Billshurv needs glasses. Sorry. You must need glasses.
I hit the link. First thing I saw was 3 pcb's marked XY.
I hit the link. First thing I saw was 3 pcb's marked XY.
The series is called Lillehammer, bit off the wall but quite funny.
Yes, you are right, but you have to do something else as well. Look at the red lines on the picture. This point is the earth reference for the feedback loop. This needs to be connected to the grounding point for the loudspeaker return current.
But DIY is not just to follow recipes, it is to learn something. Therefore, I recommend studying the following article:
http://www.hypex.nl/docs/papers/The%20G%20Word.pdf
If you understand the principle that Bruno discusses here, you also understand the fix I suggested for the XY-board. Better, you will be able to apply it yourself in your future projects.
Before this nit gets picked: this fix still not makes for a perfect board and technically I expect Toms board to be better. The reason is that upper and lower ground plane of the XY board form a capacitor, which I calculated to be roughly 33pF. That seems like a lot, but it is just 1/4 meg @ 20kHz, so nothing to worry about. But it doesn't win the beauty contest.
Last edited:
vacuphile
Thanx man!
Oh yes! Lillyhammer rules! Good show!
Will read the article.
And now I see what u mean by pcb's acting as caps.
A better layout (like Toms for instance) take routing on top and bottom of the PCB's in consideration as well as placement of components etc I guess.
Devil is in the details.
Thanx man!
Oh yes! Lillyhammer rules! Good show!
Will read the article.
And now I see what u mean by pcb's acting as caps.
A better layout (like Toms for instance) take routing on top and bottom of the PCB's in consideration as well as placement of components etc I guess.
Devil is in the details.
ebay being ebay, each time you click the link you get different results as new auctions appear. The latest auctions do clearly show 'XY' on the PCB, the others are too small without opening the auction. My mistake was assuming that 'XY' was a selling point and people would advertise them as such. Same mistake anyone not aware of them before reading this might make.
And yes there is a 3 board set BIN For $6.54. I have no idea how they can even post them for that little. Thank you for clearing that all up. Andrew could have course done it in a single post this morning.
And Koperberget: Yes I wear glasses. It's considered very bad form to make fun of someone when they have a limitation in one of their abilities. Viewing ebay pages on a mobile device when you have less than ideal eyesight can be a challenge, not something to make fun of.
And yes there is a 3 board set BIN For $6.54. I have no idea how they can even post them for that little. Thank you for clearing that all up. Andrew could have course done it in a single post this morning.
And Koperberget: Yes I wear glasses. It's considered very bad form to make fun of someone when they have a limitation in one of their abilities. Viewing ebay pages on a mobile device when you have less than ideal eyesight can be a challenge, not something to make fun of.
Me neither. I can't get the boards made for that. Never mind making any kind of profit that would pay the eBay and Paypal fees and leave me even a single red cent for my efforts.
+1
Tom
I could really not see that from here. And I did not know that u use glasses either. I did not make fun of you. Its an expression. I could call you daft or stupid for not noticing the XY logo, but clearly you aren't. So I didn't.
Sometimes I wish conversations like these were face to face. Little easier not to appear offensive and so on. Not easy to tell the true meanings of some written messages.
Sorry! Did not mean to offend anyone.
Agreed. The Signal Return and the Power Ground are two different circuits. Each has the be complete in it's own right.
The small pcb (561945) you have shown is slightly different from the bigger pcb (561946) shown later
You describe "See the red lines. You have to cut all the connections between ground of the attachment point of the 22 uF cap," this is exactly the same as shown in the next pic.
Why are you saying that the first pic is correct "cut all the connections" and in the next para you are saying "Now for the other mod I have seen: This is not advisable."?
It is exactly the same modification
Yes, the output side has to be referenced to the input side. That is what your red line shows.
or you can use HBRR (inserted into the red line connection) in one channel and HBRL in the other channel to attenuate the loop current created when two interconnects are linking the two power amplifier channels.
I can't understand why you say one is wrong and the other is right, when both modifications (cutting the three little bridge connections to the bottom plane) are the same !
Have you overlooked that at the Speaker Return terminal where the top plane is bridged into that return terminal?
Are we to cut those three bridge links as well?
Last edited:
So the conclusion is:
The LM3886 with white print + XY pcb performs not equally good as Done Right with genuine parts.
We suspect that the white LM3886 could be less good performing than a genuine part.
We haven't tested the XY amp with modified pcb neither with a white nor a genuine LM3886.
We haven't tested the white LM3886 with the Done right pcb.
I'll bet that the pcb layout is crucial to get top performance.
The LM3886 with white print + XY pcb performs not equally good as Done Right with genuine parts.
We suspect that the white LM3886 could be less good performing than a genuine part.
We haven't tested the XY amp with modified pcb neither with a white nor a genuine LM3886.
We haven't tested the white LM3886 with the Done right pcb.
I'll bet that the pcb layout is crucial to get top performance.
I approve of this message. That's basically my lesson learned from ten years of working in the industry on precision analog circuits. The layout is the circuit.
Tom
I work with rather young people and they don't realize that a milliohm here and a milliohm there is important. They simulate too much and regards copper as a pure connection which has no influence on the whole circuit.
Very true Peranders and very well delineated by Christiansen in his treatise to subject of harnessing the most out of the LM3886: Taming the LM3886 Chip Amplifier
I find reading and re-reading that document to be terribly important in fostering a more meaningful understanding on the vagaries of layout design. This is something I had not considered previously for precisely the reasons you ascribe.
Best,
Anand.
Thank you for high quality measurements on state of the art equipment. Interesting to see the rise in THD at about 4 volts output. Wonder what component is responsible for that rise. I have built to much veroboard circuits to blame the routing of the circuit at that low frequency/voltage/current. (1 amp at 5 volts is what USB can provide)
An inexpensive way of making hobbyist amp load for measuring is to buy 1000 100 Ohm 5 %,Then make 100 in parallell for 1 Ohm blocks. Use crimp connectors (old type car ones) to connect the 1 Ohm blocks. Then make parallell and series for the right load. Good for measuere up to 100 watts with ordinary homestudio soundcard. (Measuere over the last 1 Ohm block for voltage deviding and not destroing the soundcard input)
Good enough for down to -100 dB measurements.
I of cource asume everybody have access to a voltmeter to calibrate and can convert between volts,amperes and watts🙂
Regards Torgeir
An inexpensive way of making hobbyist amp load for measuring is to buy 1000 100 Ohm 5 %,Then make 100 in parallell for 1 Ohm blocks. Use crimp connectors (old type car ones) to connect the 1 Ohm blocks. Then make parallell and series for the right load. Good for measuere up to 100 watts with ordinary homestudio soundcard. (Measuere over the last 1 Ohm block for voltage deviding and not destroing the soundcard input)
Good enough for down to -100 dB measurements.
I of cource asume everybody have access to a voltmeter to calibrate and can convert between volts,amperes and watts🙂
Regards Torgeir
I bought 1000 1r0 1% 600mW metal film for about £10 including taxes and delivery.
I also bought 1000 20r 5W 5% for high power duties. and a few hundred of 10r 5W and 50r 5W
Multiple groups for higher power use works very well for many applications.
Yes, then you have an full measurement lab;-) Usefull for impedance measurements also.
The most difficult for low noise measurements when approaching -100 db, is often to get rid of hum. So using the recomended grounding techniqes in this thread is usefull.
And stop the measurement when you smell the smoke🙂
Well its getting there!
Connected all the dots now and the XY just went pop and buzzzz. He he.
Reconnected the cut traces for signal ground and it was fine.
Will cut them again as soon as the amp is in a chassis with proper ground.
Guess a separate ground cable must go from the point of cut traces?
Anyway - it sounds great! Turned the volume to max whit no input signal - no noise. Got <0,6 mv dc on output.
Gonna compare it to the Juanjing gainclone when everything is in place.
Connected all the dots now and the XY just went pop and buzzzz. He he.
Reconnected the cut traces for signal ground and it was fine.
Will cut them again as soon as the amp is in a chassis with proper ground.
Guess a separate ground cable must go from the point of cut traces?
Anyway - it sounds great! Turned the volume to max whit no input signal - no noise. Got <0,6 mv dc on output.
Gonna compare it to the Juanjing gainclone when everything is in place.