The data sheet doesn't mention it, but if you look up 2SD2226K on Rohm's website, they list it as a muting transistor. Hence, the high Vebo.
Then there's 2SD2153, with a perfectly regular Vebo of 6V, but hFE still in the 1200-2700 range (rank W) AND - if you can trust the spice model - r_bb of .47 ohm!
Attachments
A good 'conventional' (ie not Marcel's supa version) preamp with MM cartridge shows a somewhat 'whiter than pink' noise spectrum above 1kHz. A good MC cartridge with SOTA noise preamp will be 'redder than pink' above 1kHz and this is less objectionable. Common MC cartridges with SOTA preamps will also show significantly smaller 'meaningless' numbers cos they usually have high power output to go with their low voltage output
But I want to see the noise spectrum of Marcel's beast with a good MM cartridge
It should look something like this. Mind you, this is based on a calculation that treats the amplifier's current and voltage noise as white, so in reality you will see more noise at low-frequencies - but it were mainly the high frequencies you were interested in anyway - and it is, of course, without record surface noise. I haven't a clue if you regard a Shure V15-III as a good cartridge, but it is the only cartridge I have enough information about. The attached zip file contains an Excel sheet that should do the calculation, but as I use LibreOffice Calc, there could be compatibility issues.
When measured with a 600 Ω source, like Stereophile apparently does it, the LT1028 amplifier would come out of the test much better than the NE5534A amplifier and my amplifier would be the worst.
Last edited:
Somewhere in the 1,800+ posts on this thread are my circa 1980 noise spectrum measurements of a good MM preamp + cartridge compared to a MC with my humble circuit. As a beach bum today, I no longer have B&K 1/3 8ve filter set and 2307 chart recorder and even if I had, I can't get the steam to run them
I want to compare Marcel's noise performance to what I remember from those days.
His simulation puts stuff into perspective. The biggest difference between his orange 'conventional' noiseless amp and yellow 5534A appears to be at 1kHz ... about 2.3dB.
So improving a 'conventional' amp is unlikely to get you more than 2dB compared to 5534A and even less compared to NJM2068.
I've also measured LT1028 designs and also a supa $$$ by JC with similarly yukky noise in da previous Millenium. I'm not allowed to identify JC cos he's a pseudo guru on this forum and will get me banned.
The big gain is with Marcel's 'noiseless termination' which turns the 'whitish' MM noise into a nicer 'pinkish' MC type noise.
That explains Vebo and the low Rbb, it also makes high gain desirable to allow a reasonably large base resistor.
No real need for low 1/f noise though.
Interesting post.
Maybe 10 years ago, A friend got some things from an unhappy, stranded, soon to be x-wife of an engineer who had worked for the Mark Levinson Co - at that time in Middletown CT. I ended up with a lot of "these things" which included a number of parts from their original series - JC-x; ML-1; LNP-2; and other things up thru a bit of their current stuff at the time. Seems this engineer, maybe as part of a severance package, brought home stuff lying around on the workbenches and closets. Unfortunately no data tho as ML kept no files.
There were two little perf boards as part of the stash. Reverse engineered them and they are the same circuit as the patent but with 2SC1844's and 2SA981's for transistors. Unfortunately, I threw out the boards. Per the notes I took, believe the circuit was used as part of their ML-26 circa 1980. However, I found that there was also an optional head-amp module offered in the early/mid 70's around the time of their JC-2/ML-1 which may have been the actual "production version" of the patents' circuit. All the electronics in those early units were built in potted, plug-in modules. Part of the "stash" was a box of module covers and pins. The pins stuck out of the bottom of the PC boards which were then potted into the cover, with the pins sticking out.
Maybe 10 years ago, A friend got some things from an unhappy, stranded, soon to be x-wife of an engineer who had worked for the Mark Levinson Co - at that time in Middletown CT. I ended up with a lot of "these things" which included a number of parts from their original series - JC-x; ML-1; LNP-2; and other things up thru a bit of their current stuff at the time. Seems this engineer, maybe as part of a severance package, brought home stuff lying around on the workbenches and closets. Unfortunately no data tho as ML kept no files.
There were two little perf boards as part of the stash. Reverse engineered them and they are the same circuit as the patent but with 2SC1844's and 2SA981's for transistors. Unfortunately, I threw out the boards. Per the notes I took, believe the circuit was used as part of their ML-26 circa 1980. However, I found that there was also an optional head-amp module offered in the early/mid 70's around the time of their JC-2/ML-1 which may have been the actual "production version" of the patents' circuit. All the electronics in those early units were built in potted, plug-in modules. Part of the "stash" was a box of module covers and pins. The pins stuck out of the bottom of the PC boards which were then potted into the cover, with the pins sticking out.
Last edited:
I've updated my calculations with what I believe to be reasonable estimates for 1/f noise:
Theoretical noiseless amplifier: no 1/f noise
Amplifier with a 47 kΩ termination resistor that generates thermal noise, otherwise noiseless: no 1/f noise
NE5534A and noisy 47 kΩ: 1/f component of the noise current of the NE5534A taken into account, as 1/f noise of bipolar transistors is mainly noise current
LT1028 and noisy 47 kΩ: 1/f component of the noise current of the LT1028 taken into account, as 1/f noise of bipolar transistors is mainly noise current. Guessed that the 1/f corner frequency of the common-mode noise current is the same as for the differential noise current.
My amplifier: 1/f component of the noise voltage estimated, as 1/f noise of JFETs is mainly noise voltage. I could calculate the noise density at 1169 Hz from my measurements, subtracted the main known white noise sources and assumed the rest to be 1/f.
You now also see the disadvantage of JFETs: more noise voltage at low frequencies, as 1/f noise of JFETs is mainly noise voltage. This has very little impact on the weighted integrated noise due to the sharp roll-off at low frequencies of A- and ITU-R 468-weighting filters.
Theoretical noiseless amplifier: no 1/f noise
Amplifier with a 47 kΩ termination resistor that generates thermal noise, otherwise noiseless: no 1/f noise
NE5534A and noisy 47 kΩ: 1/f component of the noise current of the NE5534A taken into account, as 1/f noise of bipolar transistors is mainly noise current
LT1028 and noisy 47 kΩ: 1/f component of the noise current of the LT1028 taken into account, as 1/f noise of bipolar transistors is mainly noise current. Guessed that the 1/f corner frequency of the common-mode noise current is the same as for the differential noise current.
My amplifier: 1/f component of the noise voltage estimated, as 1/f noise of JFETs is mainly noise voltage. I could calculate the noise density at 1169 Hz from my measurements, subtracted the main known white noise sources and assumed the rest to be 1/f.
You now also see the disadvantage of JFETs: more noise voltage at low frequencies, as 1/f noise of JFETs is mainly noise voltage. This has very little impact on the weighted integrated noise due to the sharp roll-off at low frequencies of A- and ITU-R 468-weighting filters.
Thanks for this Mike. And thanks to Marcel for his updated analysis. It looks like 5534A is even closer to 'noiseless' than 2.3dB
I think with conventional 'noisy termination', SOTA noise for MM preamps is NJM 2068 closely followed by 5534A.
For even less audible noise, perhaps NJM 2068 in Marcel's Fig 6.
IIRC, Wayne's supa ultra MM preamp does Marcel's trick too.
There are less 'noisy' MMs than V15, particularly the later ADC and the Grados. The ADCs have always rivalled the V15s for 'best' MM but IMHO, not the Grados.
I've done a lot of theoretical and practical work on this in Jurassic times and have found NO evidence of ANY practical or audible effect. Got some to show?
The trick is not really mine. As far as I know, the electrically cold resistance was invented by W. S. Percival and W. L. Horwood in 1939, see W. S. Percival, "An electrically "cold" resistance", The wireless engineer, vol. 16, May 1939, pages 237...240. It is used all over the place in low-noise amplifiers for radio receivers, but you don't see it very often in phono preamplifiers.
The earliest use in phono preamplifiers I know of was described in Jean M. Hoeffelman and René P. Meys, "Improvements of the noise characteristics of amplifiers for magnetic transducers", Journal of the Audio Engineering Society, vol. 26, no. 12, December 1978, pages 935...939, see also Ernst H. Nordholt, "Comments on "Improvement of the noise characteristics of amplifiers for magnetic transducers"", Journal of the Audio Engineering Society, vol. 27, no. 9, September 1979, pages 680...681. My figure 5 is based on this.
The 1939 paper is in this document
https://worldradiohistory.com/UK/Experimental-Wireless/30s/Wireless-Engineer-1939-05.pdf
https://worldradiohistory.com/UK/Experimental-Wireless/30s/Wireless-Engineer-1939-05.pdf
This thread is from right here at diyAudio,
https://www.diyaudio.com/community/threads/allowable-dc-current-for-a-phono-cartridge.229794/
That thread is mostly conjecture and a warning NOT to use i/o caps. But even that caveat shouldn't be a problem with simple 5532 circuits ... but could be with some of da crap Golden Pinnae discrete circuits.
Similarly for the MC thread it links to.
My pontificating is with my Transducer Designer hat on .. IIRC, you can find the full story in one of Wayne's threads where I used to be active.
Probably a topic for a separate thread, but I tried it, and in the spirit of not thread-jacking will put it to rest here.
My guess that the leakage of C1, an electrolytic I suppose, is playing havoc with the measurements. Too bad I don't have a chart recorder anymore! Have also been informed that the AoE3 setup is similarly beset with issues.
Elvee has a more complex solution: https://www.diyaudio.com/community/...omatic-rbb-extractor-the-black-knight.290332/
So let it rest and stay tuned.
This is a most informative thread, and I am in awe of the encyclopedic knowledge dispensed every day!
It's the nature of DC coupling. Once the DC coupling is compensated in the circuit, you can have a signal source at a short and be noise free. When the source goes below the input z the divice is at cut off so yes you are not going to have any noise at idle.
This is also duplicated across tubes as well. One example would be a U47 mic circuit.
I'm surprised you guys still messing with 553X op amps. Since there are now so many better choices out there. Like a TLE2074 as a front end device.
Last edited: