To be precise, gain = R28 / R29 + 1
In this particular case, 15k / 0.56k + 1 = 27.8 (times)
20 * log(27.8) = 28.9 db
Cheers,
Valery
Yes this is true, I offered the simplest - though not exact - explanation.
@Terry - The '20 log (Av)', where Av is the voltage gain, converts the number from a linear scale into decibels of voltage gain, which is a logarithmic scale. For instance 0dB of gain is what we call unity or a gain of 1, and 20bB of voltage gain is a gain of 10, and 40dB of voltage gain is a gain of 100. For every 20dB of voltage gain the 'linear' gain is multiplied by a factor of 10.
Taking the 'Voltage Gain' as calculated by:
(Feedback Resistor/Shunt Resistor)+1=Av (aka 'Voltage Gain')
Then taking the base10 logarithm multiplied by 20 gives us the voltage gain in decibels. Sometimes thinking of things that change geometrically isn't always natural, but it allows us to speak of very large or small quantities with more manageable numbers. I hope that helps more than it clouds the concept.
Hi Terry, log is "logarithm" - I put a link to wikipedia here - Logarithm - Wikipedia, the free encyclopedia - they explain mathematics better than I would do
With this firmula you convert "times" (linear scale) to "decibels" (logarithmic scale) - in most cases decibels are more convenient.
Your calculation for 22k is correct - that's the gain you had (equal to 32.1 db).
De-facto standard gain for the power amp (also THX standard for home theaters gear) is 29 db. So 15k resistor brings you very close to it.
Cheers,
Valery
OK, wrote simultaneously with Jason
Just a few examples, letting you convert more easily.
You normally add "decibels", but multiply "times".
6 db means "multiply times by 2" = twice as much
20 db = 10 times more
Then you can combine those:
40 db = 20 + 20 db = 10 * 10 times = 100 times
26 db = 20 + 6 db = 10 * 2 times = 20 times
and so on...
Last edited:
Hi Valery and Jason,
Thanks for the detailed explanation. I looked up the 20*log(Av) and realized I have the log function on my calculator so I was able to do the math. So, with the 22k in R28 I get this.
22k/560+1=40.28
20*log(40.28)=32.1db
That doesn't seem too bad but when looking at the scope on the high gain IPS it latches up at clipping. I can't tell if it is doing that now that I lowered the gain because my sine wave generator only puts out 1.2v and that isn't enough to make the amp clip.
Thanks, Terry
Thanks for the detailed explanation. I looked up the 20*log(Av) and realized I have the log function on my calculator so I was able to do the math. So, with the 22k in R28 I get this.
22k/560+1=40.28
20*log(40.28)=32.1db
That doesn't seem too bad but when looking at the scope on the high gain IPS it latches up at clipping. I can't tell if it is doing that now that I lowered the gain because my sine wave generator only puts out 1.2v and that isn't enough to make the amp clip.
Thanks, Terry
jarkaa, you will have, may-be to try various solutions with the input ground, depending the way the board is designed.
If the input stage has an isolated ground (with a resistance) you can try the star point for grounds from input plugs and input stages. (A strap between the two plug grounds, and only one wire between-it and the star point).. Use a shielded cable from the plug and the boards, shield connected only on the plugs...
Or... so many solutions ;-)
If the input stage has an isolated ground (with a resistance) you can try the star point for grounds from input plugs and input stages. (A strap between the two plug grounds, and only one wire between-it and the star point).. Use a shielded cable from the plug and the boards, shield connected only on the plugs...
Or... so many solutions ;-)
Hi Esperado,i believe all these would be more clear if you can post some draft schematics instead of many words.
Hi,
I am getting frustrated with this. I have had problems with grounding before but nothing like this.
The following attachments describe by latest tries without success.
The way I see this it is impossible to have a ground loop with proposed layouts.
My IPS boards have the signal return at the same point as input gound as BV noted few posts earlier.
Thanks for your help so far.
J
I am getting frustrated with this. I have had problems with grounding before but nothing like this.
The following attachments describe by latest tries without success.
The way I see this it is impossible to have a ground loop with proposed layouts.
My IPS boards have the signal return at the same point as input gound as BV noted few posts earlier.
Thanks for your help so far.
J
Attachments
Did-you ensure one of your amps is not oscillating ?
Did-you ensure all the the rails are clean enough ?
If no to the previous questions, is-it quiet when you shortcut the inputs, unconnected to other equipment ?
Still quiet if you connect just the ground between your input and the preamplifier ?
Quiet if only one side is connected to the preamplifier ?
Are-you sure your signal from preamp is hum-free ?
Don't be frustrated, there is always a solution.
This is NOT the board that used the ground plane. I repeat NOT, the board that used the ground plane. That board seriously failed and suffered from TERRIBLE, I repeat TERRIBLE, induced distortion. Why this was so I don't really know as the high quality and dirty grounds were still separated. So here's your warning, do not bother with a ground plane type design for a power amplifier. It is not needed and all it does is give you a headache. If you don't have the capabilities to measure what you're doing properly then there's more reason to simply stick to what we know works.
NO GROUND PLANES!
Just so no one misses it!
Where is the layout for that board? I'd like to study it and find out why it had such bad results. Applying ground planes in an audio context is going to take some translation because most ground plane advice is geared towards RF where the resistance of the plane itself is insignificant. The behavior of a ground plane has to be understood and adapted for audio use, which means developing a different and possibly contradicting set of rules.
I will also point out that you are already using a ground plane, it's the heatsink! And no, it's not too far away to have an effect. Look at figures 7 and 8 here. This is a wire 2cm above a plane!
Fundamentals of EMC Design: Our Products Are Trying To Help Us
So, why doesn't the heatsink cause the same problems? Maybe it does, or maybe the heatsink just doesn't have more than one connection to the PCB? If the heatsink does cause induced distortion, then the solution would be a partitioned ground plane on the PCB next to the heatsink, to short the magnetic fields in a benign way before they reach the heatsink.
Last edited:
+1.
In fact, better avoid any big current from Rails and output stage. It can mean a 3 layer board if it is impossible to design power ground tracks on the other side.
As far i'm in concern, i prefer to design my boards without, assuming i'm aware to understand what's happen with currents and evils. Quite often, i have two parallel ground tracks aside ;-)
I remember, in the 70s, we reduced by a factor of 10 the distortion of a power amp, just by taking the feedback point right at the Speaker + instead of somewhere else between the two power transistor's emitters and making the tracks equal on the two sides.
Little details can change a lot.
There is a well written article here. I have used the loop breaker in figure 4. On one amp I needed to remove the 10R resistor to kill the hum. It uses an inexpensive 30A bridge and mounts easily to the case. It goes in place of the 10R in the above drawing.
EDIT: Something I meant to mention. Make sure that the source and amp are both plugged into the same outlet. I took me a while to realize that my work shop has two separate circuits. They must be from two separate sides of the panel because in some cases it will cause a hum if the pre and amp ate plugged into separate outlets.
EDIT: Something I meant to mention. Make sure that the source and amp are both plugged into the same outlet. I took me a while to realize that my work shop has two separate circuits. They must be from two separate sides of the panel because in some cases it will cause a hum if the pre and amp ate plugged into separate outlets.
Attachments
Last edited:
Well indeed, the first arrangement is the better idea. If you have things connected like in the second one then there is no connection to ground for the input stage except through the 10R resistor when there are no cables plugged in. If the 10R resistor wasn't there the amplifier could explode. In the first example the input stage always has an excellent connection back to a clean ground.
I agree though, I cannot see where a ground loop could being created. You say the amplifiers are completely silent when the input cables are disconnected and the problem is only there when they are plugged in?
I agree though, I cannot see where a ground loop could being created. You say the amplifiers are completely silent when the input cables are disconnected and the problem is only there when they are plugged in?
Yes, the same as if you forget any other important component or wire.. It simply must be connected, it is innevitable condition, no mistakes allowed.. And here is nothing such as "clean" ground..
10R resistors should decrease circulating ground currents, and so decrease voltage drop at signal cable shield non zero resistance (which is added than to signal in input as hum and noise), If you use those resistors, voltage drop at sheld resistance is divided in ratio (10R/shield resitance), if NFB ground arm is connected to input ground. Imagine ground loop, if you connect signal cable in both inputs and in signal source. And grounds for both channels are also connected at PSU star point..You disconnect input cable, ground loop is away.. 10R are needed, if you use suboptimal arrangement with common PSU for both channels and unbalanced connection form signal source to amp..
It is recomended for safety (to avoid burned 10R..) to bypass 10R resistors with two antiparalelled diodes, rated at least at 1A.
Hello Esperado and others.
Oscillation is the first suspect in this case I need to do more careful examination on this.
Rails are clean enough, very low ripple at main caps.
Even with a long rca connected shorting the inputs makes the hum go away completely.
unplugging rca wires kills the hum completely.
If I plug just rca cable into a channel the hum starts. It does not have to be connected to anything. Just a cable is enough for initiating the hum.
Preamp is hum free (Classé ssp-50).
The case earth is exactly how it is described in bonsai´s document. A resistor together with a hefty bridge rectifier.
I also tried removing the psu to chassis connection for testing purposes -> did not help
It seems that I need to take the amp into pieces once again and start from the scratch. When I tested the IPS and OPS separately I noticed nothing special.
The way BV presented the grounding was my first version.
Oscillation is the first suspect in this case I need to do more careful examination on this. I also need to recheck all the components...
Thanks for your help and support.
J
Last edited:
I don't have the board any more, nor it's CAD design as I modified it to do away with the ground plane. Needless to say it wasn't exactly very complicated. The clean signal ground covered beneath the input stage and only had the clean stuff connected to it. The dirty decoupler ground was poured beneath the output stage. It is possible that a component was defective within this board as I had problems setting it up and problems managing the bias, every component I tested mind you checked out fine and all transistors did too. The amplifier performed well without a load connected, but was terrible with a load connected in a way that screamed extremely bad rail induction distortion.
Of course that is the idea behind the isolation circuit as the 10R resistor was catch fire if the chassis went live, but it wont solve the ground loop problem.
If you wire the amplifier up like I proposed with my drawing you do not need a separate power supply for each channel, nor do you need 10R isolation resistors anywhere within the design (except for the safety earth which I consider a different problem). A loop is also not created by connecting the signal source to the amplifier.