Digital/Analog noise problem

[ohall]

Dear All,

Please excuse this post if it's really easy stuff, but I'm more at home with software, and very much the novice with hardware design (but I'm keen to learn!).

I have a small project that is essentially an RS485 slave device connected to a domestic HiFi amplifier. I send messages to the device and it controls the amp (source selection, volume control, etc.) all via i2c.

The problem I have is that if I power my controller board from the 5v supply within the amp, I get an audible "bleep" on the output audio. This bleep coincides with my device responding to a message on the RS485 bus. If I disconnect the RS485, or stop sending messages, the bleeps stop - as soon as I send messages again, the bleeps recur. I think I've narrowed this down the a fluctuation in power when my micro enables the RS485 send enable (and draws additional power). I should say, if I power the board from an external 5v supply, there's no bleep.

I've also found that, when powered from the external supply, if I connect the amp's ground, the bleeps remain silent, but if I connect the amp's 5v supply (effectively in parallel with my external supply) the bleeps come back.

So - my question is, what should I be doing to stop whatever noise is getting down the 5v line and causing the amp to play back my power fluctations (as musical as they are! Actually, it gets a bit repetetive :wink

I thought a decoupling capacitor might do the trick, but either that strategy is wrong, or I used the wrong value/type? Suffice to say, I'm stuck!

I've managed to overcome all the other issues so I'd hate to fall at the final fence - can anyone advise on what I should do?

Many thanks,
Oliver

 

[Prototyp_V1.0]

Hi.

Without further knowledge of your slave device and how it interact with your amplifier, it's very hard if not impossible to lay any solution down on the table.

 

[thannara123]

have you connectf your computer earth /ground connection .please check it .

 

[KerimF]

Perhaps I am wrong, it seems to me that when you supply your controller from the amplifier you use the ground of the RS485 as the negative return. If this is the case you need to take the two supply wires (+5V and ground) from points as close as possible to the regulator output terminals (on the HiFi board) so that the controller current will not pass in any PCB tracks of the HiFi amplifier.

In case you have a higher voltage than 5V on your HiFi board, a suitable series resistor is added followed by a decoupling capacitor (both not far from the source). The two supply wires (taken from the capacitor terminals) will first supply a voltage regulator as 7805 at the controller side. But it is better, since the coming wires are not short, to add another decoupling capacitor at the regulator input (say 100u).

Hope this help a little bit.

Kerim

 

[ohall]

Hi.

Without further knowledge of your slave device and how it interact with your amplifier, it's very hard if not impossible to lay any solution down on the table.

Without you giving me a hint as to what additional information I need to give you, it's very hard from me to give you that information :wink:

 

[Prototyp_V1.0]

Without you giving me a hint as to what additional information I need to give you, it's very hard from me to give you that information :wink:

Oh, very well. I'm sorry I cant' help you furter on. However it seems that other forum member know about this more than I do.
Good luck

 

[ohall]

Perhaps I am wrong, it seems to me that when you supply your controller from the amplifier you use the ground of the RS485 as the negative return. If this is the case you need to take the two supply wires (+5V and ground) from points as close as possible to the regulator output terminals (on the HiFi board) so that the controller current will not pass in any PCB tracks of the HiFi amplifier.

Hi Kerim,

Many thanks for your help - this is just the sort of advice I was hoping for.

You're absolutely correct - I do have the RS485 ground connected to the ground plane of the controller board (which is then also connected to the amplifier's ground) though I tried disconnecting the RS485 ground and the bleeps continued unchanged, and I appreciate I was fortunate that the device continued to work without an RS485 ground!

Now - the amplifier has it's own digital circuit (for the front panel display etc.) and I'd taken my 5V supply directly from the same supply that's powering this board, and not used a voltage regulator or capacitor on my controller's board - so would adding a voltage regulator and capacitor to my board (possibly) do the trick?

I know this is probably all utterly naive for many of you, but thanks so much for the help - I very much appreciate it!

Oliver

---------- Post added at 12:27 ---------- Previous post was at 12:20 ----------

However it seems that other forum member know about this more than I do.
Good luck

Who knows! You haven't really given me anything to go on - though yes, other's have made an attempt.

Many thanks for the good wishes though.

I'm using a PIC16F88 and a 75176AP transceiver if that's the kind of info you wanted.

 

[KerimF]

Hi Oliver,

Is it possible for you to know the maximum supply current of your controller?

By the way, since there is a 5V regulator on the display there must be a source of a higher voltage behind it. Could you measure it?

Knowing both answers will help a lot to build a good supply for your controller with no interference... I hope it will be done, since I don't know all the details yet to be sure 100% ;-)

Kerim

 

[ohall]

Hi Kerim,

I'll try and measure the supply current consumed by the controller, and I'll try and trace the supply - shouldn't be too hard as the amp circuits are pretty easy to trace.

Many thanks again for the help and I'll report back when I've been able to do some more investigating

Cheers,
Oliver

 

[ohall]

I read 24mA between the Amp +5v and my controller's +5V input terminal.

I've also investigated the amplifier itself a bit more, and found that the display PCB (which brings together an LCD display, and another micro - and very little else) doesn't have a voltage reg on it at all, and the micro, LCD, IR receiver etc. all take their power directly from the same trace that I've hijacked.

This power comes from the main amplifier board which appears to have all the power amp bits and pieces, and I assume the DC voltage rectification and regulation. This board is powered by the amp's trasformer which is putting out approx 31v and 11v AC. This board also powers another pre-amp board with + and -8V and an "analogue" ground. I'm pretty sure I'm not affecting the pre-amp board as my noise isn't amplified in any way as the amp's volume is increased, and it appears there isn't any seperate gain control in the power amp.

I was hoping to avoid messing around with the power amp/power board, but if I understand your previous advice correctly, are you suggesting that I should :-
1) Identify the 5V reg on the main power board
2) Take my power feed from the input side of this voltage reg
3) put it through another voltage reg with a 100uF cap on the input side
4) Feed my controll from the output of this

I just want to check I'm on the right track - I've got a few spare 7805's and caps, so I should be able to try that out when I get another spare hour :wink:

Again, many thanks for the help!
Oliver

---------- Post added at 17:17 ---------- Previous post was at 16:42 ----------

Right - I've found the voltage regs on the main power board. From the transformer, there's a bridge rectifier arrangement with approx 14v DC on the output - this goes to a series of voltage rectifiers ( 7805 and 7808 ) with caps on the input and output, which then feed the other two boards.

 

[KerimF]

I see, so you will try to locate the 5V reg on the main board.
you will likely see a coupling capacitor at its input and perhaps a power resistor (or ordinary one) between its input and the high voltage supplying it. This resistor helps decreasing the input voltage and the heat dissipation of the regulator. I am saying this as it is in the general case... but you know there are always exceptions.

Kerim

 

[ohall]

Correct again - two big power resistors from the +/-11v AC input leading to the rectifier.

Many thanks Kerim.

 

[KerimF]

If you will connect the supply wires to the 14V, try to choose the ground closest to main supply of the amplifier.
Since the controller current is only 24 mA, the power dissipation (controller regulator) will be about (14-5) * 24 = 216mA ... it is not high, no heatsink is needed.
But I still recommend to add a cap at the input of the controller reg.

 

[ohall]

It seems so obvious now that we've had the conversation :wink:. Is this (i.e. giving each section it's own regulator and smoothing capacitors) the "usual" way to isolate noise between different sections of a design?

 

[KerimF]

I think also the seperation of grounds play an important role in isolating possible interferences among the main parts of a system.

edited:
As possible, the ground tracks form a star with the ground node of the main supply.

Added:
You got it well... but sometimes I forget this even in my designs :smile:
... this is important to me because I have to build my circuits using 1-layer PCB.

 

[ohall]

I think I understand - so each section's ground goes directly to the main supply ground, rather than taking a route through/near some other section?

 

[ohall]

... this is important to me because I have to build my circuits using 1-layer PCB.

Me too! :smile:

 

[ohall]

Hi Kerim,

I took power and ground from as close as I could to the rectified 14v DC supply, put it through a 5v voltage reg with a 100uF cap on the input and... just the same :-(

I have to admit, the noise is fairly minimal, but it's annoying me that I can't eradicate it. It's only perceptable at very low listening volumes though, so I guess I may just have to live with it.

I hate being beaten by stuff like this! Many thanks for all the help though - I feel like I've learned something even so.

Cheers,
Oliver

 

[KerimF]

Let me see... I wonder if you will get a difference if we use a coaxial cable (as for audio) from 14V to your controller where the 5v regulator is added.

Added:
Perhaps there is a portion of the fast rising/falling currents on your controller that still finds a path from the controller 5V trail back to the 14V ground but via the data ground (RS485).

 

[FvM]

I reviewed the discussion and got the impression, that the contributions are mainly guessing about the exact interference path. Some general meaningful suggestion have been made, but they are only valid, if the underlying assumptions about the crosstalk mechanism are correct.

I would expect a circuit diagram and at best a photo of the setup to be able to make precise suggestions.