LM1875 nested Gainclone

[pacificblue]

Originally posted by linuxguru
The overall closed-loop gain of both op-amps is 28, and the closed-loop gain of the TDA is 23. That forces the swing at the output of the NE5532 to be fairly low - less than 1V peak. However, the closed-loop gain of the combination is not 5, it's still 28.

You're right. The NE5532's contributions is not 28-23=5, it is 28/23=~1,2.

[anti]

Hi - thanks again for the informative answer.

For the "TDA" combos, I meant actually if the likes of TL07x would be "okay" (not trying to "repair" their spec with a "super" opamp). I guess bad wording.

I have a couple of lm386 I'd like to use (and a TDA2009 in an old amp I'd like to revamp). This approach seems like a good solution.

What I also considered - for the lm386 - would it be possible to simplify the "nested opamp" approach and use a single FET or a BJT nested/wrapped around the chip-amp (instead of the opamp - similar to some composite RIAA pres but without the frequency correction curve), taking the neg. feedback signal from the output? Has this been done yet? Any pointers perhaps?

[gfiandy]

Hi linuxguru

Thanks for your feedback.

The 27K was a typo, I have also added the HF filter which I had left off.

Thanks for the info on the 0.22R, I thought it was just to isolate the output from cable capacitance. Do you know the mechanism by which it reduces H3 or was that discovered impirically.

Your right 1M probably isn't required with an amp like the NE5532 and with the TDAs gain being defined. However I worked on a compound opamp design in which the origonal designer had not controled the open loop gains of either of the opamps, relying entirely on the single loop to control the behaviour and they went unstable all the time as the loop characteristics changed. So I am a bit over sensitised to leaving opamps open loop locally, even if the entire loop gain is controlled. I have added a location for the 1M resistor, I don't have to fit it if its not required or detrimental.

However I have found in the past that controling the open loop gain of internal parts of an amplifer so that it is flat accross the audio bandwidth can improve the sound quality. So it may be worth me doing some experiments to see what effect it does have.

Thanks for your help,
Regards,
Andrew

[linuxguru]

I plugged a gain-limiting resistance in parallel with C4 into LTSpice, and there is minimal impact on distortion and sonics between 470k to 1M, and up to 10 dB higher distortion at 100k. On the whole, the impact is minimal, so one might as well include a 330k..470k gain-limiting resistor for repeatable stability, especially with faster or higher-gain opamps like the LM4562 or LT1469. I'll post an updated schematic with the resistor and a slightly optimized compensation network in due course.

Using an LT1208 with a slightly tweaked gain and compensation network actually shows greater phase margin (up to 80-85 degrees) compared to ~70 degrees with the NE5532. I don't think it matters much to the sonics, but it might be worthing rolling op-amps and auditioning.

[anti]

Hi Linuxguru!

Some more questions:
- would it be much different if a NE5534 was used instead of a NE5532 (f.e. in a "monobloc" board?)
- would it be possible then to take the signal from NE5534's "5-pin" - the compensation pin, bypassing the output stage (since the feedforward was omitted)?
- if a NE5532 was used, would it be possible to use the second opamp as a servo and WHERE should it be placed for stability and precision?
- Would it be much different if NE5532 was A-class biased (f.e. 6k8 to "+" supply)?

Just a question: this way, we are getting closer and closer to the Jung's circuit, no? But the 0.22R resistor is used as another "error" sensor (Jung has different config)? (link)

[AndrewT]

Is it possible to compensate the range of National Chipamps when used at a gain = 2?

[linuxguru]

> would it be much different if a NE5534 was used instead of a NE5532 (f.e. in a "monobloc" board?)

Should work, since there is external compensation C4.

> would it be possible then to take the signal from NE5534's "5-pin" - the compensation pin, bypassing the output stage (since the feedforward was omitted)?

Depends on the drive capability from the VAS (?) of the NE5534, but perhaps possible if the resistor network at the input(s) of the chipamp is made higher impedance, say of the order of ~10k.

> if a NE5532 was used, would it be possible to use the second opamp as a servo and WHERE should it be placed for stability and precision?

I haven't thought about a DC servo yet.

> Would it be much different if NE5532 was A-class biased (f.e. 6k8 to "+" supply)?

Not much - at low swings (< 1V), the NE5532 seems to have fairly low distortion already. I'll plug in the 6.8k resistor later and check - it's an inexpensive improvement.

> Is it possible to compensate the range of National Chipamps when used at a gain = 2?

In the sims, the LM1875 can be used below the recommended gain of 10 with suitable compensation. However, I'm not sure that it extends all the way to a gain of 2. For unity gain or similar operation, the TI/Burr-Brown OPA549 family may be more useful, since they're unity-gain stable.

[AndrewT]

Hi,
the NE5532/4 is not optimum for a DC servo.
The LF411 is excellent for DC servo duty.
There are a number of more modern FET input opamps that can do even better, but is better needed?

[stocktrader200]

I have been using a similar circuit for several years now, but with the
differences of the op amp runs on +- 15 volts, output +- 36 volts
I am wondering if this cct will still work as well if the op amp voltage is reduced and the input divider network to the LM1875 reduced by half.
This is optimal for me because the output is not limited to the op amp's
maximum supply voltage.

[linuxguru]

The zeners ensure that the NE5532 runs on +/- 15V in this circuit as well, but it's probably an improvement to use 78L15 and 79L15 voltage regulators instead. The rails for the chipamp can be higher, depending on the chipamp.