JGreenslade wrote on Fri, 26 November 2010 07:44 |
John Roberts {JR} wrote on Wed, 24 November 2010 22:39 | but the simple discrete transistor preamp design has delivered adequate service in good executions like the Bozak 102DL, so your unit probably just needed some bench love. .
JR
|
It's funny you should mention the Bozak, John. Rebuilding them is a hobby of mine (I've probably rebuilt 30+ of them now and have FFTs for each one, so the FFTs are an accurate representation). See attached FFTs. Note the 2nd harmonic around -64dB... The bizarre thing being, that people often prefer them to modern preamps, which have all harmonics well below -100dB (see comparison)...
I've also attached an FFT comparing the RIAA accuracy of the Bozak against a modern preamp. To the untrained eye, it looks pretty wild... However, are you going to hear 0.6dB boost at 20k? There's a boost around 50Hz as well, which *could* explain why DJs like the Bozak (i.e. it may give kick drums a slight boost).
The other FFT shows the noise floor difference between a re-capped standard Bozak and one fitted with an external linear PSU. The Bozak's PSRR isn't great - nor is its regulation: you can see a couple of mV of ripple on a Bozak's DC rail. EMI from the EI transformer doesn't help, either.
I'm not knocking the Bozak btw - things we take for granted today such as toroidal t/formers and RIAA filter design software weren't around then (although I have a Sony preamp from the same era - TA-E88B - which has RIAA accuracy better than any modern unit I've seen; mind you, it was fiercely expensive. Again, the Sony's twin EI transformers don't aid noise levels...with an external PSU it would really kick ass. Interestingly enough, when you measure the Sony, it spits Sony's quoted 0.002% THD figure straight back out - they didn't lie!).
Justin
BTW - all measurements taken via Lipshitz reverse RIAA filter - it's very accurate.
|
I guess I was trying to be gracious about the Urie design, which from a superficial inspection looks similar to the Bozak (front end). I have never listened to a Bozak myself, but recall that it was highly regarded in DJ circles (perhaps for other reasons than absolute fidelity).
I am also trying to move beyond my phono preamp purist days where as a designer I carried the designs way past the limits of the medium. That said I can imagine a number of audible differences between preamps.
In order...
#1 input capacitance- Typical loading is 100-150pF but this must be added to phono cables and tone arm capacitance to figure total loading on the cartridge. This can result in multiple dB variation at top of audio band.
#2 LF skirt tuning- As already mentioned there was an unresolved difference between RIAA and NAB about how to handle 20Hz and below. IIRC NAB added a real pole at 7950 uSec, RIAA just left it up to the designer how to handle below 20 Hz. (originally they stopped at 30Hz but later extended it to 20 Hz). Different designers dealt with this different ways, but the result was much variation between designs at 20 Hz and the octaves below, where unfortunately there was information from out-of-round records and other mechanical input.
#3 EQ Topology- RIAA defines the top octave EQ as a real pole at 75 uSec. The popular non-inverting topology will have a zero in the response approximately a decade above 20KHz, This causes a modest error at 20kHz and more significant error above, where again RIAA is unconcerned.
#4 Distortion- The common practice of characterizing phono preamp linearity using simple THD, IMO understates the nonlinearity. Since the RIAA is delivering an almost 40 dB gain change between 20 Hz and 20kHz, IMD at HF that generates LF IM products can be far greater than THD measurements would suggest. I rolled my own two-tone IMD test using 19 and 20kHz 1:1, and found it most revealing of preamp purity. Since the 1 kHz IM product from 19 and 20 kHx was getting 20 dB MORE gain from RIAA EQ instead of simple THD that is getting rolled off by RIAA and falling GBW.
Note: with the exception of #4 the above items are expected to result in simple frequency response variations at both extremes of the bandpass expected to be audible even without blind listening tests.
I spent a couple decades exploring this as a designer. That said I stand by my original advice. IMO the OP will be well served to buy some inexpensive modern (opamp based) preamp, and replace the cheap opamps with some newer, better, off the shelf opamps. Perhaps upgrade the RIAA caps with film if they aren't already.
I will add one more suggestion. As I already mentioned MM carts are sensitive to load capacitance. Back in the day I used to sell a small add-on PCB with gold plated dip switches and 4 small caps that could be switched to dial in a range of load capacitance with 25 pF resolution. These days I would just suggest experimenting with some low and normal capacitance phono cables to tweak the high frequency loading. Note: you can add more capacitance by making the cables longer, shorter for less.
Sorry if all this esoterica is confusing to the OP but this is a design forum not audiophile advice.
JR
PS: I don't know if he was "the" design engineer, but I met one Bozak design engineer while working with Bozak on a consumer time delay product I did for them back in the '70s. He was a Swiss engineer, and seemed pretty knowledgeable from my brief conversation. As I recall he was dealing with a production problem on the 102DL that day related to device beta variance. I don't recall which particular circuit in the unit he was working on as it was unrelated to my discussion with him.