tmuikku

Isolating DUT from signal source after transient
« on: June 13, 2019, 07:14:48 AM »
Hi!

I've been trying to figure out Gunnes Focusing https://www.fulcrum-acoustic.com/assets/pdf/whitepapers/improving-loudspeaker-transient-response-with-dsp-2005.pdf . One scenario mentioned in the paper is to remove some LTI reflections from compression driver and horn combination. I had an idea if some of the horn reflections Gunnes is fixing could be measured with the compression driver itself, since I can't imagine I could measure them (or identify them from a measurement ) with a measurement microphone. 

Anyway, I've been using a simple 1.5v battery + toggleswitch  to provide a transient signal to a compression driver and record the impulse with mic pre connected to the driver terminals. The measurement shows all kinds of resonances, since there is no damping for the driver other than the mic leads attached (with U-pad). Some of the resonances measured must be the driver diaphragm resonating, but there might be the horn reflections recorded as well.

My procedure has been like this: switch the battery on, start measurement (ARTA) procedure and switch the battery of to get an impulse measured. The driver is not in normal position when doing it like this, since there is DC potential while starting the measurement / transient.

Tried to use amplifier to provide a proper signal from computer, but cannot measure simultaneously, amplifier dampens the whole thing.

I was wondering if there was a simple (or complicated) electrical circuit that would enable me to switch the signal generating part effectively out from the circuit (DUT) automatically after the transient signal is over and before the reflections happen so that the measurement of the reflections could be taken?

Round trip from diaphragm to horn throat and back would be maybe ~10cm, ~0,3ms, so the transient should be over and "signal generator" out from the circuit before that.

Signal could be provided through a power amp, or with a battery + some circuitry to get DC free impulse.

I've found out that a Monostable multivibrator could provide a transient, but cannot figure out how to best utilize it in this application. I tried quick simulation of simple battery + capacitor + switch jig with TinkerCAD for DC free transient, but the capacitor would be in parallel with the driver while taking measurements so probably not worth trying. Also TinkerCAD has only piezo speaker in the library,  impedance of which is unknown, and didn't have time to make electrical model of a speaker driver with passive components :) So, thought to ask from the GroupDIY for tips.

Thanks, any feedback welcome!
« Last Edit: June 13, 2019, 08:31:55 AM by tmuikku »


Dan Kennedy

Re: Isolating DUT from signal source after transient
« Reply #1 on: June 13, 2019, 09:49:56 AM »
How about the output of your monostable drives an open collector transistor tied to the speaker motor.  When the output transient of the monostable is low it switches on a cmos switch like a DG408 to connect the speaker motor (now measurement mic) to the input of your data collection equipment?

Put some transient limiting diodes to protect stuff, run it all off +/-12 volts you should be fine.

JohnRoberts

Re: Isolating DUT from signal source after transient
« Reply #2 on: June 13, 2019, 10:15:33 AM »
IIRC David Gunness rolled his own DSP to handle the processing required to linearize the speaker interactions.

JR
Don't only half-ass tune your drums. Visit https://circularscience.com to hear what properly "cleared" drums sound like.

tmuikku

Re: Isolating DUT from signal source after transient
« Reply #3 on: June 14, 2019, 10:20:27 AM »
Thanks guys!
 
Jeah I propably never get this sorted out, Gunness seems to be very clever and has a lot of experience on the field. In contrary, I don't have any experience on this topic. It has been very much fun to try to figure out the stuff, I feel at least 10 years younger after having studied this and done some experiments at home. I haven't had time for DIY audio at least in 6 years :)

Anyway first step would be to try to measure and then identify such reflections from the measurement. I don't know if I'm able to do the DSP to fix those thought, but it would be very cool to see what is the frequency response of a reflection compared to the initial transient for example. I might be able to do that in GNU Octave (similar to Matlab), which is another fun exercise I'm looking forward to.

Dan,
thanks for the suggestion! I probably did not understand all of it. Did you mean something like this (simplified)? https://www.dropbox.com/s/nunhe9r9iqlt18y/example1.jpg

Seems complicated and open collector switch probably hooked wrong?

I think the battery 1.5v is plenty for the driver impulse signal, and measurement probes can always be attached.  Only the signal generator should be switched out.

Could something like this work? https://www.dropbox.com/s/rlwd8a1zcxeguz2/example4.jpg?dl=0

Eliminate monostable if battery + capacitor + switch could act as a impulse source? Multiplexer could also go since probes cab be always attached. I can build monostable if it is required. Trying to get it as simple as possible.

I've got no clue yet if the open collector switch can be used like in my illustrations?

edit: can't get more than one image here.. added the two linked images combined in one.
« Last Edit: June 14, 2019, 10:40:54 AM by tmuikku »

tmuikku

Re: Isolating DUT from signal source after transient
« Reply #4 on: June 14, 2019, 10:45:18 AM »
Hah, the simplified, the open collector switch is of no use now that I look at the picture.
Anyway, perhaps ditch the battery thing and build proper thingamajig where I could use signal from computer, through an power amp.

Dan Kennedy

Re: Isolating DUT from signal source after transient
« Reply #5 on: June 14, 2019, 02:20:23 PM »
I was thinking a pnp open collector to the top of the speaker from 5v or 15v depending on level wanted. The CMOS gate would disconnect the input of the data collection gear during the impulse, reconnect after so as to not overload/saturate and slow response.

The pnp when off disconnects the speaker from the generator, ie the 5 or whatever volt supply.

JohnRoberts

Re: Isolating DUT from signal source after transient
« Reply #6 on: June 14, 2019, 03:12:55 PM »
I'm not sure I want to encourage this pursuit but you should probably drive the speaker from a low impedance coming and going. If you hit the loudspeaker driver with a pulse and then let it flop around unterminated it is likely to ring adding artifacts to your measurement.  or not...

I am not a speaker guy but Gunness is one of a small handful operating at that level these days.

JR
Don't only half-ass tune your drums. Visit https://circularscience.com to hear what properly "cleared" drums sound like.

tmuikku

Re: Isolating DUT from signal source after transient
« Reply #7 on: June 14, 2019, 06:06:08 PM »
Dan, thanks for the details, I'll try what happens, I've got an L-pad for the driver, dampens it as well. John, duly noted.

Thanks both of you!


 

Related Topics

  Subject / Started by Replies Last post
3 Replies
1455 Views
Last post March 15, 2006, 05:29:53 PM
by Svart
2 Replies
1111 Views
Last post September 21, 2006, 01:40:05 PM
by guavatone
5 Replies
1749 Views
Last post October 10, 2006, 01:03:12 AM
by dukasound
4 Replies
2398 Views
Last post April 04, 2008, 05:11:35 AM
by clintrubber