As above+
Your SCLK line crosses a split in the ground plane, not a good idea for a clock line.
Digital lines to mux need a return path under them.
Your ground is split all over the place. I would re think the placement and try different placement to see if you can simplify the power deliver for a two layer design, or go 4 layer.
Your SCLK line crosses a split in the ground plane, not a good idea for a clock line.
Digital lines to mux need a return path under them.
Your ground is split all over the place. I would re think the placement and try different placement to see if you can simplify the power deliver for a two layer design, or go 4 layer.
Thank you again for your help. Yes, this makes perfect sense to me. I will update layout with improved grounding. One question though. What is the proper way to evaluate signal path through opamp? Opamp doesn't have power supply grounding pin, only supply rails. I see you draw line white from the input pin 2 to the 5th pin of opamp but it is just grounded non inverting input of servo amp. Can you please explain why is that way? I want to look at other signal paths also so I need to understand it.
Thank you for your comments.
Can you please explain what effect split under SCLK line might have?
As for return path for digital lines, do you mean that digital ground plane should be extended under them? DG409 has only one grounding pin and it is connected to analog ground so return path anyway will be through connection of analog ground plane to digital ground plane. I haven't decided yet whether it will be on this board or outside this board. Can you please give more details here.
I tried various placement for components and this one is the best I could find to keep signal paths as short as possible and at the same time be able to split analog and digital grounds. Having dual supply rails make things complicated. I will update power supply traces as gfiandy recommended and that should reduce ground plane splitting significantly. I'm not saying that this is the best components placement, but it's the best that I could find.
Hi,
Well thought, you are right the ground on the pin is not the source ground. I did this quite quickly and obviously wasnt checking closly enough. The source ground for the signal from the opamps will be the ground on the decoupling caps for the opamp.
It is better if high speed signals do not cross plane splits as they form a sort of dipole anttena and radiate noise, also as the high speed current will want to return to its source under the outwards track, as this is the path of minimum inductance , it can cause noise on the analogue plane. This is the sort of situation where a coupling cap to return the HF return currect can help.
However I have assumed you will shut down the SCK when not talking to the parts and hence during listening there will be minimal noise on these lines as they will be static. In this case the poor return path might cause noise but hopefully not whilst you are listening. The problem then becomes will you get transmision errors due to the poor return path, I think it is unlikely. But if you do the the ground plane resistor option you have added will probably be enough to fix it.
Regards,
Andrew
Well thought, you are right the ground on the pin is not the source ground. I did this quite quickly and obviously wasnt checking closly enough. The source ground for the signal from the opamps will be the ground on the decoupling caps for the opamp.
It is better if high speed signals do not cross plane splits as they form a sort of dipole anttena and radiate noise, also as the high speed current will want to return to its source under the outwards track, as this is the path of minimum inductance , it can cause noise on the analogue plane. This is the sort of situation where a coupling cap to return the HF return currect can help.
However I have assumed you will shut down the SCK when not talking to the parts and hence during listening there will be minimal noise on these lines as they will be static. In this case the poor return path might cause noise but hopefully not whilst you are listening. The problem then becomes will you get transmision errors due to the poor return path, I think it is unlikely. But if you do the the ground plane resistor option you have added will probably be enough to fix it.
Regards,
Andrew
I have also assumed you have used two layer for cost reasons, Keruskerfuerst is right, you would undoubtly get a better result with four layer. However if you work through all the issues a two layer soultion should work well. It is also a much better learning tool, as on four layer boards you can get away with all sorts of issues that would have been a problem on a two layer.
Regards,
Andrew
Regards,
Andrew
Another useful visulisation tool for signals is something said to me on a couse on high speed digital design but its ture for analogue as well, “The signal does not flow down the track it is in the electric feild between the track and the return ground” this becomes more and more obvious at higher speeds but even at DC there is a field between the outward conductor and ground. Hence if you stand under a DC high voltage line with a fluorescent tube, the field strength will light it up!
Dear, this is not mine circuit. It is an arrangement to improve cut-off frequency response. It makes a sharper frequency response in cut-oof frequency.
The attached circuit was done by Salas and it is done around AD823.
You can use C5 and C4 with 470nF to lower frequency cut-off.
Some designer create an extra R-C-R in output of servo (operational amplifier) to isolate any residual AC to be introduced in audio feedback (if done you have a third order filter) in place of R14. C must connected to clean or analog GND.
I am more interested in digital control as I told I am not familiar with arduino, but I suppose there is no digital control yet.
I know well how to route a PCI and it is good enough now and do not need to become more complex of even use more than two layer, but the circuit is missing some care.
Regards
Ronaldo
Attachments
Last edited:
No it is not true. I use it in amplifier projects for a long time. It do not create instabilities as it only do DC polarization. You must take care of cut-off frequency to have an very good low frequency response.
I also did you a suggestion to your secondary filter. It also can be improved.
The stability of an amplifier or pre-amplifier do not depends of DC polarization, but from AC frequency and Phase response.
And I ask you what about put an amplifier in output short circuit during start up. Do it works well?
Ronaldo
Last edited:
You need to join the two grounds, if you look at the evaluation board the two grounds are joined by J3...
http://www.ti.com/lit/ug/sbou012/sbou012.pdf
http://www.ti.com/lit/ug/sbou012/sbou012.pdf
DC servos can motorboat (low frequency oscillation sounds like a motorboat through the speakers) if you are mot careful with the phase. It is already second order as it has a post noise filter. He has added 47R to the output of the opamps in the later design.
The origional poster is not an experienced designer so is not going to be able to modify the servo without someone doing the design work for them. I am not going to re- enginer this servo because it comes from an existing successful comercial design and I know it is stable and sounds good.
So if you belive it needs redesigning please show how you think it can be improved. Please note the design you showed in your previous post is for non inverting application, this aplication requires an inverting design.
The origional poster is not an experienced designer so is not going to be able to modify the servo without someone doing the design work for them. I am not going to re- enginer this servo because it comes from an existing successful comercial design and I know it is stable and sounds good.
So if you belive it needs redesigning please show how you think it can be improved. Please note the design you showed in your previous post is for non inverting application, this aplication requires an inverting design.
I am not interested in make a good design to you. If I want to I will open a new thread. I do not know what you are cloning, but there is lots of products in audio area that do not perform well and all people say it is good just because they do not have a good comparative reference.
I suppose you read something wrong about DC servos and I explained well to you how it works. If you have a low cut-off frequency and a sharp cut-off behavior from your servo you will have only DC to make first stage polarization. Cut-off frequency must be lower than 1Hz in servo circuit. If you add R-C-R like to wrote you, you will avoid further more any signal to depreciate the DC polarization.
It is very easy to improve your circuit following my suggestions. I showed you an schematic sample to you know what I was talking about. You can see lots of comercial designs using inverting circuits in servo, you only need to add an simple inverter circuit. Look to any old Onkyo Amplifier Schematic or service manual. Onkyo uses it a lot in very good amplifier.
By the way there is lots of cheap Chinese boards using PGA2310 without any care concerning buffers amplifier in output or input of PGA2310. They also works. This thread was supposed to make a better solution and now you wrote me the designer do not know what to do or even follow suggestion?
More one suggestion to isolate noise signal from Microprocessor Control - Use Opto-Couples and avoid totally Digital and analog loop GND.
During DC start-up no one can predisse what will happens with operational amplifier with very strong capacitive circuit. It is not a good practice to put operational amplifier output connected to ground in this time. Is it from the circuit you are cloning?
I am out of this thread and of course it is my last answer.
Ronaldo
Last edited:
If your not interested in helping why are you posting on a DIY audio forum?
I think your proposal is not necessary and needlessly complex but if you belive in it then a circuit example would help the origional poster understand the possible options, then he could decide if he wants to try it.
Opto isolators could help but again they would increase complexity, it looks like this is the first design for the origional poster. I would propose that this is also an uneccesary complication but if you want to do a circuit for the origional poster to understand that would also be interesting for him to learn from.
Regards
Andrew
I think your proposal is not necessary and needlessly complex but if you belive in it then a circuit example would help the origional poster understand the possible options, then he could decide if he wants to try it.
Opto isolators could help but again they would increase complexity, it looks like this is the first design for the origional poster. I would propose that this is also an uneccesary complication but if you want to do a circuit for the origional poster to understand that would also be interesting for him to learn from.
Regards
Andrew
I have one question left about evaluating of signal paths. How the signal that enters op amp should be treated? Should I assume that signal reaches op amp and returns back to the source and the signal that goes from the output of the op amp is a different signal that op amp emits and it should be returned back to the op amp? Or should I assume that signal goes through the op amp, volume control, etc and then goes all the way back to the source?
From your drawing I guess the first way is the correct one.
I won't have enough time during Christmas and new year to update the design but after that I will post a new updated layout.
Yes, I've decided to go with 2 layer for cost reasons but also to keep things simple for my first design. (I think professional engineers would think that 4 layer is simpler
because you can have just separate planes for ground and for each supply rail)I did not layout micro controller board yet because there are some things still to be resolved the main one of which is the control of DAC. I can post it when it will be done but I don't really think it will be much of a use directly because it will have entire Arduino nano soldered in it rather then just micro controller IC. Also it will have LCD soldered on it and bunch of other stuff for powering integrated amplifier in proper sequence and so on. Besides compiled code will have preprogrammed IR codes that most likely you'll want to change so source code that I've posted should work the best for you. You can install Arduino studio and try to compile it yourself if you want to produce binary.
I absolutely understand that. Otherwise it won't just work. Initially I was planning to connect grounds outside of the board because I will have also DAC board running from the same power supply and was thinking if grounds will be connected both on the preamp board and DAC board, it will for a ground loop. However in the latest layout I've added an option to solder 0 ohm resistor to connect ground planes.
I still need to find out how to measure which approach is better though
This thread was supposed to make a better solution and now you wrote me the designer do not know what to do or even follow suggestion?
Exactly. You're right I'm not experienced designer and doing this just for fun. That's why I post all my questions on DIY forum rather then on professional one. And unfortunately I can't understand what was said about 2nd order filtering in DC servo and startup behaviour. However I'm still trying to get to know and understand as much as I can so thank you again for all your help.
Last edited:
I saw designs using those for digital lines but this would require one more 5V power supply so I decided to keep it simple and use only one.
Hi, I hope you had a good Christmas. Yes your first assumption is correct. The signal can be treated as terminating at the opamp and being regenerated at the output.
For a first build a two layer PCB is a good option. You really need more experience with CAD and a CAD package where the netlist is transferred from the schematic for a multilayer design.
I will try to track down an example of the different servo options so you can see what a second order servo means and does. I remember now sone of the reasons the designer went with first order was as the filter had to trade off frequency cut off with settling time and noise generated by the large resistors and group delay in the filter. He spent quite a long time on this part of the design and this was his best compromise in this application.
Regards
Andrew
For a first build a two layer PCB is a good option. You really need more experience with CAD and a CAD package where the netlist is transferred from the schematic for a multilayer design.
I will try to track down an example of the different servo options so you can see what a second order servo means and does. I remember now sone of the reasons the designer went with first order was as the filter had to trade off frequency cut off with settling time and noise generated by the large resistors and group delay in the filter. He spent quite a long time on this part of the design and this was his best compromise in this application.
Regards
Andrew
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.