AR-3a - Impedances seen in bi-amp configuration

[onplane]

Well, it’s been one of those days! Reminds me of the hours I spent working over breadboards that just would NOT behave. (I guess this frustration is why I ended up in management rather than in the labs.)

This afternoon I devoted to measuring the impedance of the AR-3a by itself and then with the xovers split. Purpose in doing this was to see what each amp would see in a “passive bi-amp” arrangement.

Naturally, the results I got are NOTHING like I would have predicted (so they must be wrong). In any event, here is what I computed as the frequency dependent impedance of the AR-3a (with the strap in place).

(Can anyone confirm whether this looks correct? If this is "way off" then so will my other graphs.)

[onplane]

For some reason the image did NOT attach ...

It's supposed to be here:

http://www.classicspeakerpages.net/dc/user_files/1516.jpg

[onplane]

Now, below is a graph of just the woofer alone. Notice that the impedance of just the woofer by itself is LOWER than the impedance of the AR-3a!

I predicted just the opposite! My logic was something along the lines that higher impedance for the woofer by itself would begin to explain why my amp is running cooler. In short, the lower the impedance the more power dissipation in the amp (for the same output power).

Some observations:

1. Below 100 Hz there is no difference between the AR-3a and the woofer alone

2. Clearly the impedance of the woofer by itself is better behaved than the AR-3a - fewer peeks

3. I’ve long contented that I got “tighter bass” in the bi-amp arrangement – well all things being equal, more power is going to the woofers under the bi-amp arrangement

4. I also mentioned that I got more volume at the same power level and now it’s clear why that is. My power meters are volt meters (and they just assume a speaker impedance of 4 ohms). Under bi-amping for the same voltage the woofers are drawing more current and thus the increase in perceived sound.

5. Lastly the woofer xover causes it’s impedance to rise very rapidly after 600 Hz

Regards,

Jerry

http://www.classicspeakerpages.net/dc/user_files/1517.jpg

[onplane]

Now for the complete and total mystery. Below is the graph for the tweeter/mid-range. I really can’t begin to explain what’s going on here.

If anyone has some ideas, please jump in.

Observations:

1. Above 800 Hz the tweeter/mid-range pretty much track the AR-3a. Now, there is a slight difference, but I don’t think it’s significant.

2. Below 800 Hz the tweeter/mid-range is much LOWER than the AR-3a. Further, while the impedance is very low, there is very little sound! Below 450 Hz there is no sound at all!

So the huge question is where is the current going?? Further, how is it this extremely low impedance doesn’t “drag down” the impedance of the AR-3a?

Regards,

Jerry

http://www.classicspeakerpages.net/dc/user_files/1518.jpg

[onplane]

So, what can we conclude from all of this?

Well, from the beginning I’ve claimed that my AR-3a’s under “passive bi-amping” provide:

1. better transient response

2. far better stereo imaging

3. bass that is better defined

4. clearer/cleaner high frequencies

5. amps running much cooler than before

Many folks have speculated that the reason for this improvement is simply power. That is, had I a more powerful amp I would get these same results without the need to bi-amp. That argument never satisfied me, because my normal listen volume is UNDER 1 watt.

1 watt with a 4 ohm load translates to an average rms voltage of 2 volts with an average peak voltage of 2.6 volts. The rail voltages on the AR1500 are +/- 43 volts. I mean it just doesn’t “feel right” at these moderate power levels that my amp lacks an adequate reserve.

What these graphs show is that something else is going on. No matter what amps you connect to the AR-3a in a passive bi-amp arrangement, you WILL get a different portfolio of sounds from a single amp arrangement.

Now, whether it is better, is strictly a matter of taste.

For example, it’s clear that the woofers get more power when they are separated from the rest of the network. Now, I never claimed that I get MORE bass. On the contrary, I think I get LESS bass, but a “better defined” bass. It’s hard for me to explain this, but prior to bi-amping I always felt the bass was too “boomy” and so had the AR-3a’s on stands. When I noticed the nice “tight” bass (I think I mean shorter duration but higher amplitude??), I removed the stands. They are simply no longer needed.

Some people may feel that the changes that come about with “passive bi-amping” are just plain wrong. That is, we have moved further rather than closer to the idea. I can not dispute that.

What I can say is this hobby is a lot of fun and experimentation is to be encouraged. Try anything ... try everything, then share with us what you hear.

Regards,

Jerry

[rrcrain]

I'm equally as baffled. How did you go about measuring the impedance curve of the speakers? I'm curious enough to measure my AR90s the same way if I can.

Regarding the power debate...

A few weeks ago, I listened to a pair of old Ohm omnidirectional speakers being driven by a small Jolida tube amp. They sounded fine, everything was there but nothing to write home about. The shop owner switched to a larger Jolida tube amp and replayed the same piece. I was stunned by the difference in sound. The speakers were suddenly very dynamic with excellent bass. The small amp was capable of driving the speakers without any apparent distortion, but it could not produce enough current to drive the bass properly.

If you know someone who owns a large amp and is willing to loan it for a day or a few hours, it should prove or disprove the power idea.

[soundminded]

I measured the two 5" midranges sealed in a sub enclosure in AR2a as having a combined DCR of 1.6 ohms. I figured two 4 ohm speakers in parallel. The crossover network consists of a 16 ohm pot in series, not as a voltage divider, and a 4 mfd capacitor also in series and that is all (unless there is something else in that sub enclosure like a transformer I don't know about (they are sealed with that same unbelievable putty that held the woofers in so I'm not going to try to find out.) Now here's the puzzle, when I rotate the pot from 0 to 16 ohms, the volume of the drivers doesn't change, only their high end rolls off at 0 ohms. I was so surprised, I figured the pot must be defective (or my meter) so I substituted two 8 ohm resistors in series for the pot and got the same result. The second speaker system gave identical results. I do not have any explanation for this. Isn't it wonderful how after 40 years ago, there is still mystery and intrigue? If only it were so true with women.

[rrcrain]

I never fail to be mistified by my wifes thought process. She's a darling and pretty much a typical women.

Your far wiser than I if you completely understand women.

[soundminded]

Don't be rediculous, no man can ever understand women. I was being facetious. OTOH, they seem to read us like an open book. It always amazes me at the circuitous process their thoughts take. Entirely unpredictable. They make a statement and you never know where it's going. All you can do is wait and hope to find out.

"Are you angry at me? What are you angry about?"

"If you don't know, I'm not going to tell you."

[JoeB]

>Now for the complete and total mystery. Below is the graph

>for the tweeter/mid-range. I really can’t begin to explain

>what’s going on here.

>

>If anyone has some ideas, please jump in.

>

>Observations:

>

>1. Above 800 Hz the tweeter/mid-range pretty much track the

>AR-3a. Now, there is a slight difference, but I don’t think

>it’s significant.

>

>2. Below 800 Hz the tweeter/mid-range is much LOWER than the

>AR-3a. Further, while the impedance is very low, there is

>very little sound! Below 450 Hz there is no sound at all!

>

>

>So the huge question is where is the current going??

>Further, how is it this extremely low impedance doesn’t “drag

>down” the impedance of the AR-3a?

>

>

>Regards,

>Jerry

>

Hi Jerry,

It looks like you may be measuring a series resonant effect. Taken alone the 50 UFD mid cap in series with the 0.5 Ohm resistor and L1001#4 (taken at .880 mH frtom Tysontoms chart below) would be resonant at around 770 Hz with a very low impedance. The mid driver, pot, and L1001#1 complicate this quite a bit. Your graph is a real eye-opener!

Joe

http://www.classicspeakerpages.net/dc/user_files/1519.jpg

[onplane]

>I'm equally as baffled. How did you go about measuring the

>impedance curve of the speakers? I'm curious enough to measure

>my AR90s the same way if I can.

Hi, Richard!

I should have shared my method for measuring impedance, so that obvious flaws in my technique can be discussed. Equipment used:

1. old audio amp – no concern if I fry

2. audio signal generator – you can use the sound card on your computer, but I happen to have an old tube generator

3. audio patch cord to get signal from generator into amp

4. AC scope - DC scope will work as well – dual trace would reduce the time to measure by a factor of 4!

5. know resistor – any value will work, but I used 4.5 ohms as it gave me pretty good resolution on the scope screen

The known resistor is connected in series with the speaker and then directly to the amp output (see diagram). The procedure is rather simple. It just takes forever!

So let’s start at 1000 Hz. We first set the scope inputs across test points 1 and 3. (actually you can leave TP 1 in place during the entire test). Then we adjust the generator to 1000 Hz and next we vary the generator and the amp levels until we get an even deflection on the scope (maybe 4 cms, etc). Then we move the probe at TP 3 to TP 2 and we measure the % of voltage drop across the known resistor.

If the voltage drop across the known resistor is 50% of what was initially across TP’s 1 and 3, then we know the impedance of the speaker at 1000 Hz is 4.5 ohms. That is, they are equally sharing the load.

If the voltage drop across the resistor is 60%, then the speaker impedance is 3 ohms.

If the voltage drop across the resistor is 40%, then the speaker impedance is 6.75 ohms.

The formula is:

Z = R(1/x -1)

Z = speaker impedance

R = know resistance (measure it at least three times and with different instruments)

x = percentage of original voltage measured across know resistance

The hard part is we have to keep adjusting the initial input voltage to compensate for non-linear amp/load combinations. That’s why a dual trace would really cut the time significantly.

Once we have measured for the speaker system, we start all over for the pieces (woofer, mid-range/tweeter). Then when all done on one of the pieces, I’d go back to where my measurements differed significantly from the entire speaker system. Double check my reading and then manually insert a strap to verify that there really is a difference.

Richard, does this sound reasonable to you?

Regards,

Jerry

http://www.classicspeakerpages.net/dc/user_files/1520.jpg

[onplane]

>I measured the two 5" midranges sealed in a sub

>enclosure in AR2a as having a combined DCR of 1.6 ohms. I

>figured two 4 ohm speakers in parallel. The crossover network

>consists of a 16 ohm pot in series, not as a voltage divider,

>and a 4 mfd capacitor also in series and that is all (unless

>there is something else in that sub enclosure like a

>transformer I don't know about (they are sealed with that same

>unbelievable putty that held the woofers in so I'm not going

>to try to find out.)

Soundminded, if the cap is in series, the puzzle for me is how you got any DC resistance reading at all!! This should be an open circuit.

When I measured the DCR of my AR-3a's, I only get a reading on the woofer. The mid-range and tweeter are capacitively coupled and all I measure is an open circuit.

Regards,

Jerry

[soundminded]

"We measured their output level peaks to about 115 dB. Is this what you call acceptable sound level?"

Simple, I removed it. It was worthless anyway.

[Carlspeak]

>I measured the two 5" midranges sealed in a sub

>enclosure in AR2a as having a combined DCR of 1.6 ohms. I

>figured two 4 ohm speakers in parallel. The crossover network

>consists of a 16 ohm pot in series, not as a voltage divider,

>and a 4 mfd capacitor also in series and that is all (unless

>there is something else in that sub enclosure like a

>transformer I don't know about (they are sealed with that same

>unbelievable putty that held the woofers in so I'm not going

>to try to find out.) Now here's the puzzle, when I rotate the

>pot from 0 to 16 ohms, the volume of the drivers doesn't

>change, only their high end rolls off at 0 ohms.

Problem solved - Perhaps?

I don't have an original Aetna-Pollak rheostat but used one of my new 25 watt, 15 ohm units (I normally use to replace the original AR rheostats when they are unrepairable) in a little experiment. What I did was rotate the dial from 15 ohms to almost zero in 2 ohm increments and measured the inductance at each increment. What I found was inductance indeed increases with impedance - which may explain the roll off you heard. However, I'm a bit puzzled as to why the volume didn't change as well. A rheostat is nothing more than a length of wire. Rotating the dial varies the active length. An inductor is also a length of wire. Resistors are used all the time to pad tweeters in many speakers. Most audio grade wire wound resistors are pretty much noninductive, however. So, therein may lie the difference. Below is a plot of my results. Inductance tables I have indicate the xover frequency should be reduced by about 1/2 from 0-15 ohms. That's pretty significant.

>I was so surprised, I figured the pot must be defective (or my meter)

>so I substituted two 8 ohm resistors in series for the pot and

>got the same result. The second speaker system gave identical

>results. I do not have any explanation for this. Isn't it

>wonderful how after 40 years ago, there is still mystery and

>intrigue? If only it were so true with women.

Remember, it's all about the music

Carl

Carl's Custom Loudspeakers

http://www.classicspeakerpages.net/dc/user_files/1521.pdf

[JoeB]

?

>I don't have an original Aetna-Pollak rheostat but used one of

>my new 25 watt, 15 ohm units (I normally use to replace the

>original AR rheostats when they are unrepairable) in a little

>experiment. What I did was rotate the dial from 15 ohms to

>almost zero in 2 ohm increments and measured the inductance at

>each increment. What I found was inductance indeed increases

>with impedance - which may explain the roll off you heard.

>However, I'm a bit puzzled as to why the volume didn't change

>as well. A rheostat is nothing more than a length of wire.

>Rotating the dial varies the active length. An inductor is

>also a length of wire. Resistors are used all the time to pad

>tweeters in many speakers. Most audio grade wire wound

>resistors are pretty much noninductive, however. So, therein

>may lie the difference. Below is a plot of my results.

>Inductance tables I have indicate the xover frequency should

>be reduced by about 1/2 from 0-15 ohms. That's pretty

>significant.

Hi Carl,

I took a look at an Aetna-Pollak pot and counted about 61 turns of nichrome wire. It looks like the diameter is about .21 inches (.53 cm) and the length of the coil is about 1.9 inches. Using the calculators at http://hyperphysics.phy-astr.gsu.edu/hbase.../indtor.html#c1 for a solenoid the inductance calculates to about 0.0018 mH and using the toroid calculator about 0.001mH. I was a little suspicious of the calculator so I used the u(0)u®N^2A/l formula for a solenoid using the permitivity of free space u(0) as 4*pi *10e-7 H-M. This gave a result of 2.3e-6H or about .0023 mH. The Pollak calculated inductance seems to be 100 times less than measured on the replacement. It seems unlikely that the replacement pots would be that different (it would require 10 times as many turns, a much larger loop area, or a combiniation of the two to increase the inductance that much. Is it possible the resistance confused the instrument you used to measure inductance?

Joe

[Carlspeak]

>?

>>I don't have an original Aetna-Pollak rheostat but used

>one of

>>my new 25 watt, 15 ohm units (I normally use to replace

>the

>>original AR rheostats when they are unrepairable) in a

>little

>>experiment. What I did was rotate the dial from 15 ohms

>to

>>almost zero in 2 ohm increments and measured the

>inductance at

>>each increment. What I found was inductance indeed

>increases

>>with impedance - which may explain the roll off you

>heard.

>>However, I'm a bit puzzled as to why the volume didn't

>change

>>as well. A rheostat is nothing more than a length of

>wire.

>>Rotating the dial varies the active length. An inductor

>is

>>also a length of wire. Resistors are used all the time to

>pad

>>tweeters in many speakers. Most audio grade wire wound

>>resistors are pretty much noninductive, however. So,

>therein

>>may lie the difference. Below is a plot of my results.

>>Inductance tables I have indicate the xover frequency

>should

>>be reduced by about 1/2 from 0-15 ohms. That's pretty

>>significant.

>

>

>Hi Carl,

>

>I took a look at an Aetna-Pollak pot and counted about 61

>turns of nichrome wire. It looks like the diameter is about

>.21 inches (.53 cm) and the length of the coil is about 1.9

>inches. Using the calculators at

>http://hyperphysics.phy-astr.gsu.edu/hbase.../indtor.html#c1

>for a solenoid the inductance calculates to about 0.0018 mH

>and using the toroid calculator about 0.001mH. I was a little

>suspicious of the calculator so I used the u(0)u®N^2A/l

>formula for a solenoid using the permitivity of free space

>u(0) as 4*pi *10e-7 H-M. This gave a result of 2.3e-6H or

>about .0023 mH. The Pollak calculated inductance seems to be

>100 times less than measured on the replacement. It seems

>unlikely that the replacement pots would be that different (it

>would require 10 times as many turns, a much larger loop area,

>or a combiniation of the two to increase the inductance that

>much. Is it possible the resistance confused the instrument

>you used to measure inductance?

>

>Joe

>

Hi Joe.

I used a LCR and VOM meter that are not calibrated. However, I did take measurements separately with each instrument so as not to mess up the readings. Perhaps you could repeat my experiment with your Aetna-Pollak rheostats - assuming you have the instruments. Or, anyone else is welcome to do the same and let me know what they find. I could be completely off base. I did the experiment quickly and at the time the data seemed to make sense in answering Soundminded's mystery.

It is also worthy of note that my rheostats are constructed of larger wire of greater length than the A-P's.

Remember, it's all about the music

Carl

Carl's Custom Loudspeakers