For reference and future posterity, I've loaded Richard Lee's head amp schematic and his commentary about it up on hifisonix.com - its there if you need it.
Any questions, please post them here - maybe Richard can chime in.
Richard Lee’s Ultra-Low Noise MC Head Amp
And here is the updated (03 July 2019) MC Head Amp Compendium:-
MC Head Amp Circuit Compendium
Any questions, please post them here - maybe Richard can chime in.
Richard Lee’s Ultra-Low Noise MC Head Amp
And here is the updated (03 July 2019) MC Head Amp Compendium:-
MC Head Amp Circuit Compendium
Last edited:
I've collected a compendium of 15 or so MC head amp circuits with quick sim results. Here's the link to the pdf file (c. 1.3 MB)
MC Head Amp Circuit Compendium
MC Head Amp Circuit Compendium
At a quick look, you can parallel JFETs on the page 15 circuits until the cows come home; the 68ohm source resistor will limit the noise to some 1.1nV/rtHz.
It’s quite easy to adjust the gain on his design by changing the load resistor.
Transistors are an issue because of the required complementary. There are some ultra-simple designs in the compendium that circumvent this, but noise is 3dB higher and distortion is higher as well. Something that needs a bit of dev work.
Transistors are an issue because of the required complementary. There are some ultra-simple designs in the compendium that circumvent this, but noise is 3dB higher and distortion is higher as well. Something that needs a bit of dev work.
Only differential stages come with a 3dB noise penalty, single ended complementary push-pull stages come with a 3dB gain in noise (assuming n/p transistors with the same noise).
Thanks for posting this Bonsai.
I've just uploaded another version to Yahoo MicBuilders. You have to join. I try to keep that version my most updated 'public' version.
It includes more optimisation for different cartridges and also my comments about Diodes Inc ZTX851/951 which are now the lowest rbb' BJTs I'm aware off.
Their advantage is for cartridges with DC less than 4R3 where you can run more than 3mA up to 5mA. (The older Unobtainium d786/b737 et al don't improve noise above 3mA cos their rbb' becomes dominant.)
Best of all, ZTX851/951 don't come from Ye Olde Unobtainium Shoppe
For best real life performance, it is important to match the amp to the cartridge. This holds for other designs too though they are more difficult to match.
The document also has my notes on MM noise .. all a bit moot now as in Mar this year, I got rid of my entire record collection and bits of record playing gear which I have not touched this century to the Op Shop (Oz speak for Charity Shop).
In the week that followed, I had to be physically restrained from going to the Op Shop to buy back some of my records
If you are doing a commercial design, I would look at Opto power source as in the TODO section at the end. If you do this, please let me know as I would love to have more info on these SFH206K devices.
At these noise levels, PSUs are a real pain and its easiest to measure noise on the battery versions.
I've just uploaded another version to Yahoo MicBuilders. You have to join. I try to keep that version my most updated 'public' version.
It includes more optimisation for different cartridges and also my comments about Diodes Inc ZTX851/951 which are now the lowest rbb' BJTs I'm aware off.
Their advantage is for cartridges with DC less than 4R3 where you can run more than 3mA up to 5mA. (The older Unobtainium d786/b737 et al don't improve noise above 3mA cos their rbb' becomes dominant.)
Best of all, ZTX851/951 don't come from Ye Olde Unobtainium Shoppe
For best real life performance, it is important to match the amp to the cartridge. This holds for other designs too though they are more difficult to match.
The document also has my notes on MM noise .. all a bit moot now as in Mar this year, I got rid of my entire record collection and bits of record playing gear which I have not touched this century to the Op Shop (Oz speak for Charity Shop).
In the week that followed, I had to be physically restrained from going to the Op Shop to buy back some of my records
If you are doing a commercial design, I would look at Opto power source as in the TODO section at the end. If you do this, please let me know as I would love to have more info on these SFH206K devices.
At these noise levels, PSUs are a real pain and its easiest to measure noise on the battery versions.
All I’m saying is that noise adds stochastically in the ‘push-pull’ versions like Richard’s so you get cancellation, just like you do when you parallel devices. The single ended designs don’t have this benefit unless you parallel input devices.
Last edited:
I've added a slide - slide 9 - showing a single ended design c. 16mA current draw, 0.02% distortion and 350 pico Volt/rt Hz noise. This is a refinement of the slide 8 concept with parallel ZTX851 devices. Its possible to go below 300 pico volt/rt Hz but I need to do a bit more work on it - the main thrust of these SE designs is to get the current consumption down of course.
You need to add, for all schematics, an (at least rough) dynamic range analysis. Any flat (frequency range) gain stage before the RIAA correction is prone to severely limit the whole pre dynamic range. That's in particular critical for open loop stages, that may have gains higher than optimum (for the dynamic range purposes). The gain has to be carefully balanced across the gain stages, to avoid early overloading. Call dynamic range "overload margin" if you prefer.
Also be aware that very few devices model correctly model the noise performance. Also, there are much more noise sources beyond the obvious, in particular the power supply noise injection, the electrolytics(which can also be annoyingly microphonic when in the signal path) ESR. Expect a 20%-30% degradation in noise when implementing any design that simulates great.
Agreed - these schematics are simply to explore potential ideas and get a discussion going. I've done a 10x overload test (so 20 dB) and none clip - but they really need to handle 30+ dB IMV to cover all eventualities. Job for another day . . .
Cables are also very microphonic - I suspect better quality noes with decent insulators are better in this regard. If you move them around they squeak and make all sorts of popping sounds at these amplification levels.
Cables are also very microphonic - I suspect better quality noes with decent insulators are better in this regard. If you move them around they squeak and make all sorts of popping sounds at these amplification levels.
Solid state 30+ dB overload margin is probably a stretch to design for (been there, done that) anything over 22-24dB makes sense.
IME this problem is grossly overstated. Simply taking a flat G = 100 amp into a decent field recorder will bring some reality into the discussion.
Scott, we’ve discussed this before. I don’t think the music coming off a record (or CD) demands huge overload margins and 10-14 dB is sufficient for a given cart and a given LP.
If you are designing for a commercial application then you have to cater for a bigger spread: a hot recording - add 6dB, hot cartridge add another 6dB, space for crackles and pop recovery add 3-6 dB.
That’s why 30 dB OL is required. If you don’t like 30 dB OL, then you need gain switching. Dialing down the volume pot is easier in most cases.
If you are designing for a commercial application then you have to cater for a bigger spread: a hot recording - add 6dB, hot cartridge add another 6dB, space for crackles and pop recovery add 3-6 dB.
That’s why 30 dB OL is required. If you don’t like 30 dB OL, then you need gain switching. Dialing down the volume pot is easier in most cases.