Low Q Woofer in OB with High Zo Amp

[CLS]

I put this post here based on a reason that I see it as a loudspeaker system design issue, instead of discussion on amplifier's circuit.

A little background:

Power Transconductance Amp with Tubes, How?
(In the first post of that thread, you may see THE background of this background.... )

That was the era I used tube amps for OB bass and thought about the current feedback concept. As usual, my action in building things is quite slow. I haven't done anything in that discussion yet and things went other direction.

And then I ran into 'T'-bass drive for OB LF drivers. which, for me, was an easier build than messing with (whole new) amps, and it delivered excellent results.

T-bass does one thing very well - drawing current from amp where most woofers can't do themselves around the low end resonant peak. Most amps are constant voltage, and designed for under-damped speakers (or simplied as resonant boxed speakers). When using those mid to low-Q woofs (designed for boxed) on open baffle, a lot of EQ is needed, not only for baffle loss, and for the too early roll off due to the low system Q.

Ordinary amp with a lot of EQ in front is trying to deliver more voltage gain on where the system impedance is high. But the impedance is often so high that the EQ can not catch up with. And the EQ is bringing some phase problem. Most people do this, I did it, too, but don't like it.

While T-bass circuit largely reduces the system impedance by the turn ratio of coupling transformer, thus much more current is delivered into the driver effectively. No (or less) EQ involved makes it superior in phase repsonse. I do love the bass sound from it. Very coherent, jumpy, tuneful, and delivered with a sense of effortless. (technically, the 'effortless' is a mistaken impression, because the amp is driving a hefty load as low as 1 Ohm.... )

Afterwards, by some post around here (I forget which, sorry), someone brought up Variable Amplifier Impedance, and it looked simple enough (I'm mostly a speaker guy, then a tube guy. SS is somewhat difficult for me ) OK, only 3 resistors are needed, or one of the original feedback resistor is kept, then I need only 2 per side

Mixed feedback with more 'current' content should be a very ideal way to work with low Q OB system. It 'automatically' delivers proper amount of current into the driver, which is a LOT more compared to the ordinary constant voltage amp - just what we need.

By the very simple (and doable) guideline in the article above by Mr. Rod Elliott, I modified my existed amp (a poor old Hafler) to have 20 Ohm output impedance.

How is this 20 Ohm determined? The measured Q's of my wooder on OB are Qes=0.31, Qms=5.25, Qts=0.29. I want a new system Q of 1.1, so I added extra 20 Ohm to the Re to have a new Qes of 1.25, and then a Qts of 1.1.

This mod on my amp was quite simple, with only one little different from the article - I need a cap to isolate DC in the feedback path.

Previouly, in my OB bass, T-bass circuit contributed about 10~12dB of 'EQ' ( 5~6dB boost on the low end and 5~6dB cut on the MB). And extra digital EQ in the upstream contributed another 7~8dB (+3 in the low end and -4~-5 around 80~120Hz... ). I think if I had tuned the T-bass better, or have a more capable amp, less EQ could do fine. Anyway, this low Q system needed 16~20dB of 'EQ' in total.

Now with a 20 Ohm Zo, it eliminates most of those EQ, now 2~3dB in total is enough to make it very flat

How is the sound? Simply put, GREAT ! I can not compare it with the T-bass side by side, but I feel their sounds are almost the same -- low bass sounds are delivered effortlessly with coherent, jumpy, and tuneful qualities. ( a little bonus: the amp is now running much cooler, now it's really effortless )

I should've tried this much earlier. It's so elegant, no brute force is needed, and the result is so beautiful.

There might be a lot of mixed-up technical terms in my writing which were far from conscientious and thorough because of my excitement. Please correct me if something wrong.

[CLS]

Seems no interest, pity

I made some spreadsheets and charts for visualizing the effects. The driver's (system's) impedance data were exported from the measurements by WT3. Then I added series R (output impedance, Zo) to simulate the 'loss' along the combo of this fixed series R and the measured impedance curves.

This is the current setting with 20 Ohm Zo, by calculation, system Q=1.1, and the deepest loss between 20Hz and MB is 10dB.




The real in room response needs extra 3dB or so to fulfill my taste, so maybe I'll increase the Zo to 32 Ohm, system Q=1.5, and the max loss comes to 12.5 ~ 12.7 dB, then it'll look like this:

(a mysterious peak around 160Hz, appeared after the mod of phase plug... )


OTOH, I'm also working on the center 'sub', which accommodates 2 woofers in parallel, by the same 1.5 system Q and total loss of 12.5 dB, I'll add 24 Ohm Zo to the amp, it'll look like this:

Before that, I'll have to trace out the plate amp to find its feedback circuit...

[Pano]

Very interested here! Sorry I did not see the thread before. I will study it.
one note: I am surprised that you get a total Q of 0.29 on baffle. That's rather low for on baffle. What is the driver?

[Samuel Jayaraj]

I tried this idea once and was very briefly pleased with the results. But the amp overheated and the output stage went up in flames.

Is there any consequence to this approach, when the loudspeaker (impedance) is disconnected? Does it affect amplifier stability, gain, phase margin etc?

Thanks,

[ScottG]

Originally Posted by CLS

Seems no interest, pity

Oh, there's interest all right.. some have suggested this and done this before..

Tube amps with higher output impedance react this way to a certain extent, but it's a matter of matching the resonant impdeance to the freq. response curve. (..one reason I often advocate a driver's resonance at around 70 Hz for this application in conjunction with an appropriate sized baffle.)

Suggestion (Isuzu commercial): GO FARTHER!

In this respect consider that many don't want to give-up the tonality of their "reference" amp for an active setup.. OR don't want such a reactive setup broad-band. What's a designer to do? *Dual-VC's*. In fact the gain provided from the secondary VC IN CONJUNCTION with the primary will often off-set the need for any additional eq. depending on how you utilize a low-pass filter for the second VC.

Note that there are issues with headroom - so series connect your power output devices (..or parallel with higher output devices and add-in more resistance).

[CLS]

Hi panomaniac,

My woofers are Eminence Sigma Pro18 (older version with smaller magnet). Measured Q's in free air and on baffle were very close. Qes and Qms were slightly higher on baffle, but the differences were really small, like 0.30 vs 0.32 / 5.24 vs 5.29... and similar case in the fs -- 28 vs 29 Hz, or the likes. Maybe because the main baffle is very small - 55cm wide by 60cm high for a single 18".

In the case of center sub OB, the baffle is larger, and closer to the wall. It's measured fs dropped to high teens to low 20's. And the Qes even dropped to 0.17. Maybe I should measure it again to make sure....

Originally Posted by Samuel Jayaraj

...

But the amp overheated and the output stage went up in flames.

Quite the opposite in my case. My amp is running much cooler than before. Cooler than when it was not modified and driving a normal load, let alone the 1 Ohm load by the T-bass circuit afterwards.

The voltage of the power supply rails in the amp is +/- 53V or so, and the efficiency of the woofer is quite high, plus I usually listen quietly, so the head room for me should be OK.

I'm wondering, though, the current feedback and high Zo make the output current pretty much constant (related to input), so it would not rush up on the low impedance load like a low Z amp. While in the the mid to low bass frequencies where the impedance is so high (mostly dominated by resonant peak), current draw of both high and low Z amp would not be very high. So I thought the high Z amp must be clipped mainly by the limit of supply voltage, not by the current capability. No?

OTOH, the woofer in free air or on OB is easily driven to exceed its Xmax. No matter driven by current or voltage sources, it needs only that much (little) power to reach Xmax. There shouldn't be any problem in amps I think.

BTW, my system is basically active 2-way. The amp mentioned above only works in the bass range.

Originally Posted by ScottG

In fact the gain provided from the secondary VC IN CONJUNCTION with the primary will often off-set the need for any additional eq. depending on how you utilize a low-pass filter for the second VC.

Interesting And it makes me think of this:
Large midrange for OB??? Scott G ?

Wire two VC's in series, and shunt one of them with a large cap. Will it work?

But I'm thinking, even with (electrical) 1st order filter, the phase is still changed. So there'd be 2 forces which are not acting simultaneously. How can we deal with this?

[CLS]

For those who are not familiar with SS amp circuits (like me), here I show you how I did the mods. We don't need to deal with the entire amp, only focus on the feedback circuit is enough (for this topic only, of course )

Here is a highly simplified circuit diagram of a typical SS amp (the upper one):


Trace it out from the output stage would be easier (components and copper traces are larger and simpler). You could find out a small resistor connected between output point (bus) and one leg of a tiny transistor. (Probably much nearer than you'd expect) At the same point it connects to the tiny transistor, there'd be another small resistor in series with a cap (mostly a low voltage 100uF electrolyte) and then to ground. These are the R1, R2, and C1 shown above.

Then according to the Figure 2 of this article:
Variable Amplifier Impedance

The following actions should be easy. Now the grounding point of the cap should be connected to the resistor in series with the load (R3 as shown above in the lower diagram) The calculation according to the article above is quite straight forward and you'll get the proper values of all the components by that simple equation.

Since the R2 might be changed, so you may also need to change the C1 accordingly to maintain a proper time constant.

Don't forget to re-adjust the EQ and damping of the system if any.

Have fun

[ScottG]

Originally Posted by CLS

Interesting And it makes me think of this:
Large midrange for OB??? Scott G ?

Wire two VC's in series, and shunt one of them with a large cap. Will it work?

But I'm thinking, even with (electrical) 1st order filter, the phase is still changed. So there'd be 2 forces which are not acting simultaneously. How can we deal with this?

A 1st order electrical (passive) on the INPUT of the amp, represents minimum excess phase (..or additional group delay).

Do NOT wire the VC's together, instead one VC should be part of the loudspeaker driven from your "reference" amp, the other should be driven (actively) with an additional "beefier" amplifier with the impedance correction on the output and the passive 1st order on the input. (..your preamp should be "split-out" to each of the respective amplifiers, the "reference" amp and the impedance compensation amp.)

Look to the 12 inch DVC drivers from Pioneer (..they have foam surrounds that help with eff.). Decent eff. on *EACH* VC of around 91-92 db, and most notably a LOW FS. (..or rather a relatively low FS.) MCM electronics also has a rubber surround version that is similar. Both are cheap and should be used in "quads" per side/channel (with each driver's VC series/parallel connected to the next driver (of 4) for a net 6-8 ohm load for each of the 2 VC connections).

[Pano]

Maybe because the main baffle is very small - 55cm wide by 60cm high for a single 18".

Yep, that's it.
--

This circuit does look like a very nice idea for OB. You can dial in the system Q you want.
Looking over the circuit brings some questions to mind.

1: A constant current amp is going to be just that, CC. So what happens at the upper end of the woofer's range? The impedance is rising, so the CC amp is going to supply more voltage to keep the current constant. Will that tend to flatten the woofer response, or even cause a rising response?

2: Looking at the schematic in the article the feedback resistor at the speaker (R3 in your diagram) immediately made me think "inductor DCR." So what if this resistor were an inductor? It would be acting as a low pass filter on the woofer, but what would happen to the voltage divider (woofer+R3) that drives the feedback? It's going to change with frequency. Could it be put to good use?

I'm sure many people have thought about this before. So what are the answers?
.

[CLS]

Originally Posted by ScottG

...

one VC should be part of the loudspeaker driven from your "reference" amp, the other should be driven (actively) with an additional "beefier" amplifier with the impedance correction on the output and the passive 1st order on the input.
...

That's what I'm doing with my system! It's just not dual VC in one woofer. Instead, I use separate woofers driven avtively with different LP points. (similar to those 2.5way or 3.5way systems)

Last night I tried the modified plate amp with 24 Ohm Zo. It's driving my OB center sub (with 2 identical woofers as the main channels). The 24 Ohm Zo yields a 1.5 system Q. It seems quite high but actually adequate (by ears) for such a subwoofer working on the very bottom octave (20-40Hz).

Working together with the main channels, now the whole system needs ZERO EQ on the bass range!

The total Q of the main channels is tuned to 1.1, and 1.5 for the sub. But the whole system is not boomy at all. The sub adds a tad of (but also significant) weight and authority to the foundation. The senses, although very small, of vibrations or 'puffs' by the air flows in the sounds adds quite an awsome effect. And it doesn't need high SPL to enjoy the bursts of bass notes which are coherent with the whole spectrum.

Originally Posted by panomaniac

...

1: A constant current amp is going to be just that, CC. So what happens at the upper end of the woofer's range? ...

2: Looking at the schematic in the article the feedback resistor at the speaker (R3 in your diagram) immediately made me think "inductor DCR." So what if this resistor were an inductor? ....

According to the article by Nelson Pass (who made a CC amp), most speakers don't like 'pure' CC (infinite Zo). Instead, in his experiments, 47 Ohm works well with most high sensitivity fullrangers. (I use 20-some Ohm on my woofers.)

I think, as a whole, it's a mixed matter of system Q and intrinsic (or system) frequency response.

About the upper end, I have no experiences. Again, in the article by Nelson Pass, he used some LRC networks in parallel with speakers to tune the load impedance, thus the response.

And using an inductor as the current feedback R3 is so interesting. Yes it'll be a LP filter and maybe good for pure bass (subs) usage. Or cancelling out the rising impedance of some loads by carefully chosen inductor (or a LR combinations). I think it'd have same or similar phase behavior as (any other forms of) 1st order LP filters. Let's wait and see if anyone will try...