Compression driver protection

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Those tweeter protection circuits both look like they are designed to short out the tweeter when activated with an RC network to delay turn on.
So they both depend on a resistance in series with the tweeter or they will short out the amp and possibly damage the crossover as well.
In general short circuiting protection is reserved for line of last resort protection (such as power supply crow-bars), however this may not be a problem in this case as compression drivers have a higher sensitivity than other drivers so it would be an unusual speaker design that did not have an attenuator resistor in series with the compression driver.
Circuit A would be my choice of the two as Circuit B looks like Capacitor C will present a load across the tweeter regardless of if it is triggered or not and will damp out the high frequency response of your tweeter as a consequence.

Depending on the rectifier bridge used in either design, there may be highish capacitance from the bridge which may also roll off some top end. - although having said that, this would normally be in the order of a couple of hundred Pico-farads so shouldn't be significant (I just mention it here as a potential design consideration).

A lightbulb in series with a tweeter will provide some protection (and is commonly used) but will cause attenuation as it heats up (and a sag in top end) so the trick is to choose one that does not heat up much during normal usage but will still heat up under abuse conditions.

My usual protection circuit involves an SCR and Relay and is somewhat more complex (but occupies little room on the crossover board) but has the advantages that it will trigger quickly and does not cause high frequency sagging under normal use:

Passive%20Xover%201A.jpg
 
Like my circuit, D-fend is line (amplifier) powered - although you can also connect a local power supply.
any affect on fidelity is likely to be so insignificant as to be not worth worrying about - certainly less than an effect of speaker cables etc.
while I do not know how much D-fend draws, in my case I aimed at greater than 100 Ohms load (greater than 140 ohms when not tripped) which in the scheme of things, compared to a typical 8 ohm load is insignificant.

There are some good low power Modern DSPes and at lower speaker levels the DSP does not need to be active anyway so I would not expect D-fend to present much extra load.
 
Two good points, that's interesting. You have experience building these circuits? What does the power circuit look like? I'm assuming the relatively delicate digital processors operate on DC, and the incoming AC line is a multitude of frequencies and voltages. How do you design a power supply for that environment? I know just enough about circuit design to suspect it might be difficult, but I'm not sure.
 
Just for the sake of the exercise lets just think about what might be required.

I am thinking while the DSP does want some computing horsepower, it may not need an awful lot in the scheme of things, as it is only analysing the incoming signal, so it can be a fairly low powered device.

Having said that you probably want to go with a 24 bit ADC and processor to suit, it is only looking at a single channel so only one ADC needed.
We also want a DSP that can run stand alone (or self boot).

The USB interface and associated smarts does need more power, but this would only be needed while the DSP is being programmed so we can discount that power requirement during normal use.

So choosing a DSP at near random (I have no idea what parts are used in the D-Fend, nor am I spending a lot of time designing this at this stage) from the Analogue devices stable, I would start with the ADAU1701 which has dual 24 bit ADCes (so actually you can make it smarter and monitor Bi-amped speakers, which I think the D-fend can) and runs off 3.3v with a max supply current of 85mA, by the time we add an EEProm lets make that an even 100mA.

Throw in power supply monitoring, to ensure the DSP Resets cleanly and doesn't try starting under low voltage/low signal conditions so lets be generous and add another 50mA which should also cover anything else I haven't thought of like Op-amps on the front end.

so we need 0.495 Watts to run the brains and while a relay will draw a lot more power, this will only be applied in a 'fault' condition or mute mode so its effect on the signal is almost irrelevant.

the little bit of information I could glean from the D-fend Data is that it becomes active at 8V RMS (which corresponds with the external power supply requirement of 12V DC) so assuming an 80% efficient DC-DC convertor we need about 51mA.

if you use a current limit circuit - like R4, Q1, ZD1 and R5 in the circuit I presented above then at higher power levels this effective load will increase and Q1 (or its equivalent in the D-fend) will need to dissipate up to 16 Watts at 300 Volts input and in the tripped state probably about the same again for the equivalent current limit for the relay, but as the case is all heatsink this is not likely to be a problem.

This gives us an effective load of around 233 ohms (I rounded up to 55mA) at low power rising to something like 5K4 ohms at full power rating which is insignificant against even a 16 ohm speaker system.

As I said, this is all rough back of the envelope figures, so there may be other factors I have not considered.

While I should encourage folk to buy my crossover PCB with tweeter protection circuit off Ebay (shameless plug here - search for "Speaker crossover with active tweeter protection" ;)) even I have to admit the D-fend looks interesting.....
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.