So you wanted a speaker deflappy function. Fine. We have different goals in vinyl optimisation. My fool's errand is more foolish that yours.
Btw greetings from Sedgeford. Visiting for the weekend.
Try to get up to Brancaster and then take the Coast Road east to Wells (seals, nice beach, neat little town). There’s a very nice pub/Restuarant At Brancaster Staith called the White Horse. If you like sea food, another good one is the Crab Shack at Stiffkey.
Happy travels!
I suspect in most cases the opamp gain is rolling off before it’s a problem (older 5534 type devices) while with newer devices if there’s a zero up way beyond 20 kHz it does not seem to cause any measured errors - even if you do it directly with say a sound card. The HF post active EQ LPF is important in this regards as well.
I prefer all active because you have the best overload and noise performance - no trade-offs. If you go tube, or discrete, you might get a better result in one of the parameters, but you’d have to go to quite some lengths to get it in both simultaneously IMV (eg higher supply rails, noise)
A good opamp and 1% COG caps and MF resistors will get you the best bang for your buck by a long margin.
Not quite. I'm not arguing that you are subjectively pleased with the result you are getting, don't get me wrong. I'm merely pointing out, as others have done, that this procedure will not get you the best imaging and be all the superlatives that it is portrayed to be.
John, what about splitting the RIAA-network in 2 parts?
That means 1st gain block > 75us pole > 2nd gain block > 318us & 3180us poles (and Neumann pole) > 3rd gain block
You are giving away noise performance (second stage is amplifying front end stage noise) for overload margin (75 us turnover frequency is in the first stage).
The best solution is all active - whether you go discrete or opamp, and no matter what your supply rails are.
NB there is no Neumann pole
The best solution is all active - whether you go discrete or opamp, and no matter what your supply rails are.
NB there is no Neumann pole
Unless you consider that's part of the restauration " in its juice" of this antiquity before to expose-it in the museum ?
;-)
There were marvelous objects, the LPs, which one treated with religious precaution and respect ... their splendid covers. Before they become post stamps in CDs, then evaporate into network streams.
Nostalgia.
Last edited:
An undegenerated pair of LSK489's does about 3-4nV and is still ~90dB linear at almost 40mV peak signal, smoothly going to 1% at several 100mV. For MC use more FET's.
Overload margin IMO is overstated. A showpiece 45RPM LP from the JVC RDC series (recorded to specifically demonstrate dynamic range) has a surface noise level (with cart resonance and wobble removed) of about 8 16bit LSB's peak and the max peak on the entire side is ~23000 codes. IME the crest factor before vs after RIAA is much less a difference than one might think.
Not trying to be difficult but idea's like Bob's Vinyl Trak are fine and I seriously doubt either noise, DNR, or THD differences will have any bearing on audibility vs. the classic OPA627/37 based pre-amp (actually playing LP's of course).
I don't think we are arguing about the same thing.
I'm not arguing whether FET's or bipolars are quieter, or whether discrete or IC is better. I am simply saying that a single gain stage with active feedback around it offers the best noise and overload performance from a systems point of view. A split or passive EQ solution has trade-off's to make. With this latter approach I see higher supply rails (to meet a desired overload spec) or, if you put the HF pole in the first stage to overcome the overload issue, you then have a high gain second stage and the attendant noise penalty. Reverse it so the front end stage is high gain for the sub 500.5Hz breakpoint, and you take an overload hit at HF - ergo, you will probably need higher supply rails.
I haven't looked at the Cordell circuit, so I cannot comment on its performance re the above points - maybe I am missing a trick there, but I do DIY stuff (and commercial) so its not really an approach that lends itself to dealing with a potentially very wide range of cartridges without tweaking.
I target 20 dB overload margin as a minimum requirement and an overall gain at 1kHz of at least 100x to deliver c. 300mV output from a 3mV sensitivity cartridge. I've used a few approaches based on the all active formula, but the most successful (noise, overload and distortion) is an all active front end followed by a 3-4x flat gain stage.
Last edited:
There are several approaches to making an optimum phono stage. What we have found for the most part, it it should be composed of two gain blocks, or one super gain block and an active buffer. Every approach has its potential limitations.
The way approved by Scott can run out of headroom, because ALL the gain has to be produced by the 'super gain' stage. That means 80 dB of gain for a typical MC. 60dB for MM. Think what this means! 1mV in for MC input, is 10V out. Headroom might be 15dB.
The way approved by Richard, might have 20-30dB gain in the post gain amp so can have a corresponding increase in dynamic range. This is also true with a hybrid like the Vendetta Research, where the 75us is passive and after the first gain stage, with the other EQ in the second stage.
It is the ALL FEEDBACK, single amp phono stage that appears to have the most problems, in fact one additional that seems to be associated with the use of global negative feedback, from listening tests. This was the the type of phono stage that I first put into the Levinson JC-2. People in following years tried to show me that the 2 stage design was subjectively better, but I resisted, until about 1978, where I tried it for myself. Previously, Harmon Kardon, Electrocompaniet, and Bascom King found the 2 stage phono was better. I finally learned, and I hope that others might learn through listening as well.
The way approved by Scott can run out of headroom, because ALL the gain has to be produced by the 'super gain' stage. That means 80 dB of gain for a typical MC. 60dB for MM. Think what this means! 1mV in for MC input, is 10V out. Headroom might be 15dB.
The way approved by Richard, might have 20-30dB gain in the post gain amp so can have a corresponding increase in dynamic range. This is also true with a hybrid like the Vendetta Research, where the 75us is passive and after the first gain stage, with the other EQ in the second stage.
It is the ALL FEEDBACK, single amp phono stage that appears to have the most problems, in fact one additional that seems to be associated with the use of global negative feedback, from listening tests. This was the the type of phono stage that I first put into the Levinson JC-2. People in following years tried to show me that the 2 stage design was subjectively better, but I resisted, until about 1978, where I tried it for myself. Previously, Harmon Kardon, Electrocompaniet, and Bascom King found the 2 stage phono was better. I finally learned, and I hope that others might learn through listening as well.
Bonsai, this is where subjective listening sometimes deviates for good engineering design.
By the way, I contacted a fellow phono preamp designer who is now at the Chicago audio show (I can't spell it without looking).
I mentioned to him that we were talking about phono design here on DIY. He said that he avoids such discussions because most contributors only have a partial understanding of phono stage problems and solutions, but usually are strongly opinionated that they know more than anybody. Oh well, he is right, but I must trudge on. '-)
By the way, I contacted a fellow phono preamp designer who is now at the Chicago audio show (I can't spell it without looking).
I mentioned to him that we were talking about phono design here on DIY. He said that he avoids such discussions because most contributors only have a partial understanding of phono stage problems and solutions, but usually are strongly opinionated that they know more than anybody. Oh well, he is right, but I must trudge on. '-)
Your JC2 is hardly and example of an accurate phono stage John. I would suspect any discussions around all-feedback or passive EQ would have been swamped by the response anomalies. Granted, given the tools of the day (slide rule and less than sterling measurement gear) one could not expect better.
Each to his own. I will keep doing it the all active way, that’s for sure.
Each to his own. I will keep doing it the all active way, that’s for sure.
Last edited:
There was a journal paper, I can't locate now. They sent two gain of 50 amps, which were primitive, and 2 diff-pairs which had the emitters tied together. I might have paralleled both sides to measure, can't remember now.
At 20Hz maybe, I guess we are talking different things. I'm talking about an open-loop transconductance with the RIAA network as the load the gain goes as the RIAA. 100mV pops are no problem nothing overloads. The noise is all RTI determined by the gm. Overload margin is 30dB at 1k and noise is 3-4nV RTI where is the 10X better with a 5534? This was a fun project for folks to play with, the one I built was a few dB low on gain but that is easily fixed.
I see nothing but praise for Pure Vinyl which uses a flat pre-amp and digital RIAA. One should read their AES paper the DNR issues are overstated.
Last edited:
Last edited:
I think you need some first hand knowledge and experience of the subject matter. The subject matter being the set up procedure and results.
All the things you recommended are common things that have been done for years by many, including myself. I like the way I do it now.
Scott, obviously I overlooked the fact that you made a TRANSCONDUCTANCE gain stage, rather than an OP AMP gain stage. I, of course, do the same thing, and it DOES extend your dynamic range, because it reduces the 1KHz gain by about 20dB compared to 10Hz. You are right, your design should be OK with dynamic range. Now your challenge is enough linearity over such a range.
- Status
- Not open for further replies.