Member
Joined 2009
Paid Member
Gannaji, I took your .asc Spice file and embedded the device models and spice directives into the file so that it can be run directly. I also tried to align the part numbers with those in the schematic you posted. And I took the liberty of deleting a couple of parts that I don't think you need along with clarifying the correct wiring for the grounding (resistor with back to back diodes should go between signal ground and chasis safety earth). Hope this is OK with you.
Attachments
Hi Gareth
I like this schematic a lot and think you guys have done a great job. The schematic at post #21 is a little hard to read so forgive me if I misquote some of the labels...
You could make R2 = 68k, since it is a standard value, and eliminate R8. Without putting this through simulation, I'm not sure how much current is being pulled through the rail RC filter R16 (?) but I'd be inclined to reduce it to say 33R to improve headroom on this low power design.
I've laid out a couple of P3A boards and the output base stoppers R21 and R22 really aren't required so long as you are sensible about placement and trace routing. I've noticed that you are a Nervous Nellie when it comes to these things - I am too but only with EF3 and MOSFET stages ;-) It should be possible to reduce or even eliminate the driver base stoppers but I would definitely put pads on the boards just in case.
Personally I would increase the output cap to reduce LF distortion to say 10mF or even as high as 22mF if space permits, and then bypass it with a quality film cap.
The VAS device should be substituted with a KSA1381 and Cdom can probably come down slightly to 47p.
I want to build this one. If no one else volunteers to do the layout I'm happy to have a go; however, it might take a while with work/family commitments.
I like this schematic a lot and think you guys have done a great job. The schematic at post #21 is a little hard to read so forgive me if I misquote some of the labels...
You could make R2 = 68k, since it is a standard value, and eliminate R8. Without putting this through simulation, I'm not sure how much current is being pulled through the rail RC filter R16 (?) but I'd be inclined to reduce it to say 33R to improve headroom on this low power design.
I've laid out a couple of P3A boards and the output base stoppers R21 and R22 really aren't required so long as you are sensible about placement and trace routing. I've noticed that you are a Nervous Nellie when it comes to these things - I am too but only with EF3 and MOSFET stages ;-) It should be possible to reduce or even eliminate the driver base stoppers but I would definitely put pads on the boards just in case.
Personally I would increase the output cap to reduce LF distortion to say 10mF or even as high as 22mF if space permits, and then bypass it with a quality film cap.
The VAS device should be substituted with a KSA1381 and Cdom can probably come down slightly to 47p.
I want to build this one. If no one else volunteers to do the layout I'm happy to have a go; however, it might take a while with work/family commitments.
Member
Joined 2009
Paid Member
To be honest, I assumed R8 was there simply as an adjustment to set dc-offset for purposes of simulation at least. In reality it would likely require Select-On-Test and might be easier if it were replaced with a potentiometer - I'll leave those construction details to Gannaji.
I agree, the RC filter would benefit from larger C and smaller R for better PSRR but if he has the parts already it's not a bad place to start from.
I agree, they may be found to be necessary.
I agree, it may be found beneficial to reduce AC voltage across the cap for lower distortion. But it looks big enough as a starting point. Rather than one large cap I'd go for a small number of good quality electrolytics in parallel (as per Vladimirk)
I have found that Cdom can have a real affect on the sound, quite a profound effect. It's part of the journey for the constructor to explore this and make his own discoveries about what sounds best. Board parasitics and choice of VAS device will affect this so there may not be a universally correct value. To start with something around 100p is safe for avoiding oscillations and then it can be reduced. I remember enjoying the tinkering.
Yeah, it's looking very tempting isn't it !
I think instead of one of my ultra-modern layouts with SMD parts, I'd want to make it a through-hole easy to build layout with power supply included. Mono.
That Gareth. I took a closer look at the schematic. Why is the output node returned to the output of the NFB shunt cap? I have seen designs where the output is shunted directly to ground through a large (10k+) resistance to aide stability under no-load conditions. But what purpose does this arrangement serve?
Dear all,
Diptrace is a full fledged PCB program used by many on this forum.
I set the grid and snap to 0.1 inch and the resulting file is suitable for use on prototyping boards with holes and pads at 0.1 in pitch. The .dch and .dip files can be used to layout a pcb with all frills and whistles. (Remove the txt extension)
The ground lift resistor is usually put between the input signal ground and Power ground. The nfb is usually connected to signal ground. In this single supply case, I connected the bias of input transistor also to signal ground. This can be easily modified after discussion.
We require around 82k for R2+R3. They can be fixed after trial. I feel that if the Ib of TR1 is <= 10% of IR4, it should be ok. Can be finalized after sim.
I shall make corrections to all files as per your changes and upload them.
"And I took the liberty of deleting a couple of parts that I don't think you need along with clarifying the correct wiring for the grounding (resistor with back to back diodes should go between signal ground and chassis safety earth). Hope this is OK with you."
this is exactly the type of help I wanted. Thank you.
How to do multiquote as you did above? When I press on the multiquote symbol above, it is turning red, but nothing else happens.
--gannaji
Diptrace is a full fledged PCB program used by many on this forum.
I set the grid and snap to 0.1 inch and the resulting file is suitable for use on prototyping boards with holes and pads at 0.1 in pitch. The .dch and .dip files can be used to layout a pcb with all frills and whistles. (Remove the txt extension)
The ground lift resistor is usually put between the input signal ground and Power ground. The nfb is usually connected to signal ground. In this single supply case, I connected the bias of input transistor also to signal ground. This can be easily modified after discussion.
We require around 82k for R2+R3. They can be fixed after trial. I feel that if the Ib of TR1 is <= 10% of IR4, it should be ok. Can be finalized after sim.
I shall make corrections to all files as per your changes and upload them.
"And I took the liberty of deleting a couple of parts that I don't think you need along with clarifying the correct wiring for the grounding (resistor with back to back diodes should go between signal ground and chassis safety earth). Hope this is OK with you."
this is exactly the type of help I wanted. Thank you.
How to do multiquote as you did above? When I press on the multiquote symbol above, it is turning red, but nothing else happens.
--gannaji
Last edited:
Interesting project, i want to build something like this if someone can draw a pcb
How to do multiquote as you did above? When I press on the multiquote symbol above, it is turning red, but nothing else happens.
Press multi-quote then multi-quote(other post or posts) and finally press post reply.
How to do multiquote as you did above? When I press on the multiquote symbol above, it is turning red, but nothing else happens.
Press multi-quote then multi-quote(other post or posts) and finally press post reply.
The input Return connection is missing from the sch.
The bottom of R7 (the gain/feedback defining resistor) is actually the Signal Return connection. The bottom of R7 should have a trace connection to the bottom of C2. This completes the INPUT CIRCUIT.
Once this is made the Signal Return to Power Ground resistor in post19 can be re-inserted. Or use a wire link.
Gann noticed the error.
The bottom of R7 (the gain/feedback defining resistor) is actually the Signal Return connection. The bottom of R7 should have a trace connection to the bottom of C2. This completes the INPUT CIRCUIT.
Once this is made the Signal Return to Power Ground resistor in post19 can be re-inserted. Or use a wire link.
Gann noticed the error.
Last edited:
Thank you AndrewT. Corrected and included the change of C6 to 68p. The diptrace files are correct, and the schematic shows the connection correctly.
Attachments
Member
Joined 2009
Paid Member
This additional feedback loop was from Gannaji and I understand it to provide for some feedback AFTER the output capacitor, hence reducing the impact of capacitor distortion (whatever that is) on the sound.
Good! - time to put together something and see what works - then finalize a pcb once you have it sorted.
Please note - I don't think you've received good advice on the ground-lift, that resistor and back to back diodes is in the wrong place and should be as I recommended. Think about it....
The speaker output is a function of the current flowing through it, which depends on the voltage across it. The feedback network is a potential divider that samples a fraction of the voltage across the speaker. Therefore, the feedback network must return to the ground side of the speaker directly. And because the feedback is fed to the input device acting as a differential error amplifier the feedback ground must also return to the signal ground directly. Hence, you must not put your earth-loop breaker in the ground rail as you have drawn it or any earth-loop currents that create a voltage drop across that 10R resistor and back to back diodes will be injected into the feedback network.
Last edited:
I have not built a single ended input amplifier like this.
I have measured the volts drop across a Signal Return to Power Ground resistor/diode link in many differential input power amplifiers and cannot register any voltage, neither DC, nor AC, on a 199.9mV scale.
That does not mean that this type of amplifier will also measure near zero mV as well.
But it is worth putting in the pads and shorting them with a wire link until testing of the mono channel is complete.
Then when a stereo version is wired up, one has the option to investigate whether a wire link remains, or be replaced with a resistor/diode network.
If the resistor creates an unwanted interference voltage, then it's simple to revert to the wire link.
I have measured the volts drop across a Signal Return to Power Ground resistor/diode link in many differential input power amplifiers and cannot register any voltage, neither DC, nor AC, on a 199.9mV scale.
That does not mean that this type of amplifier will also measure near zero mV as well.
But it is worth putting in the pads and shorting them with a wire link until testing of the mono channel is complete.
Then when a stereo version is wired up, one has the option to investigate whether a wire link remains, or be replaced with a resistor/diode network.
If the resistor creates an unwanted interference voltage, then it's simple to revert to the wire link.
I am not sure.
Are you sure?
It seems to me that the capacitor is a low impedance path for interference being coupled from the supply rail to the ground rail.
If this is the case then connecting the R4/C4 combination to the signal return will inject that interference onto the signal return.
That to me seems wrong.
That is why I asked.
Is there a better place for the connection, or a better arrangement that injects less interference?
Are you sure?
It seems to me that the capacitor is a low impedance path for interference being coupled from the supply rail to the ground rail.
If this is the case then connecting the R4/C4 combination to the signal return will inject that interference onto the signal return.
That to me seems wrong.
That is why I asked.
Is there a better place for the connection, or a better arrangement that injects less interference?
See C4 in your new diagram.
That shows connection to a generalised "ground".
It could be to Power Ground.
or Main Audio Ground.
or PSU zero volts
or Signal Return.
All of these are possible. All will work to some extent.
Noise and distortion will measure differently and almost certainly different audibly.
That shows connection to a generalised "ground".
It could be to Power Ground.
or Main Audio Ground.
or PSU zero volts
or Signal Return.
All of these are possible. All will work to some extent.
Noise and distortion will measure differently and almost certainly different audibly.
Member
Joined 2009
Paid Member
For the simplest implementation I think of ground as being built up in layers. You want large varying return currents to go back to the ground side of the main power supply cap - i.e. the speaker return first. Any small voltages created in the ground return by these large currents should then be isolated from entering the amplifier. You then layer on additional ground returns, the 'dirtiest' first (e.g. zobel return) and the cleanest last (feedback return and signal ground). In practice, you can usually lump them all together at a start ground point.
Safety earth is a separate requirement - it is connected to star earth through a network of resistor and back to back diodes.
I don't know if this is the best approach, it's not the only approach, but I found it has worked well for my projects.
Safety earth is a separate requirement - it is connected to star earth through a network of resistor and back to back diodes.
I don't know if this is the best approach, it's not the only approach, but I found it has worked well for my projects.
In practice your explanation is not wrong.
But it does not help in UNDERSTANDING where current flows.
Far more important to minimising interference effects is maintaining the FLOW and RETURN halves of each circuit.
These two halves of a circuit must be close coupled.
Talking about "grounds" does not aid in getting the circuit ROUTES into the best locations.
But it does not help in UNDERSTANDING where current flows.
Far more important to minimising interference effects is maintaining the FLOW and RETURN halves of each circuit.
These two halves of a circuit must be close coupled.
Talking about "grounds" does not aid in getting the circuit ROUTES into the best locations.
Hey guys here's my first attempt at the layout...
With all the controversy over the ground lift I decided to remove it and rely on (hopefully) good ground trace routing to save the day.
I'm not sure that I'll stick with the depletion MOSFET input transistor and may substitute a JFET instead. Only drama there is that most have very low breakdown voltages necessitating a cascade, increasing complexity.
Gareth, Andrew, any thoughts?
With all the controversy over the ground lift I decided to remove it and rely on (hopefully) good ground trace routing to save the day.
I'm not sure that I'll stick with the depletion MOSFET input transistor and may substitute a JFET instead. Only drama there is that most have very low breakdown voltages necessitating a cascade, increasing complexity.
Gareth, Andrew, any thoughts?
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.