Transistor equivalent

spec said:

Hi MrAl,

Same here- measuring the actual cone movement or even the sound in the room would complete the feedback loop- nose to tail. It has been tried but, as fsr as I know, never really worked out. You can put a transducer on the stiff cone of a base driver, the old KEF B139, for example, where the cone moves like a piston up to say 200Hz. The end result would be to flatten the frequency response.

With other drivers the question is what do you measure because the cone doesn't necessarily move like an ideal piston so the problem is what do you use for the error.

But the real killer in any feedback system is loop stability. In general, the more accurate, implying high open loop gain, the more stability problems you have.

Using feedback from a microphone would seem to be the ultimate, but I think that is even more problematic because the room alters the sound.

Click to expand...

Hi,

Yes there are problems that have to be overcome, and that's the way control circuits are.

For example, for the cone movement/measurement problem you would have to do testing to find the curves for such a thing, and then incorporate that into the control law you choose to interpret the dynamics.

For the microphone pick up problem, one thing you'd have to do is compensate for the delay between the speaker and the mic pickup.

That's how control systems work...they take an imperfect system and make it better
Not always easy, but working out the problems is part of the job.
And it seems that anything like this would at least improve the audio, even if it still was not what we would want to call 'perfect'.

I recall flat response active woofers in a small cabinet using reflective foil and IR to close the loop.

By taking the 2nd integral of current signal using 2 stage LPF to convert force= acceleration to position then comparing position feedback, servo can flatten the loop and reduce resonance and extend response to limit of cone with somewhat lower distortion. Works better with high dampening factor speakers, ie. high acoustic/resistive impedance ratio. Can be tested by two cones sealed in box and one with a shorted coil or similar test. Woofers with ferro- fluidic coil cooling are best . Acoustic impedance matching with labyrinth 1/4 wavelength was also common but most woofer boom boxes dont do this, rather use brute force.

MrAl said:

Hi,

Yes there are problems that have to be overcome, and that's the way control circuits are.

For example, for the cone movement/measurement problem you would have to do testing to find the curves for such a thing, and then incorporate that into the control law you choose to interpret the dynamics.

For the microphone pick up problem, one thing you'd have to do is compensate for the delay between the speaker and the mic pickup.

That's how control systems work...they take an imperfect system and make it better
Not always easy, but working out the problems is part of the job.
And it seems that anything like this would at least improve the audio, even if it still was not what we would want to call 'perfect'.

Click to expand...

Just so, but sometimes it is an impossible task and I have known complex control loops cause more problems than they are trying to correct. With the mic approach you would need a different profile for each room and possibly if you moved the furnature or opend the curtains, and what would you do when someome walked around the room- just thinking out loud.

UPDATE

Just realised, noise-cancelling headphones must use a form of feedback. I have ofted wonderd how they work- any info?

Tony Stewart said:

I recall flat response active woofers in a small cabinet using reflective foil and IR to close the loop.

By taking the 2nd integral of current signal using 2 stage LPF to convert force= acceleration to position then comparing position feedback, servo can flatten the loop and reduce resonance and extend response to limit of cone with somewhat lower distortion. Works better with high dampening factor speakers, ie. high acoustic/resistive impedance ratio. Can be tested by two cones sealed in box and one with a shorted coil or similar test. Woofers with ferro- fluidic coil cooling are best . Acoustic impedance matching with labyrinth 1/4 wavelength was also common but most woofer boom boxes dont do this, rather use brute force.

Click to expand...

Interesting- I could imagine doing feedback at low frequencies where the waves are widely dispersed, but at the higher frequences all sorts of things are going on as the room dimensions and speaker cone get near the wavelength of the signal. I though a lot about doing overall feed back a long time ago. There was even a series of articles in Wireless World (I think) at one time.

Yeah, acoustic matching and blood-and-thunder, those are the two ways, with blood-and-thunder being the more common now.

It has always amazed me how two speakers, identical on paper, can sound so different- this is where the black art comes in.

Class A Room heater.

I used to work for Burroughs Disk Drive division as Test Engineer and Mgr. They had already designed and built large linear motor closed loop servos with 1HP coils with 2 " stroke and servo very fast and very accurate. I could imagine it driving a 1m diameter cone.

Later in life I saw one biomedical engineer use our 207 HDD linear motor to drive an air piston for a pressure assisted breathing device for quadripalegics.

Servo feedback only works on the signal for Bass signals, whereas broad EQ spectrum servo feedback works on the envelope for calibration.

Class A room heater

You know of some great gear- dont let Nikola see this. Six heat-sinks. If it is a monoblock that means output tripples. How much: many $Can I would think

You don't happen to have a shematic?

(I started life in civvy street as a test engineer- many stories)

Tell the truth, servo systems have always scared me to death- much too complicated. Even the feedback in audio power amps does my head in!

Disk drives have the most advanced and sophisticated servo control systems. Very fast and very accurate.

I understood them well enough to know the difference between a great one ( Maxtor's 1st 5.25" 100MB aka R2D2) and a poor one with overshoot or lack of thermal compensation on magnetics on gain phase margin. ( where permeability has a wide tolerance and temp sensitivity.) It was my job to reverse engineer and test them to cull out the bad OEM HDD designs from the best in class for approval for Corp Marketing to buy. I knew all HDD designs well enough to get 2 job offers, Seagate QA Mgr and Micropolis Eng Mgr in training. My (ex)wife didn't like Cali. and possibility of workaholic. I got 20 yrs experience in my 1st 10 yrs before I started at Burroughs in '82 and only thing I knew about HDD was it had platters than spun around. 2 yrs later I became HDD test mgr. since all my previous experience was used inside HDD's but never knew it. Test Engineering was my 2nd life after Electronic R&D for Aerospace, Nuclear Instr design and Telecom bleeding edge ISDN BBWAN T1 to the home.

Class A example cost previously posted. No schemas.

MrAl said:

Hi,

Yes there are problems that have to be overcome, and that's the way control circuits are.

For example, for the cone movement/measurement problem you would have to do testing to find the curves for such a thing, and then incorporate that into the control law you choose to interpret the dynamics.

.

Click to expand...

As I indicated, Since position is the 2nd order integral of acceleration and current (f=ma) ( 2nd order LPF) SNR drops as frequency rises, so effective limit is 0 to 200 Hz for cone travel and distortion feedback to flatten the acoustic response. Actually you can differentiate position feedback to velocity and integrate current to get velocity and close the loop using this for greater stability/SNR
Above this frequency, graphic EQ and spectral power is monitored by amplitude and not phase for adaptive AGC. Sony pro EQ would adjust band gain by 1dB every 10 seconds max using stage mics on speakers.

Tony Stewart said:

Disk drives have the most advanced and sophisticated servo control systems. Very fast and very accurate.

I understood them well enough to know the difference between a great one ( Maxtor's 1st 5.25" 100MB aka R2D2) and a poor one with overshoot or lack of thermal compensation on magnetics on gain phase margin. ( where permeability has a wide tolerance and temp sensitivity.) It was my job to reverse engineer and test them to cull out the bad OEM HDD designs from the best in class for approval for Corp Marketing to buy. I knew all HDD designs well enough to get 2 job offers, Seagate QA Mgr and Micropolis Eng Mgr in training. My (ex)wife didn't like Cali. and possibility of workaholic. I got 20 yrs experience in my 1st 10 yrs before I started at Burroughs in '82 and only thing I knew about HDD was it had platters than spun around. 2 yrs later I became HDD test mgr. since all my previous experience was used inside HDD's but never knew it. Test Engineering was my 2nd life after Electronic R&D for Aerospace, Nuclear Instr design and Telecom bleeding edge ISDN BBWAN T1 to the home.

Class A example cost previously posted. No schemas.

Click to expand...

Always wondered how disk drives worked so reliably... and fast... and so accurate- a real feat of engineering. One of the benifits of working for a big company is that if you need something sorted that was beyond your capabilities, someone knew the form. We had a good mathamatician and a couple of apps engineers. They could whip out a design and sim as fast as I could design an op amp circuit with a gain of 10. As a policy, I kept well away from control loops, althoughI almost failed a practical exam in the RAF when I turned the antenna on without turning the servo mag-amps on first- something tou never did! The antenna slewed from left to right and up and down. I dived for the off swich and the antenna wasn't damaged, but I could see the examiner scibbling on his pad. Passed, but with an observation.

Tell me, have you had any experiance of Hittachi HDDs. I am advised that they are one of the best types and have always used them.

Never been to California, but always wanted too- only Clearwater Beach Florida.

The ladies don't understand these things. Then I don't understand babies, chocolate, clothes, jewelry, soap operas, and flowers

UPDATE 2 (2016_01_09)
No price shown

UPDATE
I now remember there was two types of servo loops, but can't recall more at the moment, One of the guys published a book about it all. I had a copy of the book but it got lost along the way.

spec said:

Just so, but sometimes it is an impossible task and I have known complex control loops cause more problems than they are trying to correct. With the mic approach you would need a different profile for each room and possibly if you moved the furnature or opend the curtains, and what would you do when someome walked around the room- just thinking out loud.

UPDATE

Just realised, noise-cancelling headphones must use a form of feedback. I have ofted wonderd how they work- any info?

Click to expand...

Hi again spec,

Well i wasnt the one to actually mention using a mic in the first place i dont think
I was more talking about a sensor for the speaker itself. This would senses the travel and if it did not comply, the circuit would force it to. Tony mentioned something about this with an analogy to platter type disc drives, where the mechanism has to move fast and accurately. But of course here is always the cost hurdle to get over. For what i am talking about, we would need an over specified speaker, like maybe with a 100 watt rating for use in only a 20 watt system. That's because the coil would have to put up with the constantly occurring current peaks which would be required in order to drive the cone into nearly complete physical compliance.

I actually did a theoretical study (short and simple though) on something like this in the past, but using a "sort of" electrical analog: the inductor in a buck switching power supply. I noted that some of the delay in response was due to the inductor temporarily limiting the current flow, but that was due to the limited input voltage supply. Using the right feedback and a power supply with a much higher voltage (even though that would never be used in a constant mode) i found that i could obtain super fast output response, really unheard of response that starts to sound like science fiction. It really works, but there are very big requirements. For one, a higher voltage power input, which for most of the time will never be fully utilized. Second, the output transistor has to be able to handle that higher voltage. These things alone lead to cost increases.
But as we all know, sometimes the hobby side of it takes over, such as where the audio buff is willing to pay more for something really great. So it all depends how much money and effort we want to put into it for the most part.

Think about the Space Shuttle. There's an example of extreme control engineering. It seems like it cant be done (the reentry procedure takes miles and miles of travel just to reduce landing speed), but it's been done. Had it not been for the lack of materials science technology at the time, it might not be in retirement today.
If that's not enough, think about the Mar's projects. Very seemingly impossible but yet still accomplished.

I cant account for all noise canceling headphones, but some work by detecting the ambient sound waves and adding the same wave out of phase with the original so as to reduce it's volume.

High again Tony,

Well what i found in the study with the buck supply was that no matter how stubborn the inductor was it could always be driven to comply, as long as we are willing to find out how to create the right drive signal. The physical mass of the speaker is like an inductance and so should be able to be driven to match what we want it to do. Of course this means we'd need an over rated speaker in order for it to put up with the harsher environment we'd be creating for it.
The limits we have originally available always 'seem' to set the limits of the product. But I would bet that if you had 500 watts available to drive the HDD coil you'd get it to break the sound barrier
Of course there could be a physical conundrum where we get to the point where the animal is too heavy to walk on earth, so needs thicker legs, which then makes it even heavier, so with the gravity on earth there is a physical limit to the size any animal can become. We could run into that too, but we'd have to find out. The thicker the coil wire the heavier the mass, so there could be some limit.

MrAl said:

Hi again spec,

Well i wasnt the one to actually mention using a mic in the first place i dont think
I was more talking about a sensor for the speaker itself. This would senses the travel and if it did not comply, the circuit would force it to. Tony mentioned something about this with an analogy to platter type disc drives, where the mechanism has to move fast and accurately. But of course here is always the cost hurdle to get over. For what i am talking about, we would need an over specified speaker, like maybe with a 100 watt rating for use in only a 20 watt system. That's because the coil would have to put up with the constantly occurring current peaks which would be required in order to drive the cone into nearly complete physical compliance.

I actually did a theoretical study (short and simple though) on something like this in the past, but using a "sort of" electrical analog: the inductor in a buck switching power supply. I noted that some of the delay in response was due to the inductor temporarily limiting the current flow, but that was due to the limited input voltage supply. Using the right feedback and a power supply with a much higher voltage (even though that would never be used in a constant mode) i found that i could obtain super fast output response, really unheard of response that starts to sound like science fiction. It really works, but there are very big requirements. For one, a higher voltage power input, which for most of the time will never be fully utilized. Second, the output transistor has to be able to handle that higher voltage. These things alone lead to cost increases.
But as we all know, sometimes the hobby side of it takes over, such as where the audio buff is willing to pay more for something really great. So it all depends how much money and effort we want to put into it for the most part.

Think about the Space Shuttle. There's an example of extreme control engineering. It seems like it cant be done (the reentry procedure takes miles and miles of travel just to reduce landing speed), but it's been done. Had it not been for the lack of materials science technology at the time, it might not be in retirement today.
If that's not enough, think about the Mar's projects. Very seemingly impossible but yet still accomplished.

I cant account for all noise canceling headphones, but some work by detecting the ambient sound waves and adding the same wave out of phase with the original so as to reduce it's volume.

Click to expand...

Hmm! heady stuff- sounds like you have studied this quite a lot.

I see your approach about having a speaker with characteristics tailored for control. These days you dont have to worry too much about efficency and transient power demands- that can be relatively easy to sort. I'm not sure about this, but I think the speakers in the aticles in the mag I mentiond had an auxilary coil for monitoring what the speaker was doing.

Good point about not mentioning mic- I was speculating. Controlling the cone is a feasible task as far as I can tell. Also, as you imply, one of the reasons why this servo thing hasn't taken off is the cost as normal open-loop speakers can be pretty good and not fantastically expensive like they used to be- relative to salaries.

I have rewound a few speakers in the past- you have got me thinking.

Slight change of subject but still speakers- are you ready for this. A speaker chassis mounted in an airtight box and all the air pumped out. Forgeting the practical aspects, for the moment, it would have a lot of advantages; no reflections from the rear for example, so you would loose the boxy sound. And I don't think the resonant frequency of the cone would be upped like is a nomal box because a vacuum has 100% compliance.

Perhaps we could design a new speaker sysytem- vacuum and servo. A a sub-sonic servo could place the cone in the centre of its travel under quesent conditions- one of the problems of the approach.

Philips made some woofers with a transducer that was used for feedback. They worked very well but few were sold because ordinary woofers driven from a modern amplifier that had a very low output impedance for good damping also worked very well.
The first speaker I built in about 1960 used a Jensen 12" woofer and their horn tweeter. The cabinet was huge and the tweeter sounded something like a car horn. I soon sold it with the kit vacuum tubes (valves) amplifier I built.
The second speaker I built in about 1965 uses a Philips 5" long throw, low resonance, huge magnet woofer and their 1" dome tweeter in a fairly small sealed cabinet. It sounds very good and I still play it every day.

I bought a pair of Radio Shack Minimus 7 speakers (look in Google) which had a 4" long throw, low resonance woofer and a dome tweeter in a fairly small sealed cast aluminum cabinet and they sounded pretty good until one then the other stopped producing bass. The plate on the back of the speaker said "40W" but the woofer was marked "5W Korea". The replacement woofers were cheap but different. They had a much higher resonance and produced no bass. They also did not last long.

In about 1980 when Radio Shack was entering the phones business they put all their speakers on clearance pricing so I bought a pair that have 6.5"long throw, low resonance, huge magnet woofers and garbage cone tweeters in pretty solid wood sealed cabinets. I designed and built proper 18dB/oct crossover networks, replaced the tweeters with Scandinavian (Seas in Norway?) dome tweeters and ported the cabinets for very good powerful bass. I still play them every day.

Years ago I went to a trade show featuring sound systems from manufacturers all over the world. One manufacturer had many chairs facing two huge speakers that sounded very good with deep powerful bass. Then a pretty young lady removed fake covers on the speakers to reveal tiny little speakers. I looked all over for hidden sub-woofers but the deep bass was produced by the tiny little speakers and it made my hair stand up. The next day I designed and built a bass boost circuit to make my 8" woofers produce deep powerful bass like sub-woofers.

I replaced the cheap little 3" speaker in a clock radio with a 3" speaker from a Pamex very deluxe large intercom system from Norway, tweaked my bass boost circuit for it and added an LM390 amplifier. It sounds fantastic for a little clock radio but I had to retire it recently when its red LED display became too dim to see in daylight anymore. The display wore out because it was lighted (lit?) continuously for about 33 years.

Hi AG,

I enjoyed reading about your exploits. Would have liked to hear some of your creations. I have done simiar, but not on the same scale. My big thing in the late 1970s, was metal cone speakers. They had a very clean crisp sound which I liked but some didn't. I put a couple of expensive long throw versions in a smallish box, along with an Elac metal dome tweeter, which knocked the socks of other tweeters at the time. The xover was a single polly carb capacitor and one resistor. They sounded great, to me anyway. I was quite upset when they went the way of a lot of my speakers- burnt out voice coils due to my amp experiments. If you ever need a 4Mhz highly stable power oscillator, around 50W, I'm your man.

I had the same experience as you with smallish speakers producing phenominal bass. In my case it was the Epos ESP14. They blew away my enormous transmission lines - I was most upset.

Replacing the lousy amps and speakers in gear- the same. I'm going to have a go at the sound on our Panny TV some time- drives me up the wall. I know the LM390 (correction LM380) too- made a pcb layout for them and used the amp all over the place. It is a nice souding chip for what it is.

The LM390 was like an LM380 or LM386 amplifier except it had a bootstrap pin instead of a current source feeding the base of the upper output transistor so that its output voltage swing could be much higher, then a low battery voltage produced more output power. With a 9V power supply an LM386 produces 0.56W but the LM390 produces 1.05W.

Servo technology exists: https://www.rythmikaudio.com/technology.html

**broken link removed**

And there is optical technologies.

I saw the technology, probably in the 70's in a magazine. Search: servo woofer. servo speaker (different results)

Amp _____Vcc______Pop ____Remarks
LM380___10-22V____2.50W
LM384___12-26V____5.00W
LM386___4-12V ____1.25W
LM390___4-12 _____1.00W____Bootstrap

Hmm, hadn't noticed about the bootstrap.

What is the modern standard audio amp? I'm after single cell LiIon operation, say Vcc= 2V5 to 7V

Have you had any experience with LM3886T ?

KeepItSimpleStupid said:

Servo technology exists: https://www.rythmikaudio.com/technology.html

**broken link removed**

And there is optical technologies.

I saw the technology, probably in the 70's in a magazine. Search: servo woofer. servo speaker (different results)

Click to expand...

Nice work Keep- so it is practical then. Some more to read

What is PoP? Peak power is simply the real RMS power number doubled. An amplifier output power rating must include its supply voltage, its load impedance, the frequency, the distortion and the duration of the test before the IC melts.

Up to about 10 years ago, Philips and others produced about 76 amplifier ICs, mostly for cars. Not many are available today.
You will not get much output power with a supply as low as 2.5V but Texas Instruments has 42 low voltage class-AB and 36 class-D amplifiers, most are bridged, some with a voltage boosting circuit for higher output power and some have AGC.
A single rechargeable Lithium cell is 4.2V when fully charged but there is a new one (Li-Fe?) that is less.