Grega,
You will want to make the star ground point at the grounds of the smoothing caps in the power supply, probably on a stub trace protruding off of the caps' ground trace.
Therefore, on this board you will need a completely separate ground for the input signal ground, not connected to any power ground on this board, and with its own wire running back to the star ground on the power supply board. And in that case the grounds of the input RCA connectors for the whole unit will need to be isolated from the case and chassis, with their grounds only connected on these boards.
Alternatively, you could connect the RCA grounds together where the inputs enter the case (bot don't connect them to the case or chassis) and run one wire from between them to the star ground point, and then not have a signal ground wire from this board to the star ground.
In any case, the input signal ground on this board should not be connected to the power ground on this board.
You will want to make the star ground point at the grounds of the smoothing caps in the power supply, probably on a stub trace protruding off of the caps' ground trace.
Therefore, on this board you will need a completely separate ground for the input signal ground, not connected to any power ground on this board, and with its own wire running back to the star ground on the power supply board. And in that case the grounds of the input RCA connectors for the whole unit will need to be isolated from the case and chassis, with their grounds only connected on these boards.
Alternatively, you could connect the RCA grounds together where the inputs enter the case (bot don't connect them to the case or chassis) and run one wire from between them to the star ground point, and then not have a signal ground wire from this board to the star ground.
In any case, the input signal ground on this board should not be connected to the power ground on this board.
At post 27, connect Signal Star ground to Power Star ground with a 2.2 ohm resistor. . . because it is the small value resistor or a long thin trace that manufacturers Signal Star ground. While we might not keep the resistor, it is vital for assisting layout.
Let's take a look about what we know about the speaker cable. The same force, current, strength exists on speaker + as speaker - because if we disconnect either conductor, all of the current and signal is gone.
We can assume that there is a fiercely strong AC music signal on both speaker wires. If that gets into the the small signal ground, there is a high current noise bomb. The groundlift resistor (bright blue 2.2 ohms on my example) removes some of this current and none of the noise. However, the groundlift resistor "weakens" the extraneous current thus robbing the noise of its power.
Basically, this little 2.2 ohm resistor prevents bad layout and it gives you options:
1). Jumper for a more forwards sound to help dull speakers
2). Use 2.2 ohm groundlift on the amp board, for expected operation if the amp isn't a parallel amp.
3). Omitting the resistor gives you the option of cabling the signal star ground point to a centerpoint location via equally short length cables for the "master signal star" needed by parallel amplifiers, to facilitate hookup of the RCA jack and potentiometer. The resistor would also move to that location (master signal star) so that there is only one. One cable goes from this one 2.2 ohm resistor to 0v.
P.S.
How many speaker return cables are needed from running 3 amplifier boards as a parallel amplifier?
Only one cable--speaker return from the speaker jack goes to 0v output tap of power supply board.
Let's take a look about what we know about the speaker cable. The same force, current, strength exists on speaker + as speaker - because if we disconnect either conductor, all of the current and signal is gone.
We can assume that there is a fiercely strong AC music signal on both speaker wires. If that gets into the the small signal ground, there is a high current noise bomb. The groundlift resistor (bright blue 2.2 ohms on my example) removes some of this current and none of the noise. However, the groundlift resistor "weakens" the extraneous current thus robbing the noise of its power.
Basically, this little 2.2 ohm resistor prevents bad layout and it gives you options:
1). Jumper for a more forwards sound to help dull speakers
2). Use 2.2 ohm groundlift on the amp board, for expected operation if the amp isn't a parallel amp.
3). Omitting the resistor gives you the option of cabling the signal star ground point to a centerpoint location via equally short length cables for the "master signal star" needed by parallel amplifiers, to facilitate hookup of the RCA jack and potentiometer. The resistor would also move to that location (master signal star) so that there is only one. One cable goes from this one 2.2 ohm resistor to 0v.
P.S.
How many speaker return cables are needed from running 3 amplifier boards as a parallel amplifier?
Only one cable--speaker return from the speaker jack goes to 0v output tap of power supply board.
Last edited:
The problem would come from using a signal-ground reference for the input resistor that is connected to a conductor that runs back to the power supply with currents on it that are not from the signal ground input.
Even assuming that the speaker ground return is not involved and goes back to the star ground in the power supply separately from everything else, we'd still have the decoupling caps' ground return currents. They will induce significant voltages across the inductance and resistance of the wire from the board to the power supply ground, which will appear... back at the amp board's now-bouncing "ground". That would make the ground end of the signal input resistor have a bouncing voltage. That bouncing voltage would effectively be directly arithmetically summed with the differential input signal voltage across the amplifier input pins. That would be "a BAD thing".
There should be one and only one "star ground". Audio ground should not be connected to power ground in any way, except at the single star ground point, back at the power supply. To do otherwise would result in some amount of noise and or hum being added to the amplifier input signal, and would also result in a "sneak feedback path" that could cause anything from more distortion to oscillation and chipamp destruction.
Even assuming that the speaker ground return is not involved and goes back to the star ground in the power supply separately from everything else, we'd still have the decoupling caps' ground return currents. They will induce significant voltages across the inductance and resistance of the wire from the board to the power supply ground, which will appear... back at the amp board's now-bouncing "ground". That would make the ground end of the signal input resistor have a bouncing voltage. That bouncing voltage would effectively be directly arithmetically summed with the differential input signal voltage across the amplifier input pins. That would be "a BAD thing".
There should be one and only one "star ground". Audio ground should not be connected to power ground in any way, except at the single star ground point, back at the power supply. To do otherwise would result in some amount of noise and or hum being added to the amplifier input signal, and would also result in a "sneak feedback path" that could cause anything from more distortion to oscillation and chipamp destruction.
Last edited:
No it doesn't.
The spkr return cable runs twisted with the spkr feed cable all the way back to the amplifier. Then the spkr return runs twisted with the Power Ground wire to the Main Audio ground.
The spkr return wire does not go to the PSU.
I don't know what this means, but, if I read correctly the caps on the input need their values amending I think. I don't know what RC and F-3db means but I will try to read up and understand
OK - will do
sorry gootee, what is a stub trace?
ack that
I interpret this as being the speaker ground is the same as the input ground. correct?
sorry for the questions but some of these is well above me.
I have no choice other than to agree, because of the parameters. . .
The thread title says "triple parallel" which is a power boost arrangement and extra power could not be useful with hard sound results of returning speaker current into the amplifier board ground plane.
However, the thread title also says "dynamics amplifier" meaning that the amplifier would be lively and therefore it is not productive to return the speaker voltage to the 0v tap of the power board.
Result:
Any option that is either an even balance or better than both is a welcome improvement.
Estimation table, bare minimum NFB cap size, sized small enough for some mild "warm sound" distortion, so let's not do worse.
1uF input cap, 3.3k feedback shunt, 100uF NFB cap
1uF input cap, 2.7k feedback shunt, 150uF NFB cap
1uF input cap, 2.2k feedback shunt, 220uF NFB cap
2uF input cap, 3.3k feedback shunt, 150uF NFB cap
2uF input cap, 2.7k feedback shunt, 220uF NFB cap
2uF input cap, 2.2k feedback shunt, 270uF NFB cap
2uF input cap, 1.5k feedback shunt, 330uF NFB cap
If, instead, you'd like to hear some undistorted extra low pitches plenty loud then first allow more bass to pass the NFB cap with either a larger feedback shunt resistor value and/or a larger value NFB cap, and then after that, you could use a larger input cap. It is funny with LM1875 because with extra low tuning is like a house cat with the voice of a lion, so much the opposite of the table above.
1uF input cap, 3.3k feedback shunt, 100uF NFB cap
1uF input cap, 2.7k feedback shunt, 150uF NFB cap
1uF input cap, 2.2k feedback shunt, 220uF NFB cap
2uF input cap, 3.3k feedback shunt, 150uF NFB cap
2uF input cap, 2.7k feedback shunt, 220uF NFB cap
2uF input cap, 2.2k feedback shunt, 270uF NFB cap
2uF input cap, 1.5k feedback shunt, 330uF NFB cap
If, instead, you'd like to hear some undistorted extra low pitches plenty loud then first allow more bass to pass the NFB cap with either a larger feedback shunt resistor value and/or a larger value NFB cap, and then after that, you could use a larger input cap. It is funny with LM1875 because with extra low tuning is like a house cat with the voice of a lion, so much the opposite of the table above.
Last edited:
I hope not. The speaker return doesn't really belong at inputs, because that makes fiercely loud mids. Got baxandall? If you were going to use a baxandall anyway, well the dynamics of bouncing the amplifier's inputs about can be fantastic although the tone would need fixed a lot. I wouldn't do it that way.
My views on the matter are in the sketch in post #71 above, which is a practical triple parallel on a single board, so easily done and very compact. See the speaker and power on the farthest left side, but the small signal input on the farthest right side?
For a parallel amplifier, I am finding that the discussion of using 1 chip per each board, and then "stacking" was a great deal more confusing and more complicated. That got me lost too.
However, your circuits and layouts have improved dramatically in a very short time. Kudos!
Last edited:
a filter with just an RC or a CR pair is a single pole 6dB/octave slope low pass, or high pass filter.
The F-3dB is the frequency that the response is down by 3dB from the passband level. The F-3dB = 1 / [2piRC]
Sometimes rather than quoting the frequency, Members and others, will quote the filter time constant. The RC of a filter is that time constant.
eg. FM radio has a 50us or 75us filter.
Vinyl recording RIAA is a combination of RC filters of 3180us, 318us, 75us
Yes, one of those filters is the same as FM radio.
The RC is simply the resistance in ohms multiplied by the capacitance in Farads with the result in seconds (s)
eg 7500ohms (7k5) and 0.00000001F (10nF) {that should read 7 zeros after the leading zero.} gives 0.000075s (75us)
F-3dB = 1 / [2 * 3.14159 * 7500 * 0.00000001] = 1/2/3.14/75/10^-6 = That 75 & 10^-6 is 75us = 2122Hz (2k122Hz)
Since RC is already in the F-3dB equation, you can swap between Hz and seconds by using the factor 6.28 (2PI) and taking the reciprocal (1/X on the calculator) 2122 Hz * 6.28 = 13326. Reciprocal or Invert (1/X) that to get 75us.
And it works the other way. Take 3180us*6.28 and then 1/X = 50.05Hz
Your schematic is correct. The Signal Return (Signal Ground) is separate from the Main Audio Ground (MAG). The Speaker Return must go to the MAG. The SPKR Return must not go to the Signal Return.
Last edited:
OK
I've heard of baxandall, but, don't know what it is. Will change that
watch this space. i like your layout dan - lets see if i can board it
most of this is confusing, i'm getting it squared off in my head which is important - and a great learning experience - so thanks to all
thanks dan
cool - thanks andrew - thats great.
before reading that I didn't know what RC or CR is. do now
understood - time constant vs freq- horses for courses
i have a question (or three)
so we have two grounds - signal audio return and main audio ground.
is signal audio return the line level input negative? or the speaker negative?
and is main audio ground, the 0v in the 18-0-18 transformer out?
interestingly i built a board for something to do yesterday, based on the layout a few posts back (post 77) and proceeded to fry my test speaker - as i had +26vdc on the line out
not sure whether i didn't connect it correctly but, i reviewed my layout and my board and while it looks fine, i know it doesn't work. one thing of note is that the line input ground and the 0v ground (from trafo) are not connected (kept separate) - in previous builds they have been tied together and have worked fine - hence my questions on the ground(s)
many thanks all.
Sorry. AndrewT had it correct the first time. On a calculator, you enter 1 and then divide by each denominator factor, in any order.
The way you wrote it, you would HAVE to enclose the 2*3.14 in parentheses, since the evaluation order is simply left to right, otherwise. But writing it as you did and enclosing it in parentheses would be less efficient, when using a calculator, than AndrewT's version.
The way you wrote it gives 20953.95 instead of 2122.066.
Last edited:
My own test speaker has a 3,300uF series cap. This provides the opportunity to measure for DC offset without the consequences to the speaker.
AndrewT will soon mention the light bulb tester, a safety current drop test, so do look that one up.
Sometimes you'll get a rail output like that because it is a typical chip amp failure mode. Most prone to that are DC coupled amplifiers with authentic and also most fakes on any design, since that whole situation is more fragile and should have lower voltage--derating is the #1 major concern in such circumstances.
dan - thats cool. the unexpected operation = 26VDC on the speaker output - when you tie the input ground and power supply ground together, the amp (that single i built) performs flawlessly FYI
AndrewT added inverted diodes either side for protection which I will also add.
gootee - thanks. I used Excel to setup a little spreadsheet to calculate the values based on the formula - get the exact same result as what AndrewT got.
AndrewT added inverted diodes either side for protection which I will also add.
gootee - thanks. I used Excel to setup a little spreadsheet to calculate the values based on the formula - get the exact same result as what AndrewT got.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.