Relay switching and transistor question; keeping it quiet

[wizard333]

I'm trying to add a relay to switch between two 100K master volumes on a concentric pot. They share ground. I'm running it off a -15VDC supply, 1n4005 and cap across the relay coil.

I couldn't get it to STFU in terms of the pop when the switch was made, and it was much louder with any signal than without. Used various caps across the switch, which helped a tiny amount but not much. Since I'm switching two pots that share a ground, everything should be ground referenced through the pot. The inputs will see 100k to ground, the wipers (much) less. Not sure why the popping noise on this, I've done plenty of switching before and never had an issue as long any caps involved had a ground reference and signals had some reference to each other through a large resistance. Here that should all be taken care of by the fact that nothing is ever floating and there are no caps involved.

So since I couldn't figure that one out, I tried switching using a transistor. Tried it on the downside of the relay between the coil and ground, and also between the -15v supply and the coil. The problem I seem to be having there is with getting the transistor to saturate correctly, because even when the base is lifted from ground there is a large voltage drop across the coil or across the supply, enough to keep the coil charged in fact (release voltage seems to be very low on these 12v relays, on the order of a couple volts). Also, iirc there should be a resistor between the base and ground, but no matter what size resistor I try, it leaves a large enough drop across the coil to charge it and the switching doesn't work. If I ground the base directly it works, but there is a pop that is as least as loud as not using the transistor to switch.

Tried a couple different 12V relays, a couple difference little NPNs, same results.

Any ideas? Should I ditch the transistor switching anyway since it doesn't seem to be any more quiet? Any ideas why the loud pop wont go away even with everything ground referenced and caps across the diode and switch? If I keep the transistor switching, what is the proper way to saturate it so that when the base is lifted from ground there is no large drop across the relay coil? The formulas that I was using don't even look close to right in terms of the results. For reference, the coil uses 150mW (at 12V that's 12.5ma), min HFE on the transistor is 300 on the data sheet.

[loudthud]

What is the input to the two volume pots? Are the connected together with a single cap to a common source?

From your description I assume a SPDT relay is selecting the output of one pot or the other. What is connected to the output of the relay? Does it have a DC path to ground? While the relay is moving it won't be connected to either pot.

[wizard333]

DPDT; both inputs and outputs are switched, they aren't in parallel except the grounds.

I'll check for ground reference at the ends; the pot should supply ground reference but as you say, there are a few milliseconds during switching where it might not have that. Theoretically though, after a switch or two wouldn't the cap be charged and the effect decrease?

[Enzo]

No, you need to terminate any relay contact that is left hanging to keep charge on the cap. Your pop is probably from the contacts in the signal path switching unterminated caps, rather than something about the coil drive.


Ever notice on some amps they switch a cathode bypass cap in and out of a preamp stage to boost gain? And further notic they don't just switch the cap in bare, they have a largeish resistor to ground under the cap, and the switch shorts across the resistor. The resistor keeps the cap charged, but also is large enough the cap doesn't affect the circuit. SHort the resistor, the cap charge doesn't change, but now it is full involved. You need to do something like that.

Look at the Peavey 5150 for example, the FX loop jacks. There is a relay switching th loop jacks in and out. Note the send and return each have a resistor to ground to keep caps C11 and C65 charged.

[Enzo]

And you mention it is louder under signal.

The signal is a constantly changing voltage. Even if your coupling cap has some charge on it, any instant the circuit that is switching can be at some other voltage anyway.

[wizard333]

Ever notice on some amps they switch a cathode bypass cap in and out of a preamp stage to boost gain? And further notic they don't just switch the cap in bare, they have a largeish resistor to ground under the cap, and the switch shorts across the resistor. The resistor keeps the cap charged, but also is large enough the cap doesn't affect the circuit. SHort the resistor, the cap charge doesn't change, but now it is full involved. You need to do something like that.

Yes, I do that all the time without issue.

I guess I figured the ground of the pot provides reference and that a cap "unterminated" for 3ms or less isn't an issue. I'll pop a resistor in there tomorrow and see for sure.

[d95err]

I suggest posting a schematic of the circuit; it would make it much easier to help.

Most amps that switch mv has the two pots in paralell and switch only the viper. Using a dpdt seems to be inviting trouble.

As an experiment, I suggest temporarily disconnecting the relay and connecting a manual switch instead. That will indicate if the problem is in the signal switching or in the relay circuit.

Good luck!

[Enzo]

Your pot doesn;t provide a ground reference to something not connected to it, like the cap from the previous stage.

Are the two pots going the same place? Like selecting between two masters Is there just one feed to both? Are both going into the same power amp? And the same signal is routed to one or the other? COnsider an option then. Wire the two pots in parallel with the signa permanently fed to both. Then switch which wiper goes to the PA. Back to the Peavey 5150. They did this. Look at VR6, VR7 Clean and Ultra post controls. The signal from the preamp feeds both, and the relay selects one. That eliminates switching the incoming signal. Note R81 to keep C52 charged.

[kleuck]

If grounding or not the base of the transistor changes the pop, then your switching circuit is not well isolated from the preamp supply : isolate it with a regulator or resistor/cap.. Did you put a reverse diode across the relay's coil ? Mandatory. Said before, but as the "signal pop" goes, it's always better to bypass a big resistor always to ground than to truly switch between paths.

[kleuck]

REMOVE the cap across the coil.

[R.G.]

The advice you're getting has been good. Without a schematic, we can only guess.

There are several causes for popping.
For signal coupling caps there is almost always a DC voltage difference from one end to the other. Otherwise, you would not need the cap in most cases. The cap charges to a DC voltage across the cap that's the no-signal difference in DC from one end to the other, and stays there.

When you open one end of the cap by switching, the cap voltage stays the same, but starts leaking down because all caps leak, and even if they were perfect, some voltage would leak through the air around it. When you re-connect it, it's voltage has changed, and that difference, the amount it's changed, is dumped into the signal line as a sudden step in signal voltage. That causes a pop.

The cheap solution to this has been to tie the end that's opened to the DC voltage it would have in normal operation with a high-value resistor. Usually this is ground. The resistor value doesn't matter a whole lot, but in tube circuits, it's usually better if it's big, 1M to 4.7M being a good start.

For relays themselves, the coil switching signal is a big, instant step in voltage, both switching on and off. There is always some capacitance between the coil and the signal contacts. Some relays for very sensitive signals actually have metal shields between the coil and contacts that can be grounded for just this reason. The sudden change in voltage on the coil is coupled through the capacitances to the relay terminals.

How big the pop is depends on how big the voltage change on the coil is, the size of the capacitances (which we have no control over), and the impedances on the relay terminals. It works like signal coupling: the resistance on the terminal and the capacitance to the coil determine the lowest frequency that can be coupled. In tube circuits with 1M and higher impedances being switched, this is about as bad as you can get in simple circuitry. The lower the resistance to ground on the relay terminals, the more coil pop becomes a tick instead of a pop.

Since the capacitances are sensitive to the frequencies being sent through them, slowing down the edge of the coil voltage works too. This is why there are those big capacitors across the coils of many tube amp relay circuits. You can also use a simple slow-down circuitry on a transistor driver for the coil if you use a transistor driver.

When you turn the coil off, the coil inductance tries to instantly force the same current to flow by inverting its voltage and pulling on the driver. This is the inductor's flyback effect, and it's instant unless you do something about it. One thing you can do is clamp it to the power supply voltage with a flyback diode. This doesn't do anything about the speed, but does reduce it's size.

Another is to slow down the change in current by slowing down the current change on the coil at the driver transistor. Another is to "tune" it with a capacitor across the coil to make what would be a sudden step be a slow, lazy wiggle that doesn't couple through the terminal capacitors very well.

Finally, there's coil current noise. Relay contacts use lots of current, and you switch it instantly (in most cases). This is a classic case of big changes in power supply and ground current, which can cause sudden steps in voltage across the resistance of the wires to and from the coil and driver. If any of your signal circuits share these wires with the relay coil and driver, this can be coupled into the signal path audibly. Good wiring practice of not letting the relay share power or ground wires with the audio circuits can more or less completely fix this.

[wizard333]

Most amps that switch mv has the two pots in paralell and switch only the viper. Using a dpdt seems to be inviting trouble. As an experiment, I suggest temporarily disconnecting the relay and connecting a manual switch instead. That will indicate if the problem is in the signal switching or in the relay circuit.

Interesting thought to just switch the wipers. Although if I do that, I'd have to use a concentric twice the size of the original pot to get the same load right? They'll function as 2 resistors to ground so in parallel, twice the size will be required to not change the load on the previous stage. Further, the pot not being used would serve as a resistor bridging the active pot, which would result in pushing the taper in the direction of reverse audio. If using audio pots, that might result in something almost linear I guess.

Are the two pots going the same place? Like selecting between two masters Is there just one feed to both? Are both going into the same power amp? And the same signal is routed to one or the other? COnsider an option then. Wire the two pots in parallel with the signa permanently fed to both. Then switch which wiper goes to the PA. Back to the Peavey 5150. They did this. Look at VR6, VR7 Clean and Ultra post controls. The signal from the preamp feeds both, and the relay selects one. That eliminates switching the incoming signal. Note R81 to keep C52 charged.

Yes same place, they take the place of a single master that goes from the effects return recovery tube to the PI. There is only a single path in and out.

If grounding or not the base of the transistor changes the pop, then your switching circuit is not well isolated from the preamp supply : isolate it with a regulator or resistor/cap..

Not a lot of room in this puppy for a regulator; can you explain what you mean by resistor/cap used for isolation? Did you mean just a resistor from the DC power supply to the relay?

Did you put a reverse diode across the relay's coil ? Mandatory.

Yes, see my first post.

Said before, but as the "signal pop" goes, it's always better to bypass a big resistor always to ground than to truly switch between pths.

This isn't clear; can you be specific?

REMOVE the cap across the coil.

Why? I've always used them in the past on relays and never had an issue; in fact removing them almost always increases noise quite a bit.

[wizard333]

Ok so I removed the transistor switching, changed it so that the pots are in parallel and only the wipers are switched, referenced the .01uf blocking cap from the PI to ground with a 1M resistor just in case the switching time leaves it unreferenced. Still using a 22uf caps across the coil diode and the switch jack. Switched from 100k audio pots to 250kA (measure more like 200k) so in parallel the load on the previous stage is about the same.

The pop when an instrument signal is present is no more than the pop when there is no signal, which is good, but it still gives a little "click" on one side and "whoomp" on the other side of the switching.

I'm also using two 470 ohm resistors in parallel to isolate the -15v supply from the relay coil (supply actually measures over -16v). Should I be using a diode there to, and/or a separate cap to ground after the isolation resistor?

Any suggestions for quieting that down further? If I go back to the transistor switching, how can I make sure the transistor is fully saturated so that it acts as a switch?

Thanks for the replies!

[R.G.]

Ok so I removed the transistor switching, changed it so that the pots are in parallel and only the wipers are switched, referenced the .01uf blocking cap from the PI to ground with a 1M resistor just in case the switching time leaves it unreferenced. Still using a 22uf caps across the coil diode and the switch jack. Switched from 100k audio pots to 250kA (measure more like 200k) so in parallel the load on the previous stage is about the same.

Can you post a schematic? I'm getting a bit strained imagining the circuit from the descriptions.

The pop when an instrument signal is present is no more than the pop when there is no signal, which is good, but it still gives a little "click" on one side and "whoomp" on the other side of the switching.

Get out your multimeter and measure the DC on each pot wiper in both relay positions. Little clicks are what happens when you interrupt/connect signal voltages suddenly, and may not be removable. "Whoomps" are what happen when you switch DC levels suddenly. One of your relay positions is switching in/out a sizeable DC level.

I'm also using two 470 ohm resistors in parallel to isolate the -15v supply from the relay coil (supply actually measures over -16v). Should I be using a diode there to, and/or a separate cap to ground after the isolation resistor?

??
Can we get a schematic of the relay and its drive circuit?

If I go back to the transistor switching, how can I make sure the transistor is fully saturated so that it acts as a switch?

You can measure what you have. If the transistor collector is higher than the base voltage when it's "on", it's not saturated.

You can also ensure saturation by providing a minimum of about 1/10 the collector current (that is relay coil current at this voltage across the coil, determined by the power supply voltage and the relay coil resistance) into the base when you want it on. Most modern transistors will saturate at a gain of 10.

[wizard333]

Scanner isn't working at the moment but I can give you a block diagram:

Supply: [-15v supply(measures about -16.4vdc)] [235 ohms] [relay coil bridged by reverse biased 1N4005 & 22uf cap] [switching jack to ground bridged by 22uf cap]

Signal: [.047 blocking cap from effects recovery] [concentric 250k pots (measure closer to 200k) with inputs and grounds in parallel, wipers switch by relay] [.01 blocking cap to PI referenced to ground with 1M resistor]

There is NO DC on the pots.

You can also ensure saturation by providing a minimum of about 1/10 the collector current (that is relay coil current at this voltage across the coil, determined by the power supply voltage and the relay coil resistance) into the base when you want it on. Most modern transistors will saturate at a gain of 10.

Yes; right, but my calculations seem to be off and I'm not sure why. I'm using this type of configuration:



except using a NPN with a negative supply voltage. For R1, I'm using:
R1 = Supply Voltage / ( Maximum Current Required / Minimum HFE * 1.3 )

Where supply voltage after the drop is about 12v, coil current is 12.5ma, HFE from the data sheet is 300. That yields R1 of roughly 220k (closest value) but that doesn't work, in fact the only R that does work is 0. Where am I going wrong there?