Anyone built a Gainclone?

[mcs]

I've done some testing now. It works well. It's completely quiet when there's no input signal (I'm used to tube amps :wink. I have only fitted a small heatsink, and that stays quite cool even when the amp's working.

I have also made a couple of measurements. For my testing I used a 2*15V 100VA toroid. The actual output voltage was 2*18V though, both with and without load. The DC voltage was around +/-24V at idle.

The maximum output voltage without clipping was 19V peak. The DC voltage dropped to +/-22.3V at that point. My load resistor is 8.33R, so the max. output power is 22W. The output looks like a perfect sinewave until the clipping point. Above that you get a strange looking waveform caused by the protection circuit I think (the SPiKe stuff). The frequency response is good. The output started dropping slightly above 100kHz, but that should be enough I guess...

The only problem I found was that the amp oscillated badly when the output was loaded by the resistor (and scope), and the input was left unconnected. Does anybody have an idea why? If the input was shorted, or connected to a signal source there were no problems.

I will get a 2*22V transformer to find out what the maximum output power actually is. I think 48-50W should be possible.

Best regards,

Mikkel C. Simonsen

 

[dayvel]

I used a LM1875 to build a utility amp ages ago and have always been pleased with it. I stumbled on a couple GainClone forums a while back and was amazed at the mystical properties people attribute to these things. Just don't do anything dumb with the PCB layout and the circuit in National's app notes will work just fine.
A 47pF cap across the input should take care of that oscillation.

 

[Aharon]

[quote author="mcs"][quote author="daArry"]

[/quote]
What's this board for?

And it looks like the box is a recycled something?

The PCB/schematic for my version is here: http://stiftsbogtrykkeriet.dk/~mcs/LM3875/
I'll take some pictures when the boards arrive in a couple of days.

Best regards,

Mikkel C. Simonsen[/quote]



I get an error files do not begin with PDF and on top it kind of almost crashed my computer.What's going on?.
Aharon

 

[mcs]

[quote author="Aharon"]I get an error files do not begin with PDF and on top it kind of almost crashed my computer.What's going on?.
Aharon[/quote]
It's a PDF virus!

A lot of strange stuff can happen, when you try to view PDF files with the Acrobat plugin. I removed the plugin for the same reason years ago, and setup GSView as a "helper application" for PDFs.

Just save the files on your harddisk and open them from there, and the problems should be gone. Otherwise you may need a newer or older Acrobat reader - all the Acrobat applications are buggy as hell...

Best regards,

Mikkel C. Simonsen

 

[Aharon]

Just save the files on your harddisk and open them from there, and the problems should be gone. Otherwise you may need a newer or older Acrobat reader - all the Acrobat applications are buggy as hell...

Best regards,

Mikkel C. Simonsen




Thanks Mikkel,it worked.
Aharon

 

[neilium]

Hi,

Just recently put my Gainclone back into use. It's on a pair of Celestion 2-ways. It still sounds good to me.

-neil

 

[mcs]

I posted some photos and my test results here:

http://stiftsbogtrykkeriet.dk/~mcs/LM3875/index.html

Anybody interested in PCBs? - I guess I should post in the black market...

Best regards,

Mikkel C. Simonsen

 

[mcs]

I finally had some time to test the board at higher voltages.

The voltage was +/-32.2VDC at idle (2*22V transformer). The voltage at full power dropped to +/-30V. The output just below clipping at 1kHz was 17.25Vrms, which makes the output 36W.

But I couldn't get full power at high frequencies (around 20kHz I think). The output was clipped, so the power was reduced to about 31W accross the full range.

The small heatsink got quite hot, so I tried replacing it with a larger one - that helped! I could then get about 37W accross the entire frequency range. Why would a small heatsink cause problems at high frequencies only?

You can see the heatsinks here: http://stiftsbogtrykkeriet.dk/~mcs/Heatsinks.jpg
The small one has a thermal resistance of 3.4°C/W - I don't know about the big one.

The frequency responce (at about 7W) was 17Hz-110kHz at -1dB and 7Hz-220kHz at -3dB.

The output power could probably get a bit higher with a 2*24-25V transformer...

Best regards,

Mikkel C. Simonsen

 

[clintrubber]

The output was clipped, so the power was reduced to about 31W accross the full range.

Did you have a look at the waveforms ? The '3875 has quite something 'nasty' onboard for protection (the SPiKe-protection).

Don't know if you already trigger that SPiKe-protection.

If it gets too hot the SPiKe-prot. starts chopping the waveform. This sounds quite ehhm, :twisted: BUT for an application this is usually better than shutting off the whole amplifier and let it cool down before starting again (causing audio-holes).

We've been listening at this amp at work to see if their way of protecting an IC-audio amp made sense. It sure does because as said, it keeps the music playing. But it starts to sound like somebody threw sand & fine rocks into your speaker-box.

Obviously there are more elegant ways of protection (don't know if those Philips-chips are already on the market :wink: and without doubt National will have something going here as well).

BTW, there's an old but good related IEEE-article by the great & late Widlar, if anyone's interested I'll search for a link.

Peter

 

[mcs]

[quote author="clintrubber"]Did you have a look at the waveforms ? The '3875 has quite something 'nasty' onboard for protection (the SPiKe-protection).[/quote]
The waveform was a standard sinewave with "zig-zag" at the bottom of the negative halfwave - quite strange looking...

Best regards,

Mikkel C. Simonsen

 

[clintrubber]

from Mikkel:

The waveform was a standard sinewave with "zig-zag" at the bottom of the negative halfwave - quite strange looking...

Keep pushing that amp, keep pushing. If I recall it correctly then the positive halfwave might get 'ziggy' as well. :sad:

The '3875 datasheet has a figure of 'SPiKe in action', page 6 in my version of the pdf-file.

Bye,

Peter

 

[clintrubber]

BYW, note that for a protection-circuit decreasing the drive to the internal powerdevices somewhat can make things worse.

You'll have seen power-stage dissipation curves vs output power. Internal dissipation is usually the highest for half the max output amplitude.

So if you're say at 90% and it gets too hot, decreasing the drive for say 70% amplitude will make things worse - the internal dissipation increases further. So you have to cross that max. internal dissipation point all the way through, in this case by simply shutting off the drive completely for that halfcycle. And try again the next halfcycle - as I recall it the SPiKe works on a very instantaneous basis.

There might be a link to your higher high-freq distortion because of this.
You could imagine the protection introduces quite some artefacts, which may be relatively of more influence (but those higher distortions start to shift outside our hearing range).
What measuring filter did you use ?

Last but not least there's always of course the worse high-freq performance of devices like this - not unlikely this is the major influence here (worse THD @ 10k vs 1kHz, caused by less gain-margin or 'gain-headroom' between the open & closed loop curves).

Time for some sleep here now, bye,

Peter

 

[mcs]

[quote author="clintrubber"]What measuring filter did you use ? [/quote]
Que? I used my scope with a 10X probe - where do you want a filter?

Best regards,

Mikkel C. Simonsen

 

[peranders]

I have chosen inverting mode for my SMD Gainclone. How much better is inverting mode? Don't know but my idea is that you must have a good buffer in front of the LM3886 in order to drive the LM3886 optimally. If I'm right or wrong, I'll guess time will tell.

 

[clintrubber]

from Mikkel:

clintrubber wrote:
>What measuring filter did you use ?

Que? I used my scope with a 10X probe - where do you want a filter?

Sorry, I wrongly assumed there were THD-figures in your previous post.
(So was wondering about the bandwidth, but it doesn't apply now.)

Bye,

Peter

 

[daArry]

Hey

I made a LM3886 amp with a INA134 infront and a 22K att inbetween for level control - all good so far

 

[mcs]

[quote author="clintrubber"]Sorry, I wrongly assumed there were THD-figures in your previous post.
(So was wondering about the bandwidth, but it doesn't apply now.)[/quote]
I don't think I can measure the distortion with my 334a when the amp isn't clipping...

Best regards,

Mikkel C. Simonsen

 

[clintrubber]

FWIW, more pics of the '3875-protection are in:

http://www.national.com/an/AN/AN-898.pdf

 

[pstcho]

hi, does somebody have an idea to bridge 2 lm3886 based gainclone channel?

thank you

 

[peranders]

Lot's boards out there. The basic information can be found here: http://www.national.com/an/AN/AN-1192.pdf

Suitbale keywords for Google may be lm3886 bridge pcb

I have made a pcb with internal bridge connection but there are at least half a dozen more pcb's out there.