Is the EXAR XR-084 quad bi-fet opamp any good?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
A '084 is a lower-spec version of a '074, with the difference remaining somewhat mysterious.

Generally the choice among good quad opamps is quite limited. In terms of FET input, there's the OPA4132 (and its lesser spec cousin, OPA4134) plus the TLE2074. As far as bipolars go, there's LME49740, LT1359 and if you require significant output currents, MC33179.

I'd want to study the circuit a little further before determining what to use as a replacement... impedances at the opamp input pins, available decoupling, stuff like that. (Service docs would help.) Tossing an opamp into the circuit at random may result in instability or at least subpar noise and distortion performance. (DIYers like to do just that, and surprise!, they get random results.)

I'd expect a bal/SE converter among other things, an application in which even a '074 tends to be OK as it mainly demands good transfer linearity, though its voltage noise would still dominate the resistors'. The '2074 would make a good inexpensive upgrade then, a '4132 a better and more expensive one. Impedances tend to be highish, which means the '49740 with its fairly high current noise may actually turn out to be noisier than its FET input colleagues. Now the question is what they are doing with the second pair of opamps.

In any case one of the FET input models shouldn't be any worse than what's in there now at least. Note that they are about a factor of 3 faster though, so requirements in decoupling are likely to be stricter than for the old '084 and one should inspect the PCB layout in this regard.
 
I may just leave the stock opamps if they're generally regarded as OK; I'm actually impressed they used bi-fets considering the age of these amps. One early reviewer complained that the input section was noisier than the rest; I have no clue as I've never injected signal after. I always use the balanced inputs.

Thanks for the input. I have 9 of these monster amps now, and I like them much better than their old class-H amps. Early SMPS. I got my money's worth of 216 FET power output transistors in TO-3 packages alone!

I hosted the schematics:
http://mysite.verizon.net/gregorypinkowski/PR1800-schematic.pdf
The big power supply electrolytic caps seem OK on them all (the previous owner replaced them in one of the amps). I should probably replace the non-polar electrolytic coupling cap in each channel with something better (but cheap if I need 18 of them). Any other good ideas? Is it practical to consider changing the bias (and turning up the fans)? Is the output stage class B or class A/B now? What do you guys think of the amplifier sections (not power supply) in general? It sounds good to me, and has for 25 years; but there's always room for improvement. The seller confirmed very very low quantities of DC at the speaker terminals. This is good to know. But is it more a test of the power supply or the amp?
 
Last edited:
Don't even think of changing the bias. The Exar IC's should be fine also.

The only changes would be to replace the coupling and feedback DC blocking caps. You might try putting film caps across the main filters but you will need at least 50 uf to get a significant change.
 
I was going to say that it was a "professional" PA amp looking for high output and efficiency, but the PM1600 is the same exact schematic for the output section, it just deletes the balanced inputs, and was their top-of-the-line home hi-fi amp.

Of course if I increase the bias and try to put it farther into class-a I've got to upgrade the fans also, but they've never gone on 'high' in normal use.
 
Last edited:
Yes, 0.002 amps (2mA) is very low but that is not the main issue.

The power mosfets need about 1.5V to switch on. So, +1.5V for the top mosfets and -1.5V for the bottom mosfets, that makes 3V across the 180 ohms. At 0.4V, the mosfets are off when the amp is idling, hence in Class B.

The zeners and diodes CR12/13/14/15 are not for biasing. They are there to prevent the mosfet gates from exceeding the maximum limit, I believe +/- 20V. They clamp the gate voltages at +/- 12V.
 
Micheal Chua, thanks for your help. I wonder what the range of that adustment pot is (in bias volts)? One way to find out...

I have 9 of these amps now, 18 channels...and despite inefficient speakers (sealed isobaric and planar dynamic) I often run them at low average power but I use lots of EQ, dynamic range expanders, and devices that make "fake" sub-bass, all of which make large occasional demands. The sound of that "first watt" is certainly important! But then I need a lot of overhead! Sometimes I also use them for live-performance music, which also has extreme dynamic range (at least the way I do small PA).

So it's actually got both sides practically off completely for some small values very near the zero-crossing? I guess that's just another way of saying "class B"...I thought in all real implementations both sides would be on a little bit.

OK the current across the resistor isn't important (DOH!) If I increased bias to 3v across the resistor, approx. how many watts do the outputs and related resistors have to dissipate due to bucking each other? Then does the linearity suffer at some higher signal level where the push and pull stop bucking each other and it goes back to being class-B (more efficient and also running in a better-performing part of its curve)? Or does the feedback loop make up for those ills?

I also notice that the paralleled MOSFETs are connected directly to the rails and to the "S" output without any 'current balancing' resistors, so I assume the FETs must be matched reasonably well?

I'm still an idiot looking at transistor schematics, instead of the few tube circuits I understand.

Some day maybe I should build a small hut outside the house, and install the amps in a rack with a large furnace-type air handler to depressurize one side of the entire rack, and a big air filter on the air input side. And just run the balanced signal lines and speaker wires thru the wall. I used to have them in the garage in the old house, with dedicated power lines and remote relays to turn them on and off.
 
OK the current across the resistor isn't important (DOH!)

No, it is important. Just that the voltage across the 180 ohm takes precedence. If the gate voltage is not high enough to switch on the mosfets, the current of the drivers Q10, Q11 doesn't even come into the picture.

I also notice that the paralleled MOSFETs are connected directly to the rails and to the "S" output without any 'current balancing' resistors, so I assume the FETs must be matched reasonably well?

Yes, if no Source Resistors are used, the mosfets must be Matched.

I wonder what the range of that adustment pot is (in bias volts)? One way to find out...

Before you start adjusting the bias, I suggest you measure the voltage across R14 (0.27 ohms/5W) first. The voltage drop, if any, across R14 will tell you whether the mosfets are biased up and if so, by how much.
 
Measure voltage across R14 with no signal?

Interesting that that point CSA goes back to the power supply...near the "V0 trim" and "VTH Trim" pots. Is this protection circuitry or remote-sensing for the regulated supply?

Looking at the device spec sheets, what's the critical "turn-on voltage" spec called? Is that "Gate-Source Cutoff Voltage"
min -0.15 max -1.45 V for 2SJ56
min 0.15 max 1.45 V for 2SK176
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.