Phonograph pick-up in current mode

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
G

Guest

Guest
Hi all.
Normally ceramic pick - up (in high - impedance circuit) behaves
like displacement type transducer.
Someone use ceramic pick-up in low-impedance circuit.
It behaves like velocity pick up.

Magnetic pick-up (moving magnet) in open circuit (voltage mode) behaves like velocity pick up.
Is it possible, that magnetic pick-up in very low impedance circuit - working
to the short - behaves like displacement type transducer (in some frequency band) ???

If It is possible, we can do magnetic pick-up amplifier without inverse RIAA corections?

Is there some schemo or hint to do that?, help me, please.

xvlk
 
Even a crystal pickup doesn't really reproduce the RIAA pre-emphasis curve properly. I don't think it's possible for an electromagnetic pickup to be a position-driven rather than a velocity-driven device, even in current mode. I might be wrong, of course.

In any case, it's not that hard to build a decent RIAA-compensated preamp. As always, the subtleties in the last 5% are the hard part.

Peace,
Paul
 
[quote author="pstamler"]crystal pickup doesn't really reproduce the RIAA pre-emphasis curve properly.
[/quote]
Yes, but equalizer with two resonators (for HF and LF) makes proper reproducion.
[quote author="pstamler"]
In any case, it's not that hard to build a decent RIAA-compensated preamp. [/quote]
It is hard to build. If someone want good parameters, it need two stage
and passive equalisers distributed between them,
That passive equaliser is integrator - like and it boosts 1/f noise of
input stage. Pick - up must be properly terminated - R//C input network.

Problems and problems and problems.

xvlk
 
[quote author="xvlk"][quote author="pstamler"]crystal pickup doesn't really reproduce the RIAA pre-emphasis curve properly.
[/quote]
Yes, but equalizer with two resonators (for HF and LF) makes proper reproducion.
[quote author="pstamler"]
In any case, it's not that hard to build a decent RIAA-compensated preamp. [/quote]
It is hard to build. If someone want good parameters, it need two stage
and passive equalisers distributed between them,
That passive equaliser is integrator - like and it boosts 1/f noise of
input stage. Pick - up must be properly terminated - R//C input network.

Problems and problems and problems.

xvlk[/quote]

WRT problems, bo you mean problems doing it well, or just doing it? Any trick playback electronics may not accurately mirror the cutter transfer function and be worse than modest conventional approaches.

I agree that splitting up the RIAA can simplify EQ calculation but you can get damn close with one stage approach using modern high performance opamps. BPF the signal early is always a good idea, but once you have performed a LPF active feedback electronics will do fine.

I stopped tweaking on my designs back in the '80s when the writing was on the wall for vinyl. My favorite (last) version was a discrete low noise JFET input gain stage running with a current source on it's source to increase current density. The drain dumped right into the 75usec RIAA pole forming a passive real pole (at like 2 kHz), so the circuitry was literally impossible to slew limit with a square wave input. The following single opamp stage handled the 3180 uSec and 318 uSec time constants and buffered the output. A DC servo fed correction into the input stage current source to establish DC output zero. Not very simple (I even made my own low noise/low Z power supply regulators). If I were to revisit the design 20 years later about the only thing left to improve is perhaps add a first order linearization to the input open loop gain stage, but as it was the distortion was down in the dirt on my test bench back then, so any improvement would IMO be mainly for the spec sheet.

The cartridge loading (for moving magnets) is often ignored or overlooked. I used to sell a little dip switch with 4 caps that allowed you to set cap loading in 25 pF increments up to 375 pF. I may have a few of those PCBs laying around if somebody is still interested. It used common 4 pos (gold contacts) dip switch and polystyrene caps. No preamp boards left, sorry.

JR
 
All electronic design presents problems. That's the nature of the beast; you're always juggling one thing against another, and where you compromise is your option (or your employer's).

You can actually make a more-than-decent phono preamp for moving-magnet cartridges using a 5532a per channel with a passive RIAA network in between, an OP07 for a servo amp, and a shared OPA2604 for infrasonic filtering. Will it be a perfect complement to what the cutter head was doing? Probably not most of the time, since cutter heads vary greatly. But it can provide you with plenty of good-sounding music. Bandlimit the 5532s, take care with the cartridge loading, use a servo on stage 2 to implement a 6dB/octave rolloff in the infrasonic region (the beginning of a 3rd-order filter), and sit back and enjoy.

Peace,
Paul
 
Of course an inverting amp with feedback does not have an input impedance of zero unless the open loop gain is infinite and independent of frequency.

In fact, with the proper predominately capacitative feedback the input can look like a resistor, when the amp has a single dominant pole, and this has been exploited for synthetic "cooled" terminations.
 
[quote author="Samuel Groner"]Related link: www.geocities.com/rjm003.geo/rjmaudio/diy_pho4.html

Samuel[/quote]

IMO the good part of that design is the RIAA EQ executed across an inverting opamp. I've done that before and it eliminates an inherent error in non-inverting topologies. This is a relatively minor flaw as nominal gain at 20kHz is +20dB so the deviation from a true pole continuing forever is a zero that occurs a few octaves above 20kHz, but when you're tweaking to small fractions of a dB it all matters.

I would note from the small print, that link is to a moving coil pre. Even the author doesn't suggest trying this with moving magnet. I am not comfortable predicting how that circuit would act with MC cart. The output from MC is a very small voltage but still has the typical 40 dB larger at 20kHz than 20Hz characteristic.

When I designed my MC phono pre kits back in the '80s, I used a front end gain stage variant of the hybrid mic pre similar to what's in other current threads, but with the current density of the devices dialed up for the lower source impedance (10-20 ohm). FWIW my favorite low noise transistor series (737/786) were actually developed for the MC application.

JR
 
IMO the good part of that design is the RIAA EQ executed across an inverting opamp. I've done that before and it eliminates an inherent error in non-inverting topologies. This is a relatively minor flaw as nominal gain at 20kHz is +20dB so the deviation from a true pole continuing forever is a zero that occurs a few octaves above 20kHz, but when you're tweaking to small fractions of a dB it all matters.
I think that error is easily corrected by adding a RC lowpass filter at the output, at least if you accept another stage, right?

Samuel
 
Another inverting eg phono pre was the TAA, Jung White PAT 5 mod write-up from around 1979 If I am remembering correctly.
 
[quote author="Samuel Groner"]
IMO the good part of that design is the RIAA EQ executed across an inverting opamp. I've done that before and it eliminates an inherent error in non-inverting topologies. This is a relatively minor flaw as nominal gain at 20kHz is +20dB so the deviation from a true pole continuing forever is a zero that occurs a few octaves above 20kHz, but when you're tweaking to small fractions of a dB it all matters.
I think that error is easily corrected by adding a RC lowpass filter at the output, at least if you accept another stage, right?

Samuel[/quote]

If you're willing to add another active stage adding a post or perhaps better yet pre stage with a pole up at 200 kHz is far from a bad thing... Of course when you start adding stages you have the flexibility to execute the RIAA properly as defined.

FWIW I hope nobody believes that any record cutter would have an increasing characteristic that continues that high above 20 kHz, so the uncomfortable reality may be that the one opamp stage may actually better reflect reality, but many folks make purchase decisions based on a tenth of a dB variance from some spec. Another tidbit the RIAA spec specifically avoided making any requirement beyond the nominal audio band above 20 kHz and didn't even go down to 20 Hz at first. In high end design you must balance reality with customer expectations, not always the same thing.

Of more practical interest, I supported the IEC proposal to add a LPF (1 pole) at 30 Hz extending downward to define a response down there that could be compensated for above 20 Hz in encode curves, and would insure some rejection of infrasonic garbage that those systems were notorious for. The IEC amendment never gained traction and at this point is more of a historical footnote.

JR
 
In his classic work on phono preamp design, Tomlinson Holman made a strong case for an 18dB/octave HP filter around 16Hz. I've put one in most of the preamps I've built, Bessel function, and had good results.

Meanwhile, it's worth noting that a passive EQ setup, like an inverting-opamp setup, continues rolling off at higher frequencies without reaching a plateau.

Peace,
Paul
 
[quote author="pstamler"]In his classic work on phono preamp design, Tomlinson Holman made a strong case for an 18dB/octave HP filter around 16Hz. I've put one in most of the preamps I've built, Bessel function, and had good results.

Meanwhile, it's worth noting that a passive EQ setup, like an inverting-opamp setup, continues rolling off at higher frequencies without reaching a plateau.

Peace,
Paul[/quote]

I would warn casual readers that the following will delve into decades old audiophile trivia of little present utility.

Second item first: Indeed passive EQ is capable of closer agreement to the true RIAA curve than the common single non-inverting stage and has some attraction to the specification tweaks for accuracy delivered with some misguided presumption that performing that RIAA in any feedback path is flawed.

Passive eq, preceded and followed by wide band gain does not enjoy any benefit from the RIAA EQ's 40 dB drop between 20 kHz and 20 Hz. Instead it must deliver 60 dB of wide band gain and the passive EQ will only reduce noise of the front end gain stage, not the makeup gain stage following. Input overload will no longer track the inverse RIAA response and may be impacted by how gain is distributed pre and post. These are not theoretical or practical problems but not IMO an optimal solution even given their EQ accuracy constraint due to increased distortion/noise from the output makeup gain stage.

--------

OK, now for part 1. I am a fan of Tomlinson Holman's work and agree with his judgment in general. How to deal with below 20 Hz was conveniently ignored by the RIAA, and they wouldn't even engage when the IEC proposed defining a response down there. I did some research back then trying to determine if there was some defacto standard in what the popular record cutters were applying but extremely low frequencies were not of much interest in the pre-digital, magnetic tape days.

So if there is no incentive to compliment some encode transfer function what are the other considerations. Center hole concentricity could generate some very low frequency artifacts, with warped disc flatness making noise octaves higher. So there was clearly an incentive to start reducing gain below 20 Hz. The common one opamp non-inverting topology rolled that off at 6 dB/oct probably around 10 Hz or lower so to not be more than 1 dB down at 20 Hz.

Premium designs (like Holman's) combined an under damped active 2-pole section with the one real pole to get a steeper 3 pole curve below while perhaps getting back on RIAA spec at 20 Hz. I can't get too excited about fractions of a dB down around 20 Hz, and worry even less about phase response down there. The Bozak Concert Grand speakers were about the only infinite baffle box with response anywhere near that bottom octave. Any ported box already has plenty of phase funny business in that region.

There were some issues with infrasonic noise from records, interfering with companding type tape NR. If the original source contains LF content that is below the bandwidth of the magnetic tape path, the decode expansion side chain source will not track the encode compression and there will be a phantom LF modulation caused by the missing information in the tape playback. In my companding tape NR kit, I designed in a level dependent HPF so at lower amplitude levels where warps could be a significant part of the signal envelope it rolled off the LF garbage while delivering full LF response at higher levels when the envelope was dominated by music.

I will concede that one reviewer did claim to hear my IEC pole at 30 Hz, but it wasn't exactly a blind test, and that was his main reservation with my pre (I'll take that).

Sorry if this is TMI there's more but I'm even boring myself.

JR

PS: One certainly different circuit trick that I think I remember Tomlinson doing (?) was to make a long tail pair input stage with a JFET on the (+) input side and bipolar transistor on the (-) feedback side. It probably saved some pocket change in parts cost while also delivering less noise than two JFETs of the day. Clever.. but perhaps not embraced by symmetry loving audio tweaks. The bottom line is it worked well and cost less. win-win IMO.
 
With modern chips and discrete transistors the second-stage noise problems in a passive design are no longer anything to be concerned about. Yuo can even make a quiet preamp from tubes if you're so inclined, and careful. There are advantages in the loading area: neither amplifier needs to drive excessively low impedances.

Anyway, Holman had a fascinating design for including the active filtering in the same amplifying circuits used for the preamp itself. Very clever designer.

Peace,
Paul
 
[quote author="pstamler"]With modern chips and discrete transistors the second-stage noise problems in a passive design are no longer anything to be concerned about. Yuo can even make a quiet preamp from tubes if you're so inclined, and careful. There are advantages in the loading area: neither amplifier needs to drive excessively low impedances.

Anyway, Holman had a fascinating design for including the active filtering in the same amplifying circuits used for the preamp itself. Very clever designer.

Peace,
Paul[/quote]

If may last post wasn’t already TMI, here’s more. :)

Agreed modern opamps are quite good, but perhaps another way to look at this, are the exact same opamps running at 40 dB less closed loop gain at 20 kHz going to deliver the exact same performance (60 db vs. 20 dB)? Since this will arguably be spread across 2 opamp stages making it only 20 dB more gain per stage, perhaps not a show stopper, but all things being equal they aren't.

Regarding loading in an opamp's feedback path, I was also motivated to use very high quality dielectric so biggest EQ cap I used was 8200pF polystyrene for LF pole in RIAA, not very heavy lifting for most opamps.

While headroom at HF may not be a major issue, using a feedback design it can be roughly the reciprocal of the RIAA curve, while flat 30dB+30dB pre-post around a passive EQ would offer 10 DB less headroom at 20 kHz (arguably again not a huge deal but a compromise). Parsing the gain 20/40 gets you back the headroom, but now 40 dB of wide band gain in the post stage. Sure it works but I didn't care for the tradeoffs.

JR

PS: Agreed, TH is clever as I noted in my previous post.
 
Somewhat apropos: I have a need for a dip EQ circuit with fairly high Q's. The limitation, given that I want to preserve good signal-to-noise, becomes the required swing on some of the opamp outputs. With the two topologies under consideration, a modified state-variable and a GIC, the requirement is pretty demanding.

So I am looking at which opamps will tolerate added gain and higher voltage signal swing using another discrete gain stage in the local feedback. So far sims seem to indicate that the old TL072 followed with a gain of two fast stage will tolerate use even with maximum loop gain, i.e., as a feedback integrator for example. This will allow me to roll a gain-of-about-two stage with +/- 30V swings (that limit based on available power supply rails elsewhere) and apply the composite structure wherever the opamp by itself was to be used.

Wondering if anyone has done this recently with real TL072s. I may break away from the damn computer and actually breadboard something :razz:

Anyway, such an approach if practical could facilitate the multistage RIAA synthesis by reducing overload concerns a bit, without migration totally to DOAs.
 
[quote author="bcarso"]Somewhat apropos: I have a need for a dip EQ circuit with fairly high Q's. The limitation, given that I want to preserve good signal-to-noise, becomes the required swing on some of the opamp outputs. With the two topologies under consideration, a modified state-variable and a GIC, the requirement is pretty demanding.

So I am looking at which opamps will tolerate added gain and higher voltage signal swing using another discrete gain stage in the local feedback. So far sims seem to indicate that the old TL072 followed with a gain of two fast stage will tolerate use even with maximum loop gain, i.e., as a feedback integrator for example. This will allow me to roll a gain-of-about-two stage with +/- 30V swings (that limit based on available power supply rails elsewhere) and apply the composite structure wherever the opamp by itself was to be used.

Wondering if anyone has done this recently with real TL072s. I may break away from the damn computer and actually breadboard something :razz:

Anyway, such an approach if practical could facilitate the multistage RIAA synthesis by reducing overload concerns a bit, without migration totally to DOAs.[/quote]

I liked wein bridge topology for low noise boost or cut EQ. For cut you just "weiny" out the reject band from the input leg. Opamp runs at noise gain of 2x.

Is this fixed or variable? If fixed couldn't you vary the Q in a SVF by just offsetting where you set the two integrator poles instead of pushing the BP section gain (I know it works for wider bandwidths. There may be a practical limit to how narrow you can go, and hard to make variable)?

I have put voltage gain in feedback of TL07x series with good results (to extend bandwidth of precision rectifier circuit) but I always used a cap shunt to restore unity gain at HF for stability. There probably is a little phase margin to spare but I never tested the limits of that. I've been bit by enough designs when you keep it on the page to wander too far afield.

JR
 
[quote author="JohnRoberts"]

I liked wein bridge topology for low noise boost or cut EQ. For cut you just "weiny" out the reject band from the input leg. Opamp runs at noise gain of 2x.

Is this fixed or variable? If fixed couldn't you vary the Q in a SVF by just offsetting where you set the two integrator poles instead of pushing the BP section gain (I know it works for wider bandwidths. There may be a practical limit to how narrow you can go, and hard to make variable)?


JR[/quote]

There is one fixed with different dip depths and one that must be variable depth and variable freq.

For the fixed one with different depths I just played with yet another topo, this using a bootstrapped twin t. Although due to the bootstrapping the noise gain is high, if the signal is available at a high enough level to begin with the S/N ratio compared to the GIC topo is comparable or better---this because you can swing lots of volts with only unity gain. So that may be the best way to go. Of course the twin t does require 3 capacitors of accurate value/matching.

Meanwhile just got off the horn with a co-worker who pointed out that the prescription for what all of this is supposed to do is likely to change, so I shouldn't get too carried away. Nonetheless the knowledge is applicable to other things that will likely come up in the future.
 
[quote author="mediatechnology"][quote author="Gus"]Another inverting eg phono pre was the TAA, Jung White PAT 5 mod write-up from around 1979 If I am remembering correctly.[/quote]
www.ka-electronics.com/Images/Jung-White-RIAA.pdf

I always like that one with the RIAA in the inverting stage too.

John R wrote:
IMO the good part of that design is the RIAA EQ executed across an inverting opamp.

I agree. Been thinking about trying a pair of THAT1510 and an OPA2604 for a vinyl transfer application. I still have a bunch of the precision Cs and Rs for that one too.[/quote]

I'm not in love with that 1000 uf in series with 3k but my objection is more gut than brain. IIRC Walt also got into some obscure design efforts to tweak the opamps OL gain to mirror the closed loop gain so input error term was in phase. Beyond my imagination or interest. He may have designed/published more phono preamps than me, I only had 4 that were actual products.

If I were going to knock off that particular Jung preamp I'd be tempted to execute the front end with a modern low noise Bi-fet. Recall that nominal termination is 47K with 150 pF or so. Opamp input Z needs to be high and well behaved.

Amount of front end gain probably OK. I don't like to swing Bi-fets inputs too much using NI topology, might push a few dB more gain there. I'd also be tempted to finesse out the first 1000uf cap by dc coupling the 3.3k to the RIAA EQ network but cap coupling that node to the - input of the opamp (cap actually inside loop so can be smaller, even film). Perhaps use 1M DC feedback R and tweak 330K up to dial in 3180 uSec pole.

If either opamp is precision may be possible to design out the output blocking cap too.

FWIW: I wouldn't be willing to bet that any of these "improved" topologies (mine included) would register in a controlled ABX listening test against a well executed single opamp NI design, but I enjoy the pursuit of possible flaws.

JR
 
[quote author="mediatechnology"]
I'm not in love with that 1000 uf in series with 3k.

John I think that's a 100 uF. And in that era he spec'd a tantalum which I wouldn't do.

My reasoning for trying a 1510 there is that the catridge is DC-coupled into the input and the source resistance should be fairly low. But a Bifet might be a better choice due to reduced loading. Even with 5532s it wasn't bad-sounding.[/quote]

MM cart IIRC is maybe 1.5k R and rising inductive component... not low Z IMO.

JR
 
Back
Top