Sure, but the discussion is not about my ears.
The dicussion is how low THD the electronics must go to meet the best ears (like the person you met) considering that we will have electromechanical in the source (vinyl) and electromechanical in the end (speakers). These equipment have more than 0.1% separately and even more summed.
RIAA pre-amp and power amp are the electronics in the middle of the chain under discussion.
What should be the maximum THD the pre-amp+powerAMP could add to the chain in order to still meet the golden ears requirements?
0.01%, 0.001%, 0.0001%?
I'm interested on that, from the point of view of an EE.
As you said, there are costs and component types involved. Building a pre-amp or even power amp with THD 0.01% is very easy, cheap, doable with regular components.
But building things with THD 0.001% or 0.0001%, is not that easy, not cheap and/or might require special components and layout.
I agree that layout is important, but it is free 😉 In phono pre-amplifier which the input signal is very low, noise performance is more important than THD.
But harmonic profile of the distortion more important than the THD itself.
For MM cartridge:
1. Choose op-amp with very low current noise (usually JFET input).
2. Choose op-amp with low voltage noise below 4nV/sqrt(Hz).
3. If the front-end is discrete, use low noise JFET or BJT with low rBB (you can combine with op-amp to make simpler).
4. Use low impedance for negative feedback (or RIAA feedback).
5. If you care about harmonic profile of the distortion, you can use class A output or add buffer class A before negative feedback. Class A in small signal is not complicated.
6. Use low noise voltage regulator.
7. Use low noise resistor.
But harmonic profile of the distortion more important than the THD itself.
For MM cartridge:
1. Choose op-amp with very low current noise (usually JFET input).
2. Choose op-amp with low voltage noise below 4nV/sqrt(Hz).
3. If the front-end is discrete, use low noise JFET or BJT with low rBB (you can combine with op-amp to make simpler).
4. Use low impedance for negative feedback (or RIAA feedback).
5. If you care about harmonic profile of the distortion, you can use class A output or add buffer class A before negative feedback. Class A in small signal is not complicated.
6. Use low noise voltage regulator.
7. Use low noise resistor.
Sorry for the absence...
Before I forget, I thank MarcelvdG for the answers.
And I also thank all those who are participating with various interesting insights, but for the benefit of those who will read in the future, I invite you not to stray too far from the main theme (I speak in general...)
In this period I am overwhelmed with commitments and I struggle to find time for everything.
My goals would be, in order:
I am referring to the first RIAA Preamp circuit with OPA1656 revised by MarcelvdG, but I have also started to prepare the schematic of Marcel's circuit with High-Pass filter, which I attach.
If you see errors, please report them to me.
Before I forget, I thank MarcelvdG for the answers.
And I also thank all those who are participating with various interesting insights, but for the benefit of those who will read in the future, I invite you not to stray too far from the main theme (I speak in general...)
In this period I am overwhelmed with commitments and I struggle to find time for everything.
My goals would be, in order:
- Buy the smd components necessary for the circuit, so as to have them available at the time of printing.
- Complete the PCB of the circuit and print it.
- Complete the device in a metal case, with the dedicated power stage (already ready).
I am referring to the first RIAA Preamp circuit with OPA1656 revised by MarcelvdG, but I have also started to prepare the schematic of Marcel's circuit with High-Pass filter, which I attach.
If you see errors, please report them to me.
Last edited:
Looks fine to me, except that second and third order are swapped in the notes about jumper settings.
Hi MarceldvG,
You are right, thanks.
I have updated the schematic with the correct settings.
Yesterday I managed to buy most of the components from the JLCPCB stock.
I still have a couple of values missing, but I will buy them soon.
Lucky for me, the OPA2192 are back in stock, at a lower price than the OPA1656 (in theory, they should cost twice as much).
I took it as a sign, an invitation to try the modification proposed by Nick Sukhov, with the OPA2192. 🙂
In the next few days I hope to find some time to prepare the pcb of your version with high pass filter.
You are right, thanks.
I have updated the schematic with the correct settings.
Yesterday I managed to buy most of the components from the JLCPCB stock.
I still have a couple of values missing, but I will buy them soon.
Lucky for me, the OPA2192 are back in stock, at a lower price than the OPA1656 (in theory, they should cost twice as much).
I took it as a sign, an invitation to try the modification proposed by Nick Sukhov, with the OPA2192. 🙂
In the next few days I hope to find some time to prepare the pcb of your version with high pass filter.
About the input capacitors of the Pre with high-pass filter, I would also be in favor of using a decent polypropylene (for the 2.2uF one).
I attach a photo of capacitors I have at home, all in 2.2uF Polypropilene (Except the Russo K73-16B) and voltages ranging from 100V to 630V.
My preference would go to the Clarity Cap ESA 250V, for its quality and for its still small size (it is 29 mm long for about 20 mm in diameter).
In these cases, however, if it does not create problems for the circuit, I prefer to implement a multiple Footprint, which allows the insertion of capacitors of different sizes (as in the photo below).
The first question is:
- In what size range do we want to stay?
Are we satisfied with a 40 mm long footprint (Pad to Pad), or do we extend it up to 55 mm to accommodate the Russian Cap, as well?
Some people speak highly of these capacitors, I have some, but I have never tried them.
The second question concerns the third-order filter capacitor, the 220nF one.
The simplest solution would be to use an SMD capacitor, also to free up space.
On JLCPCB this is the only one available, a Murata C0G, in 1206 pack.
https://www.mouser.it/datasheet/2/281/1/GRM31C5C1H224JE02_01A-1987815.pdf
- Would you be okay with having this in your phono stage, or would you prefer a different capacitor?
To close the Saga of capacitors, there would be the two 6.8uF ones.
Doing a search on Mouser among Bipolar electrolytics, there seems to be only this one (not in stock, on jlcpcb).
https://www.mouser.it/datasheet/2/293/e_udb-3082293.pdf
- Could it be ok, or would you prefer another type?
I attach a photo of capacitors I have at home, all in 2.2uF Polypropilene (Except the Russo K73-16B) and voltages ranging from 100V to 630V.
My preference would go to the Clarity Cap ESA 250V, for its quality and for its still small size (it is 29 mm long for about 20 mm in diameter).
In these cases, however, if it does not create problems for the circuit, I prefer to implement a multiple Footprint, which allows the insertion of capacitors of different sizes (as in the photo below).
The first question is:
- In what size range do we want to stay?
Are we satisfied with a 40 mm long footprint (Pad to Pad), or do we extend it up to 55 mm to accommodate the Russian Cap, as well?
Some people speak highly of these capacitors, I have some, but I have never tried them.
The second question concerns the third-order filter capacitor, the 220nF one.
The simplest solution would be to use an SMD capacitor, also to free up space.
On JLCPCB this is the only one available, a Murata C0G, in 1206 pack.
https://www.mouser.it/datasheet/2/281/1/GRM31C5C1H224JE02_01A-1987815.pdf
- Would you be okay with having this in your phono stage, or would you prefer a different capacitor?
To close the Saga of capacitors, there would be the two 6.8uF ones.
Doing a search on Mouser among Bipolar electrolytics, there seems to be only this one (not in stock, on jlcpcb).
https://www.mouser.it/datasheet/2/293/e_udb-3082293.pdf
- Could it be ok, or would you prefer another type?
To consider:
Given that the input node is high impedance at higher frequencies, any large surface like from a huge capacitor is a nice antenna (for E-fields). You will want a well-shielded metal enclosure to avoid noise and RF pickup, and you have to consider internal noise sources as well (power supply).
And at low frequencies, where circuit gain is very high, you want to have minimized loop area formed by the input path and GND return. Any
loop is an antenna for magnetic fields, which are much harder to shield+. And again internal noise sources can become an issue.
Given that the input node is high impedance at higher frequencies, any large surface like from a huge capacitor is a nice antenna (for E-fields). You will want a well-shielded metal enclosure to avoid noise and RF pickup, and you have to consider internal noise sources as well (power supply).
And at low frequencies, where circuit gain is very high, you want to have minimized loop area formed by the input path and GND return. Any
loop is an antenna for magnetic fields, which are much harder to shield+. And again internal noise sources can become an issue.
Yes, large caps are a hum-pickup risk - the solution is a well shielded enclosure, i.e. all metal. RF pickup will come down the cables, so RF-filtering on inputs and outputs is needed to address that - larger components are more at risk from RF simply because that are better antennas.
Near-field low-frequency magnetic interference from mains transformers and so forth is the difficult one, that will sail through a few mm of aluminium. Mu-metal shielding is one way to help with this, but its awkward to use (has to be re-annealed after any bending/forming...)
Near-field low-frequency magnetic interference from mains transformers and so forth is the difficult one, that will sail through a few mm of aluminium. Mu-metal shielding is one way to help with this, but its awkward to use (has to be re-annealed after any bending/forming...)
I had some success by using multiple layers of simple tinned magnetic sheet metal, half a dozen layers or so.
Really effective (and even better than mu-metal) is solid copper "shorting ring" style enclosure, like one inch wall thickness or more. Neither cheap nor DIY-friendly.
Better fix the root problem, the loops. I'm using a mains-operated tape head demagnetizer for this as spot magnetic AC field source, modded to operate on half-waves for broader spectrum.
I think all of them will work fine as long as there are no AC magnetic fields strong enough to cause hum issues. If the phono preamplifier needs to be placed relatively close to a mains transformer, using the smallest capacitor may help to keep the area small of the loop from the centre pin of the cinch connector via the coupling capacitor, via the positive and negative op-amp input pins, 470 ohm resistor and 6.8 uF capacitor back to the ground of the cinch connector. In your layout, when you place the smallest capacitor, the wires to the unconnected alternative pads don't increase the loop area. (Electric fields are easy to shield with a conductive enclosure, as mentioned by KSTR and Mark.)
An alternative I haven't mentioned yet on this thread is to leave out the 2.2 uF altogether. When it is replaced with a wire, the filter slope will change from second order to first order below 1.3 Hz, but you already have over 40 dB of suppression there.
Very good choice in all respects: reasonably accurate, very linear and stable, small size. (It will be your phono stage, of course, even though it's my design.)
I'd rather use 5 % tolerance MKT capacitors than 20 % tolerance bipolar electrolytic capacitors there.
Until quite recently, WIMA had a series of compact 5 % tolerance MKT capacitors such as MKS2B046801M00JSSD Unfortunately, they are out of production now because the manufacturer of the PET film used in the capacitors has stopped making it. If you can still get some, they will be quite suitable.
Last edited:
I'm pretty sure a 1 inch thick Vibranium cabinet can work wonders,
but caveman DIYers like me would probably start with a regular metal case.
(Hello Mark and KSTR...!) 🙂🤚
I've often used 2mm aluminum for the walls and base, 0.5 - 1mm for the top.
Below you can see two possible configurations, and while version A (single case) is easier to build, I'm sure most people will prefer version B, with the transformer in a separate metal case.
As you can see in the (very simplified) pictures, my idea is to keep the input connectors on the PCB, very close to the RCA signal pins.
Around 2mm between the signal pins, just under 6mm between the ground pins.
Is this what we mean when we talk about minimizing the loop area?
In these pictures I've included the input capacitors with the same size as the ESA Clarity Caps, rated at 250V.
MarcelvdG, thank you very much for your answers about the capacitors...!
I will buy the 220nF SMD mentioned above.
About the 6.8uF capacitors, as you also said, in Polypropylene they are expensive and huge.
And yes, the Wima MKS_2 in PET, have still reasonable dimensions (8.5 x 7.2 x 14 mm. WxLxH) with 5% tolerance.
https://www.mouser.it/datasheet/2/440/e_WIMA_MKS_2-1139871.pdf
And as you said, they do not seem to be available, at least from Mouser.
but caveman DIYers like me would probably start with a regular metal case.
(Hello Mark and KSTR...!) 🙂🤚
I've often used 2mm aluminum for the walls and base, 0.5 - 1mm for the top.
Below you can see two possible configurations, and while version A (single case) is easier to build, I'm sure most people will prefer version B, with the transformer in a separate metal case.
As you can see in the (very simplified) pictures, my idea is to keep the input connectors on the PCB, very close to the RCA signal pins.
Around 2mm between the signal pins, just under 6mm between the ground pins.
Is this what we mean when we talk about minimizing the loop area?
In these pictures I've included the input capacitors with the same size as the ESA Clarity Caps, rated at 250V.
MarcelvdG, thank you very much for your answers about the capacitors...!
I will buy the 220nF SMD mentioned above.
About the 6.8uF capacitors, as you also said, in Polypropylene they are expensive and huge.
And yes, the Wima MKS_2 in PET, have still reasonable dimensions (8.5 x 7.2 x 14 mm. WxLxH) with 5% tolerance.
https://www.mouser.it/datasheet/2/440/e_WIMA_MKS_2-1139871.pdf
And as you said, they do not seem to be available, at least from Mouser.
Quick update.
I've done some more research, but the Wima MKS2 6.8uF seem to be unavailable everywhere.
The MKS4 series is available, but they already measure 18 x 8 mm. and cost more than two euros each, for lots of at least 10 pieces.
Considering that you need 4 of them, I'd say they're too big and expensive (...but they're still small among the film ones, which easily reach 26 x 10 mm.)
If the problem with Nichicons was mainly tolerance, it can be partly circumvented.
I always solder the THT components myself. I can measure their values and create selected pairs (or even better groups of 4).
Maybe the value itself (depending on luck) could oscillate between 10 and 20%. but the variation between right and left channel could be smaller, maybe around 5-3% or even less.
They are described as: "Designed specifically for crossover networks in Hi-Fi sound systems", which may not mean much, but they have a small footprint (12.5 mm diameter), and they cost relatively little, which is not a primary parameter, but it can help.
If anyone can think of interesting alternatives (good dielectric, THT, reasonable cost, small size), or you think differently, it's the right time to say so.
I've done some more research, but the Wima MKS2 6.8uF seem to be unavailable everywhere.
The MKS4 series is available, but they already measure 18 x 8 mm. and cost more than two euros each, for lots of at least 10 pieces.
Considering that you need 4 of them, I'd say they're too big and expensive (...but they're still small among the film ones, which easily reach 26 x 10 mm.)
If the problem with Nichicons was mainly tolerance, it can be partly circumvented.
I always solder the THT components myself. I can measure their values and create selected pairs (or even better groups of 4).
Maybe the value itself (depending on luck) could oscillate between 10 and 20%. but the variation between right and left channel could be smaller, maybe around 5-3% or even less.
They are described as: "Designed specifically for crossover networks in Hi-Fi sound systems", which may not mean much, but they have a small footprint (12.5 mm diameter), and they cost relatively little, which is not a primary parameter, but it can help.
If anyone can think of interesting alternatives (good dielectric, THT, reasonable cost, small size), or you think differently, it's the right time to say so.
Reichelt still has the automotive-qualified, 63 V version, but I don't know how many or for how long:
https://www.reichelt.de/de/de/shop/produkt/mks2_pet-kondensator_6_8_f_5_63_vdc_rm_5-172687
I'd go for the MKS4 if this leads to nothing (not just lower tolerance, but closer to ideal capacitor behaviour in all respects), but selected bipolar electrolytics should also work well. C6 and C16, in your schematic, are more critical than C7 and C17, so you could use the more accurate ones for C6 and C16.
https://www.reichelt.de/de/de/shop/produkt/mks2_pet-kondensator_6_8_f_5_63_vdc_rm_5-172687
I'd go for the MKS4 if this leads to nothing (not just lower tolerance, but closer to ideal capacitor behaviour in all respects), but selected bipolar electrolytics should also work well. C6 and C16, in your schematic, are more critical than C7 and C17, so you could use the more accurate ones for C6 and C16.
Thanks Marcel,
I had missed them, even though Reichelt is one of my favorite suppliers.
I have already added them to the cart. Tomorrow I will finalize the purchase.
And they are at 5% tolerance, as we hoped. 👍
I had missed them, even though Reichelt is one of my favorite suppliers.
I have already added them to the cart. Tomorrow I will finalize the purchase.
And they are at 5% tolerance, as we hoped. 👍
This seems like an interesting possibility.
What if instead of 2.2uF a lower value was used, such as 1uF?
When you don't change anything else, you get an extra steep roll-off between about 1.3 Hz and 3.2 Hz. When you also reduce the 220 nF capacitor to 100 nF, change a resistor value and add an extra resistor, you can get the response back to normal again. (I'll draw a schematic to show how, but I can't do that now.)
MarcelvdG, thank you, but I didn't want to push you towards further variations of the circuit (...but I don't want to limit your suggestions either 🙂).
I've already bought most of the components and the rest are in order. Apart from the 2.2uF input and the 6.8uF MKS2s, all the other components are SMD, so fixed values that cannot be changed, including the 220nF 1206 that you liked and that I bought.
The alternative suggested by you seemed to be not to insert the 2.2uF, which will have a large and versatile footprint (for the MKP), therefore capable of accommodating, if necessary, a capacitor of a lower value.
But if reducing only this value does not give any advantage, I have no problem leaving the circuit unchanged.
I've already bought most of the components and the rest are in order. Apart from the 2.2uF input and the 6.8uF MKS2s, all the other components are SMD, so fixed values that cannot be changed, including the 220nF 1206 that you liked and that I bought.
The alternative suggested by you seemed to be not to insert the 2.2uF, which will have a large and versatile footprint (for the MKP), therefore capable of accommodating, if necessary, a capacitor of a lower value.
But if reducing only this value does not give any advantage, I have no problem leaving the circuit unchanged.
Last edited:
I'll leave it at this then. An extra steep roll-off between two deep subsonic frequencies actually does no harm, while the alternative I had in mind has a small amount of extra noise at the lower audio frequencies.
Hi stocktrader200,
This is good news! 😎
By a curious coincidence (some would say: "Chance does not exist..."), a few days ago I also bought an AT-VM95C.
I have read many good opinions about this cartridge and for what it costs, the VM95C seems to have excellent performance.
As described in the first post, I planned to insert a 3x SPST Dip Switch, just to be able to select low values of input capacitance (mainly between 100 and 200 pF), as required by cartridges of the AT-VMxx line.
This is good news! 😎
By a curious coincidence (some would say: "Chance does not exist..."), a few days ago I also bought an AT-VM95C.
I have read many good opinions about this cartridge and for what it costs, the VM95C seems to have excellent performance.
As described in the first post, I planned to insert a 3x SPST Dip Switch, just to be able to select low values of input capacitance (mainly between 100 and 200 pF), as required by cartridges of the AT-VMxx line.
