, bigger one (but don't go over 1 nF). That should stop the oscillations. Also, try to make those wires shorter. You don't want antennae (especialy in input signal part of the circuit - potentiometer and input wires should be short as possible and twisted together).
Cheers
Oscillations
Dear Juma,
The 300pF capacitor is the little iridescent blue shiny blob seen in the bottom right of the "bottom view". In fact, it is the reason I built a second copy of this amp, just to make sure I hadn't stuffed up the soldering on my first attempt.
Your thoughts about antennae and loops are valid. I was hoping not to have to be TOO careful, given this is just a mock-up. Also the "antennae" must have changed positions in the weeks between my two similar failed efforts. But if that is what it takes, off I go to fiddle some more .... 🙂
Thank-you for your input.
Regards,
George.
Dear Juma,
The 300pF capacitor is the little iridescent blue shiny blob seen in the bottom right of the "bottom view". In fact, it is the reason I built a second copy of this amp, just to make sure I hadn't stuffed up the soldering on my first attempt.
Your thoughts about antennae and loops are valid. I was hoping not to have to be TOO careful, given this is just a mock-up. Also the "antennae" must have changed positions in the weeks between my two similar failed efforts. But if that is what it takes, off I go to fiddle some more .... 🙂
Thank-you for your input.
Regards,
George.
Yes, that's right.
But as all in life, you have to make compromises...
I also share that opinion but I have a different approach.
I prefer small capacitance capacitors because they are faster and better in high frequencies.
And I imagine the copper layer as a hydraulic system, you should never concentrate the pressure(read it energy) in just one place... you will stress it 😉
I also have 380uF near the chip.
( I use 6x 380uF and 2x 1000uF )
XELB said:
But they don't have enough reserve and tend to slow the lower frequencies.
The snubber deal with your worries.
When you parallel too many small caps you will end with high (trace) inductance, which is not good for... the high frequencies. 😀
This has been discussed so many times...🙄
It is possible to put 2x2,200uF caps near the chip, along with bypasses and snubbers. I don't see where's the problem.
George, you must revise all those solder joints.
And a little more practice on that art would not hurt.
http://www.stolenantiques.bravehost.com/HiFiPhotos/LM3886Conns.jpg
And a little more practice on that art would not hurt.
http://www.stolenantiques.bravehost.com/HiFiPhotos/LM3886Conns.jpg
carlosfm said:
The problem is that I prefer the small caps.... And I tried the snubber 😉
But I also reduce the ESR and the inductance is very low.
BTW, I just use near the chip the enough capacitance for my sonic taste.
I never put big caps near the chip or in the chip board, these ones are always near but in a different board.
PS: It's always a question of sonic taste, sound source, speakers, etc...
BTW, what chips to you recommend for a good DIY DAC ?
I was thinking in the Wolfson WM8740 or the Analog Devices AD1955...
XELB said:
AD1862. 😉
Unfortunately long discontinued, but a very very good multibit dac, which always sounds better to me than all those delta-sigma dacs.
oscillation
Thank-you helpers.
I will re-solder the board today. Carlos, looking at the joints to the capacitor leads in that close-up photograph do look a bit scuzzy. I thought the connections to the LM3886 chip looked pretty good though!?? What do you think about those? (Pins 8 and 10 are soldered underneath the board.) I don't want to cook it unnecessarily.
(It is a bit daunting putting your work up for world-wide assessment.
)
Regards,
George
Thank-you helpers.
I will re-solder the board today. Carlos, looking at the joints to the capacitor leads in that close-up photograph do look a bit scuzzy. I thought the connections to the LM3886 chip looked pretty good though!?? What do you think about those? (Pins 8 and 10 are soldered underneath the board.) I don't want to cook it unnecessarily.
(It is a bit daunting putting your work up for world-wide assessment.
)Regards,
George
Re: oscillation
Just resolder everything.
I also see a resistor without any solder on the top layer... it should be soldered on both sides.
Aahhh, worry not, George.
This is public service.
Someone someday will find his/her solution for the same problem here.
In that case, you are helping someone too.
GeorgeBoles said:
Just resolder everything.
I also see a resistor without any solder on the top layer... it should be soldered on both sides.
GeorgeBoles said:(It is a bit daunting putting your work up for world-wide assessment.
Aahhh, worry not, George.
This is public service.
Someone someday will find his/her solution for the same problem here.
In that case, you are helping someone too.
carlosfm said:
And where could I find the AD1862 chip ?
I was thinking in 2 DAC's per channel ( one stereo chip per channel) and a ballanced output with AD827.
The ideia is to make a DAC/PRE running on batteries 😀
This way I would retire my SSM2142 PRE ( single ended to Differential )
XELB said:
I don't know.
I have a pair, but they are in use. 😀
XELB said:
The AD1862 is mono, one chip per channel.
Oscillations ... cured, I think
Well, thank-you everybody.
For now, I say tentatively that the oscillations have gone.
I think Carlos was on the money.
I put on some magnifying loupes so I could see properly what was going on with the solders. Many only went about 180 degrees around the leads, which I thought would be OK, but I re-did them. I also double checked the3uF and 300pF capacitors with some extra solder ... I think they were the most suspicious, and even before I read Carlos tip about soldering the top surface of the resistor leads, I had done that.
And lo and behold, no oscillations that I can detect at normal listening levels on the scope. I just wonder if there might be some really low level scuzz showing up when the pot is set so that it is nearly short to ground ... but perhaps this is noise rather than something oscillating. Certainly she only gets hot when driven just off clipping for a minute or two, and then stays there without cutting out.
I also re-arranged the antennae on Juma's advice (and poked my tongue out a different way.) Unfortunately I stall cannot get Television Channel 10 🙁 .
I will let you know if it falls in a heap again!
Regards,
George.
PS I am still toying with trying a "ground plane" type PCB design for my next channel ... to see if that low level scuzz clears up.
Well, thank-you everybody.
For now, I say tentatively that the oscillations have gone.
I think Carlos was on the money.
I put on some magnifying loupes so I could see properly what was going on with the solders. Many only went about 180 degrees around the leads, which I thought would be OK, but I re-did them. I also double checked the3uF and 300pF capacitors with some extra solder ... I think they were the most suspicious, and even before I read Carlos tip about soldering the top surface of the resistor leads, I had done that.
And lo and behold, no oscillations that I can detect at normal listening levels on the scope. I just wonder if there might be some really low level scuzz showing up when the pot is set so that it is nearly short to ground ... but perhaps this is noise rather than something oscillating. Certainly she only gets hot when driven just off clipping for a minute or two, and then stays there without cutting out.
I also re-arranged the antennae on Juma's advice (and poked my tongue out a different way.) Unfortunately I stall cannot get Television Channel 10 🙁 .
I will let you know if it falls in a heap again!
Regards,
George.
PS I am still toying with trying a "ground plane" type PCB design for my next channel ... to see if that low level scuzz clears up.
Hi Carlos,
I finally got my hands on pair of LT1083 regulators (they are a bit expensive, but they handle more current, have lower voltage drop - important to me, and they seem to have better ripple rejection). I was not impressed with LM338 compared to your unregulated snubbered PS. You recommended 0R47+47uF as a snubber for LM338, but I doubt that the same would be the optimal for LT1083 (they don't give output impedance graph for 1083 in datasheet but from ripple rejection point of view it is obvious that 1083 does better than 338 - rejection is stronger and the curve doesn't fall so rapidly as frequency rises).
What would you recommend as a snubber for LT1083 ?
Also, I noticed on the other thread (don't hold it against me - after all it's done to improve complete implementation of a NI chipamp - PS and preamp are unavoidable parts of it) your new design - the UNO (single 2SK170 preamp - http://myweb.tiscali.co.uk/nuukspot/decdun/gainclonepre2.html).
How would you compare it to your AD815 pre ?
Thanks !
I finally got my hands on pair of LT1083 regulators (they are a bit expensive, but they handle more current, have lower voltage drop - important to me, and they seem to have better ripple rejection). I was not impressed with LM338 compared to your unregulated snubbered PS. You recommended 0R47+47uF as a snubber for LM338, but I doubt that the same would be the optimal for LT1083 (they don't give output impedance graph for 1083 in datasheet but from ripple rejection point of view it is obvious that 1083 does better than 338 - rejection is stronger and the curve doesn't fall so rapidly as frequency rises).
What would you recommend as a snubber for LT1083 ?
Also, I noticed on the other thread (don't hold it against me - after all it's done to improve complete implementation of a NI chipamp - PS and preamp are unavoidable parts of it) your new design - the UNO (single 2SK170 preamp - http://myweb.tiscali.co.uk/nuukspot/decdun/gainclonepre2.html).
How would you compare it to your AD815 pre ?
Thanks !
juma said:
Without the impedance curves, there is no basis for making a calculation.
Start with the same values as for the LM338 and then try a lower cab value, like 22uF.
juma said:
Juma, I'm not going to make judgements this time.
The AD815 works better for my needs (on my main system), but both are very good.
What is better you will only find out if you build both.