Best 1-3W Amp for sensitive speakers?

[fb2017]

Hi all,

I would like to ask if anyone could recommend (DIY preferable) an ultra high quality 1-3W amp kit for powering 98dBm sensitive speakers (Dallas II Fostex FE206En). I recently measured the power I usually use them at, and I'm way under 1W (actually more like in the miliWatt range).
Thanks.

 

[thiagomogi]

ACA Nelson Pass

 

[kpsthakur]

Or JLH1969...bit simpler, then upgrade to ACA


Sent from my iPhone using Tapatalk Pro

 

[scottjoplin]

Class A for such low power is tempting

 

[tiefbassuebertr]

check out this circuits
http://www.audiodesignguide.com/my/Follower_99c.gif
the Power Folllower
http://www.diyaudio.com/forums/solid-state/45460-pavel-macuras-mosfet-power-follower.html
for less output power (and heating) reducing of supply voltage is necessary
http://www.diyaudio.com/forums/soli...ng-rail-voltage-class-power-follower-amp.html
http://www.diyaudio.com/forums/solid-state/27502-low-power-class-se-mosfet-amp.html

ckeck also out this thread:
http://www.diyaudio.com/forums/pass-labs/34437-power-follower-99c-vs-zen.html

The only commercial amp in this kind i know is this one from Germany:
https://web.archive.org/web/20010425074712/http://www.greenwall.de/de/smallone.htm
unfortunately only available in used condition.

 

[kpsthakur]

Looks good


Sent from my iPhone using Tapatalk Pro

 

[vdi_nenna]

Not a kit, but if you can get a pair of 2sk82 or 2sj28, build a L'amp CCS. Article available here in Articles section. It's 5 or 6 watts. Very nice sound.

http://rothacher.typepad.com/files/lampccs.pdf

ACA sounds nice. It can drive my 90 db speakers to high levels. Amazing bass.
I still prefer the L'amp CCS. It's just magical.

 

[vdi_nenna]

There are 3 different designs of the L'amp, but the CCS version can be built in one case and it's relatively inexpensive to build, although the 2sk82 are hard to get making them expensive. There are some for sale on ebay and here as well.

http://www.diyaudio.com/forums/swap-meet/300975-fs-2sk82-2sj28-2sk63-nos.html

 

[AndrewT]

3W is equivalent to +4.8dBW (1W is equivelent to 0dBW and 2W is equivalent to +3dBW).
Add that to your 98dB/W @ 1m sensitivity and you end up with a maximum output from the speaker of ~102.8dB @ 1m
If you have a pair of these @ 8foot (2.4m) listening distance, your maximum SPL will be ~98dB @ 8feet

That your maximum SPL.
To avoid clipping on transients your average levels could be 10dB to 30db below those maxima.
i.e. you need to turn down your avearge SPL to around 68db to 88dB for your loudest listening.
I don't think you will be satifsfied with a 3W amplifier. It will sound too quiet, or you will be aware of repeated and frequent clipping ruining the sound quality.

 

[sgrossklass]

Actually the rule of thumb is that SPL @ typical hi-fi distances of 2-3 m stereo ~= SPL @ 1 m anechoic, but of course specifics would depend upon the room in question and how lossy / reflective it is.

Typical amplifier power recommendations seem to be based on ~103 dB SPL peak, more conservative estimates would even be shooting for 110 dB. If you're not a loud listener, 103 should be just fine though. You rarely see a crest factor beyond 14-20 dB these days, 10 might be more common, so that's somewhere betwen 83 and 93 dB average. If you're not a loud listener and normally averaging around the mid-60s, that would give you about the recommended 20 dB volume reserve.

So in a nutshell, 3 W might well do the job here. Plus, it's the equivalent of 60 W into something 85 dB / W / m, and nobody would complain about a combo like that, right?

I would make sure that inherent gain is not too high, or include input attenuation if need be to keep noise in check. After all, obtaining 3 W / 8 ohms from a typical 1.5 Vrms input requires little more than 10 dB worth of gain, and typical circuits tend to be around 20-30 dB.

 

[fb2017]

WOW I'm impressed by the responses, and going through the articles. Thank you all.
I'm used to design circuits for high powers, high currents, It's really interesting seeing where are the focus points for such an application.
The next question would be, is someone selling a high quality kit for such an amp (ACA or L'amp)?

 

[fb2017]

Actually the rule of thumb is that SPL @ typical hi-fi distances of 2-3 m stereo ~= SPL @ 1 m anechoic, but of course specifics would depend upon the room in question and how lossy / reflective it is.

So in a nutshell, 3 W might well do the job here. Plus, it's the equivalent of 60 W into something 85 dB / W / m, and nobody would complain about a combo like that, right?

I have measured the power used at my normal listening volume: it is around 1 miliWatt. If I'm listening loud, it is around 1 Watt. At 2 Watts it is way too loud.

 

[sgrossklass]

While a bit oversized for your needs, our Bonsai's sx-Amp would also be worth considering:
Ovation sx-Amplifier: A 15 W RMS class A Current Feedback Amplifier

No offense to the likes of the trusty Hiraga or the Amp Camp Amp, but their distortion levels are at best acceptable but not great.

Then there's also this guy.
http://www.diyaudio.com/forums/soli...0w-power-amplifier-opa2134-2n3055-mj2955.html

Really this kind of power level is pretty much screaming for an opamp + Class A CFP output stage (so basically the above, just with CFP output like the sx-Amp has). You'd think someone would have done this with good results.

EDIT:

I have measured the power used at my normal listening volume: it is around 1 miliWatt. If I'm listening loud, it is around 1 Watt. At 2 Watts it is way too loud.

That's average power though, isn't it? The amp should be sized for peak power, which is average plus crest factor (IOW, crest factor is the peak to average ratio). So you would just about get by with 10 W on relatively compressed source material when listening loud, at 14 dB of crest factor you could use 25.

 

[fb2017]

That's average power though, isn't it?

My measurement procedure was setting the volume to the listening level by using the volume pot, then replacing the input signal with a 1000 Hz signal, and measure the voltage rms (true-rms Fluke 87V multimeter). From there I calculated the power. So basically we're talking about the maximal power output for a given volume.

At 1000 Hz, the FE206En has 12Ohm, btw.

 

[TNT]

Ananview 0100.

//

 

[fb2017]

Ananview 0100.

//

Well I do have an extraordinary good 30W tube amp (the Edison 60 kit built in a dual monoblock configuration). It does sound very good, but it uses about 350 W in power, which is OK in winter, serving as heater, but not so in summer . I haven't moved it's pot more than a few degrees though while listening. The Ananview has 50W .

 

[DPH]

Then there's also this guy.
http://www.diyaudio.com/forums/soli...0w-power-amplifier-opa2134-2n3055-mj2955.html

Really this kind of power level is pretty much screaming for an opamp + Class A CFP output stage (so basically the above, just with CFP output like the sx-Amp has). You'd think someone would have done this with good results.

That was along my lines of thinking. Something like an LME497x0, OPA164x or AD4898 has a strong enough drive to power a fast, class-A, CFP output stage. With that little power, you can look at higher-power driver transistors (and their attendant bandwidth) as your OPT. Put as much of the loop gain into the outer loop as possible and enjoy a beefy opamp. Just make sure to avoid oscillations!

(I'm in the design phase of a similar design with 24VCT transformers and using a LME49860 on boosted rails to a similar effect, albeit class AB EF2 OPS; I really don't need that much power and 10-15 W gets it done)

 

[sgrossklass]

My measurement procedure was setting the volume to the listening level by using the volume pot, then replacing the input signal with a 1000 Hz signal, and measure the voltage rms (true-rms Fluke 87V multimeter). From there I calculated the power. So basically we're talking about the maximal power output for a given volume.

How did the "replacing the input signal" part work exactly? On the same source, with unchanged gain, you played a 1 kHz, approximately full scale (0 dBFS) signal? That would be correct then, and you're looking at the right thing.

That Fostex driver is interesting. The gentle rise in the impedance plot from about 500 Hz up has to be a result of acoustic coupling, basically reflecting increasing radiation efficiency, it doesn't look like a resonance. If you see stuff like that, internal losses have to be pretty low. (It's rated at "only" 96 dB, BTW, but that still is fairly high for an 8" job.)
Impedance response would seem to be lending itself to outputs with non-negligible output impedance so this may be worth experimenting with. Thankfully, adding some externally is not a big deal at all if need be.

 

[TNT]

Well I do have an extraordinary good 30W tube amp (the Edison 60 kit built in a dual monoblock configuration). It does sound very good, but it uses about 350 W in power, which is OK in winter, serving as heater, but not so in summer . I haven't moved it's pot more than a few degrees though while listening. The Ananview has 50W .

12 dB extra headroom - perfect!

//

 

[fb2017]

How did the "replacing the input signal" part work exactly? On the same source, with unchanged gain, you played a 1 kHz, approximately full scale (0 dBFS) signal? That would be correct then, and you're looking at the right thing.

Exactly - I am using an IQAudIODAC as my player. In Linux one can generate a 0dB (100% volume) sine with the speaker-test command.