A possible approach to adding a silent start/shutdown to the ACA

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Administrator
Joined 2007
Paid Member
Digital logic with its normally hard and fast rules really is a different ball game compared to analogue and all its intermediate states.

I'm not rally sure what to make of it tbh :) (hint... if you want to simulate using a sine wave output then selecting a much lower frequency of say 100Hz will speed the simulation time up no end)

The goal in the simulation is to get a flat line output when no signal is applied, meaning that the speaker sees absolutely no voltage change. That doesn't happen I'm afraid.

You also seem to have the same issue that I keep getting in that the simulated relays/switches etc allow the turn off pulse through. Some hazy thought has just appeared to me... is it because the rail has disappeared and the switch needs the supply present in order to work in the simulation i.e does the switch have zero resistance with no supply present. I wonder.
 
Member
Joined 2011
Paid Member
... is it because the rail has disappeared and the switch needs the supply present in order to work in the simulation i.e does the switch have zero resistance with no supply present. I wonder.

If you use the NC (Normally Closed) contacts of the relay to connect an X ohms resistor across the speaker, then the job of the control electronics is to de-energize the relay coil at power-off. Which is easy to do even with a collapsing power supply.

To simulate this using the LTSPICE voltage controlled switch as a handy substitute for a relay: connect the switch control circuitry to the always-present +24V output of the SMPS, upstream of the front panel ON/OFF switch.

If that disgusts you, build a little subcircuit in LTSPICE that implements a current controlled switch, using the LTSPICE voltage controlled switch and a small number of additional components. It might be as simple as a single resistor and an LTSPICE voltage controlled switch, with cleverly chosen switching threshold (model parameter Vt). This current controlled switch does not need any supply rail to operate correctly; when the input current is sufficient, the relay closes. By the way, that's how real-world relays work too! When coil input current is sufficient, the electromagnet's field is now strong enough to pull the armature.

One way to simulate NC relay contacts is to swap the + and - control voltage input pins on the LTSPICE voltage controlled switch, and change the Vt parameter appropriately.

Another way to simulate NC relay contacts is to define "Ron" to be 66 megohms and "Roff" to be one milliohm. Aha, now open means closed and vice versa.
 
Last edited:
Administrator
Joined 2007
Paid Member
Another way to simulate NC relay contacts is to define "Ron" to be 66 megohms and "Roff" to be one milliohm. Aha, now open means closed and vice versa.

Thanks Mark :) That's actually the way the sim is set up. I have found some weirdness in the simulation that I need to try and figure out. The relay works as expected with a static control voltage, however if I source the control voltage from a 'pulse generator' set to unmute at say 12 seconds then I find the relay doesn't work even though the control voltage has flipped.

Crazy thought... its behaving like a latching relay. Hmmm... is that possible ?

I've actually posed this relay question over in the LT thread so I'll see what develops.

Very true not acceptable in commercial product. Some people put up with it but I wouldn’t, especially at some of today’s prices. I think the mono ACA is a cute package though.

Very cute, yes. It would be great if this noise aspect of its operation could be sorted easily though.
 
One quirk of the hugely popular ACA that seems to regularly get a mention relates to the strange noise it can make at power on and the loud thump that occurs at switch off. This kind of behaviour is actually 'normal' for pretty much any AC coupled amplifier.

Yep. The output cap has to fill with charge somehow. Unfortunately, the charging current flows through the speaker.

Just a quick thought while skimming through your thorough work: What about putting that solid state relay in parallel with the speaker and shunt that thump to ground?

That might work in a tube preamp where the output current is limited. In a solid state power amp like the ACA, you'll end up with full signal across the cap if someone turned on the source while the relay was closed. I doubt either the cap or the amp would appreciate that.
I suppose you could add a mute relay on the amp input, but it may be worthwhile to take a step back and look at which problem you're trying to solve exactly.

I for one would not like to wait half a minute for my amps to come on... (well, I'm already annoyed by two seconds turn-on delay :rolleyes:)

In that case, an AC coupled amp may just not be for you. Or you'll have to accept the turn-on/off thump.

It's perfectly possible to build a speaker protection / turn-on thump eliminator circuit with MOSFET switches and floating photovoltaic drivers. That's what I did in my Guardian-86. I wasn't able to measure any meaningful difference in THD vs output power between the output of the amp and the output of the speaker protection circuit.

I would shy away from mechanical relays. As this guy demonstrates, they're not great for speaker protection due to their tendency to arc over, which causes contact wear (and resulting distortion) and may not actually protect the speaker. I'd certainly perform thorough testing on a batch of relays before committing to use one in production.

Tom
 

Attachments

  • Guardian-86 Rev. 1.0_ THD+N Input vs Output (1 kHz, 20 kHz BW, 8 ohm).png
    Guardian-86 Rev. 1.0_ THD+N Input vs Output (1 kHz, 20 kHz BW, 8 ohm).png
    40.1 KB · Views: 283
Last edited:
Administrator
Joined 2007
Paid Member
With many thanks to Hans Polak over in the LT thread and a little tweak of the switch simulation parameters and we have a great result. This is how the real relay would behave.

(I still don't understand why the simulated relay isn't behaving as I expect but that's one for tomorrow probably)

Anyhow, this simple version should give a silent start up and shutdown and is non invasive to the ACA.
 

Attachments

  • Capture7.JPG
    Capture7.JPG
    280.3 KB · Views: 303
  • ACA1 plus Shorting Speaker Relay Simple Version.asc
    5.2 KB · Views: 71
Administrator
Joined 2007
Paid Member
Doing something like that would certainly alter things at power up but how would you do that without greatly altering things.

Ideally you want something very small (and that generates no losses) that can be added to a working ACA with minimum effort.
 
It would alter things, but would it be for the better ? :)

A mosfet in the power line controlled by an r/c constant ?
The power switch then becomes a trigger for the mosfet soft start/shutdown system.
Could then use a relay at either end for full power supply on or disconnect.

Yes I know there's a lot of ifs, buts, maybes, etc in that and it could all be an absolute waste of pixels even typing this :D

Expanding on this, something like a cl60/50/40 ala pass labs power supplies, with relays again for full power or shutdown ?
 
Last edited:
Administrator
Joined 2007
Paid Member
No way of knowing without actually trying it or setting up a simulation.

If you leave the SMPS on all the time and ramp the rails up and down very slowly then I'm sure you would quieten things down, but it leaves the problem of the ACA producing grossly distorted output while the rails climb and fall.

You could probably minimise the on noise by adding an NPN transistor and RC delay on its base to supply the 10k (R7) with a slow ramping voltage. The off noise would still be an issue though.

Any solution should be fool proof and work under any conditions and it also shouldn't need special conditions to be present for it to work successfully (such as making sure no audio input was present at power on and power off).

To me the shorting relay is an ideal option, and its non invasive to add to the ACA. It also allows a very short settling time (so no long waiting) for the ACA to stabilise. The quality of the relay is of secondary concern... all it needs to do is present as a short circuit during mute.

The series relay (solid state for consistency and reliability) is a technically more challenging solution but it should work in practice although you have the drawback of needing to have a large initial delay for the cap to charge via a dedicated resistor.
 
Oh I agree with you entirely.
When trying to find a solution to a problem I tend let my brain go wild.
Some things are doable, some are just pure stupidity :D
There are far greater minds here than I could even begin to speculate about.

Me, I was ready to throw in a simple speaker protector on mine.
Then I saw your musings and thought I'll wait a bit to see what happens ;)

In the direction of the output shorting relay.
I'm sure using a transistor in place of the relay would make things simpler, faster, better. (edit, second thoughts, no it probably wouldn't)
Wait, didn't someone (you ?) already suggest that ?
 
Last edited:
Member
Joined 2011
Paid Member
One tremendous advantage of using the NC contacts of a relay is: it's completely out of circuit, completely, when music is playing and the output is not muted.

A solid state switch would contribute capacitance and leakage current to the output, and worse yet, since the loudspeaker output swings from 8V below ground to 8V above ground, you'll have to figure out how to bias the solid state switch OFF when the voltage across it is +8V and also when the voltage across it is -8V. Bipolar transistor? Gotta generate yourself a -9V power rail. IGBT? even worse. SCR/Triac? fuggeddabouddit. MOSFET? don't forget that pesky VGS "Absolute Maximum Rating". Oh snap!
 
mosfet speaker `relay'

Relays suitable for amplifier outputs are large and expensive. Much better to use a pair of mosfets.

Attached is a scheme to:
- Short the output to ground with m1,m2 (via R2) to charge up the output capacitor, and
- Connect the output after a delay (via m3,m4).
In order that the output is shorted at power on, a pair of depletion mode mosfets are used for m1,m2.

The vom1271 photovoltaic generators create a floating 8v supply to turn m1,m2 off and m3,m4 on. vom1271 are in short supply at the moment, but APV1121 is equivalent. Note the opposite connection for the DN2540s. M3,M4 need to be low Rds on (<15 mOhm).

The size and cost of 4 mosfets and two vom1271 is about the same as a decent relay. Drive the vom1271's with 20mA after a 1..2 sec delay.

The small white box shows what you would use for a headphone amp.

Regards,
Paul Bysouth
 

Attachments

  • Output-Relay-Test-1.JPG
    Output-Relay-Test-1.JPG
    405.1 KB · Views: 311
Member
Joined 2008
Paid Member
That might work in a tube preamp where the output current is limited. In a solid state power amp like the ACA, you'll end up with full signal across the cap if someone turned on the source while the relay was closed. I doubt either the cap or the amp would appreciate that.
I suppose you could add a mute relay on the amp input, but it may be worthwhile to take a step back and look at which problem you're trying to solve exactly.

Add a muting relay for the input then ;).


In that case, an AC coupled amp may just not be for you. Or you'll have to accept the turn-on/off thump.

I'm currently living with the thump, but I'd certainly appreciate a simple and reliable way to avoid it.
 
Administrator
Joined 2007
Paid Member
The short on the output shouldn't be a problem, even if the amp was fed with a very high input signal.

This shows the current draw of the ACA delivering its maximum voltage swing into 8 ohms. You can see the current rise and fall from the nominal bias current value. Second image is the PSU current when the amp is shorted.

The circuit topology of the ACA means it is inherently 'safe' working into a short circuit, Definitely DO NOT use this approach on a class AB output stage.
 

Attachments

  • Capture1.JPG
    Capture1.JPG
    246.1 KB · Views: 248
  • Capture2.JPG
    Capture2.JPG
    235.3 KB · Views: 254
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.