Andy,
When i mentioned placing the zobel on the speaker terminals, i was suggesting the speaker output terminals of the amplfier, not on the speakers themself.
Thankyou for all your help and suggestions
.
Yes, i plan to split the plain into two, and run two seperate paths back to the star ground.
Can you see any more effective way to remove 50hz hum from the diodes?
Rhys
When i mentioned placing the zobel on the speaker terminals, i was suggesting the speaker output terminals of the amplfier, not on the speakers themself.
Thankyou for all your help and suggestions
.Yes, i plan to split the plain into two, and run two seperate paths back to the star ground.
Can you see any more effective way to remove 50hz hum from the diodes?
Rhys
Hi Rhys,
Sorry for the misunderstanding on Zobel, on the amplifer output terminals should be fine.
Since you have asked
I would have four returns back to the star point as follows; 1. the posive caps C10 and C12, 2. the negative caps C11 and what I am gussing is C13 (it doesn't have an ID), 3. the input ground components R152 and R150 this can be shared between the two inputs but will introducts a bit of crosstalk if it is (Probably at arround - 80 to 90dB so barely measurable and not noticably audible although it will probably add to the "character"of the amp). 4. the local amplifer decoupling c100, c101, c102, c104
You coulds split it into 6 returns if you want the lowest crosstalk and have the local decoupling and the input grounds for each amplifer seperated. But this isn't necessary if you don't care about a bit of low level crosstalk.
I would also move c100, c101, c102, c104 for the left channel closer to the power pins by moving it to the other side of the amplifer. I have found these amplifers are very susptable to load induced instability if the local decoupling is not very close to the pins. This happens especially into 4R or lower loads and causes the output to clip much earlier than it would normaly.
Since this is a DIY design and you don't need the conectors to neatly match up with some output cabling I would move the two output connectos closer to the amplifers and not run the output of one channel right past the input stage of the other channel. It might be OK but it could cause crosstalk or even possibly instability.
By the way unless you are power limited in the discharge resisotors you don't need a resistor for the seperate caps C10 and C12 one resistor will discharge both. Similary on the other supply. So you could remove R3 and R4. I am not sure why you have two caps for the bulk storage on each rail unless you couldn't get the value you want in a single cap. There isn't really much gained buy splitting the caps and you havent added a filter resistor between teh two like you have on the Tube supply. (I wouldn't either as you want a low impedance supply for the power stage)
Normally I don't use ground planes on power amplifiers as it is better to define the current paths.
I don't want you to think it wouldn't work without these mods it probably will. These would just be how I would do it, another engineer may have a different approach.
Best of luck.
Andy.
Sorry for the misunderstanding on Zobel, on the amplifer output terminals should be fine.
Since you have asked
I would have four returns back to the star point as follows; 1. the posive caps C10 and C12, 2. the negative caps C11 and what I am gussing is C13 (it doesn't have an ID), 3. the input ground components R152 and R150 this can be shared between the two inputs but will introducts a bit of crosstalk if it is (Probably at arround - 80 to 90dB so barely measurable and not noticably audible although it will probably add to the "character"of the amp). 4. the local amplifer decoupling c100, c101, c102, c104
You coulds split it into 6 returns if you want the lowest crosstalk and have the local decoupling and the input grounds for each amplifer seperated. But this isn't necessary if you don't care about a bit of low level crosstalk.
I would also move c100, c101, c102, c104 for the left channel closer to the power pins by moving it to the other side of the amplifer. I have found these amplifers are very susptable to load induced instability if the local decoupling is not very close to the pins. This happens especially into 4R or lower loads and causes the output to clip much earlier than it would normaly.
Since this is a DIY design and you don't need the conectors to neatly match up with some output cabling I would move the two output connectos closer to the amplifers and not run the output of one channel right past the input stage of the other channel. It might be OK but it could cause crosstalk or even possibly instability.
By the way unless you are power limited in the discharge resisotors you don't need a resistor for the seperate caps C10 and C12 one resistor will discharge both. Similary on the other supply. So you could remove R3 and R4. I am not sure why you have two caps for the bulk storage on each rail unless you couldn't get the value you want in a single cap. There isn't really much gained buy splitting the caps and you havent added a filter resistor between teh two like you have on the Tube supply. (I wouldn't either as you want a low impedance supply for the power stage)
Normally I don't use ground planes on power amplifiers as it is better to define the current paths.
I don't want you to think it wouldn't work without these mods it probably will. These would just be how I would do it, another engineer may have a different approach.
Best of luck.
Andy.
Andy,
Thankyou for all the input, i have done a redesign on the board, you may find the image below.
Big Version
I am aware there are still a couple of clearences to sort out etc, but do you think this design would be more effective?
Rhys
Note: R1L and R1R have been added straight after the input caps.
Thankyou for all the input, i have done a redesign on the board, you may find the image below.
An externally hosted image should be here but it was not working when we last tested it.
Big Version
I am aware there are still a couple of clearences to sort out etc, but do you think this design would be more effective?
Rhys
Note: R1L and R1R have been added straight after the input caps.
Hi Rhys,
This certainly looks better than before however there are still some bits which arn't optimal.
You have connected the diode grounds back to the star point. If you do this the rectifier currents will flow though the star point and with two rectifiers and floating windings you don't need to do this.
Connect from D4 and D3 to the bulk capacitor +ve C9. This way you shorten the noisy rectifier loop. Similarly with D5 and D6 to C10 -ve.
Since the plane is still connected to the local decoupling caps you haven't really provided a seperate path for the input and local decoupling. I would seperate the decoupling ground from the plane and use the plane just for the input circuits with a seperate return to the star.
The left hand amplifer output now goes straight through the right hand amps sensitive front end. I'd bring it further out near the PSU. The output of the amplifer is not very sensitive to the noise of the PSU where as the input of the right hand amp could be upset by the high currents going to the load from the left hand amp.
I am also a bit concerned about the caps C102L and C101L coupling to the input signal. However I am struggling to see a better way without a complete redesign. Perhaps you can see something better. Are you trying to not use any vias? as I think that is how I would probably resolve the decoupling and input signal problem.
Again I don't want you to think it wouldn't work without these changes, it sounds like I am just complaining about everything your doing and that is not my intention. It is just difficult to explain stuff clearly.
Regards,
Andrew
This certainly looks better than before however there are still some bits which arn't optimal.
You have connected the diode grounds back to the star point. If you do this the rectifier currents will flow though the star point and with two rectifiers and floating windings you don't need to do this.
Connect from D4 and D3 to the bulk capacitor +ve C9. This way you shorten the noisy rectifier loop. Similarly with D5 and D6 to C10 -ve.
Since the plane is still connected to the local decoupling caps you haven't really provided a seperate path for the input and local decoupling. I would seperate the decoupling ground from the plane and use the plane just for the input circuits with a seperate return to the star.
The left hand amplifer output now goes straight through the right hand amps sensitive front end. I'd bring it further out near the PSU. The output of the amplifer is not very sensitive to the noise of the PSU where as the input of the right hand amp could be upset by the high currents going to the load from the left hand amp.
I am also a bit concerned about the caps C102L and C101L coupling to the input signal. However I am struggling to see a better way without a complete redesign. Perhaps you can see something better. Are you trying to not use any vias? as I think that is how I would probably resolve the decoupling and input signal problem.
Again I don't want you to think it wouldn't work without these changes, it sounds like I am just complaining about everything your doing and that is not my intention. It is just difficult to explain stuff clearly.
Regards,
Andrew
gfiandy said:
Andrew, thankyou very much for your comments, they are very usefull to me!
Please find my new board
here.
As you can see, i have removed the ground plane from the chip amp section, and i have then routed the signal and power grounds to the star ground seperatly.
I have also changed the diode paths like you said, but im not sure if this is what you were trying to say, can you confirm this?
Do you think my attached board is better then my last one? I still have yet to move the left output more towards the psu.
Thanks!
Hi Rhys,
Yes diodes now wired how I was suggesting.
R3 and R4 should be wired across there respective caps locally to the cap. At the moment they will be carring ripple current through the star point. As I mentioned earlier unless you need two resistors beacuse of dissipation I would just use one and accept that it will take a little longer to discharge the caps. However if you want to keep them they should be wired locally.
Amp grounding now looks much better.
Regards,
Andrew
Yes diodes now wired how I was suggesting.
R3 and R4 should be wired across there respective caps locally to the cap. At the moment they will be carring ripple current through the star point. As I mentioned earlier unless you need two resistors beacuse of dissipation I would just use one and accept that it will take a little longer to discharge the caps. However if you want to keep them they should be wired locally.
Amp grounding now looks much better.
Regards,
Andrew
Andrew, a huge thankyou for your comments and advice, i am very grateful of your help, and im sure the other builders will be once they have their amplifiers built and working!
Here is the PCB after i have changed the resistors to their local caps.
The only thing left to do now is to move the left channel output.
I have put a credit on the board for you, for your help and advice you have given me to help make this possible.
Thanks!
Here is the PCB after i have changed the resistors to their local caps.
The only thing left to do now is to move the left channel output.
I have put a credit on the board for you, for your help and advice you have given me to help make this possible.
Thanks!
Hi Rhys,
I took another look at your design and there is another modification I think you can make fairly easly that should improve performance.
At the moment the ground return path for the speaker outputs runs the far side of the PSU. This means that the field for the output signal will run across the PSU circuit (between the output track and the ground return.) This may pick up noise from the PSU.
If you move the return track so it runs between the amps and the PSU past C1 and C6 it will significantly reduce the loop area between the output and the return and the loop will no longer go round the PSU. Looks like it should be possible by just changing the track fom C12 to C101R to the top layer.
This combined with bringing the output track further over so it doesn't run through the input section of the second amp should help.
Thanks for the credit on your PCB, I hope it actually works OK. Like I said you never really know till you try it.
Regards,
Andrew
I took another look at your design and there is another modification I think you can make fairly easly that should improve performance.
At the moment the ground return path for the speaker outputs runs the far side of the PSU. This means that the field for the output signal will run across the PSU circuit (between the output track and the ground return.) This may pick up noise from the PSU.
If you move the return track so it runs between the amps and the PSU past C1 and C6 it will significantly reduce the loop area between the output and the return and the loop will no longer go round the PSU. Looks like it should be possible by just changing the track fom C12 to C101R to the top layer.
This combined with bringing the output track further over so it doesn't run through the input section of the second amp should help.
Thanks for the credit on your PCB, I hope it actually works OK. Like I said you never really know till you try it.
Regards,
Andrew
gfiandy said:
Thanks for the additional tips and hints, i will apply these to the board in the morning when i get chance. I dont think there will be any problem with this amplfier, unless it is to do with grounding. I have tested the tube section a few times now, and i have tested the tube section with the amp front end in a p2p configuration, just to check basic workings, it seems to be fine so far!
Rhys
This is exactly the type of thing I'm looking for, for my first DIY loudspeakers. I would have sent an e-mail but I am still new to the forum and therefore under moderation. I am interested in 2 of these boards, if any more are available. please contact me at spjonesksu@gmail.com. Thanks!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.