MC Step-up Xfmr and SNR

[analog_guy]

If this thread has been discussed elsewhere then let me know.

In reviewing the performance of a phone preamp under design, I am starting reach the opinion that, from an SNR point of view, there is a cartridge output value below which a step-up transformer is inherently quieter than can be achieved in a practical low-noise amplifier.

The first factor I considered is the acceptable equivalent input noise, such that when passed through the amplification chain, yields inaudible noise at the speaker. Starting with a typical +90 dB @ 1W sensitivity level and working backwards, I assumed an audibility threshold of +10 dB. For 4-ohm speakers this equates to a drive voltage of 200 uV RMS. The next step was to divide by the product of the power and line amp gain (120) which yields 1.67 uVRMS. This is equivalent to the max output noise of the phono amp.

The next step was to simulate the phono stage which yielded an equivalent input noise of 1.3 uV RMS for a flat gain of 10 over a 20-20KHz BW. When the RIAA network is connected this increased to 10 uV RMS. In other words, an SNR improvement of 10/1.67= 6.00 is required to meet the above noise requirement. The 1.3 uVRMS corresponds to 0.92 nV/sqrt(Hz), meaning that the SNR improvement would require 0.92/6.00 = 0.15 nV/sqrt(Hz). This number is lower than any I have seen.

DC gain of the phono stage was set at 60 dB, which is actually lower than is required for many low output MC cartridges. If a ~25 dB step-up transformer is included in the chain then the DC gain of the phono stage can be decreased by ~20 dB, and the noise appearing at the speaker is correspondingly reduced to ~1.0 uV RMS, which is below the 1.67 uV RMS threshold calculated above.

[analog_guy]

Since all phono cartridges have a nonzero DC resistance they also have an associated noise which may be computed by sqrt(4kTBR). This noise is further modified by the RIAA equalization curve, but in any case is a substantial, if not the dominant, factor in determining the overall amplification chain SNR.

My question has to do with the cartridge rated output voltage. Is this voltage near the maximum output or is there typically some overhead? It matters because the SNR will differ depending on the assumptions regarding the max vs. nominal cartridge output. For example, a low output MC cartridge specs 200 uV at 1.0 KHz at 5 cm/sec. What is a reasonable maximum undistorted output voltage?

[Joachim Gerhard]

50cm/sec can be cut but does not apear on any commercial record. I could imagine a 45 with halve speed mastering.
As far as i know Shure made a survay many years ago. When i am back in my lab in 2 weeks i go through my files. So theroretically output of your 200uV cartridge can be 2mV on extrem peaks.

[analog_guy]

So theroretically output of your 200uV cartridge can be 2mV on extreme peaks.[/QUOTE]


Joachim,

Thanks for the info. I recall hearing somewhere that one should allow ~20 dB headroom over the rated carridge output when designing an MC amp.

[syn08]

Originally Posted by analog_guy

So theroretically output of your 200uV cartridge can be 2mV on extreme peaks.


Joachim,

Thanks for the info. I recall hearing somewhere that one should allow ~20 dB headroom over the rated carridge output when designing an MC amp.

I would suggest at least 30dB headroom. You have to consider also the vinyl "pops", not because you want them reproduced accurately but because you need to avoid clipping. Clipping recovery in the MC stage can have a horrible impact on the overall sound. I dare to say, this is the main mechanism that significantly impacts the sound of a MC pre. Another reason is that you need to design the MC pre with some flexibility in the input level.

There's a writeup about headroom on my web site, under HPS4.1

[syn08]

Originally Posted by analog_guy

If this thread has been discussed elsewhere then let me know.

In reviewing the performance of a phone preamp under design, I am starting reach the opinion that, from an SNR point of view, there is a cartridge output value below which a step-up transformer is inherently quieter than can be achieved in a practical low-noise amplifier.

The first factor I considered is the acceptable equivalent input noise, such that when passed through the amplification chain, yields inaudible noise at the speaker. Starting with a typical +90 dB @ 1W sensitivity level and working backwards, I assumed an audibility threshold of +10 dB. For 4-ohm speakers this equates to a drive voltage of 200 uV RMS. The next step was to divide by the product of the power and line amp gain (120) which yields 1.67 uVRMS. This is equivalent to the max output noise of the phono amp.

The next step was to simulate the phono stage which yielded an equivalent input noise of 1.3 uV RMS for a flat gain of 10 over a 20-20KHz BW. When the RIAA network is connected this increased to 10 uV RMS. In other words, an SNR improvement of 10/1.67= 6.00 is required to meet the above noise requirement. The 1.3 uVRMS corresponds to 0.92 nV/sqrt(Hz), meaning that the SNR improvement would require 0.92/6.00 = 0.15 nV/sqrt(Hz). This number is lower than any I have seen.

DC gain of the phono stage was set at 60 dB, which is actually lower than is required for many low output MC cartridges. If a ~25 dB step-up transformer is included in the chain then the DC gain of the phono stage can be decreased by ~20 dB, and the noise appearing at the speaker is correspondingly reduced to ~1.0 uV RMS, which is below the 1.67 uV RMS threshold calculated above.

Your calculations are correct (although I would challenge the 10dB audibility level for white noise). Unfortunately there's more than white noise at the output. Keeping the hum below the same limits is, from a practical perspective, very unlikely. 200uV hum at the output means an equivalent (in your case) of 1.6nV hum at the input. I doubt any cartridge, arm wiring and cable can go that low in hum capturing.

[jackinnj]

Originally Posted by syn08

I would suggest at least 30dB headroom. You have to consider also the vinyl "pops", not because you want them reproduced accurately but because you need to avoid clipping. Clipping recovery in the MC stage can have a horrible impact on the overall sound. I dare to say, this is the main mechanism that significantly impacts the sound of a MC pre. Another reason is that you need to design the MC pre with some flexibility in the input level.

There's a writeup about headroom on my web site, under HPS4.1

Nice write-up -- suggest you save schematics as GIF's rather than JPEG's.

Pop's and clicks -- they cause the downstream circuitry to "hang" momentarily.

[gpapag]

Hi all
Regarding max velocity modulation in LPs and required headroom in pre design, useful data can be found here (Question #78) http://www.vandenhul.com/userfiles/docs/Phono_FAQ.pdf

Happy New Year

Best Regards

George

[Joachim Gerhard]

In that PDF 40cm/sec is the maximum, close enough to the information i had. I agree on the tick and pop problem.
When i am back home i plan to measure some recordings that are worn to find out how high those disturbing peaks are. I hope i can shine some light on that problem.

[jackinnj]

Originally Posted by Joachim Gerhard

When i am back home i plan to measure some recordings that are worn to find out how high those disturbing peaks are.

Maybe I should send some disco records from when my son was a DJ -- they have beer and potato chip residue.