Power On/Off Thumps and Plops

[abbey road d enfer]

Why not put a small electrolytic (1 or 2.2 uF) after R96 and lose C82 and R89?

For what purpose?

 

[Winston OBoogie]

I went with the expensive option when I had to worry about this in a commercial product two decades ago - Universal Audio 2108 mic amp with external supply.

Basically as others have mentioned: An audio mute relay with slow on and quick off. The relay was driven by a 555 timer for the delayed "on". Disconnecting DC then immediately (or as near as damn it) shut the relay off.

A few dollars here and there probably doesn't matter in a one off DIY project. For a commercial unit, there are less expensive ways.

I simply did it the way I did because it was a last minute add-on. I stupidly neglected to think about the possibility of a customer plugging or unplugging a hot DC supply without using its AC power switch until I'd finished the pcb layouts and we'd already sent off gerbers and ordered.
So I did a wee 1" X 2" board with the quick "fix" and mounted it at the back of the unit.

 

[dyamakuchi]

Relays (usually mechanical) on the 'main' outputs are the generally accepted way the higher end units do this. Transistor switches work too. The question was: "WHY does this happen?"

The short answer is: power on transients can cause circuits to behave in other than ideal fashion. You want to keep the audio out of the speakers until it's good. That's probably _not_ going to be during rails ramping up.

using a perfectly symmetrical circuit.

The power up transients are not necessarily symmetric tho. Often, the positive rail has a higher load from being used to drive LEDs logic, etc. and might have more filter capacitance. This can _cause_ otherwise symmetric supplies to unbalance during power up.

 

[abbey road d enfer]

The power up transients are not necessarily symmetric tho. Often, the positive rail has a higher load from being used to drive LEDs logic, etc. and might have more filter capacitance. This can _cause_ otherwise symmetric supplies to unbalance during power up.

I wrote: "perfectly symmetrical circuit". Of course it excludes unbalanced loads on the rails. There are a few proofs of the validity of the concept, mainly power amps, but also some esoteric preamps..

 

[Trouble Report]

A cap on the base of Q19A is just going to delay the turn-on of Q19. Once the cap reaches Vbe, Q19 is going to turn-on relatively quickly. So, I don't gain anything vs. just delaying the enable line assertion from my MCU.

The purpose of C82 is to slow the transition of the switch, not to delay the switch.

I thought the point WAS to delay the turn on of the relay.

 

[JohnRoberts]

As I shared back in May I have given this much thought while at Peavey designing value gear. Customers for inexpensive gear care about turn on/off behavior too.

A couple more observations; some op amps have much better turn on characteristics, wether by design or accident. IIRC the 4560 op amp was favored for it's benign turn on/off characteristic.

Another contributing factor is how do dual supply rails come up and go away. With asymmetrical loading one supply will generally collapse before the other. This can increase or decrease turn off transients. Half wave rectified supplies by definition will power up one rail half a cycle before the other.

I recall my senior mixer engineer working on a PS control circuit that would specifically sequence how the rails come up. Ultimately he abandoned it as the benefit was not worth the complexity.

Another thought, when designing a large mixer/console there can be tens of active audio outputs. It is impractical to mute every single audio output line, so on/off muting gets focussed primarily on lines typically feeding amplifiers and loudspeakers.

JR

 

[JohnRoberts]

I thought the point WAS to delay the turn on of the relay.

The general approach is delayed un-mute upon power up, and immediate mute upon power down.

I generated the instant off by feeding a dedicated unregulated PS rail using a diode and small reservoir cap that would discharge well before the main PS. Turn on delay is relatively easy using some combination of RxC.

JR

 

[abbey road d enfer]

Looking at what other designers do, I have quite often been surprized by the complexity of their solutions. I decided it was too hard for me to grok, so I went with my minimal solution using a zener and a transistor with said RC. The only issue I had was the zener noise, amplified by the transistor, that managed to pollute all the circuit. Fixed with a capacitor.

 

[Newmarket]

I thought the point WAS to delay the turn on of the relay.

There are different things being discussed here -
eg
a) delaying turn on to allow capacitors to charge / Power rails to stabilise after turn on etc.
b) ramping a 'switch' to avoid a too fast change in level that results in an audible click. Instead transitioning over a few ms eg for a Mute function.

 

[JohnRoberts]

There are different things being discussed here -
eg
a) delaying turn on to allow capacitors to charge / Power rails to stabilise after turn on etc.
b) ramping a 'switch' to avoid a too fast change in level that results in an audible click. Instead transitioning over a few ms eg for a Mute function.

a) yes
b) proceed with caution

I killed a bunch of brain cells thinking about that. When using JFET shunt mode mutes, turning on/off too slowing can introduce audible distortion. I did a bunch of bench work on this for cheap (value) noise gate designs. I found I could reduce the audibility of the distortion during transitions by feeding 50% of the AC drain voltage into the gate (old trick). Another old circuit trick to reduce severity of clicks is to apply some HF pre-emphasis to audio before the mute/switch, and complementary de-emphasis after the hard switch. This way the signal is restored to flat, while the click gets low pass filtered.

Still never going to get a smooth fade from JFET shunt, for that use a VCA (impractical for output mutes).

JR

PS: Years ago when the big IC makers asked Peavey engineers for advice I suggested building benign turn-on/off characteristics into their new output drivers.

 

[Newmarket]

I recall that the (now obsolete) SSM switches eg SSM2404 were popular at one time with some UK desk manufacturers. AMEK and DDA come to mind.
Switching over some tens of microseconds ?

https://www.analog.com/media/en/technical-documentation/data-sheets/SSM2404.pdf

 

[MisterCMRR]

You might check out this optically-isolated bi-directional switch. It holds off ±60 V, has on resistance of 7 Ω typical, and takes less than 1 mA to drive ... and costs well under $1 at one-piece. It's the IXYS CPC1006 (data sheet attached). I've used them in a couple recent designs (not audio).

 

[JohnRoberts]

I like that the price doesn't break the bank... ($0.82)

JR

PS; I've use some opto controlled triacs, for power control and that simplifies avoiding mains voltage.

 

[dom]

IXYS is a pretty neat company, too. Quite a few interesting, niche, power products. They're one of the few players still making decent power MOSFETs explicitly intended for linear-mode operation.