BTW opamp-array amps handle short-circuits without failing (assuming enough ventilation to cool them) as each opamp is output current limited for indefinite periods. I might add big diodes to the rails on the output to protect against output voltage excursion from a reactive load, thinking about it.
With lots of devices you might get a reliability issue - any opamp failing such that it shorts the rails or its inputs might be an issue, and with 256 of them the probability is going to be higher. Of course 256 times "very low" could still be "very low".
With lots of devices you might get a reliability issue - any opamp failing such that it shorts the rails or its inputs might be an issue, and with 256 of them the probability is going to be higher. Of course 256 times "very low" could still be "very low".
I finally got round to getting some PCBs made up for my attempt at an opamp array amp, using SMT NE5532s. I've half-populated one of them:
I'm seeing about 60C temperatures (according to IR camera) with +/-15V supply. The opamps are higher than the other surface mount components so adding a heatsinking metal plate is probably the simplest approach if I want to tame these temperatures.
The pcb underside is all +/- supply distribution.

I'm seeing about 60C temperatures (according to IR camera) with +/-15V supply. The opamps are higher than the other surface mount components so adding a heatsinking metal plate is probably the simplest approach if I want to tame these temperatures.
The pcb underside is all +/- supply distribution.
Nice! I'm still enjoying mine--totally worth the effort.I finally got round to getting some PCBs made up for my attempt at an opamp array amp, using SMT NE5532s.
I'm seeing about 60C temperatures (according to IR camera) with +/-15V supply. The opamps are higher than the other surface mount components so adding a heatsinking metal plate is probably the simplest approach if I want to tame these temperatures.
The 5532 has slightly higher quiescent current (I think it was around 4 mA per channel?) than the OPA1602 (2.6 mA per channel) that I ended up using, so you would expect the NE5532 ICs to run a bit warmer. A metal plate heat sink should work quite well, however, from a reliability perspective I wouldn't expect these temperatures to be problematic.
For low quiescent power perhaps OPA2145's - expensive though!! They have 445uA per amp, although less drive capability.
OPA1662's are 1.5mA per amp, and can drive 600 ohms, perhaps a suitable compromise.
However for cost alone the 5532 is very attractive.
OPA1662's are 1.5mA per amp, and can drive 600 ohms, perhaps a suitable compromise.
However for cost alone the 5532 is very attractive.
Most definitively. Have you tested the first board already? I’m eager to hear your experience. 🙂However for cost alone the 5532 is very attractive.
Bergquist (and probably others) produce various "gap pads" in all sorts of thicknesses that could be easily used to thermally connect the tops of the opamps to a metal plate. We used these quite a bit at work when chips didn´t get enough cooling through the underside.The opamps are higher than the other surface mount components so adding a heatsinking metal plate is probably the simplest approach if I want to tame these temperatures.
Then again I recall Jan Meier once built a power amp with many LM6171 all biased in class-A and he didn´t use any heatsink IIRC.
Found it: A Precision Preamplifier-Power Amplifier System with Natural Crossfeed Processing. – HeadWize Memorial
Last edited:
Then again I recall Jan Meier once built a power amp with many LM6171 all biased in class-A and he didn´t use any heatsink IIRC.
Found it: A Precision Preamplifier-Power Amplifier System with Natural Crossfeed Processing. – HeadWize Memorial
That opamp amplifier I hadn't seen before--thanks for linking it! I agree a heat sink probably isn't needed, but it won't hurt either.
And if you need more power, stick a transformer on the output. Douglas Self would love that!
(sarcasm).
McIntosh make O/P transformers with good THD specs. If you can find a supplier with the same level of skill, why not?
(sarcasm).
McIntosh make O/P transformers with good THD specs. If you can find a supplier with the same level of skill, why not?
I’m not following. An output transformer only transforms the same power level. You can change the impedance or output voltage swing, but not the power as energy is conserved.
If an inductor is used as a choke loaded CCS in the output stage of a SE class A amp, then it can store energy and release it temporally to increase the peak output power for brief period.
If an inductor is used as a choke loaded CCS in the output stage of a SE class A amp, then it can store energy and release it temporally to increase the peak output power for brief period.
Sorry - I should have worded it clearer. A transformer will step up the voltage, which will give you more 'power', in that regardless of how much current the amp can supply, unless it can generate sufficiently high enough voltage into a load, it won't realise it.
I note my image got lost in some upgrade to the forum, here it is again http://sphinx.mythic-beasts.com/~markt/opamp_array.jpg
I’ve been using this ‘op amp’ amplifier almost daily for over 5 years now! When building it, I wondered if the high number of op amps would have ramifications for the life expectancy of the amplifier, but it turns out operational amplifiers are robust little wonders of modern engineering with all of ’em happily plugging along to this day.
Looking back, this is a crazy design, but I still think it has lots of charm. You certainly don’t see these every day and the performance is out of this world—a few years ago I used a few ultra-pure fixed frequency oscillators, notch filters, an external preamplifier and the QA401 to measure the distortion at about -130dB at 1 kHz and 2 kHz and a couple dB worse at 5 kHz. Too bad no speakers exist that could possibly match that. 😉
Looking back, this is a crazy design, but I still think it has lots of charm. You certainly don’t see these every day and the performance is out of this world—a few years ago I used a few ultra-pure fixed frequency oscillators, notch filters, an external preamplifier and the QA401 to measure the distortion at about -130dB at 1 kHz and 2 kHz and a couple dB worse at 5 kHz. Too bad no speakers exist that could possibly match that. 😉
- Home
- Amplifiers
- Chip Amps
- Building a massively parallel op amp power amplifier