Multiple Small Subs - Geddes Approach

[pjpoes]

This may be a completely unimportant issue, but what about group delay. I know that group delay at low frequencies is generally a really small to non-existant issue because of the wavelengths, but it's my understanding that group delay audibility thresholds vary with frequency. How much group delay is audible at 100hz, 80 hz, 150hz, etc. Highpass filters add group delay, and the steep the filter the more delay. If you added a 4th order L-R at 100hz the group delay at 100hz would go from around 1ms with no filter (clearly inaudible) to 6-7ms. It seems that some research on group delay indicated that between 1 and 2 cycles is just audible. At 100hz, 6ms isn't a full cycle, but it's getting close. Does this become an issue? EQ add's group delay, if you eq the mains above 100hz, could the amount it adds be enough to be audible?

[Sheldon]

Originally posted by gedlee
John

Yes, this should work as long as there is no ground path from the receiver to the amps or DCX.

Sheldon

"and lift all ground lugs with an adaptor"

How is this any different than cutting off the ground lug?

Because the series path to earth is never broken, even if the RCA's are unconnected. It's only lifted by 10 Ohms (if the resistor is used) if the diodes do not conduct, and at most by a diode drop (0.6V for a pair, 1.2V if a bridge is used). In other words, the chassis limited to a maximum potential of 1.2V, relative to earth.

Sheldon

[Sheldon]

Originally posted by Patrick Bateman


Hum is a huge issue in the car stereo world. I've used a cheap $10 isolation transformer from Radio Shack and it seems to work well.

At first I didn't like the idea, since a cheap transformer will be bandwidth limited. But if it's only used on the sub channels, it seems acceptable.

The use of a transformer on the RCA inputs to the sub would eliminate the hum, wouldn't it?

(just thinking out loud - I'm working right now and can't test it one way or the other.)

The transformer method is fine. But, limited bandwidth can be seen at either end - HF or LF. At LF, in particular, source impedance can be very important - be sure your transformers are spec'd for LF extension you want, at the impedance from your source (lower source impedance is fine, higher will increase the roll off frequency).

Sheldon

[gedlee]

Sheldon

Yes, but an adapter is nothing more that a plug without the ground lug. That was my point.

Markus

In thinking about it, there is nothing to prevent the customer from using "small" for all the channels as you suggest. They would just need to provide the "mains" measurements with this filter in place and the fitting algorithm would work just fine at fitting to that situation. It's not what I would do, but my technique works either way.

[markus76]

Originally posted by gedlee
In thinking about it, there is nothing to prevent the customer from using "small" for all the channels as you suggest. They would just need to provide the "mains" measurements with this filter in place and the fitting algorithm would work just fine at fitting to that situation.

I'm moved to tears. Looks like the idea of selling more than 3 subs got you hooked
By the way I like the idea of making the mains part of the calibration. The box without a name might not be able to do so.

Best, Markus

[Sheldon]

Originally posted by gedlee
Sheldon

Yes, but an adapter is nothing more that a plug without the ground lug. That was my point.

I understand. You can cut off the ground lug, or use a 3prong to 2prong adaptor (for US, anyway). If you do that, the chassis is usually still earthed through the signal connectors. If the signal connectors are too small to carry a fault current, or are not connected, or are isolated, the chassis could float up to line potential in event of a fault. Depending on the construction and probablity of a fault that energizes the chassis, this may or may not be a significant safety hazard.

If you leave the earth connection in place, but "lift" it with a resistor, antiparallel diodes, or combination, the signal grounds are isolated from the earth connections. But the chassis still has an effective, high current path to earth, which prevents it from ever floating to an unsafe level.

Sheldon

edit: even used in some sophisticated measuring gear (page 10): http://richpc1.ba.infn.it/~atauro/in...ding/Earth.pdf

[gedlee]

Originally posted by markus76

By the way I like the idea of making the mains part of the calibration. The box without a name might not be able to do so.

Best, Markus

That would be a big mistake, since in the modal region one cannot make any assumptions about how the mains will work at the "crossover" point. They have to be part of the calibration or it isn't a calibration of the system.

[markus76]

At least it will be a calibration of the subs. We will see.

Best, Markus

[markus76]

New spectrogram with findings that might surprise some people here. This time the DVD "Slumdog Millionaire" (chapter 4, ≈4 min)—C, L and R:



Best, Markus

[markus76]

FYI—a new box with a name:
http://www.svsound.com/products-parts-subeq.cfm

Best, Markus