Switching currents

[modulay]

Reading about led indicators (peak,vumeters...) I've found some comments about need of avoid leds switch currents flow through gnd...So, some led driving circuits I've seen use switching bjt's or simply wire the diodes to supply rails and not to gnd.

And this is my question...is this absolutely needed?

I think that a good pcb routing,which separates audio gnd and switching elements gnd routings is okay for ensure a good operation and hold safe audio signals.

Then...what about relays?

I have some relays on my design, and they are driven by pnp trt's, so relay coils are directly connected to ground

 

[JohnRoberts]

You always need to account for non audio signals in proximity to audio paths.

There are lots of different strategies. One I've used to both manage noise and efficiency is to string LEDs in series and drive with a single current source with shunts across LEDs to turn on/off. Actual current draw doesn't change, but gets steered around LEDs

Another design I did drove the leds between opamp outputs. That way the current always flowed from +V to -V.

Many ways to skin that cat but yes pay attention to where the current goes and edge rates for capacitive coupling, or inductive grounds.

JR

 

[Mark Burnley]

Hi,

This is quite a big area to cover, but here are a few basic thoughts.

90% of the time we are dealing with a piece of equipment with an unbalanced audio path within the equipment. Most "balanced" in and out gear has bal>unbal and unbal>bal electronics at input and output.

This means that the return path for the signal is the common ground, which includes the 0V rail from PSU, through PCB's, interconnecting cables, cable shields etc. etc.

If you have a long skinny ground track running the length of a PCB with a sensitive preamp or gain stage, and hang one side of a relay coil returning to this ground track, then you have a high likelihood of interference when you operate the relay. This is because the current flowing in the relay will no doubt be much higher than the quiescent and operating level of the audio electronics. This will cause voltage drop across the ground trace, and this voltage difference will be "seen" by the amplifier circuit as part of the signal.

Also, in a poorly regulated power supply, a higher current being supplied can show up flaws in the regulator and increase the ripple voltage, which again will show up in the ground rail.

I always run a a separate ground/0V wire from the PSU 0V terminal directly to any "control" equipment- LED's, lamps, relays etc. If possible, I'll run the +Ve line from a separate regulator- e.g. use a 7805 for a +5V control circuit supply (handy for LEDs and small relays, also good for any logic bits and pieces in your design)

An item such as a relay also has what is called a "back-emf" which is the current which is stored within the relay coil. When you switch the supply to the coil off, the magnetic field collapses and slices through the windings inducing an opposite polarity voltage across the coil. This can play havoc in audio equipment- the voltage spike can be much higher than the Vs. It's common to connect a backwards-connected (w.r.t the supply line) diode (e.g. 1N4004 in parallel with the relay coil to clamp this back-emf. Again, good grounding which avoids a direct path through analogue-land helps a lot.

I've spent many happy hours rebuilding the grounding bus in old desks which had relay muting and patching. Everytime a switch was pressed, a CLICK came out of the monitors :roll:

Putting in a separate ground bus and supply regulator sorted this out entirely.

Check out some of the excellent pdf's knocking about the 'web on "mixed signal techniques" (texas instruments??) and any analogue-with-digital design papers.

What you're actually dealing with here is a digital (i.e. on/off signal) superimposing itself on an analogue signal, so a lot of the info for mixed signal design may seem a bit over-the-top, but makes a lot of sense- separating "analogue" ground from "digital" ground and isolating sensitive (i.e. high impedance and high gain) analogue circuitry from on/off higher-current areas.

...lots more to add here though!!

Mark

 

[JohnRoberts]

This is a bit beyond the OPs question but yes indeed signals routed within a single product's chassis are rarely balanced, but prudent design uses differential receivers or transmitters as appropriate if signals are sent any appreciable distance to forward or back reference the two local grounds to each other. Ground integrity then need only be managed as a local issue. Trying to brute force ground integrity the full length of a large recording console is an invitation to problems and I don't doubt you saw some if a designer used that approach. These differential circuits can be incorporated into existing buffers and gain stages and typically cost pennies for a few extra resistors if adequate active stages are available.

There needs to be collaboration between the mechanical packaging and circuit design to parse the circuits for ground and differential management. Experienced designers have typically figured this stuff out by doing it wrong once (or more times for some).

A good mental exercise is to think of ground traces as a string of resistors (which it kind of is)...Good design will work with modest ground traces, but robust traces and good design is even better.

JR

 

[modulay]

[quote author="JohnRoberts"] One I've used to both manage noise and efficiency is to string LEDs in series and drive with a single current source with shunts across LEDs to turn on/off.
JR[/quote]

Could you post an image that show this?

 

[JohnRoberts]

I don't have anything handy, and drawing a schematic is like work $$$. I've done several variants on this over the decades. Simply visualize a string of LEDs in series. With a current source at the top and the bottom cathode tied to ground.

A simple way to shunt the current is using something like a LM339 comparator which has open collector outputs. If all of the open collectors are hi (open) all the LEDs are on. As the 339 outputs sequentially close, the LEDs get turned off one at a time depending on how you connect them up.

There are practical limits to how many you can put in series due to voltage build up, etc. But this isn't rocket science, you can parallel a few strings for lots O' leds..

JR

 

[tomelectro]

JohnRoberts wrote:
One I've used to both manage noise and efficiency is to string LEDs in series and drive with a single current source with shunts across LEDs to turn on/off.
JR


Could you post an image that show this?

Take a look at the meter drive circuit for the DBX902 de-esser (this unit has been discussed here quite a bit - there should be a schematic floating around here somewhere).

 

[Mark Burnley]

...and many Soundcraft desks e.g Series 5- all the LEDs in the channel strips are run off two current sources- each switch is then shorting out each LED (makes good use of PSU voltage!)

Google for a simple circuit- usually one transistor with 2 resistors and 2 or 3 diodes (or a zener)

Mark

 

[PRR]

> is this absolutely needed?

Don't do brain surgery in a dance hall.

Sure, some delicate surgery has been done near battlefields.

And yeah, I have boxes which thump signal ground with switching events, and sometimes nobody ever notices.

But if you have a choice (in audio electronics, you often do), move the stomp-dance to another floor, off your fine-work floor.

 

[modulay]

Understood. Very illustrative,PRR.

Well,I toke a look at the schemo suggested by tomelectro.
It appears incomplete (I drawed the red wire,which I think must be in the original schematic).



It seems that leds are driven by each anode comparator output high level....nice solution

I have another question
What's de advantage of positive feedback on comparators?

 

[bcarso]

[quote author="modulay"]
I have another question
What's de advantage of positive feedback on comparators?[/quote]

Keeps them from chattering/oscillating at the switching thresholds.

 

[modulay]

Sorry if this is a newbbie question but then....does that feedback turn the comparator to schmitt trigger type?

 

[bcarso]

Yes---although typically we think of schmitts as having a ponderable amount of such feedback. Here you don't need much (few millivolts usually) to fix the oscillations. National Semi has some discussions about this, I think in the LM139/239/339 datasheet.

BTW those using simulators will find some of them don't handle this part well at all with respect to such feedback.

 

[ClaudeTucker]

In the SSL 4 and 6K the logic supply "hung" 5V below the +18V rail. Although it was labeled "+13V" on the panel it was actually -5V below +18. It's "ground" (actually logic return) was 13V.

One way to avoid ground currents with transient loads is to just not have them return to ground. Of course, almost everything coming off the logic supply had to be level shifted and the console was full of BC212's just for that. Almost every FET switching function had a BC212.

 

[modulay]

I've found some problems with relays.
Both (polarity relay and MIC-HI Z switch relay) make monitors CLICK.

This is the relays board schemo, which has 3 signal conectors...one for mic in, one for DI in, and one output for connecting to the pre input



And this is the board layout...there are no signal grounds on this board.Center pad of three signal connectors is used to shield the wires inside de box (the shields are not connected to nothing in the other side of the wires)

.

Relays use a psu regulator only for them,different of audio rails ones.
JP5 gnd pad is wired to ground star point (power supply gnd).

Anyway, I think that relays tied to ground may be guilty here...

:sad:

 

[JohnRoberts]

#1 consider adding some capacitance across relay winding to slow down edges.

#2 don't commingle relay ground with signal ground, if you have to share one pin for ground bring currents back to that pin on separate traces. It looks like you have two pins labeled ground, I guess one goes to switch, other to actual ground?

If this is the one place in your system where you tie power ground to signal ground, do it but be careful about layout.


If you cut a narrow slot in the ground run at the very top of image above RELE FASE, it will partition realy ground in the path to the left and signal grounds in the path to the right, and break a potential loop (albeit pretty small).

JR

 

[modulay]

[quote author="JohnRoberts"] #1 consider adding some capacitance across relay winding to slow down edges.[/quote]

How do you suggest to do it?
A capacitor paralleled to the relay would provoque high current transients when transistor switch, wouldn't it?

[quote author="JohnRoberts"] I guess one goes to switch, other to actual ground? [/quote]

That's it.
JP5 GND pin is wired to supply ground (system star point), and JP4 GND pin is wired to the switch which operates polarity relay (RELE FASE).

[quote author="JohnRoberts"]If you cut a narrow slot in the ground run at the very top of image above RELE FASE, it will partition realy ground in the path to the left and signal grounds in the path to the right, and break a potential loop (albeit pretty small) [/quote]

Ok, I will try.
But this board has no signal grounds.
GND pads on signal connectors (JP1,JP2 and JP3) are used to shield de wires...in this case,could shields induce non desired effects on signals?

 

[JohnRoberts]

I would add a cap across PS on the relay board to keep current transients local. Then caps across relay may not need to be large.

I don't know whether noise is electrostatic, magnetic, or simple common conduction.

If you have the board try stuff...and listen to see what makes a difference.

JR

 

[nickt]

JLM Audio has a neat trick where relays are driven in series by a constant current source. On Off the coils are short circuited. There's a schemo on the JLM site http://www.jlmaudio.com/JLM99v.htm. Not sure how well this works, just saying it seems like a cool idea.

 

[modulay]

Here again

Very interesting these schemos, nickt. Added to my collection :grin:

Things done,by now:

1) Cutted gnd plane track referred by John.
2) 100 uF bypassing capacitor added on PS board in.
3) 33 pF capacitors paralleled to relay coils.

The problem looks to be fixed...now, there are no clicks (thanks, John).
Anyway, I'd like to do more tests under better conditions, and I'd like to comment again about the capacitors paralleled with relays.
Are good these 33pF ones?
Wouldn't be better put resistors in series with them?