Starts with the presumption that 5mV sensitivity = 5 mV max output. 20 dB of "linear" output over the sensitivity rating is entirely possible on a phonograph record under many circumstances not including everything from mistracking to warp output. 5 mV + 60 dB is 5V rms or 14V P-P. 50 mV turns into 140V P-P, simply not happening with conventional opamps.
However it is possible that some opamps recover from overload quicker with lower feedback.
As Bill Waslo said, if you think it sounds better that way have at it. However defending those choices beyond a personal subjective basis will be very hard.
At least someone liked my work back then. I was young and innocent when I designed the DMA and DMC products.
Didn't your buddy , Walt Jung , have an article about using
a 071 open loop for some audio circuit ?
Phono cartridges are typically rated for output with a nominal "0VU" of 5cm/sec lateral, equal to 3.54 cm/sec each channel, although sometimes this is poorly explained these days. And sometimes the spec is for 5 cm/sec each channel (3dB difference).
As 1audio has pointed out, +20dB above 0VU should be expected and accomodated linearly. The hottest clean cuts T. Holman investigated in his Audio mag article would hit +30dB, (and noise peaks higher than this must be at least clipped cleanly with fast recovery). This puts a strict limit on gain for a phono stage claiming linearity from +/- 15 volt supplies.
All good fortune,
Chris
As 1audio has pointed out, +20dB above 0VU should be expected and accomodated linearly. The hottest clean cuts T. Holman investigated in his Audio mag article would hit +30dB, (and noise peaks higher than this must be at least clipped cleanly with fast recovery). This puts a strict limit on gain for a phono stage claiming linearity from +/- 15 volt supplies.
All good fortune,
Chris
Last edited:
If you make the decision that +20VU is adequate headroom, and that your max unclipped output is +20dBV (28.3 volts peak-to-peak), and you don't do anything in your gain structure that limits output prematurely, like distributed RIAA equalization, then your max output for 0VU input is 0dBV. How much input voltage do you have midband at 0VU?
All good fortune,
Chris
In a word, no.
In practice you can get away with an active eq (the F word) with 40 dB of midband gain. High output MM cartridges probably less. And this is a gross simplification.
Last edited:
The very first thing to know is how much input voltage you're designing for. The idea of a universal phono input stage isn't practical - there is too much variation in phono cartridges' output levels. So the question remains: how much input signal are you designing for, midband at 0VU? Generalizing without this number is time wasted.
All good fortune,
Chris
There are two exceptions to my earlier generalization. If you're John Curl or Demian Martin you can design for the best cartridges of the day and assume that your clients will spring for the best, (but most of us peasants will be restricted to lesser cartridges of widely varying output.)
The other exception is DIY. That's us! But we have to know what signal we're to expect.
All the best,
Chris
The other exception is DIY. That's us! But we have to know what signal we're to expect.
All the best,
Chris
1V peak single ended (2V balanced pk - Pk) will drive 90 % of the power amps on the market to full power. Start from there and work backwards.
5mV cart at 1kHz gives 50 mV approx at 20 KHz. Add 'Holman's Hot Cut' and you get 500 mV out the cart.
Use all active RIAA when using op-amps - other topologies require severe tradeoffs wrt noise and/or OLM.
1st EQ stage gets you to circa 200 mV out nominal and after this you can add a 10 to 15 dB post EQ gain stage to bring signal levels up to 1V.
You now have > 20 dB OL margin across the audio band, lowest noise from opamps and accurate RIAA. The 5534 at less than a $ still shines even though it's a 30 year old part due to it's low input current noise.
For passive EQ, go discrete or tubed with appropriately high rails.
For ultimate in noise, JFET can get you another 8-10 dB reduction in noise over an opamp, but you have to parallel 'em and an all active topology is a good way to go. See the Syn08 and Denis Colin designs for inspiration, although they may be a bit OTT in some aspects. The BF882 is still available and its better than 1nV/rt Hz
5mV cart at 1kHz gives 50 mV approx at 20 KHz. Add 'Holman's Hot Cut' and you get 500 mV out the cart.
Use all active RIAA when using op-amps - other topologies require severe tradeoffs wrt noise and/or OLM.
1st EQ stage gets you to circa 200 mV out nominal and after this you can add a 10 to 15 dB post EQ gain stage to bring signal levels up to 1V.
You now have > 20 dB OL margin across the audio band, lowest noise from opamps and accurate RIAA. The 5534 at less than a $ still shines even though it's a 30 year old part due to it's low input current noise.
For passive EQ, go discrete or tubed with appropriately high rails.
For ultimate in noise, JFET can get you another 8-10 dB reduction in noise over an opamp, but you have to parallel 'em and an all active topology is a good way to go. See the Syn08 and Denis Colin designs for inspiration, although they may be a bit OTT in some aspects. The BF882 is still available and its better than 1nV/rt Hz
Last edited:
- Status
- Not open for further replies.