The LM1875 for best

[kokoriantz]

I decided to design the LM1875 for its best. First I will study the character in detail , sort out the problems to resolve and bring solutions to get highest quality possible. It can be paralleled and bridged, it can be composite or not.
Hayk

[kokoriantz]

This amp is in fact a power opamp with open loop frequency of 70hz and 90db gain. This is measured with 2k load. If the output is a tranconductance amplifier as I can guess from the internal circuit, than the OLG/BW depend upon the load. This is not specified but what betrays is the need of output snubber to limit the gain for inductive load at high frequencies. If the output impedance vs frequency could inform the distortion vs frequency does also and it is provided.
One can see that the distortion is flat to 10khz, this means with 8 ohms load it has the same NFB on the whole spectrum despite decreasing gain shown on OLG curve for 2k load. Comparatively see that of LM3886.


If you look at low frequency distortion of LM1875 you can see it start rising bellow 70hz as the compensation capacitor is no more providing the internal feedback. If you look that of LM3886, although the OLBW is few hertz it starts distorting bellow 200hz Why?. This phenomena occurs in IC amplifiers because the temperature variations of the outputs alter the Hfe of the low power transistors provoking nonlinearities. The LM1875 not only better conceived but also the measurement is done at a point where the power dissipation remains constant, hence less internal temperature variations.


If look at 4 ohms load curve distortion starts at higher frequency and gets double at 20hz vs 1.5. This is due to variable power dissipation at measured point


conclusion, This amp can give it's best when it is running at constant temperature/dissipation. The curve for best sound is that of 8 ohms load with +/-20v supply with maximum power at 1% distortion of 17w.

[chermann]

Hi Hayk


Interesting view...

[JMFahey]

Just thinking aloud: TO220 5pin Class AB amps are being dropped, most already disappeared, LM1875 is practically the last one of a dying breed.

Isn´t it a little too late for this effort?

[Sangram]

My experience with the chip is that it is quite a bit better sounding than the 3886 (in spite of "worse" paper spec), but implementation is key. It is not as stable/reliable/indestructible as the 3886, conversely its simplicity makes it easier to get good sound.

However one must stay clear of the temptation to minimise the component count. It needs control over the turn-off thump, lacking any form of low Vs-muting. It also needs substantial number of components in the input side to ensure stability and good distortion performance. You've already covered the Zobel, this is pretty much mandatory for these chips.

I like to use it as a tweeter/mid amplifier in active systems. The current limit is a little on the low side for proper full range performance into medium sensitivity speakers. I am also keen to see how it behaves with current feedback, for more demanding loads. I have a few of these built out that run on 21V rails, and they sound superb. There is an output relay that drops out before the chip, so turn-off thump is avoided.

Keen to see where you go with this.

[chermann]

Hi Sangram


Interesting feedback. so you are going through the LM1875 completely.
Can you explain what you prefer in case of sound between LM1875 and LM3886?


I think you are using Class D for the low frequencies?


thx
chris

[Mooly]

I've been doodling around... nothing built though. Also an LM3886 version. The LM1875 would really need three rather than two chips.

[Sangram]

Chris, I'm using LM3886 for LF duty running off the same supply. While the output power into the same load is the same, it runs more relaxed during higher current passages.

I find the 3886 more neutral but the 1875 has a nicer quality in the midrange. I can't put a quantity to it though I would like to, at some point. A well-implemented 3886 can sound superb.

Mooly, that is quite the doodle It's an interesting approach, but the chips are only used to provide DC balance, correct? I used this sort of an approach to bridge TDA2030s together in the early part o the century, and repeated the 'trick' to bridge LM3886s for a subwoofer application (8 ohm driver).

[Mooly]

Thanks

It is a true bridged design. The lat fet front end is one half and the parallel chips are the other half configured as inverting voltage followers.

Nothing says bridged amps have to use identical stages

The chips run with increased noise gain for stability in the unity gain configuration which is provided by R19 and R20.

This shows the voltage output of the front end stage and the chip amp stage with its 180 degree phase shift.

The small opamp (a servo) provides active biasing of the chip stage to maintain a perfect zero volts DC offset across the load.

A FET speaker relay and photovoltaic coupler which would be driven from extra circuitry to provide a switch on/off mute function.

[adason]

Originally Posted by kokoriantz

I decided to design the LM1875 for its best. First I will study the character in detail , sort out the problems to resolve and bring solutions to get highest quality possible. It can be paralleled and bridged, it can be composite or not.
Hayk

How about nested?