My new gainclone page

[tl]

Application Note BPA-200

Do you still have the PDF?
It was on NSCs web site for about 1 month before it was retracted.

The application note still exists here

google is your friend...

Thomas

 

[pinkmouse]

Hi Dorkus

I have had similar ideas about the PSU too, I have put an extra 1000uf cap across the rails just after the rectifier( I used 10A Shottkys), and as I have room left I might try the 0.1uf before the diodes as well.

As it is currently tipping down with rain , I can't get outside to build the cases at the moment, so I might try this out today.

My current version uses a 30-0-30 transformer, made by Nuvotem Talema, from RS, and this seems like a good balance between power handling and sound quality.

I am using Panasonic electros, skinned, for my power rails, with Evox Rifa 5% MMKs for the 1uf and 0.1uf, and a generic poly 10% 2.2uf on the input, that will probably be changed later for something more tasty.

Resistors are currently 5% metal film, they seem to add a bit of sparkle at the top, with a plain 7w wirewound as the output, but this will be changed to a Caddock once RS get them back in stock!

One of my first prototypes used the dead bug style of construction, and I found this to have a hollow, empty sound, so suspecting that component resonance may be a problem, I now use a small pcb, and that seems to have cleared up the problem.( Perhaps this was the problem you heard with the Gaincard?)

I can't find my camera at the moment ( I think I left it at the party I was at last night ) to take a picture of the current construction, but below is a photo of an earlier prototype, using off the shelf components. You can possibly make out the feedback resistor soldered across the pins of the IC, I managed this by snipping off the unused pins of the 3875.

The heatsinks I will be using for the real amp will be ones I bought off Jackeh, that are standard finned, 30mm deep, 150mm high and 200mm long, cut in half, with one half per channel. This is possibly overkill, but it works well with the case I have designed.

To all those that are lurking on this thread or undecided, give it a go, these are great little amps, simple to build, cheap, robust and you can have hours of fun playing around with them and learning how all this electrickery works!

 

[MWP]

Re: Application Note BPA-200

The application note still exists here

Thanks... but thats not the same applicaton note as the one i have printed. The one you linked to looks a lot older (scanned pages made into a PDF).
The one i have has more details on thermal design and some other points... looks like its written by someone else, not just a modifed version of the one you linked to.

 

[Pjotr]

Hi,

Funny to see that a chip costing a few dollars and intended to be used in an average TV-set, outperforms many high-end amps, costing tons more.

It makes me a really good laughter. Don’t misunderstand me, NOT on you folks, but on the whole high-end scenery. Yes I also used the chips many times for multi-amping and active x-overing.

About the small buffer caps: They inherently generates much less hf noise, due to smaller (amplitude) and broader (time) charging pulses.

 

[joensd]

My current version uses a 30-0-30 transformer, made by Nuvotem Talema, from RS, and this seems like a good balance between power handling and sound quality.

Wouldn´t it exceed the absolute maximum ratings of the chip?
In the datasheet it says not to exceed 84V in total with attached input signal.

For power optimization I´d use +-35V with 8Ohms and +-28V with 4Ohm speakers (as recommended in the datasheet for single chip application).

 

[pinkmouse]

Wouldn´t it exceed the absolute maximum ratings of the chip?

Yes, but I think using the dual bridge system drops the voltage because of diode losses, maybe also more diode loss due to using Shottkys, I don't know, ( I'm still learning ), but I get a measured 30.17 volts per rail after smoothing, and under load! I am using a 300VA transormer, so I doubt the output is drooping under the load of 2 ways of amp.

 

[tl]

Re: Application Note BPA-200

Thanks... but thats not the same applicaton note as the one i have printed. The one you linked to looks a lot older (scanned pages made into a PDF).

Sorry, but I was not aware of different versions of BPA-200. But: Did you notice that page 4 of the pdf is not scanned? This page is obviously modified and is about 'Thermal Background'.

Thomas

 

[MWP]

Yeh i noticed pg4 looked different.
I also noticed its dated 1997... im sure the one i got off the NSC web site was later than this, would have been around 1999-2000.

Ill see if i can dig the printed copy up and see what info i can get from it.

On the subject of power supply for the LM3886, running mine at 40-0-40 (after recitification) was difficult to get working. It took a lot of research into managing the heat dissipation. Your normal mica and silicone impregnated rubber insulators were no where near up to the job.
I ended up mounting the ICs directly on a 6mm aluminium bar which was then insulated from the heatsink.

 

[ThorstenL]

Re: gainclones...

Hi,

> i am seriously considering using the chip in inverting
> configuration a la Thor (hence my inverting amp thread
> a few weeks ago)

Followed all of this and it did part of making me post....

> but i have some concerns with the compromise between
> feedback and input impedance.

As mentioned in the original inverting amp thread, there are ways around this in the feedback loop.

Another option which I prefer is to simply preceed the inverting mode Amp (or preamp) with a complementary J-Fet Buffer, pretty simple really. This gives you an input impedance of up to a few Megaohm (or even more if you bootstrap the input) while retaining the advantages of the inverting mode.

> i don't want input impedance any lower than 20k or so (i'm using a shunt attenuator before it)

It makes me wonder why?

How about ditching these awfull resistive attenuators that steal at least 10db subjective dynamic range and going transformer attenuator? Drop me a line to see if we can fix you up..

> i'm also unsure of the need for all the extra capacitors shown
> in the Thor schematic.

You can do without many of these. One of the things my schema does is to go "belt and braces". I don't want anyone to build this and find it to have loads of offset and fry speakers, oscillate and fry speakers and so on.

I suspect removing both the input coupling cap and the cap on the non-inverting input will improve the sound.

In case you use a low DCR volume control (like a transformer) you can simply connect a parallel connection of the two resistors for feedback on the non-inverting input to ground. This will both dynamically and statically balance the inputs, removing not only common mode distortion but also killing the distortion caused by the base current modulation with signal in BJT's (one of the reasons why J-Fet Inputs are better).

I can think even of a really interesting bridged Amp using LM1875/3875 and co which would use a variation of the Tom Evans version of what I'd call "feed along" erroro correction.

Here both Amp halves would be inverting, with each opposite halves positive input connected via a voltage divider. As in an inverting amplifier the actual appearing voltage on the inverting input is directly proportional to the difference between input signal and amplified output signal, amplifying this signal and feeding it with inverting phase into the speaker would cancel the distortion.

> i've heard good things about AC coupling the input

I have not, other than that it limits what can go wrong. And unlike Vlave Amp's, when solid state amp's go wrong they often take along the speakers, despite fitted "speaker protection" circuits (the relais have a tendency to arc and thus fuse closed when trying to disangage).

I would recommend to start with all belts and braces in place and later to simplify the circuit if you feel it neccesary. Also, careful choice of the capacitors (I like Foil & Film or metalised Film Polycarbonates for solid state best) can adjust the tonal balance (I prefer changing valves to that of course).

> i think to properly use this chip in inverting configuration, you
> will need some sort of preamp or buffer before it or
> dynamics/noise/distortion performance may suffer.

You may be right, assuming you insist on high input impedance for no particluar sensible reason....

I have attached a copy of a Linestage that uses a similar principle of buffering....

> i'm not so sure about the low capacitance supply thing.

I am. I now also use something similar in my Valve gear. Big cap's mean big problems. Small Cap's mean small problems. Many Small Cap's mean many small problem which tend worse than one big problem....

> other design touches... resistors will probably be IRC tantalum
> chip if they prove to be good-sounding

I like them okay, prefer NI wirewounds.

> i will use a zobel on the output (10ohm + .1uF), along with
> a little bit of series inductance (a few turns of 16 gauge OFC
> magnet wire).

Do yourself a favour - try it ALSO the other way. It may surprise you.

> one thing i'd like to try out is putting a small capacitor on the
> secondaries, before the rectifiers, to reduce RF noise...

Carefull. This tends to form substantial tank circuits with the secondary and/or leakage inductance that ring very well, especially if "kicked" alive by the switching pulse of the diodes starting to conduct.

> a form of AC conditioning you could say.

Yes, you could say it "conditions", though perhaps not entierly in the intended way.

> a good 0.1uF polypropylene would do the trick.

Yup, that should put the ringing nicely low in frequecy with a nice Q to make it never stop ringing.... ;-)

> the biggest obstacle at this point is getting a nice chassis for it.

;-)

> another idea i've had is use multiple LM3885's as amps for an
> active loudspeaker.

Good idea.

But in this case consider switching from driving the speakers with undistorted voltage and thus forcing a heavily distorted current (which is after all what moves the voice coil) to driving the speakers with undistorted current which kills several types of distortion and compression in the common speaker drivers quite effectivley. This of course implies the need for a non-reactive rear load to the Low Frequency driver....

> e.g. the speaker i'm using now uses very simple 1st order
> crossovers... would be a piece of cake to incorporate the
> filters using either input or feedback caps on the amp chip.

I'd say feedback for LF and input for HF....

> i think with a nice speaker, the sound could be excellent.

Yup, not that many "nice" speakers exist these days. The so-called "HiFi" drivers are universally somewhere between not very good (some really well respected names go here - all scandiavian) to awfull. And the emphasis in Pro-Drivers for high output capability before anything else compromises these units as much where ultimate fidelity is concerned, though some exceptions still exist.

Anyway, have fun and feel free to either drop me a line or to just consider to switch to antique horns and SE Valves (more milage in that anyway)

Sayonara

 

[dorkus]

Re: Re: gainclones...

hi kuei yang,

thanks for the interesting post and suggestions...

Another option which I prefer is to simply preceed the inverting mode Amp (or preamp) with a complementary J-Fet Buffer, pretty
simple really.

was thinking of doing this, but the problem is power supply complexity. the JFET buffer does not have nearly the PSSR of the amp chip so it would end up being modulated by the current demands of the amp if used on the same supply. an option is to use a regulator or a separate xformer secondary/rectifier circuit, but this makes the supply a lot more hassle. the appeal of this chip is the utter simplicity of implementation after all.

How about ditching these awfull resistive attenuators that steal at least 10db subjective dynamic range and going transformer attenuator?

i have heard a good thing or two about xformer attenuators but don't know much about them. maybe i could give one a shot.

You may be right, assuming you insist on high input impedance for no particluar sensible reason....

assumming i do not use an input transformer, i have to maintain high input impedance to avoid loading whatever's driving it. this amp is not for my own hifi rig but for a nice compact 2nd system and will most likely be driven by a portable CD player or similar device.

> other design touches... resistors will probably be IRC tantalum
> chip if they prove to be good-sounding

I like them okay, prefer NI wirewounds.

i like mills wirewound very much but with all wirewounds there is the issue of volumetric efficiency.

Carefull. This tends to form substantial tank circuits with the secondary and/or leakage inductance that ring very well, especially if "kicked" alive by the switching pulse of the diodes starting to conduct.

whoops. i knew there had to be a reason people didn't do this. maybe i'll skip this then.

But in this case consider switching from driving the speakers with undistorted voltage and thus forcing a heavily distorted current (which is after all what moves the voice coil) to driving the speakers with undistorted current which kills several types of distortion and compression in the common speaker drivers quite effectivley.

i'm not sure how you distinguish "voltage" distortion from "current" distortion. what exactly are you referring to?

or to just consider to switch to antique horns and SE Valves (more milage in that anyway)

no offense but i am not a fan of either tubes or horns. i can enjoy tubes if they are neutral (most are not, particularly SE variety). horns i can not listen to. before studying EE i was a musician (violinist), attended conservatory for 5 years (see my profile). in particular i participated in orchestra rehearsals every week so i am familiar with the harmonic structures of a wide variety of instruments. i know what the sound of live instruments is and tubes and horns distort it to my ears. to me tubes sound nasal, and horns only make that worse. i heard some SE tube/horn combos at a couple audio shows and came away with a bad taste in my mouth. sorry.

as a sanity check, go to a orchestra concert or rehearsal, then compare that to what your system sounds like. granted my system doesn't sound anywhere near realistic but it doesn't deviate from realism in the same sort of colored fashion tubes/horns do. i admit tubes can sound very alluring and euphonic but it simply isn't realistic - real acoustic sound isn't always that sort of ear candy. just my humble opinion.

 

[ThorstenL]

Re: Re: Re: gainclones...

>> But in this case consider switching from driving the speakers
>> with undistorted voltage and thus forcing a heavily distorted
>> current (which is after all what moves the voice coil) to driving
>> the speakers with undistorted current which kills several
>> types of distortion and compression in the common speaker
>> drivers quite effectivley.[/B]
>
> i'm not sure how you distinguish "voltage" distortion
> from "current" distortion.

Very simple. Apply a undistorted Voltage across a coil (say voice coil) that is filled with solid iron (say pole piece). The result is a cubic function of distortion of the current through the coil. As the force moving the diaphragm of the speaker is proportional directly and ONLY to the current through the Voicecoil, driving a conventional moving coil speaker driver from a "Voltage Source" will maximise the overall systems distortion.

In fact, if one where to compare the levels of harmanic and intermodulation distortion produced at realistic levels by the commonly found 2-Way "HiFi" speakers that seem all the rage (1" Dome Tweeter and 5.25" - 6.5" Woofer or two of these) to those of a high sensitivity system driven by the oh so awful, maligned and distorted SET Amplifier, to your surprise, most of the time a well designed SET/High Efficency based system offers much lower levels of distortion.

In fact, using Speakers with pig designed acoustic rear load, making the speaker highly reactive and thus requiring Amplifiers with a low output impedance and then pig designing the crossover to make up for driver faults and thus again neccesitating an amplifier with a low output impedance invariably maximises the distortion in the combined system.

Have you ever seen Speaker distortion measurements in the common HiFi Press? Ever wondered why not?

Soundstage has currently measurements on-line for the Ruark CL-10. They remark that it provides rather low distortion levels (which it does compared to many speakers).

So, how much distortion for 90db/2m? In the critical midrange and the format range of instruments (50Hz - 2KHz) where the ear is most sensitive the distortion peaks at -30db and averages at no better than -45db. Or turned into percentages, a minimum of around 0.6% with the worst peaks at around 3% and averages probably above 1% harmonic distortion.

No intermodulation distortion figures are given, but they look even more dire. Also not given is the spectrum of the distortion, but in most so-called HiFi Speakers third and odd harmonics (or what I call dissonant harmonics) dominate the picture above a few 100Hz.

If the Level for the example speaker is raised to 95db/2m (in my experience still not enough for realistic levels) distortion peaks at 100Hz with around -22db and and with a narrow minimum at -40db but an average close to -35db, or in percentages a maximum of around 8% THD with a minimum of around 1% and an average of around 2%.

For comparison, a decent SE Amp driving a decent High Sensitivity Speaker and correctly polarised to allow the dominant 2nd harmonics in the Amplifier to cancel the dominant magnetically and mechanically caused 2nd Harmonics in the High Efficiency Speaker will do considerably better.

I use my own system as example here. In this case the Tannoy Monitor Red 15" Speaker with around -40db/1% average 2nd Harmonics and around -50db/0.3% 3rd Harmonics for 92db/2m at which point around 1W Power Input are needed. As Amplifier a SE 300B Amplifier with around 0.9% 2nd harmonics and 0.1% 3rd harmonics (above around 300Hz, at 100Hz 2nd harmonics rise to around 1% and 3rd to 0.15%) is present.

If we connect the Amplifier and Speaker in a suitable polarity we can significantly cancel the 2nd harmonics (very much as it is done in Push-Pull Amplifiers or indeed as it is done between stages in some SE Amplifiers) to certainly significantly below 0.5% average and while the 3rd harmonic content still adds geometrically we are still only slightly above the 0.3% inherent to the driver. Maximum Power for <3% THD is BTW around 7 Watt and thus the whole system can deliver at least 100db/2m with what must be in the context of speakers "low distortion". If we accept overal 10% THD maximum we can put another db or two on top.

Thus, despite using a "high distortion" amplifier we have actually generated a system that OVERALL offers significantly lower distortion than a conventional modern system at most listening levels, all that using around 60 Years old technology, no negative feedback, no additives and colourings... ;-)

This BTW without taking refuge to current drive on either side. If we where to convert the conventional speaker to active operation and used a "constant current" operation we could basically eliminate much of the electromagnetical distortion in the driver and most of the thermal compression (another "hidden" deficiency of socalled HiFi Speakers - around 6db compression at rated RMS power are not uncommon).

The price? We must reconsider the LF enclosure. My personal opinion is to simply avoid the "around resonance" LF operation common to both Full Range Speakers and Subwoofers by operating the main ("full-range") speaker strictly above the resonace in an enclosure offering strong mechanical damping of the resonance and then to operate the Subwoofer in the "below resonance" Mode (aka Bag End ELF).

In this case "current drive" may be applied to all the drivers in the Speaker resulting in a system with much reduced distortion and compression all across the board, while retaining sane dimensions. This way we MAY actually get somewhere where the 1940's Technology of SE Triode Amplifier and High Efficiency Speaker have not already been 60 or so years ago...

Of course, SE Valves can also be operated to offer "current drive", just as any other Amplifier, so if we go active and design our rear load suitably we may reap similar benefits, as indeed done in the German Eckmiller Coaxial Speaker from the late 1930's (used for the first sterephonic magnetic tape recordings in 1943 by the German Reichsrundfunk Gesellschaft).

The Eckmiller included inherent damping in the chassis so that it could be driven by pentode or triode amplifers equally well, a series crossover and carefully designed LF and HF sections so that crossover was not required to equalise the driver response and thus drivability again by pentode and triode amplifers (SE of course).

> horns i can not listen to.

!!?? Maybe time you listen to some good ones....

> i heard some SE tube/horn combos at a couple audio shows
> and came away with a bad taste in my mouth. sorry.

I usually get that from the Krell/Wilson Rooms and those with Mark Levinson gear. Yet I have heard similar systems outside shows that where surprisingly good. That said, most commercial SET/Horn systems are equally wrong, if less unpleasantly so.

BTW, I do not recommend ANY currently produced Horn Systems with the exception of the Beauhorn Virtuoso's (need subwoofer and supertweeter though) and the Tannoy Westminster Royal. I have heard most of the other big names and recommend in general avoidance as the better course of action.

> as a sanity check, go to a orchestra concert or rehearsal,

I regulary do that. Most recently Prokovievs Violin Concerto (#2 I think) and the Mussorsky/Ravel "Pictures". I have yet to hear ANY Panel based or conventional dynamic system that can do the kind of music like "pictures" even remote justice.

> then compare that to what your system sounds like.

Yup. Within the limits of what can be recorded it is surprisingly faithful.

> but it doesn't deviate from realism in the same sort of colored
> fashion tubes/horns do.

I think many of the commercial systems in that line are indeed severely coloured, often by demand. But I thought this was DIY Audio.com? So the issue here are principles and application, not commercially produced gear, IMNSHO.

Well impemented Horns offer many advantages over conventional dynamic and panel systems, from better dispersion control (thus less room interaction) over reduced levels of distortion and compression, all contributing to a much reduced level of coloration and deviation of the acoustic output signal from the electrical input signal to the Amplifier/Speaker system (these two CANNOT be divorced - view each in isolation only at the peril of loosing the overall picture).

> i admit tubes can sound very alluring and euphonic but it simply
> isn't realistic

Given plenty of experience on both sides of the fence and given a good deal of listening to music live (and recording it on occasion) I must strongly disagree with your statement.

Tubes PER SE are not Euphonic, nor are High Efficiency (this usually means Horns) Speakers. Certainly at least no more than the commonly found socalled High Fidelity Speakers and their "low distortion, low output impedance" companion amplifiers.

Sayonara.

 

[dorkus]

interesting points. i'll surrender our discussion to the original point of this thread (gainclones) but i will say that i agree with you that most commercial design tend to be very colored, regardless of technology. it may very well be that i haven't heard a decent horn yet - they're hard to find in this country and audio shows are notorious for bad-sounding exhibits - so if i get the chance i'll give them a fair shot (i'd be interested in the Tannoys). i also agree that loudspeaker distortions are huge and relatively undocumented, though the idea of driving a speaker with a current source is new to me.

 

[fdegrove]

A WISE MAN.

Hello,

Couldn't help but notice your great post.

Having about 25 years of experience in the field myself I must say you seem to have a good understanding of what is required for faithfull reproduction of music.

I hope to cross you in the tube section and (I'm no expert there) in the speaker section as well,you may have other interests I'm not aware of.

Anyway you have a very good point and I look forward to exchange experience with you.

Regards,

 

[planet10]

Re: Re: Re: Re: gainclones...

Very simple. Apply a undistorted Voltage across a coil (say voice coil)

Thorsten,

Your moniker had me going for a bit, good one. A VERY BIG welcome to this audio community. It is nice when those with the likes of your experience join us and strengthen the whole community.

And very appropriate that your 1st posts are on this Frugal-Phile(tm) beauty that you played such a large role in.

dave

 

[ThorstenL]

Hi,

> so if i get the chance i'll give them a fair shot (i'd be interested
> in the Tannoys).

If perchance you find yourself in Londra, Capital of the Dark Empire of Granbretan, feel free to drop by my place for a listen.

> though the idea of driving a speaker with a current source is
> new to me.

Actually, it's ancient....

Sayonara.

 

[dorkus]

If perchance you find yourself in Londra, Capital of the Dark Empire of Granbretan, feel free to drop by my place for a listen.

thanks very much for the invite, i may just have to take you up on your offer one of these days... with NYC/London round trip airfares as low as $189 right now my roommate and i just might have to shoot over there for an extended weekend.

 

[mlloyd1]

Maybe Jef Rowland paid them to take the PDF down?

....Do you still have the PDF?
It was on NSCs web site for about 1 month before it was retracted. ...but they said it couldnt be given out for some strange reasons.
....

 

[Fred Dieckmann]

It was on NSCs web site for about 1 month before it was retracted

Another urban legend........


http://www.national.com/appinfo/audio/files/BPA-200_Application_Note.pdf

you kids.....
Fred

 

[pmkap]

Lectro Kaps 'n Klones

As a point of possible interest.....

Ruby-Con offers a number electrolytics of interest. While their caps are generally available only in quantiies of 1000s, Pioneer does have their 50ZL1000M16X25 in stock, no minimum, @ $0.72each.
http://mypioneer.com/onestop/pio_ca...PN&pnmatch=any&mfg_code='RBY'&display_timers=

Their specs (as if it matters) do seem to be a bit better than the Panasonic FCs. Ripple 3010ma vs. 2235-2590ma, and impedance speced at 20c 100kz( where these SMPS electros are used ) .021mOhm vs. .030-.034mOhms. The same size FCs, 16x25, as opposed to the 12.5x40, have the lower ripple rating and the higher impedance.

Dunno if these specs would translate into better subjective performance, but if anyone wants to give them a try, I'd be interested in their impressions....

 

[fdegrove]

HIPPO.

Hi,

Is it that hard for you to be a little bit lenient towards the not so well informed?

> though the idea of driving a speaker with a current source is

Statement of the year perhaps?