After the Isotope Phono, I made an Isotope preamp along the same lines. This is a very simple schematic. It uses a dual opamp per channel. I simulated it with a few low gain settings and as a buffer and it gives good results. It uses a 10k dual log pot between two stages. This way you dont have to use a high value pot and your sources are not driving the pot. It can handle higher values of pots also (needs changing only one resistor), but 10k being the least noisy will be the best. There is an RF filter at the input.
schematic -
The opamp loading will be about 2k, so that most of the dual opamps should work fine. Resistor values for loading around 600R also can be used, for the opamps that can support that. For buffer, use 100R (or even a wire) for feedback resistors and omit the other feedback resistor (to ground). Now a days, gain is hardly ever needed, so I guess a buffer-pot-buffer would make a lot of sense.
There is a gain switch on second stage. It can be used to switch on/off the gain on the second stage. It can be replaced with a wire to hard wire the gain stage.
There is an output impedance switch also. This is to use it as a headphone amp with a suitable opamp. e.g. A 100R resistor can be used for preamp use, and if the switch is turned on, a 3.3R resistor will come in parallel for use as a headphone amp. Or a 60R and 3.3R for use with 600R and 32R impedance cans. The second resistor and the switch can be omitted for standard preamp use.
The input impedance is 47k, but any reasonable value will be fine. Output impedance is 100R.
There are no input/output caps. If the dc offset is an issue, an off board cap can be used or use it at the amp's input.
There are two sets of electrolytics per rail per channel. In case one has a very good low value cap, a larger value one can support it. With a regulated supply, sometimes too much of output capacitance can become an issue. With battery use, it would be good to have some extra capacitance. I plan to use a dual shunt regulated supply with this, running at about 17v, powering both the isotope phono and the preamp in a single chassis.
PCB -
I think the layout looks decent so far. There is a grounded shield around the input stage in both the layers, the top layer pretty much acts like a metal plate over the input stage tracks. No tracks on the input stage top layer, except for the inputs pins, where I wanted to keep the parasitics low. Input ground currents dont flow through the shield and separate tracks are used for ground currents. PSU stage also looks ok and is far from the input and output stage.
I plan to make a smaller 1206 smd based design also with no switches and a very small footprint and a tight layout for tight spaces.
I am yet to order any pcbs, will be doing it soon, any suggestions and improvements are welcome.
Isotope Preamp SMD
This one is pretty much the same schematic as the other one. I took out all the switches etc. It uses all 1206 smd parts. Opamp is still through hole. I didnt want smd resistors on the top layer as these tend to become difficult to access, so left two through hole resistors (the usual 100 ohms ones). This is a very small board, 3.2"x1.3"
I also plan to use this pcb as a measurement preamp also, atleast to see and measure ripple of regulated supplies on a scope. I will use all four gain stages for this.
Schematic.
PCB -
This one I have split into two separate layers, other wise its difficult to make out things in a single image.
I am thinking of getting the smd boards done first.
schematic -
An externally hosted image should be here but it was not working when we last tested it.
The opamp loading will be about 2k, so that most of the dual opamps should work fine. Resistor values for loading around 600R also can be used, for the opamps that can support that. For buffer, use 100R (or even a wire) for feedback resistors and omit the other feedback resistor (to ground). Now a days, gain is hardly ever needed, so I guess a buffer-pot-buffer would make a lot of sense.
There is a gain switch on second stage. It can be used to switch on/off the gain on the second stage. It can be replaced with a wire to hard wire the gain stage.
There is an output impedance switch also. This is to use it as a headphone amp with a suitable opamp. e.g. A 100R resistor can be used for preamp use, and if the switch is turned on, a 3.3R resistor will come in parallel for use as a headphone amp. Or a 60R and 3.3R for use with 600R and 32R impedance cans. The second resistor and the switch can be omitted for standard preamp use.
The input impedance is 47k, but any reasonable value will be fine. Output impedance is 100R.
There are no input/output caps. If the dc offset is an issue, an off board cap can be used or use it at the amp's input.
There are two sets of electrolytics per rail per channel. In case one has a very good low value cap, a larger value one can support it. With a regulated supply, sometimes too much of output capacitance can become an issue. With battery use, it would be good to have some extra capacitance. I plan to use a dual shunt regulated supply with this, running at about 17v, powering both the isotope phono and the preamp in a single chassis.
PCB -
An externally hosted image should be here but it was not working when we last tested it.
I think the layout looks decent so far. There is a grounded shield around the input stage in both the layers, the top layer pretty much acts like a metal plate over the input stage tracks. No tracks on the input stage top layer, except for the inputs pins, where I wanted to keep the parasitics low. Input ground currents dont flow through the shield and separate tracks are used for ground currents. PSU stage also looks ok and is far from the input and output stage.
I plan to make a smaller 1206 smd based design also with no switches and a very small footprint and a tight layout for tight spaces.
I am yet to order any pcbs, will be doing it soon, any suggestions and improvements are welcome.
Isotope Preamp SMD
This one is pretty much the same schematic as the other one. I took out all the switches etc. It uses all 1206 smd parts. Opamp is still through hole. I didnt want smd resistors on the top layer as these tend to become difficult to access, so left two through hole resistors (the usual 100 ohms ones). This is a very small board, 3.2"x1.3"
I also plan to use this pcb as a measurement preamp also, atleast to see and measure ripple of regulated supplies on a scope. I will use all four gain stages for this.
Schematic.
An externally hosted image should be here but it was not working when we last tested it.
PCB -
This one I have split into two separate layers, other wise its difficult to make out things in a single image.
An externally hosted image should be here but it was not working when we last tested it.
I am thinking of getting the smd boards done first.
Last edited:
I added the PCB images for the SMD version in the first post.
Here is a simulation I ran with this circuit as a buffer-pot-buffer.
Source impedance - 500R, signal - 2.83vpk, 1khz, volume - 50%
Output - about 1.24V into 10K+1.5nf load.
Click on the image to see a larger version.
The second harmonic is -125db, third is -170db.
With full volume, it gives 2.7V into the same load. the second harmonic is -116db and the third is -158db
Here is a simulation I ran with this circuit as a buffer-pot-buffer.
Source impedance - 500R, signal - 2.83vpk, 1khz, volume - 50%
Output - about 1.24V into 10K+1.5nf load.
An externally hosted image should be here but it was not working when we last tested it.
Click on the image to see a larger version.
The second harmonic is -125db, third is -170db.
With full volume, it gives 2.7V into the same load. the second harmonic is -116db and the third is -158db
The PCBs arrived. battery is for size.
I made a build and tested it a bit. I have so far only tried NE5532. PSU is +-16v shunt reg. There is no gain, so its working as a buffer-pot-buffer. Pot is 10k panasonic. Electro Caps are 47uf/25v elna silmics. SMD caps are np0/c0g AVX.
SMD side -
I did take some measurements using my pc and the asus essence st soundcard. The 100hz ripple is just not there, shunt reg seems to be doing good. The 50hz hash and its harmonics are there, probably due to the proximity to the pc power wiring or maybe due to the ei trafo. 2nd harmonic is at -104db, 3rd at about -115db
I did some initial listening, sound is quite good. Will post more after I spend some more time listening to it. I want to try some other opamps too.
Soundcard baseline -
An externally hosted image should be here but it was not working when we last tested it.
I made a build and tested it a bit. I have so far only tried NE5532. PSU is +-16v shunt reg. There is no gain, so its working as a buffer-pot-buffer. Pot is 10k panasonic. Electro Caps are 47uf/25v elna silmics. SMD caps are np0/c0g AVX.
An externally hosted image should be here but it was not working when we last tested it.
SMD side -
An externally hosted image should be here but it was not working when we last tested it.
I did take some measurements using my pc and the asus essence st soundcard. The 100hz ripple is just not there, shunt reg seems to be doing good. The 50hz hash and its harmonics are there, probably due to the proximity to the pc power wiring or maybe due to the ei trafo. 2nd harmonic is at -104db, 3rd at about -115db
I did some initial listening, sound is quite good. Will post more after I spend some more time listening to it. I want to try some other opamps too.
An externally hosted image should be here but it was not working when we last tested it.
Soundcard baseline -
An externally hosted image should be here but it was not working when we last tested it.
You've linked the same distortion plot twice. Bit of ground loop crap in there as well.
A 5532 buffer should obviously perform quite well, though the TI part you're using is quite likely to show measurable common-mode distortion.
BTW, you may not find an all-unbalanced circuit like this very useful for measurement purposes - better go for an instrumentation amplifier kind of topology at the input then, using balanced cabling to attach to the DUT.
When testing opamps, do not overlook NJM2068 and NJM2114. The latter is kind of a dual 5534, the former has very low voltage noise for an inexpensive opamp geared towards mid-low impedances though it's not such a great output driver.
A 5532 buffer should obviously perform quite well, though the TI part you're using is quite likely to show measurable common-mode distortion.
BTW, you may not find an all-unbalanced circuit like this very useful for measurement purposes - better go for an instrumentation amplifier kind of topology at the input then, using balanced cabling to attach to the DUT.
When testing opamps, do not overlook NJM2068 and NJM2114. The latter is kind of a dual 5534, the former has very low voltage noise for an inexpensive opamp geared towards mid-low impedances though it's not such a great output driver.
sorry, my mistake, here is the essence loopback baseline.
I will start a new thread for the measurement preamp.
How do I get rid of all the 50hz hash and its harmonics. Is it just being picked up due to the proximity to the power wirings. The build is still temporary, chassis is open and wiring is still not great or neatly tucked in.
I will start a new thread for the measurement preamp.
How do I get rid of all the 50hz hash and its harmonics. Is it just being picked up due to the proximity to the power wirings. The build is still temporary, chassis is open and wiring is still not great or neatly tucked in.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Just think about how any of these loopback setups (hint!) is wired:
You have an audio cable going out to your DUT.
Then from the other end of the DUT you go back into the input.
If that's not a ground loop I don't know what is.
You can limit its effect by:
1. keeping loop area to a minimum
2. using cabling with low shield resistance
3. making sure the DUT's power transformer exhibits minimum coupling to mains
Getting rid of it entirely is not easy. You can try making a cable that interrupts the shield on the way back. The other option is using a balanced input, like in this preamp:
Soundcard Interface
It's not perfect - for one thing, I wouldn't make the protection resistors as high as 49k9, there's arguably better suited opamps than the OPA2604 for the buffer, and trying to get closer to a real balanced input would be another item on the to-do list - but you get the gist.
It gets really funny once you get into DUTs that are mains earth referenced. Guess why pro audio analyzers employ galvanically isolated inputs and outputs. Otherwise you may end up having to operate the DUT on an isolation transformer that does not connect PE through (I wouldn't do this with any SMPS-equipped device like a PC).
You have an audio cable going out to your DUT.
Then from the other end of the DUT you go back into the input.
If that's not a ground loop I don't know what is.
You can limit its effect by:
1. keeping loop area to a minimum
2. using cabling with low shield resistance
3. making sure the DUT's power transformer exhibits minimum coupling to mains
Getting rid of it entirely is not easy. You can try making a cable that interrupts the shield on the way back. The other option is using a balanced input, like in this preamp:
Soundcard Interface
It's not perfect - for one thing, I wouldn't make the protection resistors as high as 49k9, there's arguably better suited opamps than the OPA2604 for the buffer, and trying to get closer to a real balanced input would be another item on the to-do list - but you get the gist.
It gets really funny once you get into DUTs that are mains earth referenced. Guess why pro audio analyzers employ galvanically isolated inputs and outputs. Otherwise you may end up having to operate the DUT on an isolation transformer that does not connect PE through (I wouldn't do this with any SMPS-equipped device like a PC).
I tried reducing the loop area and did manage to take all power wires away from signal wiring and it helped. This is the new baseline for essence st loopback.
The last time I used the ne5532, it was the TI NE5532P. I tried two pieces and got the same results. This time I tried with the TI NE5532AP and KA5532 and i got much better results.
I have included the essence loopback as the overlay and is in red color. Opamps are in green color.
With NE5532AP:
With KA5532:
An externally hosted image should be here but it was not working when we last tested it.
The last time I used the ne5532, it was the TI NE5532P. I tried two pieces and got the same results. This time I tried with the TI NE5532AP and KA5532 and i got much better results.
I have included the essence loopback as the overlay and is in red color. Opamps are in green color.
With NE5532AP:
An externally hosted image should be here but it was not working when we last tested it.
With KA5532:
An externally hosted image should be here but it was not working when we last tested it.
Some from OPA series. Red is essence and green is opamp.
OPA2134 -
OPA1642 -
I also measured these opamps. You can click to see the images. They all look quite similar.
AD8599
OPA1688
OPA2227
LME49725
So far it looks like with most of the opamps, the preamp is hitting the limits of the Essence's measurement capability in terms of distortion numbers or noise floor.
Essence distortion data - 2khz - -116db, 3khz, -111db, 4khz - -124db, 5khz - -138db.
OPA2134 -
An externally hosted image should be here but it was not working when we last tested it.
OPA1642 -
An externally hosted image should be here but it was not working when we last tested it.
I also measured these opamps. You can click to see the images. They all look quite similar.
AD8599
OPA1688
OPA2227
LME49725
So far it looks like with most of the opamps, the preamp is hitting the limits of the Essence's measurement capability in terms of distortion numbers or noise floor.
Essence distortion data - 2khz - -116db, 3khz, -111db, 4khz - -124db, 5khz - -138db.
So it looks like only OPA2134 and AD8599 have somewhat higher 2nd-order distortion, the rest seems to be pretty much "wire with(out) gain". Granted, it's not overly surprising, they're followers and not heavily loaded here, and 1 kHz THD is hardly the most taxing test. What sort of signal level did you get, it says -2 dBFS, so about 1.6 Vrms or thereabouts? The volume pot was turned to max? Then impedance imbalance would have been little more than 50 ohms or basically nothing, I wonder what bothered the OPA2134 here.
Interesting to see that TI's NE5532AP did a lot better than their NE5532P, could you try one of those again just to rule out the differences in your measurement setup as a possible cause? If the difference remains, the cheaper Ps are probably best avoided where performance matters, and further investigation (Samuel Groner style) would be needed to find out what the problem is. I mean, TI's 5532s didn't fare too well for common-mode distortion and noninverting high-frequency linearity in his tests either, but at 1 kHz? He did get about 0.002% (-96 dB) distortion for a buffer at 10 kHz, but that was at +20 dBu (7.75 V), too - extrapolating that down, I would expect about 34 dB less or -130 dB, immeasurably low in your setup even with two of 'em cascaded.
I would try a few tests that are a bit more demanding. Maybe a 19+20 kHz IMD, or 10 kHz THD (the ST will do 192 kHz so bandwidth should be sufficient for that). More advanced distortion measurement techniques may be needed here, possibly a bridge based approach or a notch filter.
Interesting to see that TI's NE5532AP did a lot better than their NE5532P, could you try one of those again just to rule out the differences in your measurement setup as a possible cause? If the difference remains, the cheaper Ps are probably best avoided where performance matters, and further investigation (Samuel Groner style) would be needed to find out what the problem is. I mean, TI's 5532s didn't fare too well for common-mode distortion and noninverting high-frequency linearity in his tests either, but at 1 kHz? He did get about 0.002% (-96 dB) distortion for a buffer at 10 kHz, but that was at +20 dBu (7.75 V), too - extrapolating that down, I would expect about 34 dB less or -130 dB, immeasurably low in your setup even with two of 'em cascaded.
I would try a few tests that are a bit more demanding. Maybe a 19+20 kHz IMD, or 10 kHz THD (the ST will do 192 kHz so bandwidth should be sufficient for that). More advanced distortion measurement techniques may be needed here, possibly a bridge based approach or a notch filter.
You are right about the voltage levels. The volume pot was full. For load, maybe I should solder some load resistor across the preamp output for measurements.
When I tested the NE5532AP, I also retested the P version, one in both channels, got the same results. Will check again with some other P pieces.
I did the 19+20khz IMD and the 10khz distortion test.
Essence ST loopback 10khz test
NE5532AP 10khz test -
Essence loopback IMD test -
NE5532AP IMD Test -
When I tested the NE5532AP, I also retested the P version, one in both channels, got the same results. Will check again with some other P pieces.
I did the 19+20khz IMD and the 10khz distortion test.
Essence ST loopback 10khz test
An externally hosted image should be here but it was not working when we last tested it.
NE5532AP 10khz test -
An externally hosted image should be here but it was not working when we last tested it.
Essence loopback IMD test -
An externally hosted image should be here but it was not working when we last tested it.
NE5532AP IMD Test -
An externally hosted image should be here but it was not working when we last tested it.
- Status
- Not open for further replies.