Does anyone know of any published schematics for phono preamps with a differential input? I'm looking for solid state no transformer. I know it's not common but I've been told such an animal does exist. I'm mostly interested in whether it's done as a doubled up single ended topology all the way through or whether a differential front end followed by the RIAA. Parts count isn't too much of a concern.
Differential Input Phono Preamp
- Thread starter Gold
- Start date
Help Support GroupDIY Audio Forum:
This design was not only differential but balanced input when originally published in a Popular Electronics magazine kit article, but some guy in TX managed to patent the concept of a balanced input phono preamp and threatened to sue me, so I changed my "official" design to unbalanced to avoid the legal dance. I also sent him a copy of an old tube preamp with transformer input that was balanced input. ;D
I find it remarkable that he was able to get a patent for something this obvious, but the patent office has never been endowed with strong critical engineering judgement, and mostly focusses on black and white prior art to disallow claims without argument. Differential and balanced input topologies were never very common in phono preamps because it increases the cost of what is generally a price sensitive product (differential with modest noise requires an additional active gain stage which was pricey in the bad old days), and the benefits are fleeting, as the turntable and tonearm are engineered to support single ended wiring, with a grounding strategy where the turntable gets it's well managed clean signal ground telescoped out from the preamp.
Note: I did one later preamp kit, and returned to single legged input, not because of the patent, but balanced input was just not worth the trouble and extra complexity IMO.
That schematic above is from my P-10-MM the moving magnet version. I also sold a P-10-MC with low noise bipolar input devices (and more front end gain) for low Z moving coil carts.
Revisiting this design a few decades later, I would use more recent lower noise JFETs, perhaps a dedicated dual with tight Vgs specs. I selected JFETs for matched Vgs for my original design since Vgs mismatch increases distortion. I may have some old matched pairs sitting around, but those old JFETs I used were around 4 nV/rt Hz, so you can do a lot better these days with modern parts. The JFET I used in my later, but also old preamp design (2sk117) was 10dB+ lower noise but a single, so I would look for newer well matched duals.
http://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=423
here is thread on Wayne's forum where he investigated using a newer JFET and a ranging discussion about MM preamp design.
I invested many years into this pursuit, so feel free to ask any questions you have, but check the thread on Wayne's forum in case they have already been asked and answered.
JR
Thanks John. The reason I want to try it is to have ground isolation between the cart and the pre. In a studio there could be distance between the cart and the pre or they could be powered from separate mains circuits. I think it would avoid a lot of interfacing problems.
I don't know if we've had this exact conversation before (it sounds a little familiar), but if we look at the problem, it is that the (MM) cart is high source impedance, so even while a balanced interface should help reduce hum and interference, you still have the capacitive loading of a long cable run hanging off the high Z cart than can mess with the top octave loading and response (for MM carts).
I would be tempted to build a small preamp into the base of the turntable, and send a low Z output from there to wherever you want to go. One could even split up a design like mine, where the front end flat gain stage is located at the turntable, and the balanced differential output from that gets sent to a differential receiver for the RIAA eq and final gain elsewhere.
If you wanted to KISS, you could make the turntable gain stage flat, and just boost it up for the ride. You could then pad it down and send it to your favorite old school RIAA pre... I suspect people might have their favorite RIAA preamp they are used to.
JR
PS: AFAIK the turntable ground telescopes out from the preamp, and is connected to the turntable mains circuit ground. Over distance it seems perhaps a heavier gauge wire for the ground could be useful. Likewise a second shield wrapped outside the phono cables normal shield might reduce noise pickup over distance.
I would be tempted to build a small preamp into the base of the turntable, and send a low Z output from there to wherever you want to go. One could even split up a design like mine, where the front end flat gain stage is located at the turntable, and the balanced differential output from that gets sent to a differential receiver for the RIAA eq and final gain elsewhere.
If you wanted to KISS, you could make the turntable gain stage flat, and just boost it up for the ride. You could then pad it down and send it to your favorite old school RIAA pre... I suspect people might have their favorite RIAA preamp they are used to.
JR
PS: AFAIK the turntable ground telescopes out from the preamp, and is connected to the turntable mains circuit ground. Over distance it seems perhaps a heavier gauge wire for the ground could be useful. Likewise a second shield wrapped outside the phono cables normal shield might reduce noise pickup over distance.
+1 to that. The capacitance of even the usual 1m of cable between turntable & preamp is important. The better pickup arm manufacturers would state that explicitly. The better preamp manufacturers like JR, had facilities to adjust to suit the cartridge/turntable/arm/cables.JohnRoberts said:
But with too long a cable, you need -ve capacitance .. which is only available from the Unobtainium Capacitor Co. So turn your cartridge signal into a nice Line level balanced signal. Having it at the turntable means it can be better matched to the cartridge/turntable too.
If you're ready to KINSAA*, you can use a triaxial cable with the intermediate shield driven by an active buffer, which is as close as you can get to a negative capacitance.ricardo said:
*Keep It Not Simple At All
abbey road d enfer said:
That is how Neumann solved the problem in their PU74 module on the SAL74 drive rack. They use a pair of coaxial cables and drive those shields and then have an overall shield.
JohnRoberts said:
I think we have. I've been kicking the idea around for a while. I couldn't quite remember though.
I like that idea. You may have suggested it before.
The problem I see with doing that is that any noise picked up could be affected by the RIAA EQ. I know the point of a balanced line is to eliminate this but I'd rather not take the chance.
I was thinking I'd use three individually shielded pairs with an overall shield to go to the base station. I'd like to keep 0v and shield separate. I could tie together the individual shields for the pairs for lower impedance. One pair for +/- 15VDC and the others for audio L/R. Then keep the overall shield connected with a lift switch at one end.
Kingston
Well-known member
Gold said:
You must be working at a pretty inhospitably hostile environment. Either your studio is located in outer space or you are attacking a fly with a wrecking ball.
Kingston said:
RIAA eq has some 20 dB more gain at LF vs 1kHz so hum would get relative boost, but this is true in all RIAA preamps already.
JR
Kingston
Well-known member
JohnRoberts said:
Isn't that why we have balanced lines in the first place. 80dB common mode rejection isn't exactly difficult.
Anyway, your design looks great and combined with a balanced output/line driver chip to interface directly with the rest of the completely balanced environment it's something I'm also interested.
Indeed that would work fine (splitting up my design between front end gain stage and EQ). The extra flat gain there would lower source impedance and increase signal relative to noise for cleaner transmission.Kingston said:
I had a number of customers pursue my split approach, to use de-click processing before the RIAA eq which is mostly a LPF falling 40 dB from lowest to highest frequency.
FWIW my later single ended input kit is IMO an even tweakier or more esoteric design, but as I've gathered from decades of messing with this stuff, the cartridge termination is generally less dialed in than the RIAA EQ accuracy for most premium designs (so fix the weakest link).
Phono preamps were definitely a category where the preamp and EQ was over engineered compared to the cart and termination accuracy.
JR
JohnRoberts said:
Even if you dial in the termination the cart will be order of magnitudes worse than the RIAA network. That's why the only way to accurately check the calibration of a disk cutting system is with a Buchmann Meyer light beam. A cart is nowhere near accurate enough.
Kingston
Well-known member
JohnRoberts said:
I actually meant just adding a balanced line driver at the end. Could even do with just a faux-driver with one added opamp stage like the GSSL.
A preamp with RIAA sitting right next to a turntable with a short RCA cable is a sound approach to me. I don't believe in analog de-clicking personally. Probably does more harm than good compared to analytical DSP approach that understands between a click and a snare transient.
Also that thread where you and others chop the front-end into pieces was very interesting (NE5534 vs. FET input etc.)
Indeed for de-clicking in 2012 I would be tempted to pull it into the digital domain, but I would still pull it in before the RIAA EQ and apply that digitally after I de-click.. Applying the RIAA before, tends to smear and spread out the clicks in time.Kingston said:
I would even be tempted to capture the recording played at a lower speed. The valis music would scale down, the clicks, not so much. Once in the digital domain you could also play around with running the file backwards, real music would have a slow attack (actual decay) and only noise would have a fast attack playing backwards.
JR
PS Nowadays with SMD a phono preamp could be made quite small, while my old favorite through hole polystyrene EQ caps would be larger than the rest of the circuitry.. A flat gain stage could be really tiny, with no compromise in fidelity. While I suspect you can get NPO SMD caps that large nowadays. I really haven't been paying attention since I have long ago moved on from my phono preamp obsession.
I think you are confusing two different issues. One is the interference/ground potential difference on the long connection, which is taken care of by balancing the line.Gold said:
The other is the parasitic voltage induced in the low-level connection between the cartridge and the 1st active stage. You could make this connection fully balanced (as a standard only one channel floating, the other has its cold point connected to the shell), that would work, but you would still have two major limitations.
First, the HF response, which puts a direct limit on the cable capacitance.
Second, the benefits of balanced connections are directly related to the ratio of the common-mode impedance to the differential impedance AND the ratio of the common-mode impedance from perturbation source to conductors and differential impedance. The latter is inherent to the construction of the shell, tone-arm and turntable, so most of the times, you have no control of it. Being mainly constituted by the capacitance of the short length of unshielded wire from the cart to the connector, it is a very high impedance, but it is not very well controlled so may be seriously unbalanced; twisting the conductors insures best rejection.The differential impedance is intrinsic to the cartridge and again, apart from a rare exceptions, it rarely diverges from the 500mH + 1.5k model, which at low frequencies (< 500Hz) reduces at 1.5k, but reaches the "characteristic" 47k mark at 15 kHz.
To resume, you have to make the common-mode impedance quite high to take the benefits. Getting 60dB CMRR would suggest making the common-mode impedance > 1.5 Meg.
BUT. In fact, getting good CMRR is all a matter of making the bridge balanced, not a matter of absolute values, so you could get excellent CMRR with relatively low CM impedance and with some trimmer adjustment. Something like the InGenius priciple may be tempting, but dealing with millivolts and kilohms impedance is different and more challenging than line levels at 600 ohms.
Now, considering the whole SYSTEM, it just doesn't make sense. The solution advocated by JR makes sense. Install the preamp close to the cartridge, then YOU, not the cable, decide the cartridge loading and run hot signals to your listening position.
abbey road d enfer said:
I agree this is the right way to do it. The driven shield is the way to do it if the preamp is located far away. Even with the pre close to the cart I'd use twisted pair from cart to preamp. It would also be easier to re purpose the board for MC use.
I probably wouldn't want to put too much gain in the remote first gain stage to accommodate a wide variety of cartridge levels. I don't have the specs in front of me but there can be at least a 12dB difference in output level from MM carts. A Shure V15 is about 12dB lower than an Ortofon Concorde DJ cart. Then have a line stage in the base station for the rest of the gain.
Kingston
Well-known member
abbey road d enfer said:
What does a driven shield accomplish exactly? And what is "negative capacitance" here?
[edit]
The former was explained here as a way to turn capacitive noise coupling to differential mode and getting rid of it through CMRR. http://www.analog.com/static/imported-files/application_notes/41727248AN_347.pdf
The latter still eludes me.
Not "just" to be contrary (maybe a little), but I will register my vote against driving the shield to reduce cable capacitance, and even against balancing the cartridge output interface.
The legacy turntable/cartridge is designed to operate single ended, and I have even seen cartridges with one lead physically grounded to the housing and mounting screws which would unbalance (slightly) a balanced system interface.
The driven ground is good on paper, but adds complexity. Occam's razor has a variant for circuit design, where if a simpler design approach exists, that works without compromises, it is the better solution.
KISS...
JR
PS: The only time I ever drove a shield, was in an obscure console application where I needed to locate a mic preamp gain pot over a foot and a half away from the preamp circuitry.. driving the shield of the cable to the distant pot, helped keep the circuit stable by removing that capacitance from two sensitive circuit nodes. I might have gotten away with unshielded twisted pair for stability, but wanted the shielding for good crosstalk specs.
The legacy turntable/cartridge is designed to operate single ended, and I have even seen cartridges with one lead physically grounded to the housing and mounting screws which would unbalance (slightly) a balanced system interface.
The driven ground is good on paper, but adds complexity. Occam's razor has a variant for circuit design, where if a simpler design approach exists, that works without compromises, it is the better solution.
KISS...
JR
PS: The only time I ever drove a shield, was in an obscure console application where I needed to locate a mic preamp gain pot over a foot and a half away from the preamp circuitry.. driving the shield of the cable to the distant pot, helped keep the circuit stable by removing that capacitance from two sensitive circuit nodes. I might have gotten away with unshielded twisted pair for stability, but wanted the shielding for good crosstalk specs.
JohnRoberts said:
I agree that it could cause some confusion as it is a non standard way of doing it. Since I'd want this for lathe playback the ground isolation makes it worth it. Getting the cart quiet could make other things unhappy. The carts I've seen that tie one side of the coil to ground usually have a way to break the connection. I know the Stanton 681EEE is like that. I guess there is always a transformer...
A driven shield is used in the ATR 102 between the heads and the preamp. I'll grab a copy of the Neumann PUE74 this coming weekend and post it for a practical example.
Driving the shield to reduce capacitance is relatively straightforward. Buffer the input voltage, and drive the shield with a major fraction of that buffered voltage. Driving the shield with -6dB will roughly halve the cable capacitance. Driving the shield with -3dB will cut the capacitance to less than 1/3rd, etc.
On paper this is stable. ;D
I would be tempted to LPF this buffer (say 80kHz) and maybe even HPF since the cap reduction is mainly a top octave phenomenon.
If you balance the input you need 4 input buffers-shield drivers to do this right, and separate shielded cables for each input.
This is exceeding my KISS threshold.
JR
On paper this is stable. ;D
I would be tempted to LPF this buffer (say 80kHz) and maybe even HPF since the cap reduction is mainly a top octave phenomenon.
If you balance the input you need 4 input buffers-shield drivers to do this right, and separate shielded cables for each input.
This is exceeding my KISS threshold.
JR
Similar threads
Latest posts
-
-
-
-
Built custom channel for Soundcraft Boards (SSL like)- Latest: MidnightArrakis
-
-
