Hello everyone,
I decided to build my first RIAA phono stage, using OPA1656 and the schematic shown in the datasheet.
Where possible, I tried to use 0805 smd components for resistors and capacitors (exclusively Thin Film and C0G/NPO), except for the two 100nF that decouple the power lines. These are X7R. About the four 100uF capacitors, I chose Nichicon UES Bipolar.
The only change to the circuit concerns the removal of the 150 pF capacitor (for the capacitive load), replaced by three capacitors (22pF, 47pF, and 100pF) selectable or not, through a 3x SPST Dip Switch, for 8 possible combinations of capacity (0, 22, 47, 69, 100, 122, 147, 169 pF).
Considering that an average interconnect cable has a capacity of around 100pF, being able to add values between 0 and 169 pF, allows you to optimize the right input capacity in the most critical range of many cartridges (100 - 250 pF), especially those that prefer low input capacities.
I attach the schematic and some images of the PCB layout, in its current state. I have already made all the connections, but I have not yet added the ground plane on the bottom, so as not to confuse the traced tracks. I used 20 mills traces for all the signal lines, and 30 and 35 mills, only for the power lines.
But before completing the PCB I would like to hear some opinions on the layout and ask some questions:
1 - Is this preliminary rooting good enough, or are there obvious errors? Can it be improved with small changes?
2 - For simplicity I used classic 2.54 mm connectors for inputs and outputs. The input RCA connectors will be the standard through-hole type, fixed to the rear panel.
If placed at a distance of about 35 mm from each other, they will be located almost exactly above the two fixing holes of the PCB, on the input side, with the advantage of being able to use really short input cables (15 - 20 mm or even less). Slightly longer cables for the outputs I think are less of an issue.
From a sonic point of view, would it be better to use different connectors?
3 - About the Output capacitor, the last two lines of page 21 of the datasheet, state:
"... C5 is chosen to be the same value as C4; for simplicity however, the value of C5 must be large enough to avoid attenuating low-frequency information."
In my schematic, C5 corresponds to C13, C14 (100uF).
How do I check or calculate if the value is large enough, to avoid attenuation at low frequencies?
4 - Any other suggestions or criticisms, will be useful. 🙂
Regards,
Marco
I attach the schematic and images of the pcb in its current state.
Schematic From OPA1656 datasheet:
Actual layout without ground plane:
Board preview:
I decided to build my first RIAA phono stage, using OPA1656 and the schematic shown in the datasheet.
Where possible, I tried to use 0805 smd components for resistors and capacitors (exclusively Thin Film and C0G/NPO), except for the two 100nF that decouple the power lines. These are X7R. About the four 100uF capacitors, I chose Nichicon UES Bipolar.
The only change to the circuit concerns the removal of the 150 pF capacitor (for the capacitive load), replaced by three capacitors (22pF, 47pF, and 100pF) selectable or not, through a 3x SPST Dip Switch, for 8 possible combinations of capacity (0, 22, 47, 69, 100, 122, 147, 169 pF).
Considering that an average interconnect cable has a capacity of around 100pF, being able to add values between 0 and 169 pF, allows you to optimize the right input capacity in the most critical range of many cartridges (100 - 250 pF), especially those that prefer low input capacities.
I attach the schematic and some images of the PCB layout, in its current state. I have already made all the connections, but I have not yet added the ground plane on the bottom, so as not to confuse the traced tracks. I used 20 mills traces for all the signal lines, and 30 and 35 mills, only for the power lines.
But before completing the PCB I would like to hear some opinions on the layout and ask some questions:
1 - Is this preliminary rooting good enough, or are there obvious errors? Can it be improved with small changes?
2 - For simplicity I used classic 2.54 mm connectors for inputs and outputs. The input RCA connectors will be the standard through-hole type, fixed to the rear panel.
If placed at a distance of about 35 mm from each other, they will be located almost exactly above the two fixing holes of the PCB, on the input side, with the advantage of being able to use really short input cables (15 - 20 mm or even less). Slightly longer cables for the outputs I think are less of an issue.
From a sonic point of view, would it be better to use different connectors?
3 - About the Output capacitor, the last two lines of page 21 of the datasheet, state:
"... C5 is chosen to be the same value as C4; for simplicity however, the value of C5 must be large enough to avoid attenuating low-frequency information."
In my schematic, C5 corresponds to C13, C14 (100uF).
How do I check or calculate if the value is large enough, to avoid attenuation at low frequencies?
4 - Any other suggestions or criticisms, will be useful. 🙂
Regards,
Marco
I attach the schematic and images of the pcb in its current state.
Schematic From OPA1656 datasheet:
Actual layout without ground plane:
Board preview:
Last edited:
Hi!
Regarding capacitor size, the frequency response can be calculated by the following formula (it will be 6dB/Oct, so 1st order):
Fc=1/(6.28*R*C).
You have 2 high-pass points in your circuit: R4/C4 and C5/R5/R6/(impedance of next stage), but in this case, the impedance of the next stage will be much more relevant.
Let's check:
R4=127ohms, C4=100uF ->>> Fc = 1/(6.28*127*0.0001) = 12.5Hz
For C5, if you next stage has 10kohm of impedance, the frequency will super low:
R5=10kohms, C5=100uF ->>> Fc = 1/(6.28*10000*0.0001) = 0.15Hz
I would leave this 100uF since we don't know which is the next stage impedance. and it doesn't matter much if the capacitor is large as 100uF.
It's better to have lower frequency cut to not influence the overall frequency response.
Last edited:
If your turntable happens to have a muting switch that shorts the turntable outputs, you may want to read https://www.diyaudio.com/community/threads/opa1656-phono-preamp-split-from-opa1656-thread.377331/ , particularly posts #1, #16, #140 and #141. (See also https://www.diyaudio.com/community/...lifier-stability-and-source-impedance.420032/ if you are interested in the theoretical background.)
Last edited:
Hi Ron68,
Thank you very much, it is exactly the calculation I wanted to verify. 😎🤚
The 100uF tht capacitors are the only components that will be installed after the pcb printing, therefore the only ones that can be easily changed in value, if necessary.
I wanted to understand their role in the calculation of the cutting slopes.
..and Hello MarcelvdG,
Thank you for reporting the issue to me.
I tried to read the posts you reported, but to understand them I need to have a clear understanding of the whole context, also considering the many purely technical comments, well above my skills.
My circuit is nominal, it uses almost exclusively SMD components and at the moment it has a fairly compact layout, even if it can be further reduced.
Maybe it's a good thing that it is completed without particular variations, regarding the values of the components.
It could be useful as a reference, to understand if placed in the same conditions, it behaves like the JRA circuit.
I'm just making assumptions.
But if you think I misunderstood something, or missed the point, please explain further. 😉
Thank you very much, it is exactly the calculation I wanted to verify. 😎🤚
The 100uF tht capacitors are the only components that will be installed after the pcb printing, therefore the only ones that can be easily changed in value, if necessary.
I wanted to understand their role in the calculation of the cutting slopes.
..and Hello MarcelvdG,
Thank you for reporting the issue to me.
I tried to read the posts you reported, but to understand them I need to have a clear understanding of the whole context, also considering the many purely technical comments, well above my skills.
My circuit is nominal, it uses almost exclusively SMD components and at the moment it has a fairly compact layout, even if it can be further reduced.
Maybe it's a good thing that it is completed without particular variations, regarding the values of the components.
It could be useful as a reference, to understand if placed in the same conditions, it behaves like the JRA circuit.
I'm just making assumptions.
But if you think I misunderstood something, or missed the point, please explain further. 😉
Yes, it would be interesting to see if your circuit also produces a big bang when you connect a low-capacitance, high-quality shielded cable of about 1.2 m long terminated with 1 kohm in parallel with a switch to its input and operate the switch (as in JRA's post #73, https://www.diyaudio.com/community/...split-from-opa1656-thread.377331/post-6788889 ).
If my hypothesis is correct, it should produce such bangs, at least when all input capacitors are switched off. If you have a fast scope (at least 50 MHz), you could check if you see oscillations with the switch at the end of the cable closed and input capacitor bank off. You have to connect the scope to the amplifier output then, as connecting a scope probe to the input might damp any oscillations.
More pragmatically, if you don't want any bangs, I suggest you add a 15 pF-220 ohm RC series network across the input or at least reserve board space for it.
If my hypothesis is correct, it should produce such bangs, at least when all input capacitors are switched off. If you have a fast scope (at least 50 MHz), you could check if you see oscillations with the switch at the end of the cable closed and input capacitor bank off. You have to connect the scope to the amplifier output then, as connecting a scope probe to the input might damp any oscillations.
More pragmatically, if you don't want any bangs, I suggest you add a 15 pF-220 ohm RC series network across the input or at least reserve board space for it.
Hi MarcelvdG,
Thanks, now it's clearer.
I don't have an oscilloscope at the moment, so I can't do this test (I've been planning to buy one for a long time, it will happen soon), but I can certainly implement the RC network you suggest.
Do you mean this way? (bear with me...) 🙄
Apart from the correction of the oscillation noise, do you think it could have other effects on the sound?
In other words, would you just add it, or would you rather insert an SMD Jumper to include / exclude it, if necessary?
Due to a question of availability of SMD components at JLCPCB, I had to choose the 1206 pack for two values of resistors (I wanted them strictly thin film).
But I'm still having problems with them, about the price of the non-stock components and the methods of purchase.
At this point, even if it means spending a little more, I will only use 0805 components.
Thanks, now it's clearer.
I don't have an oscilloscope at the moment, so I can't do this test (I've been planning to buy one for a long time, it will happen soon), but I can certainly implement the RC network you suggest.
Do you mean this way? (bear with me...) 🙄
Apart from the correction of the oscillation noise, do you think it could have other effects on the sound?
In other words, would you just add it, or would you rather insert an SMD Jumper to include / exclude it, if necessary?
Due to a question of availability of SMD components at JLCPCB, I had to choose the 1206 pack for two values of resistors (I wanted them strictly thin film).
But I'm still having problems with them, about the price of the non-stock components and the methods of purchase.
At this point, even if it means spending a little more, I will only use 0805 components.
Exactly.
It adds 15 pF to the input capacitance, the effect of the 220 ohm series resistor is negligible below 500 kHz. I would just add it.
Last edited:
The 220R/15pF Zobel also works towards damping the otherwise unloaded "transmission line" of the coax cables connecting the turntable to the phono preamp, even without the 1656 stability issue. Stray impinging RF can have voltage maxima that are biggest unterminated, but damped lower by the Zobel.
I wonder if a Zobel dedicated to damping the (50R to 100R ?) cable could also help with the hot-pants 50MHz GPB stability issue. Doubtlessly, details matter. 75R/xpF?
All good fortune,
Chris
I wonder if a Zobel dedicated to damping the (50R to 100R ?) cable could also help with the hot-pants 50MHz GPB stability issue. Doubtlessly, details matter. 75R/xpF?
All good fortune,
Chris
If you haven't yet committed to a PCB layout you may wish to explore a superior new topology:
see schematic of post #58.
This topo combines best possible dynamic margins (for both noise and overhead) with a subsonic filter, no electrolytic caps, single stage for best power supply immunity and best battery operation if desired (I do), no servos and no output coupling cap required. Check it out.
All good fortune,
Chris
This topo combines best possible dynamic margins (for both noise and overhead) with a subsonic filter, no electrolytic caps, single stage for best power supply immunity and best battery operation if desired (I do), no servos and no output coupling cap required. Check it out.
All good fortune,
Chris
I updated the circuit to version 0.1a
Hi Chris, thanks for the suggestions.
I'm waiting to hear MarcelvdG's opinion, but the circuit with this configuration is almost complete, just add the ground plane and little else.
If the layout does not require further adjustments, it could get to printing in a relatively short time.
If instead we want to adopt the circuit you suggest, it is almost easier to create a new schematic and pcb, also in function of using different Op Amps.
In this case I wanted to try the OPA1656 in a minimal SMD circuit.
Opening to a different schematic, which could benefit from the use of different OPAmps, I think it is better to use a DIP-8.
- Fixed the schematic with the new RC network
- Now all the resistors are in 0805 pack.
- Shortened the paths and optimized the distances between the components.
Hi Chris, thanks for the suggestions.
I'm waiting to hear MarcelvdG's opinion, but the circuit with this configuration is almost complete, just add the ground plane and little else.
If the layout does not require further adjustments, it could get to printing in a relatively short time.
If instead we want to adopt the circuit you suggest, it is almost easier to create a new schematic and pcb, also in function of using different Op Amps.
In this case I wanted to try the OPA1656 in a minimal SMD circuit.
Opening to a different schematic, which could benefit from the use of different OPAmps, I think it is better to use a DIP-8.
Hi Nick,
Are you sure?
Out with the 100uF capacitors, into the OPA192 and we have a better Preamp?
(by the way, the cost of the OPA192 is similar to that of the OPA1656)
It would be enough to print the same PCB with the OPA192.
But to replace OAP1656 you will need 2 OPA192 or one OPA2192.
Also it is good to use OPA2156.
Alex.
Yes I am
OPA192/2192 has 5 uV offset (100 times less than 1656) and freely works as DC amp without any separating capacitors at the output and in the feedback loop. This eliminates a bunch of problems, including long charging when turning the power on and off.
yes they ( 1656 & 2192 ) are also pin-compatible in SOIC-8
Hi Chris,
Regarding the stability, what matters is the conductance that the RC series network adds around the frequency of the resonance that can get undamped by the op-amp. Assuming that to be close to the quarter-wave resonant frequency of the cable between the muting switch and the phono preamplifier and guessing that it is a 1 metre long cable with polyethylene dielectric, I guessed 50 MHz. I later learned that JRA's cable is a bit longer, so it might be 40 MHz instead.
Choosing RC = 1/(2 pi f) results in the largest conductance per unit capacitance, but it doesn't get worse very quickly when you take a smaller resistance:
40 MHz:
Conductance of 15 pF in series with 220 ohm: 1.852445428 mS
Required capacitance in series with 75 ohm to get the same conductance: 21.31001964 pF
Required capacitance in series with 68 ohm to get the same conductance: 22.21336249 pF
50 MHz:
Conductance of 15 pF in series with 220 ohm: 2.354662796 mS
Required capacitance in series with 75 ohm to get the same conductance: 19.65523989 pF
Required capacitance in series with 68 ohm to get the same conductance: 20.43857696 pF
All in all, theoretically, 68 ohm and 22 pF should have about the same effect on stability as 220 ohm and 15 pF, but 220 ohm and 15 pF has been tried in real life, while 68 ohm and 22 pF has not. Besides, the single-sided transmission line termination for very high frequencies gets messed up again as soon as you turn on any part of the input capacitor bank.
Best regards,
Marcel
The layout looks nice and compact to me. When you put it in an enclosure, you have to keep hum fields away and keep the loop area from the input connector to the amplifier input and back small, as is always the case with phono preamplifiers. (The alternative circuit Chris referred to is my design. I think it's the simplest way to add a fixed subsonic filter, if you should want one.)
I also suffered for a long time with the instability of phono preamplifiers, until I used a transadmittance cascade at the output, which solved many problems at once
Hi MarcelvdG,
Thank you.
And to answer your question, I had already considered building Rod Elliott's subsonic filter, once the Phono pre was completed, but the possibility of incorporating this functionality into the pre phono circuit is a special added value.
And also the modification suggested by Nick (Hi Nick...!🙂) is definitely intriguing, despite the higher cost and the less easy availability of the OPA2192 (which on JLCPCB, always translates into a higher cost).
But I'm very curious about the possibility of comparing two very similar circuits, with and without a 100uF Bipolar electrolytic capacitor on the signal path.
They are three nice versions, each with interesting strengths.
What can I say, I would like to have all three available, to be able to compare them , more with my ears, than with the oscilloscope, at least until I buy it. 😎
.So MarcelvdG, if you don't mind and you agree... 🙄
... I could also prepare the PCB for your circuit that incorporates the subsonic filter.
Then we decide which ones to print, also depending on the availability of the parts.
The most boring and delicate phase is the selection of components, to balance costs, availability and specifications.
Basically, apart from particular needs, I would only use thin film resistors, 0805, 0.1% tolerance and no more than 25ppm.
Ceramic capacitors only C0G/NP0, always 0805, possibly at 2% or better.
Thank you.
And to answer your question, I had already considered building Rod Elliott's subsonic filter, once the Phono pre was completed, but the possibility of incorporating this functionality into the pre phono circuit is a special added value.
And also the modification suggested by Nick (Hi Nick...!🙂) is definitely intriguing, despite the higher cost and the less easy availability of the OPA2192 (which on JLCPCB, always translates into a higher cost).
But I'm very curious about the possibility of comparing two very similar circuits, with and without a 100uF Bipolar electrolytic capacitor on the signal path.
They are three nice versions, each with interesting strengths.
What can I say, I would like to have all three available, to be able to compare them , more with my ears, than with the oscilloscope, at least until I buy it. 😎
.So MarcelvdG, if you don't mind and you agree... 🙄
... I could also prepare the PCB for your circuit that incorporates the subsonic filter.
Then we decide which ones to print, also depending on the availability of the parts.
The most boring and delicate phase is the selection of components, to balance costs, availability and specifications.
Basically, apart from particular needs, I would only use thin film resistors, 0805, 0.1% tolerance and no more than 25ppm.
Ceramic capacitors only C0G/NP0, always 0805, possibly at 2% or better.
Last edited:
Just built the circuit on a perfboard with njr2068. The RIAA EQ is spot on when tested with reverse RIAA circuit.
Will test with cart soon.
Will test with cart soon.