Crossover checking - Measurement vs. simulation

[nasinn]

Hi all,

I recently posted this on the Lansing section and did not get much traction in my thread there, so I thought I would give it another go here.

Essentially I picked up a pair of old JBL L65As and I am trying to evaluate if the crossover need some parts replaced or not. I recently got an oscilloscope / signal generator and had a bit of a play with one of the speakers. What I did was feed through my amp a sine wave of different frequencies and recorded the peak to peak voltages of the input terminals and of each of the driver terminals. The scope does not have one of these automatic response plot generator so it was a bit of a manual changing of frequencies and noting the Vpp values. With this Vpp information of the input vs. the drivers I calculate the voltage gain https://en.wikipedia.org/wiki/Gain_(electronics) plot, see below. One note is I don't think I was quite at the center of the brilliance / presence pads, but I can redo this all if there is any value in this sort of analysis.



In addition to those measurements I also set up the crossover schematics in KiCad and ran a spice simulation of the frequency response and added to the plot results. I was a bit unsure on the L4 connection as the original schematic has some strange inductor symbol (I am not an electrical engineer - but my googling tells me its a tapped inductor but it was not clear to me where to connect the R7 to, but the sim did not ran when I connected it at the bottom of L4).



Not sure what the value of all this is, or if I was just keen on trying out the oscilloscope haha... My hope would be that this could give some indication of the state of the capacitors and the crossover in general so I can take an informed decision on what to replace if that is needed, so I am open to suggestion on how to interpret if at all possible the state of the crossover from the above.

Cheers,
Jonas

 

[grindstone]

Curiosity is good. Hands-on is good. An XO with an autoformer (tapped inductor) is not the simplest place to start. Your road ahead depends on what's interesting to you--follow that In the mean time, listen and then ask old Lansing-heads over at the heritage site what to do for your parts & speaker. To measure each of your parts at the component level will require more gear (or at least more software) and you will ultimately need to measure speaker responses to know what it's really doing. Keep it fun and congratulations on your oscilloscope! You are metaphorically standing on the edge of an enormous and slippery slope--one false move and you will have 3 new sub-hobbies and you may not come-out for years (if ever). If you just want to restore the speaker and move on, that's another thing--do that, sell your scope, and run

 

[elnaldo]

Hello. I use ARTA software suite, it will send a pink noise or sweep to the speakers (through an amplifier), and capture the output with a microphone, giving you a freq response graph. Attached is a graph I made to check some crossovers last week, to do the same you are doing, and concluded the capacitors are in good shape.

To arrive to that conclusion, I've also measured the capacitors with a multi component tester, and in addition I compared freq response measurements at each individual driver using the original capacitors and using new capacitors. Even if the capacitance drifted a bit from the nominal value, ESR was OK and the freq response was virtually identical, so I concluded the original capacitors are still fine.

If you are doing voltage measurements, I'd focus on the -3dB, -6dB and -12dB points (you could calculate them based on the peak value, and then just sweep the freq until reaching those values). That way you can draw an approx crossover point and crossover slope without taking too many samples. Also, if you have new capacitors at hand, you could compare the readings using old vs new capacitors (same thing I did)

To simulate the response, you need the FRD and ZMA data files for each driver. I'm not sure how accurate a simulation could be without that data.

But, IMHO, the microphone-software measurement will be easier and more accurate.


(TECHNICS SB-1970 CROSSOVER, SPL NOT CALIBRATED, measurements are approx)

 

[gadget73]

One wrinkle with sims vs reality is the driver load isn't constant so the turnover points and levels are likely to not match perfectly.

Fair bet if you sub dummy load resistors for the drivers the numbers will match better.

 

[w3jn]

Your generator likely has a source impedance of 50 ohms, and an amp provides a much lower source impedance. KiCAD likely presumes a low source impedance as well. Source impedance will certainly impact the response of the crossover.

Put an 8 ohm resistor across your generator output (ie the input to the crossover) and see how the curves change.

Another potential gotchya here is the X2 relationship between power and voltage, when measuring dB. Were you to plot power curves, the dB drop would be half your voltage measurements, so your differences (and dB curves) show a more dramatic difference.

 

[rothwellaudio]

Your generator likely has a source impedance of 50 ohms, and an amp provides a much lower source impedance. KiCAD likely presumes a low source impedance as well. Source impedance will certainly impact the response of the crossover.

That's not what he did.

What I did was feed through my amp a sine wave of different frequencies and recorded the peak to peak voltages of the input terminals and of each of the driver terminals.

 

[nasinn]

Thanks for all your inputs,

Some takeaways if I can summarize the above + my thoughts on it:

• Watch out for the rabbit hole - will do, I need to make sure that this audio hobby does not take away too much time from the other ones
• Microphone measurements would be the next thing if I were to go crazy with the gear, although I think that would be a bit far at the moment for me. My first priority is just to make sure that the crossover is not completely out of wack, not optimizing the speakers in the room... yet?
• Good points on the drivers not being ideal, so maybe substituting plain resistors would bring it closer to the sim (because in the sim I am just running the drivers as pure resistors)
• On the impedance of the source, like mentioned above I ran it through my regular Audiolab 6000a amp so hopefully not too far off the sim, but for sure not 1:1 match in terms of characteristics

So what I think I will do is try to just measure the capacitance directly if my multimeter does that correctly, the scope / signal generator is actually just one of these cheap handhelp OWON HDS200 series things, so need to see if it can do that.

Other than that to see if it matches spec I just need to confirm the the crossover frequencies are at 1 and 6.5kHz, is there a way I can calculate that from the graph above? Can the crossover frequency be defined as a drop X in dBV from the peak of each driver?

Cheers

 

[w3jn]

That's not what he did.

Thanks! I missed that. My comment on voltage dB vs power dB stands however.

 

[Refugee]

I would think that changing C1,2,3 would benefit you in peace of mind and longevity of service. I have never tested the reality of a crossover against the design spechs. I am sure they are different and commend you for the effort and curiosity. Either way change out the caps while you are in there. If C1,3 are not film caps then replace them with film. Do One speaker at a time so you can listen to the difference after you have the first one done.
Enjoy!