The TL072 H does show a faster slew rate than before (now 20, was 13) but the same poor voltage noise and the same inability to effectively drive anything below 10KΩ. OPAs 1642 and 1656 MUCH MUCH better. I like the OPA1642 due to its MUCH reduced bandwidth compared to the 1656 (11MHz vs, 55 MHz)---thus much less chance of oscillation in vintage circuits.
OPA1642 is the allrounder that always goes well except as first opamp in DAC output stages.
It would be surprising if the audiophile discrete opamps come close to industry produced stuff.
It would be surprising if the audiophile discrete opamps come close to industry produced stuff.
Last edited:
I'm no expert, but some of those discrete ones sound superior in some cases. Why I have no idea. Specifically, in the I/V stage for AD1862, PCM63, and line preamp, I got the best sound so far with the Burson V6 Classic. I've tried others (Sparkos SS3601/3602, Burson V5, V7 Vivid), but this one is somehow subjectively superior. They are mostly not for low noise applications.
And when I was making a Whammy headamp, I got by far the best sound with the Muses 02. In that case, for example, the Sparkos SS3602 sounded very sharp, the OPA1612 oscillated like crazy, and the OPA2604 and TL072 were very dark sounding. I didn't investigate further, and I didn't have the OPA1656 nor the 1642 to try. I had other opamps, but only in the single version.
For OPA rolling, no one should attempt it without a function generator and an oscilloscope, whatever the device is. Overall, the TL072 is the champion here, it hasn't oscillated anywhere or asked for any compensation so far. But if you try something else instead, you may find yourself in trouble.
And when I was making a Whammy headamp, I got by far the best sound with the Muses 02. In that case, for example, the Sparkos SS3602 sounded very sharp, the OPA1612 oscillated like crazy, and the OPA2604 and TL072 were very dark sounding. I didn't investigate further, and I didn't have the OPA1656 nor the 1642 to try. I had other opamps, but only in the single version.
For OPA rolling, no one should attempt it without a function generator and an oscilloscope, whatever the device is. Overall, the TL072 is the champion here, it hasn't oscillated anywhere or asked for any compensation so far. But if you try something else instead, you may find yourself in trouble.
Last edited:
Not saying it is the case here but in the past decades I found stuff that distorted more was preferred by many. So many say they want neutral but prefer the non neutral in listening sessions. The best measuring devices were certainly not always the most preferred.
OPA1642 is the better performing TL072 replacement with the ease of use of the latter. Not sure about EMI rejection.
OPA1642 is the better performing TL072 replacement with the ease of use of the latter. Not sure about EMI rejection.
Last edited:
Nixie62: "....the OPA1612 oscillated like crazy...."
Did you use the recommendations for opamp upgrades?
Did you use the recommendations for opamp upgrades?
- Put a 100nF X7R cap across the power pins 4 & 8
- Put 22 pf COG caps across the feedback resistors (pins 1&2, 6&7)
EMI also very good on the 1642. It’s my go to opamp for line level work. The 1656 is also good, but the 1/f noise is high compared to the 1642. The 1642 uses a JFET input stage while the 1656 IIUC uses mosfets.
The 1642 is pretty much ideal for example as a pot buffer. You can even parallel 4 sections of a 1644 and get a 2.5nV/rtHz JFET buffer with lots of drive (mixing outputs with 10R breakouts) and still very low and linear input capacitance.
The only thing to remember is the rather large headroom penalty on positive input CMRR, Vs+ - 3.5V, which calls for at least +-7V supplies for 2Vrms signals.
Yes, the CMRR headroom can be an issue, but for line level I always use +-15 or +-18 volt rails depending upon the opamp I am using.
I tried all of that and something else I could think of. It's not that simple.
LME49720 & OPA2604 do not oscillate without C2&C7, no compensation is required. When even 22pF is added for compensation, the LME49720 oscillates.
The OPA1612 does not oscillate only if a Zobel network 10ohm+47nF is added to the output. Compensation even with 22pF causes oscillations with or without a Zobel network, so this OPA should be used without compensation.
The OPA1656 is probably another candidate for oscillation in this assembly (but not tested because I ran out of them).
The TL072 and OPA2604 do not exhibit any instability or need for compensation.
Muses 02 needs 68pF compensation.
Last edited:
Have you got a circuit and a picture of the layout?
A phase lead cap across the feedback resistor may help claw back a bit of phase margin or shape the overall response but it can make some situations worse.
In high GBP opamps, make sure the capacitance from the feedback node ie inverting input to ground is low. No ground plane around the feedback node and short tracks to the junction of the feedback resistors. For this reason, on modern opamps, SMD resistors are preferable. See the data sheets for layout examples.
A phase lead cap across the feedback resistor may help claw back a bit of phase margin or shape the overall response but it can make some situations worse.
In high GBP opamps, make sure the capacitance from the feedback node ie inverting input to ground is low. No ground plane around the feedback node and short tracks to the junction of the feedback resistors. For this reason, on modern opamps, SMD resistors are preferable. See the data sheets for layout examples.
Last edited:
You mean Whammy headamp? There is a whole topic.
https://www.diyaudio.com/community/threads/whammy-pass-diy-headphone-amp-guide.317803/page-258
It plays very well, just needs a bit of work to achieve stable operation for various opamps. And for low-ohm headphones, I think the voltage gain should be reduced so that more of the potentiometer's travel is available.
https://www.diyaudio.com/community/threads/whammy-pass-diy-headphone-amp-guide.317803/page-258
It plays very well, just needs a bit of work to achieve stable operation for various opamps. And for low-ohm headphones, I think the voltage gain should be reduced so that more of the potentiometer's travel is available.
Last edited:
Seems a peculiarity of that particular device. Just had a quick look and it has no input filtering and no protection relay either. Decoupling caps for the opamp are 220 uF electrolytic caps also too far away from the supply pins which may a explain a lot. Many opamps* will not accept that but it is a design imperfection of that device not the opamp. Such design details normally have tech oriented persons skipping connaisseur type of devices.
OPA1612 is normally not a problem opamp. When applied right many opamps aren’t. Except AD797 🙂
*Possibly a 10 nF film cap in 5 mm pitch from the + supply pin to the - supply pin helps.
OPA1612 is normally not a problem opamp. When applied right many opamps aren’t. Except AD797 🙂
*Possibly a 10 nF film cap in 5 mm pitch from the + supply pin to the - supply pin helps.
Last edited:
It wasn't problematic in the line preamps I was making, here it is. 10pF or 22pF compensation gets the job done.
All this can be solved, but whoever does it must have an oscilloscope and a signal generator. And I notice that many people just change opamps without measuring. That's not smart. There are a lot of those cheap Chinese oscilloscopes. 100MHz and 2 channels, enough for audio. For me personally, the Hantek 6254BD completes the job for amateur needs.
All this can be solved, but whoever does it must have an oscilloscope and a signal generator. And I notice that many people just change opamps without measuring. That's not smart. There are a lot of those cheap Chinese oscilloscopes. 100MHz and 2 channels, enough for audio. For me personally, the Hantek 6254BD completes the job for amateur needs.
Last edited:
To fix the oscillation
1. Place a 47 to 100 pF cap directly from the opamp output to its inverting input. You should remove the cap that is sitting across the feedback resistor.
2. I would add the Zobel as suggested
3. To isolate the amp output from the headphone cable capacitance, place a small inductor of about 2uH with a 2.2 ohm resistor in parallel after the Zobel network so it sits between the amplifier output and your headphone socket. There are some nice 3A SMD inductors at t for this. Using this approach preserves the low output impedance of the amp at audio frequencies which is especially important with large diaphragm headphones (you will lose bass if Zout is high with these types of headphones).
#1 is the most important.
The OPA1612 and the LM devices are both very high GBP and will easily oscillate if not compensated correctly in a system like this.
1. Place a 47 to 100 pF cap directly from the opamp output to its inverting input. You should remove the cap that is sitting across the feedback resistor.
2. I would add the Zobel as suggested
3. To isolate the amp output from the headphone cable capacitance, place a small inductor of about 2uH with a 2.2 ohm resistor in parallel after the Zobel network so it sits between the amplifier output and your headphone socket. There are some nice 3A SMD inductors at t for this. Using this approach preserves the low output impedance of the amp at audio frequencies which is especially important with large diaphragm headphones (you will lose bass if Zout is high with these types of headphones).
#1 is the most important.
The OPA1612 and the LM devices are both very high GBP and will easily oscillate if not compensated correctly in a system like this.
Last edited:
The issue you have won’t arise in a line amp if you are driving the output directly from the opamp via a 47 to 100 ohm resistor. This issue is arising because there is additional phase shift arising in the discrete output stage which means you have to compensate the opamp stage to get the overall system stable.
Normally one can replace opamps without too much trouble. One must use a scope to verify. And let’s be honest: most of the times things work out OK when PCB design and decoupling are done right. Hence the opamp rolling fashion by laymen. Of course oscillation sometimes occurs but without an oscilloscope no one knows 😀
It is this device that is intolerant to various opamps by design. The recommendations of Bonsai are normal design practice and should have been included in the design. The 47 pF and missing Zobel for sure.
It is this device that is intolerant to various opamps by design. The recommendations of Bonsai are normal design practice and should have been included in the design. The 47 pF and missing Zobel for sure.
Last edited:
I think if you at least implement #1 from post 36 above, you should brcablectobswap opamps without problems. I would add 1uF 0805 35V X7R caps from wash rail to 0V directly under each opamp as well.