op amp based power amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi guys,
I quickly put this amp together, its more or less a copy of rod's project 76...

Will this amp work? do i need to change anything? is the voltage ok?
thanks,
Kinser
 

Attachments

  • op-amp power amp copy.gif
    op-amp power amp copy.gif
    6.5 KB · Views: 945
You need to place resistors from Q2 collector to Q1 emitter, and from Q4 collector to Q3 emitter. Otherwise any bias current flowing in Q2/Q4 will cause Q1/Q3 to turn fully on, making proper biasing impossible. Somewhere around 1K will do, but the optimum value depends on the particular transistors.

I would recommend using additional 0.1uF ceramic bypass capacitors near the op-amp supply pins.

Lastly, there seems little point using +/-35V rails for the output stage since it lacks voltage gain and thus can never give an output voltage greater than +/-15V at best. If you want the higher power from higher voltage rails then you will need to either use an output stage with gain (which has potential stability problems), or use a high voltage op-amp, such as OPA445 (I've used those to make power amps before and they work well up to +/-45V).
 
There are better ways of getting more power and more voltage swing from an op amp based amplifier.

Choosing a really good audio opamp will have a big effect on the overall sound and quality of the result too... as will properly trimming the feedback network... and you may or may not want to add a small output inductor.

If you poke around the various op amp mfrs app notes you'll doubtless come across one or more of them that are far more attractive than the darlington output stage shown in this schematic! :cool:

Of course, for any first project, any schematic and plan is better than none, so no harm in trying this one.


_-_-bear :Pawprint:
 
In order to get any voltage gain from the output stage:

Q2 and Q4 should be referenced to ground. Recomend emitter resistors (maybe 22 Ohm) on Q2 & Q4; A feed back resistor (maybe 1k)from the collectors of Q1 & Q3 to the middle of the two emitter resistors of Q2 & Q4. Connect a resistor (maybe 100 Ohm) from this juction point to ground. Also the outputs, Q1 & Q3, need to have current limiting resistors (maybe 0.4Ohms) from each emitter to the rails. A resistor from the base of the outputs (also the collector of the drivers) to the rails is needed, a pull up resistor. R1 and R2(pot) should then be replaced with a current source (likewise on the negative side) that need to bias Q2 and Q4 on and for stability.

Connect the point where the bias diodes come together in the middle to the output of the op-amp. Maybe place a resistor in paralell with each diode. Might also want to put a pot accross the bases of Q2 & Q4 so bias can be adjusted.

This will make the output topology a feedback pair instead of darlington. Feedback pair is not the most stable type of output arrangment, but if PROPERLY biased, it will give much voltage gain. You might consider an emitter follower as output stage.

Sorry to re-arange your circuit so much, but as Mr Evil states, there isn't any voltage gain beyond the op-amp.:D
 
Truly ancient thread, but either I'm searching in all the wrong places or there's little discussion of op amp based power amps---the contributed project hosted by Rod Elliot which the OP mentions is basically it. I'm surprised by this as the 10-20W RMS space accessible by single ended op amp designs without voltage gain in the output stage is plenty of power for many applications (enough for roughly 100dB clean SPL given halfway efficient drivers, particularly if biamping or triamping) and it can be nice to have higher fidelity than is usually offered by class D or chipamp parts without the complexity of a fully discrete design. Granted, the LME49811 and LME49830 also address this space and are nice parts which have been exhaustively discussed over in the chip amp forum, but op amps offer some different tradeoffs around cost, GBP, open loop gain, CMRR, PSRR, and so on which I happen to find interesting. Also notable is the LME49274, which makes a nice controller for a fully differential amp that'll do 40W into 8 ohms (80W for 4 ohms).

tomchr and I have been discussing offline and he suggested roughly the attached design, which I happen to like better than project 76 for various reasons, mainly greater stability and lower supply coupling on the bias from the LM334 current source and improved thermal tracking from the Sanken STD03s. Obviously, many variations with different op amps, output stages, and placement of the LM334 on Vcc instead of Vee are possible here but I'm specifically curious what folks' experience with the bias stabilization capacitor (C4) has been. I wouldn't go smaller than 100nF and bigger than 10uF seems like overkill. But I have a board I'm laying out where it turns out the choice of stuffing an 1206 C0G 100nF versus a 10uF Elna RFS (Silmic II) versus, say, 100nF || 10uF controls the width of the board and has a significant effect on panel yield. :p Thoughts?
 

Attachments

  • STD03xAmp.png
    STD03xAmp.png
    26 KB · Views: 483
Last edited:
Thanks. Happen to know of any data about 1uF which would help me reason about whether, say, 100nF would be fine too? My rule of thumb is the cap should be a good bit larger than the capacitance of the STD03s and enough to absorb the 334's transient response. Sanken doesn't spec the STD03 input capacitances but, since they're BJTs and Darlingtons at that, I would expect them to be pretty small---O(10pF)---so the main criteria's probably maintaining a flat voltage through the 334's transient response. Back of the envelope is suggesting to me 100nF should be plenty---it'll store about 120nC and a 334 transient's around 100pC at the 2.5mA set current for the Sankens---but it's not clear to me if there are reasons to want more than 60dB "attenuation" here.
 
Last edited:
there is the "floating ps" style
Hmm, interesting. Nothing quite comparable, but some choices around the 2.2nF range. The 334 should have a couple hundred pA/sqrt(Hz) current noise, equivalent to a few nA integrated out to 1MHz and rather higher than plain transistor noise, so I'd take that up to 10+nF.

If you poke around the various op amp mfrs app notes you'll doubtless come across one or more of them that are far more attractive than the darlington output stage shown in this schematic! :cool:
Actually, I haven't been coming up with much. Analog AN-211 is one I'd seen some time ago but had forgotten about; uses a 10uF apropos of my question of the moment. National AN-272 has a few. (Quoted for context, not because I think bear cares about a six year old post.)

Oh, and I see artu is using 1uF, albeit in a rather different topology.
 
It's not a bias stabilisation cap it's a speedup cap to help the output devices switch quicker at high frequencies. 1uF as shown in your schematic is fine.

We may be saying the same thing, but here's my understanding of the Vbe multiplier. The Vbe multiplier (or diode stack in case of twest's schematic) is supposed to present a DC bias to the output devices. It should be an AC short circuit. Of course, that only works until the frequency of operation gets within some fraction of the ft of the Vbe multiplier device. The cap across the Vbe multiplier will help maintaining that AC short at high frequency.

In AN-1850 (app note for the LME49830), National uses 47 uF || 30 pF. Granted, that's a MOS output stage, but still...

Parasound uses two 100 nF polypropylene caps in parallel in their A23.

In my LME49811 + STD03 build, I recall measuring THD at 1 kHz at relatively low output powers. I forget the exact details, but I think I tried 100 nF across the Vbe multiplier and observed a measurable improvement in THD+N with 10 uF. I stopped there...

~Tom
 
Kinda fun class H sort of thing. With a 330nF bias stabilization cap. :p Would be interesting to try sometime, though personally I don't need more than 10W RMS or so.

I was able to get 1206+1210 on my board so I'll probably try permutations around 100nF C0G || 10uF X7R. This is a side project whose boards are piggybacking on another order so it'll likely be quite some time before I have data.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.