I'm thinking of using an op-amp as a phase splitter in a tube amplifier. I read an interesting article that mentioned op-amp quality and suitability for audio use. The author stressed that op-amps continue to improve over time and last years go-to op-amp might be eclipsed in sound quality by a newer model. He mentioned upgrading op-amps in CD players and even suggested that it might be good to be able to switch between op-amps so as to change the character of the sound to suit a particular recording.
Well, one can certainly her the difference between a 15 cent cap and a 15 dollar cap, so I'll bite.
I'm wondering if those who have built with op-amps consider part selection as terribly important. The only ones I have in my junk drawer are lm4562na. What are some good part numbers to try?
Well, one can certainly her the difference between a 15 cent cap and a 15 dollar cap, so I'll bite.
I'm wondering if those who have built with op-amps consider part selection as terribly important. The only ones I have in my junk drawer are lm4562na. What are some good part numbers to try?
Let an SSM2142 (or one of it's derivatives) do the splitting and driving for you. At +-18v rails there'll be plenty of swing to drive most output valves/tubes.
If you doubt it's fidelity, probably 50% of what you listen to has passed through at least 2 of these.
Of course there are better opamps by specification, but few have this heritage and potential simplicity.
If you doubt it's fidelity, probably 50% of what you listen to has passed through at least 2 of these.
Of course there are better opamps by specification, but few have this heritage and potential simplicity.
What is the source impedance of the input? Current drive available? Voltage swing at input? What are you driving out? How much current do you need? How much voltage swing out do you need? What will be the gain?
LM4562n from NatSemi or TI is a very powerful quiet fast IC. It requires close power supply decoupling with ceramic caps, and usually ceramic caps parallel to the feedback resistor to kill the gain at high frequency. As a $2 part, its advantage over $.38 part LM33078 is the increased current drive available. Both these are low noise at 50x gain where I'm using them. Both have high slew rate which allows low distortion on high frequency portions of the audio band. Does your amplifier include 20 khz, or is it a guitar/bass amp limited to 5 khz?
Another quiet op amp is NE5532, which is ancient and very cheap. It also has +-18 v rails from some manufacturers. Its slew rate is not so high, and doesn't have as much tendency to oscillate.
Other op amps have advantages of FET input with different sources.
The ones mentioned are dual op amps, and require passivation of the unused one if you are only using one part of it.
DC stability could be a factor. Audio op amps as above are usually coupled with capacitors. Phase splitters are usaully DC coupled, and wandering of the stable point could cause problems. Alternate versions are available with DC op point adjustment pins. A 553x variant comes to mind.
LM4562n from NatSemi or TI is a very powerful quiet fast IC. It requires close power supply decoupling with ceramic caps, and usually ceramic caps parallel to the feedback resistor to kill the gain at high frequency. As a $2 part, its advantage over $.38 part LM33078 is the increased current drive available. Both these are low noise at 50x gain where I'm using them. Both have high slew rate which allows low distortion on high frequency portions of the audio band. Does your amplifier include 20 khz, or is it a guitar/bass amp limited to 5 khz?
Another quiet op amp is NE5532, which is ancient and very cheap. It also has +-18 v rails from some manufacturers. Its slew rate is not so high, and doesn't have as much tendency to oscillate.
Other op amps have advantages of FET input with different sources.
The ones mentioned are dual op amps, and require passivation of the unused one if you are only using one part of it.
DC stability could be a factor. Audio op amps as above are usually coupled with capacitors. Phase splitters are usaully DC coupled, and wandering of the stable point could cause problems. Alternate versions are available with DC op point adjustment pins. A 553x variant comes to mind.
Last edited:
Pete Millet has some boards for an amp he calls the uniamp. I bought them at a yard sale believe it or not! Anyway, I’ve never done anything with them because they require sowter 3575 transformers to do the phase splitting before the first stage of amplification. They are not cheap. I’ll post the schematic below.
I built a couple of OpAmp circuits just to learn more about them and realized they could substitute for the sowters. I was thinking of either using it as strictly a phase inverter with unity gain and use the PCB as is. Or I could use it with a gain of 2 or 3 thus eliminating the need for the first stage of tube amplification. Like you say, the CD player is full of em and I’d still have the 6SN7 there for some tube mojo.
So to answer your questions: max swing of 3v, very high output impedance of a grid resistor, almost no current requiremed, digital source. I might be able to power the OpAmp from zener regulated cathode current if I’m real cleaver and use a 12v OpAmp.
Schematic
I built a couple of OpAmp circuits just to learn more about them and realized they could substitute for the sowters. I was thinking of either using it as strictly a phase inverter with unity gain and use the PCB as is. Or I could use it with a gain of 2 or 3 thus eliminating the need for the first stage of tube amplification. Like you say, the CD player is full of em and I’d still have the 6SN7 there for some tube mojo.
So to answer your questions: max swing of 3v, very high output impedance of a grid resistor, almost no current requiremed, digital source. I might be able to power the OpAmp from zener regulated cathode current if I’m real cleaver and use a 12v OpAmp.
Schematic
Last edited:
Okay, the load is 47k plus a few picofarad for the grids. The voltage swing doesn't have to go over 3. The output is DC coupled to the grids, so stable DC operating point matter. That points to the 5534, which has potentiometer inputs to control the DC operating point. Buy them quick, the DIP version P suffix from TI is going obsolete shortly. Flat pack versions, you have to hold them still while trying to solder a couple of pins. I'm not up to that, takes 4 hands IMHO. DIP, you can buy a board that spreads the pins out to where you can install point to point wires. Board is 21-4575 (small) or 21-4580 (medium) from Newark. $1-$2 each.
You need one op amp inverting, one straight through for the other 6sn7. Or follow one inverting op amp with another for the other tube.
You can run op amps less than the maximum rated supply voltage. I'm running 33078 at +-7.5 v in my disco mixer that is the hub of my stereo. Take the filament voltage rectify to DC, dual diodes using the center tap gets you +- 7.8 using silicon diodes and +-8.1 using schottky diodes. The op amp will swing 47 kohms within a couple of volts of the two power supplies, or +-5.8 or 6.1. You can salvage schottky diodes from PCAT switcher supplies or any other modern switcher supply as comes in dead flat panel TV's for example. They are the ones that have .3 v drop on the 2000 ohms scale. Or MBR1100 have long leads and 100 v piv.
Gain of the op amp is Rf/Ri where Rf is the feedback resistor and Ri is the input resistor. Look at any mixer schematic to figure out where to put the input resistor. You can use the capacitor on the input to keep sources producing DC voltage (like cell phones) from throwing the operating point off. 4.7 uf is usual, although 2.2 uf mostly works. You can make the gain adjustable with a pot parallel to the feedback resistor, or can cut the input down with a pot, to make the output swing no more than +-3 v. Paragraph 9.1 of the TI datasheet (I got from datasheetcatalog.com) shows where to put the DC operating point pot, and the necessary compensation capacitor. Don't put the feedback resistor pot on the front panel, long leads into this high gain point will inject noise. With 400 v swinging around in a tube amp, you may need to put the op amp in its own grounded steel box to prevent oscillation from feedback from the output tubes.
You need one op amp inverting, one straight through for the other 6sn7. Or follow one inverting op amp with another for the other tube.
You can run op amps less than the maximum rated supply voltage. I'm running 33078 at +-7.5 v in my disco mixer that is the hub of my stereo. Take the filament voltage rectify to DC, dual diodes using the center tap gets you +- 7.8 using silicon diodes and +-8.1 using schottky diodes. The op amp will swing 47 kohms within a couple of volts of the two power supplies, or +-5.8 or 6.1. You can salvage schottky diodes from PCAT switcher supplies or any other modern switcher supply as comes in dead flat panel TV's for example. They are the ones that have .3 v drop on the 2000 ohms scale. Or MBR1100 have long leads and 100 v piv.
Gain of the op amp is Rf/Ri where Rf is the feedback resistor and Ri is the input resistor. Look at any mixer schematic to figure out where to put the input resistor. You can use the capacitor on the input to keep sources producing DC voltage (like cell phones) from throwing the operating point off. 4.7 uf is usual, although 2.2 uf mostly works. You can make the gain adjustable with a pot parallel to the feedback resistor, or can cut the input down with a pot, to make the output swing no more than +-3 v. Paragraph 9.1 of the TI datasheet (I got from datasheetcatalog.com) shows where to put the DC operating point pot, and the necessary compensation capacitor. Don't put the feedback resistor pot on the front panel, long leads into this high gain point will inject noise. With 400 v swinging around in a tube amp, you may need to put the op amp in its own grounded steel box to prevent oscillation from feedback from the output tubes.
Last edited:
Thanks Indianapolis. That’s a lot of helpful advise. I had wondered about a noise shielding enclosure. I note that the transfos used in that application always have a mu metal enclosure. I suspect it would be necessary. Perhaps one should install the phase splitter in the CD player, haha.
I like your idea of running them off filament voltage. Are they sensitive to ripple?
I like your idea of running them off filament voltage. Are they sensitive to ripple?
In my disco mixer with the 33078, I had some residual hum until I improved the power supply to a pi filter. I used an external 18 vdc wall transformer, with inside the mixer, a hash filter (inductive choke), a 1000 uf cap, two stacked zener diode regulators (22 ohms, 1n5534 (7.5 v 5 w zener), 1n5534 22 ohms stacked) followed by two stacked 220 uf capacitors. the center of the two zeners was the analog ground. This hummed a little. when I preceded the initial cap by another 1000 uf capacitor followed by a 2 ohm resistor, the hum went away. CRC, that's a pi filter. That op amp circuit had a 50x gain RIAA circuit in it. At gain of 2 to 4 you'll be using, hum may be less of a problem. Since you have a center tap in the filiament voltage, you would start with two stacked 220 to 1000 uf caps at the input. btw the rf hash filter was a toroid with 11 turns on it - salvaged from a PCAT power supply.
I don't know how mu metal could be bought. This op amp circuit, the little DIP board would fit in a Altoids candy steel box, although the 1000 uf filter caps wouldn't. Maybe put the power supply and op amps in a 2" x 2" x 4" steel box? Hard to find, although index card steel boxes are still available at the charity resale shop sometimes. I have bent up steel boxes out of steel mesh from Home depot, which bends okay by hand over the back steps. I cut the mesh with Wiss metal shears, then tie the seams together with twists of steel wire used for hanging pictures or tieing concrete reinforcement together. The insulating boards can be held off the steel enclosure with standoffs & #6 x 1" screws.
the steel box inside the outer enclosure is a common feature of high grade electronics. My dynaco PAS2 preamp has a steel bulkhead inside the main box for the wiring to the high gain RIAA circuit. And my PV-1.3k amp, the power transformer and 120 vac entry are behind steel bulkheads, away from the gain components in the regular enclosure. Both power transformers have copper sheet eddy current shorters around them already.
I don't know how mu metal could be bought. This op amp circuit, the little DIP board would fit in a Altoids candy steel box, although the 1000 uf filter caps wouldn't. Maybe put the power supply and op amps in a 2" x 2" x 4" steel box? Hard to find, although index card steel boxes are still available at the charity resale shop sometimes. I have bent up steel boxes out of steel mesh from Home depot, which bends okay by hand over the back steps. I cut the mesh with Wiss metal shears, then tie the seams together with twists of steel wire used for hanging pictures or tieing concrete reinforcement together. The insulating boards can be held off the steel enclosure with standoffs & #6 x 1" screws.
the steel box inside the outer enclosure is a common feature of high grade electronics. My dynaco PAS2 preamp has a steel bulkhead inside the main box for the wiring to the high gain RIAA circuit. And my PV-1.3k amp, the power transformer and 120 vac entry are behind steel bulkheads, away from the gain components in the regular enclosure. Both power transformers have copper sheet eddy current shorters around them already.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.