Relays popping in a 3 loops looper

[arma61]

*** edited again (sorry ) ***


Hi guys! hope you are all well!!

so, before to declare myself defeated by relay switching.....

Got this problem of relays popping and all the tests I've done didn't solve it. Here what I've done till now.

a) Switching section
this one from Geofex.. thx R.G
http://www.geofex.com/article_folders/fxswitchr/fxswitchr.htm

b) the relays section is a mix of crk found on the net, is GGG looper with dip-switch and relays, driven by an ULN2803.

a) is battery powered 9v, then 7805 for logic switch
b) is powered by a) with a 10uF coupling(?) cap on b) +/- pad.

Power in section b) is powering
1 LED x dip-switch
1 ULN2803
1 LED x relay
3 relays

Signal ground through the relays/loops is completely disconnected by the power of the circuits.
Signal in relays is grounded when loop is off.
All relays have flyback/snubber diode
** Relays are rated 6V, I'm switching them at 5V

** EDIT
No effects in the loops, just 3 jack patch connecting sends and returns

*** EDIT 2
all test are done with following setup
a signal generator (again thx RG - battery powered
an LM386 based headphone amp - battery powered

Still I can hear pops in the audio path, tried to

get rid of the 2 led of one loop, still popping the same as the other loops.
connecting signal ground to circuit gnd, popping!
put a 10k in serie with "+" of one relay and a 10uf to ground, popping!
switching loops on/off several time, no change in popping quality/quantity
** edit **
1M resistor in or out or snd or rtn to gnd

Testing signal/ground voltage with a meter set on mv, I can see 0.001 when switching the loops.
I've noticed that the OUTs of the ULN2803 when activated by the logic on INs don't get 100% to ground, I can see till ~0.7V

May be bad relays?
What else could I try?

Thanks for help
Cheers
Armando

[Hatredman]

Did you try 1 Mohm resistors to ground on one of the relay contacts?

[arma61]

Did you try 1 Mohm resistors to ground on one of the relay contacts?

yes, tried also that (edited my post with this and other info)

Thanks for reply!

[Groovenut]

If there is DC on the signal line that will cause a pop. You need to find how the DC is getting into the signal path and fix that. Also you dont want to leave the signal ground floating while you test. The main signal output needs to be connected to an amp or other pedal to insure the signal ground is made.

[R.G.]

Relays, just like hard-metal-contact footswitches will never be perfectly silent, for reasons having to do with the suddenness of the contacts making and breaking.  But the pops are small and intermittent from just the contact. Relays do have problems with induced voltages caused by the large voltage swings on the coil being coupled to the audio path by capacitance between the coil and the audio-carrying metal in the relay. There are also other sources of pops, like grounding and power supply issues.

I didn't notice any mention of coil-induced noise suppression in your post. Is there any? Can you post a schemo of the drive and relay coil connection for one of your relays?

Relays vary a lot in how much crosstalk from coil to contacts they have. The exact relay will matter.

[arma61]

If there is DC on the signal line that will cause a pop. You need to find how the DC is getting into the signal path and fix that. Also you dont want to leave the signal ground floating while you test. The main signal output needs to be connected to an amp or other pedal to insure the signal ground is made.

in deed this is my problem, cannot think where DC come from being the signal hot+ground not connected to circuit's power.

Also you dont want to leave the signal ground floating while you test. The main signal output needs to be connected to an amp or other pedal to insure the signal ground is made.

I meant that all GNDs IN,OUT,SNDs, RTNs are connected to eachother and with the amp gnd, and not to circuit's power, so signal gnd is't floating

[arma61]

Relays, just like hard-metal-contact footswitches will never be perfectly silent, for reasons having to do with the suddenness of the contacts making and breaking.  But the pops are small and intermittent from just the contact. Relays do have problems with induced voltages caused by the large voltage swings on the coil being coupled to the audio path by capacitance between the coil and the audio-carrying metal in the relay. There are also other sources of pops, like grounding and power supply issues.

I didn't notice any mention of coil-induced noise suppression in your post. Is there any? Can you post a schemo of the drive and relay coil connection for one of your relays?

Relays vary a lot in how much crosstalk from coil to contacts they have. The exact relay will matter.

Thx RG, I'm happy you join the thread

Will post a schemo tonight, the relays are driven by an ULN2803 pushing one coil connection to gnd, the other connection is going to +v shared by a resistor for each LED, and there's a  flyback/snubber diode on each relay, isn't this latter something to avoid coil noise?

Thanks!

[R.G.]

A snubber diode is a beginning, at least. What's really going on is that the relay coil is an inductor, and in some cases a big one.

When you turn the voltage on to the relay coil, current ramps up at a rate of di/dt = V/L, somewhat tempered by the relay coil resistance. As current rises the portion of the applied voltage "eaten" by the coil resistance gets bigger and bigger until all of the voltage is across the resistance. There is a time constant to this, t = R/L, and the time til the current gets stable is about five time constants, about 5*R/L. The voltage is applied suddenly across the coil, so any capacitive effects allow the "on" transition to be coupled to the audio path from the drivers, independent of the coil.

When you turn the drive to the coil off, the inductance expends the energy stored inside the M-field on trying to keep the current flowing without a sudden change. That is what causes the voltage across the inductor to suddenly invert and have it pull on the driver to try to suck the same current out of the driver. The math says that the voltage across the inductor goes to **anything** needed to make the current not be discontinuous. For an inductor with one terminal connected to a positive voltage and turned on by an NPN pulling the other terminal down to ground, turning the coil drive off makes the transistor end flip up above the positive voltage to any voltage needed to make the current not change suddenly. If the transistor is all that's keeping current from flowing, the inductor voltage rises until it breaks the collector junction.

Hooking a reverse-biased diode across the coil lets the inductance keep its current flowing by jumping suddenly to the voltage that turns on the diode, saving the transistor. The current then ramps down at a rate of di/dt = V/L, but in this case, "V" is the diode forward voltage, not the power supply voltage. Diode snubbing keeps the current flowing longer, perhaps by many times longer than the relay "on" conditions. So the relay opens more slowly than it closed because the current changes more slowly.

Reverse diodes are first of all a protective device, keeping your driver transistor from dying. They reduce the voltage transition size as a side effect, but only on relay opening. They don't help you on the relay closing. And with latching relays, you get one of each operation on each relay action.

There is also a minor contribution to noise of the coil magnetic field inducing a voltage in the actual contact metal. Varies with the relay, like all the rest of this.

The real culprit seems to be the capacitance between coil terminals and the signal-carrying metal parts of the internal switch and the external resistance of the signal-carrying lines to ground. The actual capacitance from coil to audio signal contacts is a few to a few tens of picofarads. We artificially force the resistance to ground on the audio lines up to a meg or so, so the capacitance and line resistance make a high pass filter of a meg and some picofarads.

Doing a little math, 1M and 10pF is a nominal 15.923kHz highpass. So any signal through that filter would be attenuated by 6db at that frequency, less attenuated above that, and with increasing loss below it, losing another 6db for every octave going down. Since a vertical edge on a signal contains harmonics of every frequency, a truly vertical edge will cause a tick that has components matching the attenuation curve of the 15.9kHz filter. For a nominal 5V change on the coil contacts, you'll get an instantaneous blip of between 2.5V and 5V from turning the relay on.

If we slow down the edge of the relay "on" signal, it removes high frequency components from the "on" signal. So if we ramp up the voltage across the coil, the "on" signal contains less high frequency content to couple into the audio path. A cap across the coil helps a lot. But the driver gets whacked HARD by a cap across the coil because the cap acts like an instantaneous short circuit to the driver. My preferred solution is to slow down the drive to the driver, or to slow down the driver itself. That's what that cap from driver capacitor to base is for on my switching diagrams. The drive transistor feeds the coil a slower ramp, and the coil capacitance and line resistance highpass  can't couple the remaining lower frequency components of the coil drive into the audio path. Some people use a resistor and cap in series across the coil, some just put on a cap and tell the driver to deal with the current pulses at turn-on.

Another way of dealing with this is to lower the resistance to ground in the AC signal lines. If you terminate them in 100K, not 1M, the high pass rolloff point moves up to 159khz, and this cuts the "tick" power by 20db. This works if you're not dealing with raw guitar signal, which will lose treble to the 100K. Mostly in a looper you're dealing with raw guitar, so this is probably out unless you put a buffer in front of the loops to keep the guitar's treble happy. If you put in a buffer (effectively meaning you don't use a Fuzz Face type distortion circuit in the loops) then you can use 10K pulldowns and the ticks are history.

How's that for a complicated answer to your simple question?       

[amz-fx]

Armando,

It is most probably contact bounce in the relay. The springy metal that is being pulled by the coil is hitting the contacts together and they rebound, then come together again, and maybe repeat the process two or three more times.  This produces the noise in the audio signal path.

Pulldown resistors are not going to do much for this. The best solution is to have a circuit that mutes the output for a few milliseconds while the relay settles down, though this can introduce other problems. The Boss jfet switching is good solution all around.

In my experience, if you are playing a big amp at stage volume, any true bypass 3PDT or relay is going to make some level of noise when it switches. It probably is not audible at bedroom volumes but when you crank up, the noise becomes apparent (to one degree or another).

Best regards, Jack

[arma61]

A snubber diode is a beginning, at least. What's really going on is that the relay coil is an inductor, and in some cases a big one.
.....

How's that for a complicated answer to your simple question?       

Indeed lots of things to think and study on....

Thanks RG, will have a deep read and make test to see if those 100k/10k will do any better.

Indeed I may switch back to the transistor version (so to slow down relay activation) instead of the ULN2803.

Thanks again

[arma61]

Armando,

It is most probably contact bounce in the relay. The springy metal that is being pulled by the coil is hitting the contacts together and they rebound, then come together again, and maybe repeat the process two or three more times.  This produces the noise in the audio signal path.

Pulldown resistors are not going to do much for this. The best solution is to have a circuit that mutes the output for a few milliseconds while the relay settles down, though this can introduce other problems. The Boss jfet switching is good solution all around.

In my experience, if you are playing a big amp at stage volume, any true bypass 3PDT or relay is going to make some level of noise when it switches. It probably is not audible at bedroom volumes but when you crank up, the noise becomes apparent (to one degree or another).

Best regards, Jack

Thanks for reply Jack, in deed this is the base for an 8 loops looper, I'm trying to build for a friend of mine who will use it almost on stage, and i cant immagine the pop I'm hearing on the headphone amplified on stage .

Will study the jfet solution later on, would like to stay on this version having already "lost" months on it

Thanks

[Groovenut]

I'm still thinking the DC on the signal line needs to be eliminated before any other remedies are tried

My 2 cents...

[R.G.]

And as noted, DC on the signal will itself be heard as a pop. GN is correct - all DC has to be eliminated to get rid of that source of noise.

Switch bounce is much the same issue with hard-contact footswitches and relays. Switch bounce makes the pop string out in time, but does not itself cause noise. It does make any pop that's already there sound worse, but creates no (electrical) noise of its own, except in certain very odd circumstances. Switch bounce does make DC-caused ticks and pop worse by making more of them. It doesn't do anything to capacitively coupled coil transients.

The last generation of pedals I designed for the company I work for has relays in it, just to make the folks who keep asking for "true bypass" happy, and it contains noise suppression muting too. But this set of relays (relays vary!) and slowed-transient drivers has almost no audible pop/tick even without pop/tick suppression. We tested the same circuit (and same boards) both ways. We left the suppression in just in case we have to change relays or get a funny batch of them.

There are many sources of noise. They all have to be suppressed for quiet operation.

[amz-fx]

Thanks for reply Jack, in deed this is the base for an 8 loops looper, I'm trying to build for a friend of mine who will use it almost on stage, and i cant immagine the pop I'm hearing on the headphone amplified on stage .

Armando,

Some players are more tolerant of switch noises than others. It doesn't bother me a lot but I can see that if you have a pop that echoes through a delay pedal that it could be a problem. 

Definitely try the pulldown resistors but I think you will find that at loud stage volume, the relays will still make some level of noise, however small. Different models, types, or brands of relays are more/less susceptible to this problem.

Best regards, Jack

[arma61]

Thx everybody for replies!

did again some test yesterday, 10k or 100k as pulldown res on the loops did eat some signal and tone and a little bit of pop, but still present.

Went back to the old plan with transistors instead of ULN (based on GGG simple looper http://www.generalguitargadgets.com/effects-projects/boosters/ss-switcher/), I've noted that the relays don't even activate all the time despite the switching is perferct, (they do acrivate if I short the cap, but then they pop!).

So I'm thinking my relays, rated 6V, are the problem (or at least part of it) as, I think,  they are the only mean to get DC on the audio path, there's nothing connected in the loops, as I said, just a jack path each loop shorting snd and rtn. 

Should get soon some Tianbo relays rated for 5v 2A.... will see.... and report.

Thanks again for your help.

Cheers
Armando