gordonl@microsoft.UUCP (Gordon LETWIN) (05/14/91)
I'm building a house which will have good quality ceiling speakers in most rooms. They'll play music at low volumes and act as alarm/PA/Intercom speakers at higher volumes. Hooking say 20 pairs of speakers in parallel provides too low an impedance for the amplifier. I can't series/parallel them because each has it's own PAD for volume management, so their effective impedance may vary. Contractors have proposed buying N expensive hifi amps, each amp driving just a few speakers in parallel. I've come up with another idea that should work well, but as an amateur EE I'd like an opinion on the scheme from more knowledgable people. My idea is basically to take any amplifier that I please, but just feed it's output into a 8 ohm dumy load and build a high current voltage follower that follows the amp output. The follower would use some very high current MOS transistor. The theory is that this would impress the "high musical quality" signal produced by the hifi amp onto all of the speakers in parallel; the high current transistor would gate however many amperes this took. As for getting that current, I'm thinking of a big gell battery on a voltage regulated power supply, with caps, chokes, etc., as needed. The power supply would provide enough amps for the normal low-level background music, and when the alarm system wanted high volume the battery would provide the necessary current for the necessary minutes. So the battery acts as a giant power supply "capacitor" so that the power supply only provides average, not peak, current needs. Is there any reason that it's not just as simple as this? I.E., use a high current transistor and a car battery to provide a very high current voltage follower; use that to "amplify" the current capacity of an audio amp so that it can drive 20 or 30 speakers in parallel. Pls advise thanks gordon letwin
jeh@cmkrnl.uucp (05/15/91)
In article <72314@microsoft.UUCP>, gordonl@microsoft.UUCP (Gordon LETWIN) writes: > I'm building a house which will have good quality ceiling speakers in > most rooms. They'll play music at low volumes and act as alarm/PA/Intercom > speakers at higher volumes. > > Hooking say 20 pairs of speakers in parallel provides too low an impedance > for the amplifier. Gordon, you need to talk to someone who knows about PA systems... NOT sound- reinforcement systems, and NOT home audio... and ask about 25-volt and 70-volt lines. I don't know much about them myself... but I do know that that is how this problem is addressed in, for example, school and office paging systems. As I understand it, the PA amp has a speaker output terminal labelled "70-volt". You run this to all of your speakers, all at once. Each speaker picks power off the line via a 70-volt-to-8-ohm transformer. Your L-pad goes between the transformer and the speaker, just as you would expect. You can add as many speakers as you like (each with its own xformer) without worrying about impedance matching. Of course, the amp must still have enough power to run all those speakers at once. No free lunch. But I said "as I understand it". Ask someone who works in the field. --- Jamie Hanrahan, Kernel Mode Consulting, San Diego CA Internet: jeh@dcs.simpact.com, hanrahan@eisner.decus.org, or jeh@crash.cts.com Uucp: ...{crash,scubed,decwrl}!simpact!cmkrnl!jeh
whit@milton.u.washington.edu (John Whitmore) (05/15/91)
In article <72314@microsoft.UUCP> gordonl@microsoft.UUCP (Gordon LETWIN) writes: >I'm building a house which will have ... speakers in most rooms. >Hooking say 20 pairs of speakers in parallel provides too low an impedance >for the amplifier. I can't series/parallel them because each has it's own >PAD for volume management, so their effective impedance may vary. >Contractors have proposed buying N expensive hifi amps, each amp driving just >a few speakers in parallel. I've come up with another idea >My idea is basically to take any amplifier that I please, but just feed it's >output into a 8 ohm dumy load and build a high current voltage follower >that follows the amp output. The follower would use some very high current >MOS transistor. The theory is that this would impress the "high musical >quality" signal produced by the hifi amp onto all of the speakers in parallel; >the high current transistor would gate however many amperes this took. The high power output stage of an amplifier is just as you state, a follower. In fact, a push/pull AB amplifier has two of 'em, one to source current and one to sink it (so the output impedance is HALF the output impedance of a simple follower). What does your scheme really gain? Good power amplifiers aren't terribly expensive (especially if you get 'em as modules from one of the kit-building houses). The main problem with using a MOSFET is that as a follower, its output impedance varies with drive (so you'll get distortion unless you use some sort of feedback). A minor problem is that the output impedance is the inverse of the transconductance; for a reasonable power MOSFET, that's about 10 Ohms (i.e. worse than just driving the speaker). As for using a bipolar transistor, the output impedance there is low; it's Rout = (kT/q)/Ie = (.020 Volts)/Ie at room temperature For 1A quiescent current, this will give you a very low output impedance, all right; 0.02 Ohms. That's too low (among other problems, you will likely get RF oscillation due to stray wiring inductances). The transistor soon burns up, because this is now a class A amplifier (the careful biasing in most amps is intended to keep waste power to a minimum). Last problem: a high-quality high-power audio amplifier has at least +/- 30V power supplies (and can deliver undistorted output to about 30V on peaks); it's going to be cheaper to get amplifiers than to buy batteries that can handle that voltage. This assumes, of course, that your speaker choice is similar to the 4/8 Ohm units most common in home audio gear. That said, if your speakers ARE IDENTICAL, you can just put L-pads on 'em and put 'em in series in pairs, THEN hook the pairs up in parallel. Any stereo amp would drive eight speakers that way, easily. L-pads are a two-potentiometer attenuator that have the same input impedance at all settings (so the other speaker in series won't react to the attenuation of its sibling). L-pads are easy to find (Radio Shack #40-977 or #40-980). If the speakers are non-identical, series connection cannot be recommended because the frequency dependence of the impedances will not match. John Whitmore
kludge@grissom.larc.nasa.gov ( Scott Dorsey) (05/15/91)
In article <1991May14.164451.17@cmkrnl.uucp> jeh@cmkrnl.uucp writes: >In article <72314@microsoft.UUCP>, gordonl@microsoft.UUCP (Gordon LETWIN) > writes: >> I'm building a house which will have good quality ceiling speakers in >> most rooms. They'll play music at low volumes and act as alarm/PA/Intercom >> speakers at higher volumes. >> >> Hooking say 20 pairs of speakers in parallel provides too low an impedance >> for the amplifier. > >Gordon, you need to talk to someone who knows about PA systems... NOT sound- >reinforcement systems, and NOT home audio... and ask about 25-volt and >70-volt lines. I don't know much about them myself... but I do know that >that is how this problem is addressed in, for example, school and office >paging systems. 70V systems use an amp with a constant voltage output driving a large number of high impedance speakers. Usually the speakers are stock 8-ohm beasts with a transformer whose primary is in the 1K-20K ohm range, and whose secondary is 8 ohms. Speakers with high Z voice coils are also available. These systems sound godawful because of the distortion induced by the transformers. It would be possible to make good quality transformers with reasonable power handling too; I might consider using output transformers from tube amps. But this will cost you more money than you want to pay. Paralelling all the speakers is a bad idea, because you'll wind up with an impedance in the sub-ohm range and it's difficult to build an amp that will drive low impedance loads properly. It can be done, but it will cost you a whole lot of money. Putting all the speakers in series is an improvement, because this way you'll have a very high impedance, so your amp will need a very high voltage output and a reasonable current level. (In fact, you could probably build a tube amp that didn't have an output transformer, which would be kind of interesting). The problem here is that your speakers had better all be identical. If the efficiency varies between speakers, so will the sound level, and it'll be astoundingly different. Your third alternative is to wire several banks in series, and then wire those banks in parallel (or vice versa). This way you can get an impedance of your choosing (and I suggest you choose 8 ohms or so). Not only that, but you can pad down each bank individually and give yourself zone controls for the levels in each part of the house. Since your banks won't have 8 ohms each, your pads will be somewhat nonstandard, but that's not an impossible thing to build (especially if you banks consist of series speakers... high resistance pots are easy to get). It'll still sound better if you use an individual amp for each speaker, because you won't have each speaker influencing the amount of current available for the other ones. When you put speakers in series or parallel, not only are you increasing or reducing the resistance across the line but also the reactance. Still, it'll be a lot less expensive and might be satisfactory. --scott
jgk@osc.COM (Joe Keane) (05/16/91)
In article <72314@microsoft.UUCP> gordonl@microsoft.UUCP (Gordon LETWIN) writes: >My idea is basically to take any amplifier that I please, but just feed it's >output into a 8 ohm dumy load and build a high current voltage follower >that follows the amp output. The follower would use some very high current >MOS transistor. The theory is that this would impress the "high musical >quality" signal produced by the hifi amp onto all of the speakers in parallel; >the high current transistor would gate however many amperes this took. I think this is a good idea. The only part i don't like is that you're not using the old amplifier to do much of anything. Especially in hi-fi, simpler is better. Having a power stage that doesn't do anything is just asking for more distortion and noise. Even if you just build a current booster, i'd say you want feedback anyway to keep the output voltage accurate. At this point it's only a couple more components to add some voltage gain. So, i think that you might as well just build a high-power amplifier, which works directly from line level. If your old amplifier has features like tone control that you don't want to do yourself, you can tap into your old amplifier _before_ the power stage. I've seen lots of schematics for high-power amplifiers. Typically they use a current feedback configuration, and there is one stage of bipolar transistors between the op-amp output and the output MOSFETs. One nice feature they note is that, unlike bi-polar transistors, you can parallel MOSFETs arbitrarily to get as much power as you want. So if you're running out of power and you want to add more speakers, you can just drop in some more output transistors, with heat sinks of course. Note that even though MOSFETs have very high input resistance, power MOSFETs have can have a hefty input capacitance, so it takes a decent amount of current to drive a bunch of them in parallel. As long as the previous stage is designed with this in mind, it shouldn't be a problem. -- Joe Keane, amateur electronics hacker jgk@osc.com (...!uunet!stratus!osc!jgk)
gordonl@microsoft.UUCP (Gordon LETWIN) (05/21/91)
In article <72314@microsoft.UUCP>, gordonl@microsoft.UUCP (Gordon LETWIN) writes: > I'm building a house which will have good quality ceiling speakers in > most rooms. They'll play music at low volumes and act as alarm/PA/Intercom > speakers at higher volumes. > > Hooking say 20 pairs of speakers in parallel provides too low an impedance > for the amplifier. Thanks again to all the people who emailed a response to this posting. One person pointed out that the voltage that I want to follow goes from -V to +V, not 0 to V, so this will complicate the follower circuit somewhat. Also, since it's running class A I'll need some considerable heat sinking capacity for when in alarm mode. Most people mentioned 70v PA systems. I'm aware of these systems but I'm not interested in PA, as per above these is primarily a "hifi" music system. My concern is that the 70v transformers may not to too well in the bass. Also, I'll need a transformer for each speaker, two per stereo pair, which will run into major bux if I get hifi ones (i.e., big), assuming that "hifi" stuff exists in the PA market. The most interesting idea I received is to use those cheap single-chip audio amplifiers, one for each stereo pair. The National data book describes units which, for about $5, will drive about 3 watts a channel. The chip's specs look fine; I'll verify via blind testing that I can't hear the difference between a chip and a "high end" hifi amp through these "nice but not amazing" wall speakers. Three watts should be plenty loud; the THD is about 10% at that volume, but only alarm signals would be played that loudly. Another advantage of this is that I can now use those single chip DC-controlled volume/tone control chips. I can supply the DC via either a digital pot chip or via PWM from a microprocessor. My house will be heavily automated and each room will have a touch screen. I can then write software to control the volumes via the touch screens. Also, I can set the house into "party mode" and remotely command the volume levels in all the rooms. The micros are already there as part of the house automation; the cost for the digital pot chip, the amplifer, and the volume control comes to about $15 a pair; this is a lot less than an amplifier, transformer, and pad, so the idea looks like a winner. I'll post when I've tried it out... gordon letwin
stever@mcdham.UUCP (Steve Reinhardt) (05/23/91)
In article <4785@osc.COM>, jgk@osc.COM (Joe Keane) writes: > In article <72314@microsoft.UUCP> gordonl@microsoft.UUCP (Gordon LETWIN) > writes: > >My idea is basically to take any amplifier that I please, but just feed it's > >output into a 8 ohm dumy load and build a high current voltage follower > >that follows the amp output. The follower would use some very high current > >MOS transistor. The theory is that this would impress the "high musical > > I've seen lots of schematics for high-power amplifiers. Typically they use a > current feedback configuration, and there is one stage of bipolar transistors > between the op-amp output and the output MOSFETs. > > One nice feature they note is that, unlike bi-polar transistors, you can > parallel MOSFETs arbitrarily to get as much power as you want. So if you're > running out of power and you want to add more speakers, you can just drop in > some more output transistors, with heat sinks of course. > Unless you have money to burn, don't do this. MOSFETs do have the requisite temperature coefficient to permit paralleling, but only in switched mode systems. When they are used in the linear region, they may not react quickly enough to save themselves; that is, the transconductance will not decrease from the increase in temperature rapidly enough to promote current sharing across the group of devices. In addition, the VGS(th) (threshold voltage for turn on) can vary by a factor of two for a given type of device, and failure can occur very quickly. Imagine driving an 8 ohm load from 100 volt supply rails, when only one device turns on. Messy stuff. Of course, if the device doesn't short ( unlikely), then the remaining devices will start to share, unless one of them is higher gain, etc. There are lots of designs published out there, in applications notes from Motorola, Siliconix, PMI and others. In addition, the Audio Amateur publication has many full construction articles that will help avoid these pitfalls. You might want to ring up your local Motorola sales office, and ask for DL135, the Power MOSFET transistor data book. The first section is a good tutorial for learning how these beasties work. > Note that even though MOSFETs have very high input resistance, power MOSFETs > have can have a hefty input capacitance, so it takes a decent amount of > current to drive a bunch of them in parallel. As long as the previous stage > is designed with this in mind, it shouldn't be a problem. > -- > Joe Keane, amateur electronics hacker > jgk@osc.com (...!uunet!stratus!osc!jgk) ---- Pereant, inquit, qui ante nos nostra dixerunt. "Confound those who have said our remarks before us"- Aelius Donatus Steve Reinhardt, Motorola, Wallingford, Ct. 06492 (203) 284 0810 UUCP: stever@mcdham.UUCP or stever@ct.mcd.mot.com -- Pereant, inquit, qui ante nos nostra dixerunt. "Confound those who have said our remarks before us"- Aelius Donatus Steve Reinhardt, Motorola, Wallingford, Ct. 06492 (203) 284 0810 UUCP: stever@mcdham.UUCP or stever@ct.mcd.mot.com