Supra 2.0 - High End Preamp for Record Player [150616-I]

More than 2^5 years ago a circuit was published called SUPRA, a "superrauscharmer" MC/MM-Vorverstärker" for record players. The low noise behaviour comes from paralleling 8 cheap low noise transistors (BC550/BC560) which reduces noise by √8 = 2.82 resulting in a big board with 20 transistors (per channel!!!).

Today there are integrated circuits which are so noise optimized that a discrete aproach would make no more sense. LT1028 for example has a noise figure of 0.9 nV/√Hz. If you use four of these chips in parallel noise will be halve of this! This is so low that a lower noise behaviour is nearly impossible, because the resistors - even the low Ohm - have a higher thermal noise. One of this Opamps costs between 5 an 10 € which results in about 100..150 € for a preamp in stereo. Seems not so much because people who hear still records instead of MP3s or CDs are prepared to pay a bit more for their hobby.

But combining this 4 Opamps per channel is not all. If you build such expensive hardware it makes sense to optimise the remaining circuit too. Why not integrating a true passiv RIAA correction? Most manufacturers only speak about it. Here they are: buffered RC combinations. Have a look at the design. I made layout files (EAGLE 7) if somebody wants to test this preamp in reality. The values are for a MM preamp. A MC preamp needs some higher gain and lower impedances: R1 = 470Ω, R2...R5 = 10Ω, R14 = 3k9

PS:
(28.12.2015)
And just for the ones who would prefer a concept with "really true" passive filters (not in the feedback loop of an opamp) I made a version with two RC low-pass and one RC high-pass filters in series (each buffered). But without the 10 Hz subsonic filter of the previous solution. Have fun...

NOTE: the kit is back in stock ! Buy the kit now and make your own Supra !

Project Elements

supra2-s.png (PNG, 34.88 KB)

supra2-b.jpg (JPG, 77.33 KB)

Supra2.sch (SCH)

Supra2.brd (BRD)

Supra2_L.txt_.zip (ZIP)

Board 2.1.zip (ZIP)

Schematics and board of a version with true passive filters

Discussion (14 comments)

Pascal Hibon 8 years ago

In which Elektor magazine did this article apear in?

Hedwig 8 years ago

150616-I-D 1606 German issue
150616-I-EN1607 English issue
150616-I-F 1606 French issue
150616-I-NL1607 Dutch issue

Pascal Hibon 8 years ago

Hi Thomas,

I did try the search function but didn't find what I was looking for.
Thanks for the link; I now found the answers.

Dr. Thomas Scherer 8 years ago

If you use the search function you get this:
https://www.elektormagazine.com/search?query=Supra+2.0

3 Comment(s)

orion65 8 years ago

Hi Alfons,
If you are still interested in adaption for moving coil, please let me know.
My daughter developed a MC adapter as "Maturaarbeit" and I adapted for her the Supra 2.0 for MC. Everything works fine.
BR,
Manfred Dietrich

Dirk67 5 years ago

Hello Manfred,

how did you adapt the Supra 2.0 for MC,
I'm interested to do this as well ...

In the Authors text above there's written:
>> "A MC preamp needs some higher gain and lower impedances: R1 = 470Ω, R2...R5 = 10Ω, R14 = 3k9 "
but I think you did maybe some more ?

you may contact me via "supra_preamp [ät] 24max.de"

Dirk Boesche
(Germany)

elabos 4 years ago

Hello Dirk,
I'm looking to adj the gain for MC on the latest version of SUPRA Phono ( 4*LT 1028 + 1 LM833)
I cannot recognize Your Resistor # reference to change.

Can You confirm me the change for the R Value You did in your experience ?

Thank You in advance
Best REgards

OSK

20190819123025-150616-11-page-0001.jpg (231kb)

1 Comment(s)

Telefunken 8 years ago

Hello Mr. Scherer. I just ordered the PCB for this project. I would also be interested to build the MC preamp.

Also, do you know if the PCB for the analog power supply will be available again in the online store ?

Thanks !

Mathew 5 years ago

Hi,

I built that preamp and I got on the input of LM837 2MHz sinus.
The LT1028 working like oscilator. It doesn't matter if power is coming from battery or transformator.
Assuming that missing decoupling capacitors has an impact on oscilation.
Looking into supra 2.0 preamp (big schema), every OPA is decoupled and has frequency compensation (https://www.elektormagazine.com/news/cool-summer-free-download-of-the-week-supra-2-0-super-low-noise-mm-md-phono-preamp)

Any idea how to remove oscilation?

best regards

M.

Found the problem. The pcb (brg file) has missing path, sum isnot linked with in of lm837

Hedwig 8 years ago

The PCB 150616-2 for the analog power supply wil be available again in the online store soon.

Telefunken 8 years ago

Thank you !

Dr. Thomas Scherer 8 years ago

Oh, I don't know that. Please ask the people in the Elektor Store, they know it for sure. But in case of it is not available: You can easily build your own power supply. I personally prefer the analo variants. All you have to check is that the voltages are okay and it can deliver enoug current. And: hum & noise should be low. This part you simply can check with any scope (some mV noise and hum is okay, but not in the 10mV ranges).

4 Comment(s)

Dr. Thomas Scherer 8 years ago

Alfons: I don't know, sorry.
But your comment will give motivation to this...
:D

0 Comment(s)

Alfons Ziegler 8 years ago

When will the second part of rhis article (adaption for moving coil and LME49990) be published? I'm eager to build the preamp, but have an ortofon rondo mc system.

Bernd Soulier 8 years ago

I have also a rondo mc (bronze) playing on a original LinnLP12 with Akito ,the best result I have with a very low impedance input of two lasertrimmed transistors on a array from That each channel ,this is the optimum for the correct impedance .Inputs are symetrical after going to LME 49710 (TO-Metallcase) ,my favourite audioopamp ,also good in sound are AD797 or OPA2134 ,only this is cheaper .Seeing the price of the Ortofonsystem (about 900EUR ,the Linn is with no money to pay) is no discussion to use a few opamps with about 8 EUR .The deal is to drive complete without DCoffset and NO couplecondensator ,otherwise use the best audiocap (price up to 100USD) ,or NO cap ,this is the best cap for low ,all other is BULLSHIT

1 Comment(s)

TonGiesberts 8 years ago

Apparently the DC/DC converter used in the original publication (JCE0612D24 from XPPOWER) is hard to come by. We didn’t find modules at other manufacturers with the same specifications (6 W, 12 V in, 2 x 24 V out and the same footprint). There is an easy solution to this problem. Instead of an output voltage of +/-24 V we use +/-15 V. A good alternative is module TEL 5-1223 from Traco Power. It’s maximum output power is 5 W which is more than enough, due to the lower output voltage. The outputs can deliver 200 mA, also more than enough. Input voltage is 12 V nominal (9-18 V). The voltage drop across the regulators should be more than 3 V. The voltage drop across the 1N4007 in the power input is close to 1 V. So the voltage output of regulators LM317/LM337 should be set to deliver +/-11 V. All it takes is to change R60 and R62 to 1.4 kΩ, that’s all. To be sure we performed new measurements and got the same results with the lower power supply. With this modification the power supply voltage of the opamps is roughly +/-10 V (he voltage drop across the filters (T1…T4) is also around 1 V).

Hedwig 8 years ago

picture of the alternative DC/DC converter for PCB 150646-1

alternative DC/DC conterver for PCB 150646-1 (62kb)

1 Comment(s)

Dr. Thomas Scherer 8 years ago

Andreas, please read agein the part with the reduction of uncorrelated noise (noise/n^.5)
IIIIIII, they are just RC combinations, no magic. Lowpass fg = 1/(2 Pi RC) etc.

0 Comment(s)

lllllll 8 years ago

Where are the RIAA component calculations pointed to in the article.

0 Comment(s)

Bernd Soulier 8 years ago

final touch with my experiences ,
the LME 49990 is definitly the better sound ,much more better is the LME 49710 TO-99 with symetrical Input ,Impedance about 30 kOhms .Shit like 5534 is NOT in acceptance ,the Inputimpedance for MC MUST be aktiv low and there is no other way to do this with a Transistor Basisschematic and NOT with resistors .My original Linn LP12 Vynilplayer with a original Akido and MC Ortofon Rondo Bronze shows this all ,the best Kondensator in the way is nothing ,that means to do the hard way without any Offset .Digital Supplys destroy any good analogsound ,the best way are akkus and the charging must work another time as listening .I worked about more then 30 years as a Soundengeneer ,my ears are more than perfect ,offcourse I like the really good sound and this way is not too much difficult .

0 Comment(s)

Andreas von Ow 8 years ago

If you want to achieve optimal noise performance for MM phono cartridges I strongly recommend to only use one LT1028 opamp on the input stage. MM pickups have a frequency dependent impedance in the range of a few hunderd to several kOhms. With an input noise current of 1pA/sqrt Hz, the current generated noise gets higher than than the voltage noise. With 4 LT1028 opamps and a impedance pickup impedance of 1kOhm, the current generated noise is 2nV/sqrt Hz. For a one opamp version (only IC 1), R16 should be reduced to 1.7kOhm. MC pickups are low impedance and the 4 opamp configuration makes sense.

0 Comment(s)

TonGiesberts 9 years ago

We completed our prototype and put it inside its enclosure. The amplifier PCB is placed in the lowest slots. Make sure all leads have some distance from the bottom. We used the DC/DC converter as a power supply inside the amplifier enclosure. This way only a standard 12V/1A AC adapter is needed externally. What needs to be done first is the mechanical work. Easiest is the led in the front panel. Drill a 2.5 mm hole in the middle and file out the hole until the 3 mm led can be push through with some force. It can always be glued for safety. On the rear panel we placed the four audio connectors as far as possible on the upper left side of the panel. The holes for the Neutrik connectors should be 10 mm to ensure they’re insulated. We kept the distance between the connectors 19 mm. Keep enough distance from the top edge so the top panel can fit and enough room from the side edge so the side of the enclosure will still fit. Mounted on the right upper corner are the on-off switch and power socket. Here the same rule is applied. Keep enough room for the top panel on the upper edge and enough distance from the side edge for the side of the enclosure to fit. The photos give a good idea of what is meant. The switch is placed close to the side to make it easy accessible when reaching alongside the enclosure from the front to the back to turn the preamplifier on or off. To ground the record player (usually a separate wire) we used a 4 mm screw and nut in the middle (horizontally) just above the amplifier PCB. A washer and a metal threaded spacer are used to connect the ground wire to the case. The M4 spacer is big enough the wire can be connected manually. Internally a solder lug is placed to connect the enclosure to the ‘0’ on the amplifier board (terminal pin next to C53/C56).

First connect the inputs with good quality equal length shielded cables to the input connectors to ensure same capacity. The two bottom ones are the inputs. Then connect the outputs to the upper two connectors. Here also two cables of equal length should be used. We left the front and back panel lying two have better access in the future in in case something needs to be changed or replaced. The cables will be longer then strictly needed. In a metal enclosure there’s also the option to simply use twisted wires for in and output, but the use of shielded cables is advised.

To mount the DC/DC converter we used two leftovers from our PCB milling machine. A single piece of PCB is maybe even a better option. Minimum dimensions are 100 mm and 62 mm then. Place the DC/DC-converter module alongside the front panel the twisted wires for the 12 V can go in shortest way alongside the inner side of the enclosure, from on-off switch/power socket on the rear panel to the screw terminal block K2 on PCB 150464-1. The length of the wires for the 24 V supply lines (K4 on DC/DC-converter PCB to K5 on the amplifier PCB) can best be assessed when the DC/DC converter is put in place in the enclosure. Next step is to connect the ‘0’ of the amplifier PCB to the M4 grounding lug on the rear panel. After all these connections are made only the led remains. Use two thin flexible stranded wires and solder them on to the amplifier PCB. Cut them long enough so they can be soldered to the led on the front panel without stress on the wires.

After this maybe a check is in order to see if all is well before closing the enclosure. Connect a 12 V power supply to the power socket and measure if all voltages are correct:
-          +/- 24 V from the DC/DC-converter (K4 on 150464-1), is the green led on?
-          +/-15 V on the amplifier PCB (next to C53/C56)
-          Check power supply voltages on all opamps (should be around +/-14 V or so).
-          Place 560 Ω resistors at the inputs and measure the input offsets (across R1 and R30), should be 0.0 mV. It’s up to you if activating P1 and P2 is in order.
-          Measure output offsets from IC1..IC4 and IC6..IC9 (can be 10s of millivolt max.)
-          Check the output offset of IC5 and IC10 (a few millivolt)

If everything checks out, close the top and put all screws in place. Connect the record player and the preamplifier to your sound system and enjoy the music.

The LT1028 used in this project is around for a long time and there are probably better opamps available today. What to think of a LME49990. This particular opamp only comes in a SO8 package. Especially for SO8 opamps (single and dual) we designed a little SO8-to-DIP8 adapter PCB (10 by 10 mm). It will have an extra capacitor to decouple the power supply of a single or a dual opamp very locally. But more about this tiny adapter is to be found in a separate project (soon).

View on the frontpanel with the power led in place (215kb)
12 VDC power supply connected (384kb)
All Wires connected before finalizing (421kb)
Top/front view on the finished preamplifier (236kb)
Top/rear view on the finished preamplifier (246kb)
Another top/rear view on the finished preamplifier (284kb)
View on the back panel after mounting all parts (370kb)
Side view on the DC-DC-converter and amplifier placed inside the enclosure (398kb)
Top view on DC-DC-converter after sliding it in the enclosure (after wiring) (537kb)
Top view after mounting the front and back panel (699kb)

10 Attachment(s) 0 Comment(s)

TonGiesberts 9 years ago

For those who don’t like the idea of having a DC/DC converter as a power supply for the phono preamplifier and like to go analog all the way we designed a PCB to hold a family of low profile transformers from Block (FL series). Two footprints are combined. The smaller one is designed for transformers of 2, 4, 6 and 8 W. The bigger footprint can hold 10, 14, 18, 24 and 30 W transformers.

On the PCB next to the fuses are white text areas. Here the appropriate fuse rating must be written, since these depend on the power of the transformer used. For the phono preamplifier a 6 W version will do (FL6/18). A net filter is placed at the input of the transformer consisting of a common mode choke and two X1 capacitors (from our EPP list, Elektor Preferred Parts). The transformer has two primary windings and two secondary windings. For a mains voltage of 230 V a single jumper wire must be placed for JP1. For a mains voltage of 115 V two jumper wires must be placed. Three jumper wires are not possible. The secondary windings are each connected to a 4-way screw terminal block (K2). To place the two secondary’s in series a jumper wire can be placed for JP2. The schematic shows the configuration for the phono preamplifier. Use a separate enclosure for the transformer. If the same enclosure as advised for the phono preamplifier is used the two will form a nice pair. How the wiring between the two enclosures must be made is something we leave to your creative minds (fixed length of wiring from screw terminal to screw terminal with strain reliefs or use solid connectors or…).

Bill of materials (configuration for 150616-2)

Capacitor

C1,C2 = 100 nF, X1, 1 kV, Polypropylene, 10,12.5,15 mm pitch

Inductor

L1 = B82724J2142N1, choke, common mode, 2x27mH, 1.4A

Other

F1,F2,F3 = Fuse holder, 20 x 5 mm, 500 V, 10 A

F1,F2,F3 = Cover for fuse holder 20 x 5 mm

K1 = Terminal Block, 2way, lead spacing 7.62 mm, 500 V

K2 = Terminal Block, 4way (2x2way), lead spacing 5.08 mm, 250 V

F1 = Fuse, 100 mA, time delay, 250 V, 20 x 5 mm

F2,F3 = Fuse, 160 mA, time delay, 250 V, 20 x 5 mm

TR1 = 6 VA, 2x115 V primary, 2x18 V/166 mA secondary, Block FL6/18

JP1,JP2 = jumper wire, see text

Misc.

PCB 150616-2 v1.0

pcb-150616-2-v10-top-view.jpg (672kb)
150616-2-transformer-pcb-schematic-v100.png (28kb)
Topoverlay of single sided PCB 150616-2 v1.0 (copper bottom vissible) (181kb)
Copper on bottom of single sided PCB 150616-2 v1.0 (60kb)

4 Attachment(s) 0 Comment(s)

TonGiesberts 9 years ago

We made some changes in the schematic of Thomas. All parts are through hole, which makes experimentation with other amplifiers and other parts easier. Ceramic capacitors are the worst for audio so we chose 1 % polystyrene capacitors. These are expensive but one of the best for audio filtering. The footprint for this capacitors has multiple lead spacing’s in case anyone wants to use different capacitors. We removed the summing amplifier and the inverting output stage. Less is more. Apart from 4 input amplifiers only an extra dual opamp is needed. The passive 75 µs correction is combined with the summing of the four input amplifiers. To keep the noise of this passive correction minimal the capacitance is increased to 44 nF. Together with four 6.81 kΩ resistors in parallel the deviation from the ideal 75 µs is only 0.12 %. The tolerance of the parts is 10 times greater. Advantage of the use of an extra summing amplifier is an additional gain of four. The correction network has a gain of one assuming all four input amplifiers have equal gain. The noise of the four 6.81 kΩ resistors has little or no influence on the S/N ratio coming from the four input amplifiers. The noise of the feedback (2.2 kΩ in parallel with 47 Ω) is 0.882 nV/√Hz. The input of the LT1028 has 1 nV/√Hz. So total input noise (input shorted) is 1.333 nV/√Hz. At the outputs of the amplifiers the noise is 47.8 times larger, 63.74 nV/√Hz. Adding the noise of the 6.81 kΩ resistor will only increase the noise to 64.6 nV/√Hz. The total noise on C15/C16 is half of this and we didn’t even take the resistance of the MM element into consideration. 200 Ω produces 1.84 nV/√Hz. So total input noise is 2.272 nV/√Hz and the influence of the 6.81 kΩ resistors is even less. In all of this current noise is ignored.

Removing the summing amplifier means less sensitivity. Loss of the gain of four is compensated in the 3180 µs/318 µs stage (IC5A/IC10A). There are two ways to calculate the correct parts. The first one is a simple view on the matter. Since the two time constants are exactly a factor ten apart, so should the gain variation be. With increasing frequency the gain should change from 40 to 4. Assuming R24+R25=R1, R22+R23=R2, C17+C18=C1 and R20+R21=R3 the following equations can be derived from the circuit:

C1*(R1+R2)=3180 µs

R1//R2=3*R3

R1=39*R3

Where C1 is fixed at 44 nF.

The resulting theoretical values are 66713.2867 for R1, 5559.44 for R2 and 1710.597 for R3. Two E24 resistors in series give a more accurate result and it’s easier to calculate values needed using nothing but mental arithmetic. Looking at E24 values R1 is 62 kΩ + 4.7 kΩ, R2 is 5.1 kΩ + 470 Ω and R3 is 1.6 kΩ + 110 Ω. The gain change with these E24 values is 9,98582, which is only a deviation of -0.14 %. A more accurate result is possible with E96 resistors. And for those who like a mathematical challenge: is there a gain where the theoretical deviation is at its smallest or even 0 (using E24 and/or E96)?

A more correct approach is of course the use of complex arithmetic. The objective is to get an equation which looks like X*(1+sT1)/(1+sT2) where T1 = 318µs and T2 = 3180µs.

Formula’s: see Formula’s 150616-1.png

The selected gain defines the relation between R1 and R3: R1 = 39*R3 and C1 is still fixed at 44 nF. From here R1,R2 and R3 can be solved. The result will be the same as in the first method.

The low cut off at 10 Hz is placed at the output buffer (IC5B/IC10B). The 2.2 µF capacitors (C19/C44) are replaced by polypropylene versions, which have better properties for audio than standard polyester ones. If possible select two capacitors with equal values since the cut-off frequency of 9.6 Hz influences amplitude and phase in the audio band. And we want this influence to be the same for both channels. The footprint for these capacitors has multiple lead spacing’s (5/7.5/10/15/22.5/27.5 mm) to accommodate a variety of different types.

The power supply is pretty standard with a LM317 for the positive 15 V and a LM337 for the negative 15 V supply voltage. With the adjust pin decoupled, ripple suppression is about 80 dB. By placing a rectifier (four 1N4007 diodes) at the power input a variety of options are possible. A single secondary 18 V/6 W transformer can be used. Connect the secondary voltage to the 0 and one of the ~ inputs of K5. This way each regulator has a half-wave input signal. We don’t recommend this because of the higher input ripple. The standard solution is to use a transformer with two secondary’s. A suitable candidate is type FL6/18 from Block which has two primary windings, making it applicable in countries with a mains voltage of 115 VAC and 230 VAC. One jumper wire (JP1) must be placed for 230 V and two for 115 V. The PCB layout is made in such way that three wires is not possible. The transformer PCB 150616-2 has its own contribution.

Another option is the use of an external symmetrical DC power supply. The positive and negative supply lines are then each connected to one of the ~ inputs of K5 (so two possibilities, doesn’t matter which way). To suppress any residual noise and power supply ripple even more, time constants of the power supply filters for T1…T4 are increased to 0.1 s. To suppress any common mode interference a common mode choke (L1) is added. This can be particularly advantageous if the external DC power supply is a switched mode module like in our “Universal power supply with switched mode modules” (project 150464). A type that can be used for the isolated DC/DC converter is a 6 W ±24 V module from XP Power: JCE0612D24. This module needs a single 9 to 18 V as an input voltage. Apparently there are no standard versions with output voltages between 15 and 24 V so we used a ±24 V module that will also fit on the 150464-1 PCB. The enclosure chosen for this preamplifier is high enough to place this power supply inside. This way the amplifier can be powered by a standard AC adapter or something similar (12 VDC/0.5 A minimal, 1 A preferred). The JCE0624D24 needs an input voltage between 18 and 36 V. If you want to use a standard notebook adapter, module JCE0624D24 should be used. In case one of the mentioned DC/DC converters is used as the power supply remove capacitors C60 and C62 on the amplifier board. The maximum capacitive load of these DC/DC converters is only 47 µF for each output. These capacitors are mounted on the DC/DC converter PCB (150464-1). The LM317 and LM337 are sufficiently decoupled by C59/C61. Several parts don’t need to be mounted on the 150464 PCB as they are of no use: K1 (replace by wire, see photo), C7, C8, C9, D2, D3, MOD1 and K3.

The PCB of the preamplifier is made big enough to fit exactly in the slots of the enclosure. To connect the ins and output signals use isolated RCA sockets (we used types from Neutrik, see BOM). Do not mount the sockets on the same metal side panel without isolation. Keep them isolated! If standard sockets are used consider mounting them on a non-conducting piece of PCB or other material first and have the sockets protrude far enough through big enough holes so connectors will fit without making accidental contact with the panel. Keep the power plug and its wiring as far away as possible from the audio signals and amplifier PCB.

At the input the capacitor C1 for tuning the MM-cartridge consists of 2 capacitors in parallel (C1/C1’ and C26/C26’). This way the capacitor can be accurately built up by two capacitors in parallel if needed. The 100 pF in the schematic and the polysterene capacitor in the BOM are intended mainly as an example or starting value. Look for the manufacturers specification of the cartridge used. Don’t forget to take the capacitance of all cables into account. Some audio cables have 100 pF per meter or more.

Bandwidth of the input amplifiers (IC1..IC4,IC6..IC9) is limited by additional capacitors in the feedback. The LT1028 has a GBW of 70 MHz. Even at a gain of 48 this means a bandwidth of almost 1.5 MHz. Since the input is very sensitive it’s essential to limit the bandwidth to a more practical maximum for audio. So we added 470 pF capacitors (also polystyrene) to limit the bandwidth to about 150 kHz. The 47 Ω resistors in series make sure the amplifiers stay stable (they are internally compensated for a gain of +2). Despite this precaution the input section will still oscillate when the input is left open and no input capacitor is present. Keep the input resistance below 1 kΩ when testing the DC levels of the opamps. In case of oscillation the DC output can be virtually the maximum output voltage!

The LT1028 has an internal bias compensation but according to the datasheet the bias current at the input can still be ±180 nA maximal (the A version ±90 nA). With four amplifiers the DC current flowing through the MM cartridge is theoretically ±720 nA maximal. To compensate this current a trimmer is connected to the two supply lines. Two very high resistors are placed in series with the slider and connected to the input by way of a jumper. C2 suppresses any residual interference and noise. If you don’t like this or think the current doesn’t influence anything remove the jumpers (JP1/JP2). Without anything connected to the input measure the voltage across the input and adjust P1/P2 until it is minimal (JP1/JP2 placed). P1 and P2 are not meant to correct the output offset of the input amplifiers! Some say this will only add additional noise, but you be the judge of that.

The LT1028 is an expensive opamp and there are eight of them in this design. So if you want to keep it a bit more economy priced you could use something like a good old fashioned NE5534 (10 times cheaper). Bias current is much higher, 800 nA maximal and 1500 nA over the specified temperature range. In that case you will have to lower the values of R2/R3 and R31/R32 (2.2 MΩ). A NE5534 is internally compensated for a gain of +3. So change R6/R9/R12/R15/R35/R38/R41/R44 to 100 Ω. But there are many many other opamps out there…

Some measurements from our prototype

Input signal 2.5 mV (source is 250 mV and divider 56 kΩ/560 Ω)

Power supply is transformer PCB 150616-2 or DC/DC converter 150464-1

Signal to noise > 68 dB (transformer)

> 70 dB (DC/DC converter)

> 78 dB(A)

> 88 dB (input shorted)

> 95 dB(A) (input shorted)

Output level 468 mV

THD (without noise) < 0.001 % (B = 20 kHz)

Deviation from RIAA < 0.1 dB (100 Hz…10 kHz)

< 1 dB (20 Hz)

< 0.15 dB (20 kHz)

We also recorded some plots.

The most interesting one is of coarse the deviation from the theoretical RIAA correction. Plot A (Deviation_K2-K4.emf) shows the amplitude at the output K4 when applying a RIAA standard equalized sine wave to input K2. At 20 Hz the amplitude was -0.872 dB in the first channel and 0.905 dB in the second. At 20 kHz the deviations were -0.137 dB and 0.129 dB. The plot for the other channel is visually the same.

To get an idea of any harmonics we made a FFT analysis of a 1 kHz signal. Plot B show a FFT from 10 Hz to 130 kHz with the fundamental suppressed (FFT_K1-K3_1kHz_16avg.emf). No harmonics are visible (16 measurements were averaged to get a better view) and so we can conclude that total harmonic distortion is less than 0.001 %.

To make sure this is also the case for higher frequencies plot C shows a FFT of a 10 kHz signal (FFT_K1-K3_10kHz_16avg.emf). Only a second harmonics is visible at a level of -107.2 dB, which is 0.00044 %.

Bill of materials for PCB 150616-1 (v1.0)

Resistor

R1,R30 = 47 kΩ, 1 %, 0W6, metal film

R2,R3,R31,R32 = 10 MΩ, 5 %, 0W25, carbon film

R4,R6,R7,R9,R10,R12,R13,R15,R27,R33,R35,R36,R38,R39,R41,R42,R44,R56 = 47 Ω, 1 %, 0W5, metal film

R5,R8,R11,R14,R34,R37,R40,R43 = 2.2 kΩ, 1 %, 0W6, metal film

R16,R17,R18,R19,R45,R46,R47,R48 = 6.81 kΩ, 1 %, 0W6, metal film

R20,R49 = 110 Ω, 1 %, 0W6, metal film

R21,R50 = 1.6 kΩ, 1 %, 0W6, metal film

R22,R51 = 5.1 kΩ, 1 %, 0W6, metal film

R23,R52 = 470 Ω, 1 %, 0W6, metal film

R24,R53 = 62 kΩ, 1 %, 0W6, metal film

R25,R54 = 4.7 kΩ, 1 %, 0W6, metal film

R26,R55 = 7.5 kΩ, 1 %, 0W6, metal film

R28,R29,R57,R58 = 1 kΩ, 1 %, 0W6, metal film

R59,R61 = 180 Ω, 1 %, 0W6, metal film

R60,R62 = 2.0 kΩ, 1 %, 0W6, metal film

R63 = 10 kΩ, 5%, 0W25, carbon film

P1,P2 = 1 MΩ, 20 %, 0W15, trimmer, top adjust

Capacitor

C1',C26' = leave open, see text

C1,C26 = 100 pF, 2.5 %, 160 V, axial, polystyrene, 12.9 x 5 mm max.

(C1 and C26 should to be corrected for MM cartridge used)

C2,C27 = 220 nF, 10 %, 100 V, lead spacing 5/7.5 mm

C3,C6,C9,C12,C28,C31,C34,C37 = 470 pF, 2.5 %, 160 V, axial, polystyrene, 12.9 x 5 mm max., lead spacing 5/7.5/10/14.6 mm

C4,C5,C7,C8,C10,C11,C13,C14,C20,C21,C29,C30,C32,C33,C35,C36,C38,C39,C45,C46 = 100 nF, 10 %, 50 V, X7R, lead spacing 5.08/7.62 mm

C15,C16,C17,C18,C40,C41,C42,C43 = 22 nF, 1 %, 63 V, axial, polystyrene, 17 x 6.5 mm max., lead spacing 5/7.5/10/14.6/19 mm

C19,C44 = 2.2 µF, 10 %, 420 V, polypropylene, lead spacing 5/7.5/10/15/22.5/27.5 mm

C22,C24,C47,C49,C51,C52,C53,C56,C59,C61 = 100 nF, 10 %, 50 V, X7R, lead spacing 5.08 mm

C23,C25,C48,C50 = 100 µF, 20 %, 50 V, diam. 8 mm max., lead spacing 2.5/3.5 mm

C54,C55,C57,C58 = 10 µF, 20 %, 50 V, diam. 6.3 mm max., lead spacing 2.54mm

C60,C62 = 220 µF, 20 %, 50 V, diam. 10 mm max., lead spacing 5.08 mm

C63,C64,C65,C66 = 10 nF, 20 %, 50 V, Y5V, lead spacing 5.08 mm

Inductor

L1 = SU9V-01100, common mode choke, 2 x 10 mH, 100 mA

Semiconductor

D1,D2,D3,D4 = 1N4007, DO-41

LED1 = LED, blue, 3 mm, high intensity, T-1

T1,T3 = BC337-25, TO-92

T2,T4 = BC327-25, TO-92

IC1,IC2,IC3,IC4,IC6,IC7,IC8,IC9 = LT1028CN8, DIP-8

IC5,IC10 = LM833NG, DIP-8

IC11 = LM317, TO-220

IC12 = LM337, TO220

Other

K1,K2,K3,K4,PC1,PC2,PC3,PC4,PC5,PC6 = Terminal pin, diam. 1.3 mm, round

K5 = Terminal block 5.08 mm, 3-way

JP1,JP2 = 2-way pinheader SIL, pitch 2.54 mm

JP1,JP2 = Single shunt jumper 2.54 mm spacing

Misc.

PCB 150616-1 v1.0

HAMMOND 1455N1602 Enclosure

2 x Neutrik NYS367-0, RCA Connector, Socket, Black

2 x Neutrik NYS367-2, RCA Connector, Socket, Red

Bill of materials for power supply PCB 150464-1 (v1.1)

Resistor

R1 = 4.7 kΩ, 5%, 0.25W, 250V

Capacitor

C1-C6 = 2u2, 50 V, 20 %, Ceramic Y5V, lead spacing 5.08 mm

C10,C11 = 47 µF, 50 V, 20 %, lead spacing 2.5 mm, diam. 6.3 mm max.

C12,C13 = 100 nF, 50 V, 10 %, Ceramic X7R, lead spacing 5.08 mm

Inductor

L1 = 4µ7, 3.05 A, 80 mΩ, 10 %, radial, lead spacing 5 mm (Epcos B82144B2472K000)

L2 = 600 µH, 2 A, 2x50 mΩ, common mode choke, 17.5x14 mm (Kemet SC-02-06G)

Semiconductor

D1,D4,D5 = 1N4007, 1000 V, 1 A

LED1 = LED, green, 3 mm

MOD2 = JCE0612D24 XP Power (6 W/+/-24 V)

Other

K2 = Terminal block, 2-way, 5.08 mm

K4 = Terminal block, 3-way, 5.08 mm

F1 = Fuse holder, 20 x 5 mm, 500 V, 10 A

F1 = Cover for fuse holder 20 x 5 mm

F1 = Fuse, Cartridge, 1 A Time Delay, 20 x 5 mm

2 jumper wires!

Misc.

PCB 150464-1 v1.1

pcb-150616-1-v10-top-view.JPG (969kb)
pcb-150464-1-v10-mounted-for-phono-preamp.JPG (570kb)
150616-1-supra-2-schematic-v100.png (162kb)
Topoverlay of the pho preamplifier (PCB 150616-1 v1.0) (1089kb)
Copper on top and bottom of the phonopreamplifier (PCB 150616-1 v1.0) (966kb)
deviation-k2-k4.png (9kb)
fft-k1-k3-1khz-16avg.png (11kb)
fft-k1-k3-10khz-16avg.png (11kb)
formulas-150616-1.png (55kb)

Dr. Thomas Scherer 9 years ago

Yes great work. I'm impressed Ton! Even the PCB is a masterpiece. One can see how many aspects you thought about. Congratulations!

Dr. Thomas Scherer 9 years ago

Chaninging gain and input impedance would be the more high end solution. The circuit should have way enough potential for this. But this is a principle view. If you have this transformers you too have the oopportunity to test this out. Comparing both solutions with the same gain of your main amplifier will show you the noise difference. The sound aka technical data with transformers then can not be better than the transformers alone (datasheet). But even with transformers RIAA would be very good.

soundman111 9 years ago

Great work! Is it possible to use it with a MC System by switching gain? Or is it better to use Transformers 1:10? I have two Beyerdynamic Audio transformer 1:10 TR/BV 352.0.10.005

9 Attachment(s) 3 Comment(s)

MDV92 9 years ago

Error on schematic Supra.2 :

feedback R18 // C4 => pole 50Hz, no Zero 500Hz.

Replace 68K // 47 nF to :

1) (67K32 // 47nF) + 680R

2) 68K // (680R // 47nF)

Dr. Thomas Scherer 9 years ago

Hi MDV92, okay, the section around IC5C is a bit tricky because of the inversion respectively because the network is in the feedback path. Just have a look at the gain which is: (R17 + Z(C4||R18) / R17 a) with very high frequences IC5C has nearly unity gain. Example for 20 kHz: (6k8 + 1k7||68k) / 6k8 = 8k46 / 6k8 = 1.24 b) gain for 50 Hz pole (= RIAA 3,18 ms): (6k8 + 6k77||68k) / 6k8 = 13k / 6k8 = 1.9 nearly the 3 dB point compared to 1 kHz c) gain for 500 Hz pole (= RIAA 318 µs): (6k8 + 67k7||68k) / 6k8 = 34k / 6k8 = 5 Looks like the RIAA curve, doesn't it? Kind regards, Thomas PS: But I don't want to claim that there is no better way to make this filters. If you have any idea I would appreciate it.