If you're driving cables, NJM4580 comes in SOIC & swings full 13.5 volts (rail - 1.5v) down to 600 ohms.
MC33078 swings 11.5 volts on 600 ohms (rail -3.5 v), and comes in a quad 33079. I've used 33078, like the sound at 50x gain. No hiss. 33078 needs power bypass cap (0.1 uf + to - rail, one for 2 packages) and feedback resistor bypass cap (33pf) to not oscillate.
Both are under $1.
MC33078 swings 11.5 volts on 600 ohms (rail -3.5 v), and comes in a quad 33079. I've used 33078, like the sound at 50x gain. No hiss. 33078 needs power bypass cap (0.1 uf + to - rail, one for 2 packages) and feedback resistor bypass cap (33pf) to not oscillate.
Both are under $1.
Last edited:
.
No thanks. I had a good laugh on myself. I made up some little boards to parallel two of the AD8229 in-amps and a four resistor plus op-amp summer to make a single output. When I populated it with a 797 it oscillated.
Hence the reason several of us are pushing the 1644, as it's altogether a better replacement for the TL and OPx134 series (especially in light of the soic14 package constraint). None of these guys are in love with a 600r load, but the x134 and 164x series have similar current delivery capabilities, and the latter is lower distortion across the board. If you need to drive a 600 ohm load at the end of the crossover, perhaps a buffer would be worthwhile.
Ok im going to construct rod elliots p09 and he uses tlo series for the crossover but suggests using a better opamp for the final buffer which has a gain of 2. The tlo devices are readily available here and cheap, I can order any of the better devices suggested and it appears this is a far better option. I have built 2 stereo class d amplifiers which i wish to put a high pass crossover in one of them and a low pass in the other one, so 1 quad device gives me an input buffer, 2 filter stages and an output buffer per channel. The specs for the class d board state the input impedance is 3.3k but suggest a driver capable of driving <=600r. Greg
TL07x are extremely wimpy by today's standards (much like NJM4558s). You can parallel four of 'em and still not get near a single 5532 in lower-impedance driving. When you need high levels (several Vrms), you may want to keep output loading to at least 22k even. At least their integrated series resistance makes them easily stackable. They also exhibit a lot of common-mode distortion and the usual amount of input impedance distortion for JFET input jobs.
How were you even supposed to use them, you ask? Inverting, that's how. Then suddenly their performance characteristics start making some kind of sense, including voltage noise. By the standards of the time (the mid-1970s), they're perfect for use in a mixer or something. I suppose the reasoning behind OPA(2)604 was similar, just for higher dynamic range (higher supplies, less noise).
Good output drivers often aren't available as quads. I believe it's a die size, power dissipation and manufacturing volume issue. For example, look at how the dual LM833 had a regular complementary output stage while its quad companion LM837 had to make do with a (seemingly slower) quasicomp.
NJR/JRC has one quad that might be of interest, the NJM2060V (quad version of the trusty NJM4560, so rather dated, but still better than a 4558 class part). There also is the NJM2745V, which looks closer to NJM4580 territory (so really pretty good output driving), but the datasheet can't seem to make its mind up about whether abs max supply is supposed to be +/- 9.5 or +/-15.5 V (first it says the former, but then there are lots of graphs for +/-15 V operation, and the online selection tool lists +/-15.5 V for the part).
One potentially annoying thing about SMD devices is power dissipation - even idle current is eating up a good part of the power budget. (10 mA on +/-15 V is 300 mW already.) I'd make sure there's a decent amount of copper on the power supply and output pins, should help that out somewhat.
How were you even supposed to use them, you ask? Inverting, that's how. Then suddenly their performance characteristics start making some kind of sense, including voltage noise. By the standards of the time (the mid-1970s), they're perfect for use in a mixer or something. I suppose the reasoning behind OPA(2)604 was similar, just for higher dynamic range (higher supplies, less noise).
Good output drivers often aren't available as quads. I believe it's a die size, power dissipation and manufacturing volume issue. For example, look at how the dual LM833 had a regular complementary output stage while its quad companion LM837 had to make do with a (seemingly slower) quasicomp.
NJR/JRC has one quad that might be of interest, the NJM2060V (quad version of the trusty NJM4560, so rather dated, but still better than a 4558 class part). There also is the NJM2745V, which looks closer to NJM4580 territory (so really pretty good output driving), but the datasheet can't seem to make its mind up about whether abs max supply is supposed to be +/- 9.5 or +/-15.5 V (first it says the former, but then there are lots of graphs for +/-15 V operation, and the online selection tool lists +/-15.5 V for the part).
One potentially annoying thing about SMD devices is power dissipation - even idle current is eating up a good part of the power budget. (10 mA on +/-15 V is 300 mW already.) I'd make sure there's a decent amount of copper on the power supply and output pins, should help that out somewhat.
The TL074's may be perfectly fine for the filters, for all I know, save the 600r expectation at the output. And maybe that's not even audible.
My point was more to the fact that we've got a lot of much better-performing (if nothing, lower noise!), modern chips, and for a diy'er, the incremental cost and time invested of a couple opamps is a pretty dang low percentage of the overall project, so worth its while.
Sgrossklass covers the advantages and challenges more clearly.
Edit -- ohhh, I get it. No need for battling semantics. I was agreeing with you on the utility of an JFET input opamp, but pointing to using an updated part.
Last edited:
Not exactly
The LM833 is available with a quasi comp NPN output stage.
http://www.diyaudio.com/forums/solid-state/274575-about-op-amps-use.html#post4361276
Drive ability of these two versions:
http://www.diyaudio.com/forums/solid-state/274575-about-op-amps-use.html#post4390796
More drive ability testing of opamps:
http://www.diyaudio.com/forums/chip...32-power-amp-thoughts-anyone.html#post4390787
With low supplies TL07x is not a good choice in non-inverting mode because common mode input range is very asymmetric (VEE+3V to VCC). MC33078/9 is better here.
For cable buffers I like BUF634/LME49600 Diamond Buffer without(!) feedback, best solution I've found so far. The slightly increased THD and offset is a non-issue, mostly.
For cable buffers I like BUF634/LME49600 Diamond Buffer without(!) feedback, best solution I've found so far. The slightly increased THD and offset is a non-issue, mostly.
You could put the quad opamp and all the resistors on one side of a two sided PCB and place all the capacitors on the other side. The frequency select capacitors could be inserted into DIP sockets to allow changes later.
You may find that with through hole caps and sockets that you can make this a small single sided PCB.
Interesting thought, was already thinking double sided board but hadn't considered simple freq changes. i will work on layout this weekend.
Greg
the frequency caps could be assembled onto a dip header. Then it becomes a plug in swap.
Dip headers are available in 8pin, 14pin, 16pin and there are narrow headers and wide headers if you need big, or small capacitors. There's also SIL plugs and sockets in 0.1" pin pitch. All of these would solder through to the smd side.
Dip headers are available in 8pin, 14pin, 16pin and there are narrow headers and wide headers if you need big, or small capacitors. There's also SIL plugs and sockets in 0.1" pin pitch. All of these would solder through to the smd side.
I'll throw the OPA1654 into the mix. Still a high impedance input type (CMOS rather than JFET though) and lower noise than the OPA4134 and OPA1644 (the broadband noise on the OPA1654 is around 3.8nV/rtHz) but with a much more robust output stage than the TL0 devices. It's also cheaper than the OPA4134 and OPA1644.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.