I just had a look at the datasheet.
If your input cap and R is 2u2 and 22K, shouldn't you include the 1M resistor aswell. It seems that you used the datasheet schematic. Why not use all of it?
Sorry it's Friday so my head can't seem to remember all the calcs.
Maybe one of the old hands can help out.
If your input cap and R is 2u2 and 22K, shouldn't you include the 1M resistor aswell. It seems that you used the datasheet schematic. Why not use all of it?
Sorry it's Friday so my head can't seem to remember all the calcs.
Maybe one of the old hands can help out.
Can you get/solder SMD Resistors? It's easy to solder, you just get a pair of tweezers and wet one of the pads, place the resistor and then solder the other side.
Doing this you can use a smd Res for the feedback resistors (R3 and R5. Reducing a lot of track and freeing more space to get the caps close to their pins.
Doing this you can use a smd Res for the feedback resistors (R3 and R5. Reducing a lot of track and freeing more space to get the caps close to their pins.
Hi Pg,
the Negative Feedback (NFB) RC time constant is 22mS (R9*C7). This will roll-off the bass too early. It should also be lower frequency (minimum half an octave) than the input filter bass roll-off. The input is set to 48mS so the NFB>=68mS. Change the C7 to a non polar 68uF or 100uF.
The schematic makes the classic mistake of showing all the grounds/returns as the same symbol.
They all do a different job and deserve a different symbol. Unfortunately simulators and PCB design software cannot cope with this so we regularly see the same mistake on many schematics.
here goes:-
input ground
amplifier front end ground
power supply 0volt reference
speaker return
decoupling common
Zobel return
transformer centre tap
mains safety earth.
Just giving them names helps to understand the different function.
EACH one should have it's OWN return to a central star ground. But this is uneconomic.
So part two follows:
Combine the RCA input ground + volume control ground + input filter ground + NFB ground. Take this trace off the board as a separate wire to central star ground.
Take the decoupling off the board as a separate wire to central star ground.
Take the speaker return AND the Zobel return direct from speaker terminals to central star ground.
Take the transformer centre tap to the central star ground.
Take the PSU 0v to central star ground.
DO NOT put the central star ground on the smoothing capacitors!!!!!
Take the mains input earth wire direct to a permanent chassis fixing and NEVER remove it. Not even for maintenance. A welded connection (NOT soldered) would be ideal but a permanently bolted connection will do as second choice.
Above the permanent connection add a second nut and secure a wire running to a disconnecting network.
The disconnecting network can consist of any or all of the following all wired in parallel.
Power diode bridge (2 by 25A) wired in inverse parallel,
Power resistor (2r2 to 22r),
High frequency capacitor (22nF to 220nF ceramic 100V),
Power thermistor,
Take a wire from the disconnecting network to either:-
a. the RCA input ground or
b. the input ground connection on the PCB.
Choose which is quieter with your sources.
R11 and R12 will run very hot and increase the ripple on the PSU.
If you decide you really need discharge resistors then change them for a higher value, try 2k2 to 10k.
You could also wire in a series combination of LED and resistor across each pole of the PSU to show LIVE power.
Finally the volume control is going to screw all your input and source impedances and change the input filter turn-over frequencies. It will act like a variable tone control as you adjust the volume. If you set the filters high enough to avoid audible HF effects then one does not effectively filter RF interference. Some amps don't like RF at the input.
the Negative Feedback (NFB) RC time constant is 22mS (R9*C7). This will roll-off the bass too early. It should also be lower frequency (minimum half an octave) than the input filter bass roll-off. The input is set to 48mS so the NFB>=68mS. Change the C7 to a non polar 68uF or 100uF.
The schematic makes the classic mistake of showing all the grounds/returns as the same symbol.
They all do a different job and deserve a different symbol. Unfortunately simulators and PCB design software cannot cope with this so we regularly see the same mistake on many schematics.
here goes:-
input ground
amplifier front end ground
power supply 0volt reference
speaker return
decoupling common
Zobel return
transformer centre tap
mains safety earth.
Just giving them names helps to understand the different function.
EACH one should have it's OWN return to a central star ground. But this is uneconomic.
So part two follows:
Combine the RCA input ground + volume control ground + input filter ground + NFB ground. Take this trace off the board as a separate wire to central star ground.
Take the decoupling off the board as a separate wire to central star ground.
Take the speaker return AND the Zobel return direct from speaker terminals to central star ground.
Take the transformer centre tap to the central star ground.
Take the PSU 0v to central star ground.
DO NOT put the central star ground on the smoothing capacitors!!!!!
Take the mains input earth wire direct to a permanent chassis fixing and NEVER remove it. Not even for maintenance. A welded connection (NOT soldered) would be ideal but a permanently bolted connection will do as second choice.
Above the permanent connection add a second nut and secure a wire running to a disconnecting network.
The disconnecting network can consist of any or all of the following all wired in parallel.
Power diode bridge (2 by 25A) wired in inverse parallel,
Power resistor (2r2 to 22r),
High frequency capacitor (22nF to 220nF ceramic 100V),
Power thermistor,
Take a wire from the disconnecting network to either:-
a. the RCA input ground or
b. the input ground connection on the PCB.
Choose which is quieter with your sources.
R11 and R12 will run very hot and increase the ripple on the PSU.
If you decide you really need discharge resistors then change them for a higher value, try 2k2 to 10k.
You could also wire in a series combination of LED and resistor across each pole of the PSU to show LIVE power.
Finally the volume control is going to screw all your input and source impedances and change the input filter turn-over frequencies. It will act like a variable tone control as you adjust the volume. If you set the filters high enough to avoid audible HF effects then one does not effectively filter RF interference. Some amps don't like RF at the input.
Hi,
the 0v common connection between the smoothing caps carries all the high peak pulses that charge the caps. This is a very bad modulation to insert into the power rail of an amplifier.
Add this modulation to the ground reference and you are almost guaranteed to get buzzing at the amplfier output.
Look carefully at how the current gets from the transformer to the rectifier and then to the smoothing caps and how it returns to the transformer.
This route from and all the way back should be SHORT if possible and the LOOP AREA of the flow and return cables should be SMALL (no if possible this time).
When the amplifier is taking no current, the transformer centre tap does not inject any current into the 0v common. For this reason I solder the two centre tap wires together BEFORE I take them to the central star ground. The wire from the centre tap to the central star ground only carries the same current as the amplifier puts into the ground. If the amplifier exactly matches the +ve rail current and the -ve rail current then no current flows in the ground connection to central star ground and no current flows from transformer centre tap to central star ground.
When the amplifier sends current to the load or draws unmatched current from the supply rails then the transformer centre tap injects a matching current into the central star ground.
It follows that the current flowing from the centre tap into the central star ground is exactly matching the load current (smoothly changing audio bandwidth signal) +- a small ClassA bias for the unmatched amplifier currents.
So keeping the charging currents in that SMALL SHORT loop prevents any current and voltage effects on the amplifier ground.
The lack of voltage modulation on the central star ground and the signal grounds running to the amplifier front end is what makes the amplifier completely hum and buzz free. Some don't believe me, Try it.
BTW. this all came from this FORUM.
I was in the wilderness for 30years before I came hear/here to learn.
the 0v common connection between the smoothing caps carries all the high peak pulses that charge the caps. This is a very bad modulation to insert into the power rail of an amplifier.
Add this modulation to the ground reference and you are almost guaranteed to get buzzing at the amplfier output.
Look carefully at how the current gets from the transformer to the rectifier and then to the smoothing caps and how it returns to the transformer.
This route from and all the way back should be SHORT if possible and the LOOP AREA of the flow and return cables should be SMALL (no if possible this time).
When the amplifier is taking no current, the transformer centre tap does not inject any current into the 0v common. For this reason I solder the two centre tap wires together BEFORE I take them to the central star ground. The wire from the centre tap to the central star ground only carries the same current as the amplifier puts into the ground. If the amplifier exactly matches the +ve rail current and the -ve rail current then no current flows in the ground connection to central star ground and no current flows from transformer centre tap to central star ground.
When the amplifier sends current to the load or draws unmatched current from the supply rails then the transformer centre tap injects a matching current into the central star ground.
It follows that the current flowing from the centre tap into the central star ground is exactly matching the load current (smoothly changing audio bandwidth signal) +- a small ClassA bias for the unmatched amplifier currents.
So keeping the charging currents in that SMALL SHORT loop prevents any current and voltage effects on the amplifier ground.
The lack of voltage modulation on the central star ground and the signal grounds running to the amplifier front end is what makes the amplifier completely hum and buzz free. Some don't believe me, Try it.
BTW. this all came from this FORUM.
I was in the wilderness for 30years before I came hear/here to learn.
I see, but this is quite interesting because it is different from what I've read anywhere else. I will definately give this a try. I would imagine this would be more of a problem on Class A designes because of the current consumption? I've not built one, so I haven't had too many problems with hum (nothing that wasn't quick to resolve 100%)
Cheers
Gert
Cheers
Gert
Hi Gb,
Also the ClassA bias is usually balanced between +ve & -ve rails so little or no ClassA current goes into or out of the central star ground.
Finally ClassA currents are usually DC currents. If they did put a voltage modulation on the 0volt reference it must by definition be a DC modulation. I think one would be hard pressed to reproduce it and even harder to hear it.
no, due to sensible use of smoothing to keep the ripple at an acceptable level.
Also the ClassA bias is usually balanced between +ve & -ve rails so little or no ClassA current goes into or out of the central star ground.
Finally ClassA currents are usually DC currents. If they did put a voltage modulation on the 0volt reference it must by definition be a DC modulation. I think one would be hard pressed to reproduce it and even harder to hear it.
More on grounding..
Bhat,
Sorry for intruding into your post, but I thought I ask this question on grounding.
Andrew,
Thanks for that great post on the star grounding "rules".
Two questions:
1) How would this apply to multichannel amps? Would each have their own star grounds? Or a common star ground for all amps?
2) You mentioned that the mains earth would be bolted to the amp case. Would the amp star ground have to be insulated from the case (else they would be connected with mains earth anyway). Could the star ground be on the same pcb but connected by wires?
Bhat,
Sorry for intruding into your post, but I thought I ask this question on grounding.
Andrew,
Thanks for that great post on the star grounding "rules".
Two questions:
1) How would this apply to multichannel amps? Would each have their own star grounds? Or a common star ground for all amps?
2) You mentioned that the mains earth would be bolted to the amp case. Would the amp star ground have to be insulated from the case (else they would be connected with mains earth anyway). Could the star ground be on the same pcb but connected by wires?
Hi,
I have only built single or two channel power amplifiers.
So I cannot comment from experience on multi-channel amps.
However, the principles work for two channel with a single star ground serving boths channels, I would expect this to also extend to multi-channel.
The major difficulty, as I see it, is selecting a locaton for the star that minimises the loop area from the transformer and yet keeps the amp grounds reasonably short.
Note however that the order of bolting the grounds together matters. Eg. all the signal grounds should be grouped at one end of the stack (of solder lugs), and all the 0v from the multiple PSUs should be at the other end, next to the centre tap lug.
There would be nothing wrong with putting a star ground on a PCB and bringing all the ground wires to the PCB star. But it is less flexible to make changes to the bolting order. Since the ground reference should have no noise on it, it would be possible to locate the PCB star on the amplifier PCB, this would simplify some of the wiring but still suffers that lack of flexibility preventing experimentation.
The safety earth must be separate from the central star ground.
Go back and read how to make this arrangement safe in event of a mains related failure.
I have only built single or two channel power amplifiers.
So I cannot comment from experience on multi-channel amps.
However, the principles work for two channel with a single star ground serving boths channels, I would expect this to also extend to multi-channel.
The major difficulty, as I see it, is selecting a locaton for the star that minimises the loop area from the transformer and yet keeps the amp grounds reasonably short.
Note however that the order of bolting the grounds together matters. Eg. all the signal grounds should be grouped at one end of the stack (of solder lugs), and all the 0v from the multiple PSUs should be at the other end, next to the centre tap lug.
There would be nothing wrong with putting a star ground on a PCB and bringing all the ground wires to the PCB star. But it is less flexible to make changes to the bolting order. Since the ground reference should have no noise on it, it would be possible to locate the PCB star on the amplifier PCB, this would simplify some of the wiring but still suffers that lack of flexibility preventing experimentation.
The safety earth must be separate from the central star ground.
Go back and read how to make this arrangement safe in event of a mains related failure.
Hi,
Just to confuse the issue a little. I've built a few amps with different earthing schemes and have found having the star earth smack in the middle of the filter caps causes little or no problem. In fact my best sounding amp has this arrangement. So it might be a matter of horses for courses. So don't be two worried about trying it. You might like it, you might not.
I have read the books many times and seen the advice given on the forums, so I have implemented this with my eyes open. Douglas Self mentions it will add some distortion.
What is the mechanism of the pulses between filter caps doing a left turn and travelling up the wire to the amp?
regards
Just to confuse the issue a little. I've built a few amps with different earthing schemes and have found having the star earth smack in the middle of the filter caps causes little or no problem. In fact my best sounding amp has this arrangement. So it might be a matter of horses for courses. So don't be two worried about trying it. You might like it, you might not.
I have read the books many times and seen the advice given on the forums, so I have implemented this with my eyes open. Douglas Self mentions it will add some distortion.
What is the mechanism of the pulses between filter caps doing a left turn and travelling up the wire to the amp?
regards
Hi Andrew,
I seem to have a very simpliest view of the star earth and unfortunately don't seem to have the mental capacity to see other's understanding.
To me, the star earth is the zero volt reference by definition. It always remains zero because its is the reference and we set it by connecting it to ground. If we assume copper had zero resistance then every point connected to the star would be at exactly the same potential. But copper has resistance so we control the flow of current by adjusting the resistance by making tracks/wires thicker and/or shorter. Now the charging pulses of the filter caps go from the cap to the star earth because it is the shortest thickest path and not up the wire to the amp.
In the negative supply I assume the opposite happens.
regards
I seem to have a very simpliest view of the star earth and unfortunately don't seem to have the mental capacity to see other's understanding.
To me, the star earth is the zero volt reference by definition. It always remains zero because its is the reference and we set it by connecting it to ground. If we assume copper had zero resistance then every point connected to the star would be at exactly the same potential. But copper has resistance so we control the flow of current by adjusting the resistance by making tracks/wires thicker and/or shorter. Now the charging pulses of the filter caps go from the cap to the star earth because it is the shortest thickest path and not up the wire to the amp.
In the negative supply I assume the opposite happens.
regards
Hi,
Remember a few posts ago I said the charging pulse go around the loop formed from the transformer/rectifier/smoothing caps/back to the transformer. If the amp has balanced current requirements from the +ve and -ve rails then NO charging pulses nor ANY current passes along the transformer centre tap connection to the star ground. If no current travels along this wire then no voltage drop can be generated. That was also referred to when I said connect the centre tap wires together BEFORE reaching the central star ground.
I cannot see how one can engineer a central star ground on the 0volt wire/plate/PCB of the interconnection between the smoothing caps.
There is a physical way of achieving a mounting near the 0volt plate but the actual mounting uses the mounting bolt itself as the connecting wire and the central star ground is a few mm of bolt length away from the 0volt plate. This is electrically very different from mounting the central star ground ON the 0volt plate.
no. The charging pulses go back to the transformer.
Remember a few posts ago I said the charging pulse go around the loop formed from the transformer/rectifier/smoothing caps/back to the transformer. If the amp has balanced current requirements from the +ve and -ve rails then NO charging pulses nor ANY current passes along the transformer centre tap connection to the star ground. If no current travels along this wire then no voltage drop can be generated. That was also referred to when I said connect the centre tap wires together BEFORE reaching the central star ground.
I cannot see how one can engineer a central star ground on the 0volt wire/plate/PCB of the interconnection between the smoothing caps.
There is a physical way of achieving a mounting near the 0volt plate but the actual mounting uses the mounting bolt itself as the connecting wire and the central star ground is a few mm of bolt length away from the 0volt plate. This is electrically very different from mounting the central star ground ON the 0volt plate.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.