LM3886 based 8 Channel amps for linkwitz Orion

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi friends,
This is my first (and very ambitious) DIY project so please excuse me and point at me if I am doing something stupid :)


I am planning to build an 8 channel DIY amplifier (each channel having 2 LM3886s, so total of 16 LM3886 chips).
This amp will be used to power the Linkwitz Orion DIY stereo speaker, (planning to build next).


Orion is a quad-amped active speaker. Each speaker tower has
2 Seas Millennium Tweeters wired in parallel (net impedence 2.5-3ohms)
1 Seas W22EX001 Mid range speaker (8ohm impedence)
2 Peerless 10" XLS Woofer (powered independently) - (8 ohms impedence each)

60w per channel is the minimum power required, woofers can do with more power albeit at risk of over excursion below 33 hz.

The planned amp configuration (per side) is
Channel 1 -> 2 LM3886s in Parallel (PA100) for the 2 tweeters
Channel 2 -> 2 LM3886s in Bridge (BR100) for the Midrange
Channel 3 -> 2 LM3886s in Bridge (BR100) for the Woofer 1
Channel 4 -> 2 LM3886s in Bridge (BR100) for the Woofer 2


I am planning to run all amps at 28V, this would produce about 100W from tweeter channel and about 130W from the other 3 channel. Overall about 980W.
I am not sure, but if possible, I would also keep an option of reducing the voltage to 20V by transformer tapping (I am told it is possible to get custom built like this). That way I can keep the average heat dissipation down from 470W to 245W and still turn up the system when required.

I am planning two use two 1 KVA torroidal transformers, one for right side channels and one for left side channels. If I understand correctly this would help in quality improvement over a single 2 KVA transformer.


Linkwitz Orions are considered one of the excellent DIY speakers and here my objective is to get the best sound out of these amps. So, am I on the right track?
 
I can't myself see any advantages in paralleling the tweeters and also paralleling the amps to drive them. Perhaps this is a saving on cable, but if so that to me would be spoiling the ship for a ha'peth of tar. Drive them separately, each with a dedicated amp. Do you know where the clip points are for each amp? Seeing as so little energy goes into the tweeter on average it would be worth checking that its not going to clip first.

How did you arrive at your average heat dissipation figures? They do sound a tad extreme! One single 2kVA toroid with separate windings for each channel will generally beat two 1kVA toroids as regulation is better on the larger trafos.
 
Two tweeters should be driven by two independant amplifiers, Do not parallel the tweeters and thus need to complicate the chipamps by paralleling them as well.

In the speaker design Linkwitz mentions to run the speaker in parallel, as one faces front and one faces the back wall. The rear facing tweeter has a resistor attached to it in series to attenuate by 3db.

While I guess it should be possible to run them independently by their own amplifiers, it would make level adjustment more difficult. More wires to manage between the amp and the speaker.
Also since these are expensive tweeters (and actively powered) I am planning to use some kind of speaker protection circuit.

So more tweeter channels would mean more complexity over complexity of paralleling the amps. I guess sound wise paralleling will not have a negative impact.


Similarly for two woofers. Use independant amps for each woofer.
That comes to 5 chipamps. Is 68W of maximum power into each of the 5 drivers sufficient for balanced peak SPL capability from the Orion drivers?

Unfortunately I have never heard an equivalent (or any active) system before so I have no idea.

Each Tweeter has sensitivity of 89db and is rated at 90W.
Midrange has sensitivity of 88db and is rated at 120W
Each Woofer has sensitivity of 88.4db and is rated at 200W.

And I will be playing this in a 400 sqft living room.


One 5channel amp dedicated to each Orion is far more likely to sound better than 12 or 16chipamps in a common chassis.

Personally I would build stereo amplifiers.
One stereo amp to the tweeter pair. One stereo amp to the woofer pair.
One monoblock to the Mid driver.
Total amp requirement for the 10driver Orions: 4 stereo amps and 2 monoblocks.

What do you mean by chassis. Is it only the separate transformer or complete separate case?

I was thinking of only separating the transformer onwards for right and left channel. But do guide me what really works for good sound, without wasteful expense...

Thanks for your help.
 
I can't myself see any advantages in paralleling the tweeters and also paralleling the amps to drive them. Perhaps this is a saving on cable, but if so that to me would be spoiling the ship for a ha'peth of tar. Drive them separately, each with a dedicated amp.

Repeated from above post...
In the speaker design Linkwitz mentions to run the speaker in parallel, as one faces front and one faces the back wall. The rear facing tweeter has a resistor attached to it in series to attenuate by 3db.

While I guess it should be possible to run them independently by their own amplifiers, it would make level adjustment more difficult. More wires to manage between the amp and the speaker.
Also since these are expensive tweeters (and actively powered) I am planning to use some kind of speaker protection circuit.

So more tweeter channels would mean more complexity over complexity of paralleling the amps. I guess sound wise paralleling will not have a negative impact.

Do you know where the clip points are for each amp? Seeing as so little energy goes into the tweeter on average it would be worth checking that its not going to clip first.

How did you arrive at your average heat dissipation figures? They do sound a tad extreme! One single 2kVA toroid with separate windings for each channel will generally beat two 1kVA toroids as regulation is better on the larger trafos.

I don't know the clipping point of LM3886.

The heat dissipation is calculated based on heat dissipation formula given in LM3886 datasheet and AN1192 application note. It is total of all 16 amps at 28V and 20V rail voltage.


kuldeepsingh said:
Big project buddy... all the best...
Proper grounding will be essential in your case and you have to use proper equilization also.

and use LM4780 in place of 2xLM3886, it will simplify the construction.
Thanks Kuldeep
I already have 16 LM3886 amps. (also have 2 LM4780's)
 
Repeated from above post...
In the speaker design Linkwitz mentions to run the speaker in parallel, as one faces front and one faces the back wall. The rear facing tweeter has a resistor attached to it in series to attenuate by 3db.

All the more reason to run each tweeter from its own amplifier then. A resistor in series with a tweeter will always be a compromise.

While I guess it should be possible to run them independently by their own amplifiers, it would make level adjustment more difficult.

I must be missing the elephant in the room here then. If you feed each tweeter from its own dedicated amp, then the 3dB adjustment will be a pot on your crossover board. As it is, level adjustment is practically impossible when its a resistor in series with only one tweeter. That would by necessity place the resistor on the back of the tweeter itself - try choosing the right value for your particular set up from that location :D

More wires to manage between the amp and the speaker.

Indeed - but if wires bother you, you're not really an active speaker guy. Go passive.

Also since these are expensive tweeters (and actively powered) I am planning to use some kind of speaker protection circuit.

So you'll need another channel of protection. No big deal.

So more tweeter channels would mean more complexity over complexity of paralleling the amps. I guess sound wise paralleling will not have a negative impact.

What do you base your guess on? Do you have any experience of getting paralleled amps running? They're not a walk in the park (I have some experience). In my experience, getting amps to share well is a little bit tricky, and it gets harder still as the frequencies go up. I'm with AndrewT on this.

I don't know the clipping point of LM3886.

That is easy to guesstimate, but not what I meant. I was thinking of when your XO is presented with a transient (you should try Bob Cordell's recommendation of the Rickie Lee Jones track from 'Flying Cowboys') does the tweeter clip? If so then ideally you'd increase the rails to its amp(s).

The heat dissipation is calculated based on heat dissipation formula given in LM3886 datasheet and AN1192 application note. It is total of all 16 amps at 28V and 20V rail voltage.

But they have no idea about your active XO in that app note. How the XO divides up the signal is going to be important to the heating effect.
 
Hi,
Thanks for finally posting the sensitivities of the drivers from your original question.

I see a maximum deviation of 1dB between them +-manufacturing tolerances.

If each of these drivers is sent the same maximum peak transient then all drivers will reproduce that peak transient at similar SPL +-tolerances - power compression +0-1dB sensitivity deviation.

I would still recommend 5 amplifiers to power 5 drivers without any use of paralleling nor bridging.

The sensitivity differences of ~1dB +-tolerances can be accommodated by adjustable sensitivity controls on the outputs of the active filters.
The special -3dB +-tolerances can also be accommodated at the output of the active filter.

It will be far easier to adjust the sound "balance" of the drivers by using the filter sensitivity controls. rather than trying to match SPLs with passive resistors etc.

Five amps each of ~68W into 8ohms will result in ~ 300W per channel of active speaker power.
This will probably sound as loud, or louder, than a single 300W amplifier driving a 3way 5driver set up with passive crossover.
I would predict the maximum sinewave SPL from each driver with 68W available as
Treble 107dB
Mid 106dB
Bass 112dB
all +-manufacturing tolerances.
 
Last edited:
What do you base your guess on? Do you have any experience of getting paralleled amps running? They're not a walk in the park (I have some experience). In my experience, getting amps to share well is a little bit tricky, and it gets harder still as the frequencies go up. I'm with AndrewT on this.

Ok, you have my attention on this one. I have no experience in any amp making :eek:

But they have no idea about your active XO in that app note. How the XO divides up the signal is going to be important to the heating effect.

If you mean the crossover setup it is at 92 Hz and 1440 Hz, both LR4 (24 dB/oct).
 
I would still recommend 5 amplifiers to power 5 drivers without any use of paralleling nor bridging.

...

Five amps each of ~68W into 8ohms will result in ~ 300W per channel of active speaker power.
This will probably sound as loud, or louder, than a single 300W amplifier driving a 3way 5driver set up with passive crossover.
I would predict the maximum sinewave SPL from each driver with 68W available as
Treble 107dB
Mid 106dB
Bass 112dB
all +-manufacturing tolerances.

Hi Andrew,
Isn't max output of LM3886 only 60W at 35V rail?

The active amplification should also improve the overall efficiency, but this is an open baffle speaker, which would need some power for low frequencies.

Also is 107dB enough? Sorry I can't translate it to what it would feel like :(
What is the average music listening levels for most people?

How much double the power (if bridging), i.e. 3 db, mean while listening to music?


One 5channel amp dedicated to each Orion is far more likely to sound better than 12 or 16chipamps in a common chassis.

Personally I would build stereo amplifiers.
One stereo amp to the tweeter pair. One stereo amp to the woofer pair.
One monoblock to the Mid driver.
Total amp requirement for the 10driver Orions: 4 stereo amps and 2 monoblocks.

What do you mean by chassis. Is it only the separate transformer or complete separate case?

I was thinking of only separating the transformer onwards for right and left channel. But do guide me what really works for good sound, without wasteful expense...

Thanks for your help.
 
If you mean the crossover setup it is at 92 Hz and 1440 Hz, both LR4 (24 dB/oct).

Had a quick look at Siegfried's site and the block diagram of his ASP looked considerably more complex than just this. Do you have the schematic? If you do and you sent it to me I'd be interested to simulate it in LTSpice. Is there not a high pass filter for the bass too? I saw mention of 20Hz HP on one page. The gains of the respective channels are necessary data too.
 
Hi ST7677,
I don't really want to be a buzz-kill here but you're plan would be considered a major undertaking for an experienced builder. Getting bridged and parallel amps working correctly is no easy task. I agree withAndrewT on this, build single-chip amps in separate chassis. It will be easier for a first time builder and much more likely to be successful, and as Andrew pointed out, will probably be sufficiently powerful for adequate volume level. Also, do you have the proper equipment, tools and supplies for this? At minimum you'll need a good quality soldering station and solder supplies; hand tools like diagnal cutters, long-nose pliers, screw drivers, etc.; power drill and bits (drill press is better); a good quality digital meter; and I really consider an oscilloscope essential for testing the final product.
I think it would be better to start off with a simpler plan, build one simple chip amp to learn what works and gain some experience. After that you'll have a much better idea of how to proceed.

Mike
 
Thanks Micheal, for the heads up, really appreciate it.
I am fully aware that it's not a simple first timer project. However that sad I believe in doing things the right way than doing things the easy way and I also believe in doing thing right the first time itself.

So yes, it may take some time for me to learn, may be even a few months but this will be taken to it's planned end and I will need help of all the community members here. I believe that's what this wonderful place is all about :D

PS: I have most of the equipment except say the diagonal cutter and the Oscilloscope.
 
Had a quick look at Siegfried's site and the block diagram of his ASP looked considerably more complex than just this. Do you have the schematic? If you do and you sent it to me I'd be interested to simulate it in LTSpice. Is there not a high pass filter for the bass too? I saw mention of 20Hz HP on one page. The gains of the respective channels are necessary data too.

No I have not bought the construction plans from Linkwitz yet. I thought designing an amplifier for an active speaker should be easier, considering that there are no unknown behavior of passive crossovers to deal with. Each channel will be driving it's own predefined speaker with frequency band-limited easy loads.

For amplifier specification Seigfried Linkwitz recommends any decent amplifier with >60W output per channel with low THD. Specially, low distortion at low output levels (<5W where amplifier spends most of it's time during music reproduction).
The crossover/equalizer for the ORION was designed under the assumption that all power amplifiers have the same voltage gain from input to output.
 
Build one monoblock now.
Understand what it does. Understand what happens when you change things a little, or a big bit. What effect does that have on the sound?
Understand what can go wrong and how to recognise that you have a problem.
Learn how to solve those problems. Don't expect to do that all by yourself. The Community is here, use us as you learn. Asking the right question is a big part of the learning.

Once you have your monoblock working and you have adjusted it so that it sounds good through your speakers in your house, then you are ready to start building a stereo (two channel) amplifier. It will take a while to get it working as well as the monoblock.
Learn why multiple channels cause extra build problems.

Only after all that, can you make an informed decision on whether a 3 channel or 5 channel or 16 chipamp version of the amplifier can be built to meet the Orion's requirements.

You already know the direction I would go in, so I'll not repeat.
 
Sounds good Andrew,
However the problem here is the availability of good quality components in retail. I will have to order components from US / UK. The international shipping might be more expensive than the parts itself so I can't experiment too much. May be 1-2 iterations, but not too many.

Therefore, I will have to rely more on theory and community experience.

I want to first finalize the configuration with pros and cons of different approaches and then build one monobloc like you suggested.


As suggested I would drop the idea of bridging and paralleling for now and go with simple 5 channel.


To begin with can you please explain (or suggest relevant links):
1. Why "One 5channel amp dedicated to each Orion is far more likely to sound better than 12 or 16chipamps in a common chassis."

2. What is the logic of splitting
"Personally I would build stereo amplifiers. One stereo amp to the tweeter pair. One stereo amp to the woofer pair. One monoblock to the Mid driver."

Please guide.
Thanks
 
ST7677,

A couple related comments:

-SL no longer specs a resistor for the rear tweeter...it is run straight parallel (inverted) with the front tweeter.
-As you mention due to the active crossover Orion is very unfussy about amplifier pedigree as long as all channels are gain matched
-I run my Orions with a Sonance 1250MKII amp which is 12x50W with a single 50W channel feeding each paralleled tweeter pair and each midrange and a bridged 100W pair of channels feeding each woofer. This provides more than ample drive capability and sounds very sweet.
 
Here's what I think would be a good way to go. First build a normal stereo amp as I previously said to gain knowlege and experience, that amp should be built with the idea of eventually using it in the big project. After you get that accomplished and have a good feel for what it takes, proceed onward with the rest of the project. Andrew recommended building your amps in separate stereo and monoblock boxes which is generally a good idea, it will keep construction complexity down and makes the inevitable troubleshooting much easier. But one way I differ from Andrews' suggestion would be to build a two channel amp for the woofers and a three channel amp for the mid and tweeters for each stereo side, in other words, two amplifier boxes for each channel for a total of four for the whole project instead of the six that four stereo and two monoblcks would require. Here's a block diagram for clarity:

Orion.JPG

Mike
 
I can only imagine doing this by placing each amplifier (set of chip amps for each channel, with separate power supply for each channel - essentially a multi-amped monoblock) near each speaker, and running line-level signals (perhaps balanced) from the preamp/control console to each amp. You'll save substantially just on speaker cable.
 
ST7677,

A couple related comments:

-SL no longer specs a resistor for the rear tweeter...it is run straight parallel (inverted) with the front tweeter.
-As you mention due to the active crossover Orion is very unfussy about
amplifier pedigree as long as all channels are gain matched
-I run my Orions with a Sonance 1250MKII amp which is 12x50W with a single 50W channel feeding each paralleled tweeter pair and each midrange and a bridged 100W pair of channels feeding each woofer. This provides more than ample drive capability and sounds very sweet.

Hi Kevin,
Did you try running the woofer with 50W single channel before you made it in bridge? If so did you find it inadequate?
Is there a perceptible difference in running the woofer 50W (single channel) or 100W bridge?

Finally have you made any adjustment to the ASP for dealing with higher power woofer channel? I guess it would lead to 3db gain in woofer channel.

Thanks
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.