Multi-purpose Power Amplifier

[domtw]

I have this idea of building a 6 channels amplifier that could be used for different purpose, from stereo amplification to 2x 3way with active crossover amplification and eventually 5.1 systems.
I would like it to be able to drive speakers which impedance goes as low as 2 Ohms.
The output power:80W-100W @ 8 Ohms should be enough.

I will probably use some already made circuit board (1 channel amp), so 6 of them. Or 3 of them if they are stereo.
My concern, at the moment, is about the Power supply.
What would be sufficient to power up 6 channels at 2 Ohms?
How can I estimate the transformer(s) characteristics ?

As I am not an amplifier expert, all advise are welcome.
Thank you!

[pacificblue]

You are trying to achieve contradicting things. 8 Ohm speakers need high voltage, but relatively low current, while 2 Ohm speakers need high current, but not so high voltage.

Chipamps usually don't deliver 80-100 W per channel, neither are they happy to drive 2 Ohm loads.

If you want to use chipamps, download the Overture Design Guide. There you can play with the supply voltage and speaker impedance to get an idea, which output power you can expect. Use a transformer that has a higher power rating greater than or equal to the output power of the amplifier. If you feed several amplifiers from the same transformer, add up the output powers to find out the transformer power rating you need.
If you use an unregulated power supply, subtract ~20 % of the supply voltage to find the nominal voltage for the transformer. Then recheck by multiplying that number with 1,7 to make sure that the no-load voltage does not exceed the maximum voltage rating of the chipamp.

The power you need for a certain speaker does not depend on its impedance, but on its efficiency and the sound pressure level you want to achieve with it.

recommended reading

[domtw]

Thank you for your answer.
I may have made my own idea about amps and that idea is apparently wrong.
I thought that some amplifier could burn if the speaker impedance was going too low.
Continuing the same idea, I was thinking that amps that support low impedance speaker would also support higher impedance ones.
Is that totally wrong ?
In fact, what I would like to do with this amp is to have no worries what ever are the speakers behind and to be able to drive 6 speakers with them. (to prepare one of my next audio step: active crossover)
Is that possible ?
What would be your suggestions ?

[pacificblue]

Originally posted by domtw
I thought that some amplifier could burn if the speaker impedance was going too low.

Exactly.

Originally posted by domtw
Continuing the same idea, I was thinking that amps that support low impedance speaker would also support higher impedance ones.
Is that totally wrong ?

No, it is not wrong. What happens is that with lower impedance at the same supply voltage you get more power. If you assume an ideal amplifier you would get twice the output power at half the impedance. So an amplifier that delivers 100 W into 8 Ohm would give you 200 W into 4 Ohm and 400 W into 2 Ohm. That means you have to provide transformer and heatsinking for a 400 W amplifier, if you want to make a 100 W into 8 Ohm amplifier stable for 2 Ohm loads.

Originally posted by domtw
In fact, what I would like to do with this amp is to have no worries what ever are the speakers behind and to be able to drive 6 speakers with them. (to prepare one of my next audio step: active crossover)

Do you plan to connect the speakers without any crossover at all until you have the active solution? You would probably destroy the tweeters very soon. As soon as you use an adequate crossover the overall impedance will be in the range of the lowest impedance any single speaker has.

Originally posted by domtw
What would be your suggestions ?

Start with determining a budget. That will force you to make reasonable decisions at certain points of the design phase. Design each amplifier for the speaker load it actually has to drive.
Don't overestimate the power demand. Chipamps will not deliver more than ~60 W into any load with the correct supply voltage, and that is more than enough most the time. Keep in mind that the powers of each way add up in an active speaker. And make yourself aware that speaker efficiencies usually are 80 dB or higher with 1 W / m. 80 dB at a workplace usually means, you have to wear hearing protection.

[AndrewT]

Originally posted by pacificblue
.................Start with determining a budget. That will force you to make reasonable decisions

there's excellent advice.

And remember that the "amplifier" part will cost no more than 10% of the completed, boxed, ready to use amplifier.

Lets suppose you are prepared to spend $1000 on six amplifiers for a 5.1 system.
That leaves about $16.66 for each amplifier.

If you go active with 3way speakers the budget becomes $3000 using the same $16.66 amplifiers.

[domtw]

Originally posted by pacificblue

Exactly.


No, it is not wrong. What happens is that with lower impedance at the same supply voltage you get more power. If you assume an ideal amplifier you would get twice the output power at half the impedance. So an amplifier that delivers 100 W into 8 Ohm would give you 200 W into 4 Ohm and 400 W into 2 Ohm. That means you have to provide transformer and heatsinking for a 400 W amplifier, if you want to make a 100 W into 8 Ohm amplifier stable for 2 Ohm loads.

OK, I haven't thought about the heatsink. Good that you mentioned it...

Do you plan to connect the speakers without any crossover at all until you have the active solution? You would probably destroy the tweeters very soon. As soon as you use an adequate crossover the overall impedance will be in the range of the lowest impedance any single speaker has.

Of course not. I would use only 2 channels when used with passive crossover speakers. In that case, I would switch off the other channels. I think it's better not to use an amplifier without any load. Isn't it ?

Start with determining a budget. That will force you to make reasonable decisions at certain points of the design phase. Design each amplifier for the speaker load it actually has to drive.
Don't overestimate the power demand. Chipamps will not deliver more than ~60 W into any load with the correct supply voltage, and that is more than enough most the time. Keep in mind that the powers of each way add up in an active speaker. And make yourself aware that speaker efficiencies usually are 80 dB or higher with 1 W / m. 80 dB at a workplace usually means, you have to wear hearing protection. [/B]

As you said Chipamps usually don't support impedance as low as 2 Ohms. Is there a particular one that does ?
I wish to use some ready made board for the amplifier circuit if it is possible. (60W would be enough).
Once I will have found/chosen this ready made boards, I will have to decide on the Power Supply needed. One for all, one for each side (L/R) or one for each channel. This I am not sure...

I get the speaker part as I am designing my own
That is actually the whole purpose of this amp. One that could support any experiment on the speaker side.
The amplifier part is still a bit more mysterious for me, that is why I am asking for help and advise.
About the budget, as I am not extra rich, the cheaper the better but without compromise on the load side, with good musicality low distortion and a frequency response at +/-0.5db max from 20Hz to 20KHz.
I hope that the total cost of such an amp to be somewhere around 1000 USD. Is that realistic ?

Thank you for your help.
I will certainly have some more questions about this project (still at its idea stage)...

[pacificblue]

Originally posted by domtw
I think it's better not to use an amplifier without any load. Isn't it ?

It would be wasting energy.

Originally posted by domtw
As you said Chipamps usually don't support impedance as low as 2 Ohms. Is there a particular one that does ?

Well, officially not. It is a question of designing it that way, and it would be ignoring the specs. But National gives us a hint in the BPA200 project, where they list the supply voltage for 2 Ohm operation with the LM3886 as ±20 V.

Originally posted by domtw
I wish to use some ready made board for the amplifier circuit if it is possible. (60W would be enough).

There are many ready-made PCBs around. Here is an incomplete list.
http://www.audiophonics.fr/module-am...tf-p-4147.html
http://sound.westhost.com/project19.htm
http://www.elv.de/output/controller....0&detail2=2461
http://assemblycraft.com/custom.html
http://stores.ebay.com/gssound-savvatis-georgios
http://www.jlmaudio.com/JLM%20AMP.htm
http://www.jacksonaudioworks.com/amplifiers.html
http://www.nabucoeletronica.com.br/gainclone.html
http://www.diyfidelity.com.au/produc...83fe985abfddd9
http://www.chipamp.com/lm3886.shtml
http://www.audiosector.com/lm3875.shtml

Originally posted by domtw
Once I will have found/chosen this ready made boards, I will have to decide on the Power Supply needed. One for all, one for each side (L/R) or one for each channel. This I am not sure...

One for each channel will give the best results, however the difference is small. One for each side would probably be the best compromise. If you make one for all, the amplifiers would all be in one enclosure and you would need long speaker cables.

Originally posted by domtw
I get the speaker part as I am designing my own

Looks very well done. Inspired by another French speaker?

Originally posted by domtw
That is actually the whole purpose of this amp. One that could support any experiment on the speaker side.

Any would probably be asking too much.

Originally posted by domtw
I hope that the total cost of such an amp to be somewhere around 1000 USD. Is that realistic ?

You could probably do it for less.

[domtw]

Originally posted by pacificblue

Well, officially not. It is a question of designing it that way, and it would be ignoring the specs. But National gives us a hint in the BPA200 project, where they list the supply voltage for 2 Ohm operation with the LM3886 as ±20 V.

I haven't found this hint on the LM3886 datasheet.
Would that work with any design using the LM3886?
Where can I find information about this BPA200 project?

Originally posted by pacificblue

There are many ready-made PCBs around. Here is an incomplete list.
http://www.audiophonics.fr/module-am...tf-p-4147.html
http://sound.westhost.com/project19.htm
http://www.elv.de/output/controller....0&detail2=2461
http://assemblycraft.com/custom.html
http://stores.ebay.com/gssound-savvatis-georgios
http://www.jlmaudio.com/JLM%20AMP.htm
http://www.jacksonaudioworks.com/amplifiers.html
http://www.nabucoeletronica.com.br/gainclone.html
http://www.diyfidelity.com.au/produc...83fe985abfddd9
http://www.chipamp.com/lm3886.shtml
http://www.audiosector.com/lm3875.shtml

I cannot find any ready-made PCB that has 2 Ohms load in its specification...
Would a solid state amplifier more appropriate for this kind of load ?

Originally posted by pacificblue


One for each channel will give the best results, however the difference is small. One for each side would probably be the best compromise. If you make one for all, the amplifiers would all be in one enclosure and you would need long speaker cables.

Noted!

Originally posted by pacificblue

Looks very well done. Inspired by another French speaker?

Yes, inspired by another French design...
Aren't French the best ?

Originally posted by pacificblue


Any would probably be asking too much.

My concern is mostly about low impedance. At the moment, I would like to be safe on that side...
For my own knowledge, what kind of other specs would put a speaker out of the "normal" amplifier capabilities?

Originally posted by pacificblue


You could probably do it for less.

Good news !

Thank you for your help. I appreciate it a lot !

[star882]

It might be possible to parallel several TI digital power stages (with external driver) in order to get enough current for low impedance loads. Whether or not it would be trivial to get running is something I don't know. (I guess one application you have in mind is testing car speakers?)

A simpler way is to add a series resistor for low impedance operation, but that will reduce the efficiency and maximum power. (Of course, if you only rarely use it with low impedance loads and don't mind losing some power, it would be the cheapest solution. In one science lab I worked in, they simply wired some power resistors in series with a consumer audio amplifier (that was connected to a PC operating as a signal generator) for driving some linear motors used in experiments. It would withstand an accidental short circuit or a jammed motor for even prolonged times without any problems.)

If you're trying to run several parallel speakers, it might be better to wire them in series or series/parallel. (But then, if the connection to one speaker went open circuit, all the others in the same branch would also stop working. May or may not be an issue.)

[pacificblue]

Originally posted by domtw
I haven't found this hint on the LM3886 datasheet.

That is, why I wrote: officially not.

Originally posted by domtw
Would that work with any design using the LM3886?

Yes. It only means to use a supply voltage that is low enough to keep heat dissipation and output current within the IC's limits, which results in low power at normal 4 or 8 Ohm loads.

Originally posted by domtw
Where can I find information about this BPA200 project?

http://www.national.com/an/AN/AN-1192.pdf

Originally posted by domtw
Would a solid state amplifier more appropriate for this kind of load ?

If designed for it, yes.

Originally posted by domtw
Aren't French the best ?

Comme-ci, comme-ça.

Originally posted by domtw
For my own knowledge, what kind of other specs would put a speaker out of the "normal" amplifier capabilities?

The more complex the speaker impedance is, the worse is the load for the amplifier. E. g. electrostatic speakers are highly capacitive loads. Passive cross-over components introduce lots of capacitive and inductive components. Speaker coils themselves are highly complex loads that even change their characteristics during operation, and speaker coils need much higher currents during ring-in than their nominal impedance predicts. Published speaker specs are usually not precise enough to assess that behaviour in advance.