I've been providing flat balanced input MM preamps since about 2015 and balanced input MC preamps since 2019. In late 2021 Hans and I worked together to further improve the MC preamp. Thank you Hans for your many contributions and simulations.
To put it simply almost all carts are floating generators. Whether the interconnection is balanced depends on how it is terminated.
Here's a diagram I originally provided for Lathe Trolls which shows several shield connection options depending on the turntable and cart case ground. Ideally the shield(s) telescope all the way from the male XLR back to the wand and headshell with the TT frame ground separate leading back to a the preamp case ground. The telescoped shields, which drop off at the TT, carry no current unless they are linked.
https://ka-electronics.com/shop/index.php?route=product/category&path=66
This is how the original GyroHead pre-preamp can be viewed wired, whereupon the network can be perceived as battery powered floating amplifier inside the shield (actually consisting of a small die-cast box) on the right half of your image. A 5 pin din cable was created to connect to the base of a Linn tonearm with dual shielded twisted pairs feeding XLR's at the GyroHead input end, seemingly in like manner to what you have indicated.
The original pre-preamp was configured as a voltage gain amplifier using low noise Zetex devices with XLR in and out. Because the network is floating an XLR to RCA adaptor can (and was) connected to the output that shorts ground pin1 to the -ve signal pin3 on the XLR output. This creates an SE output for RCA stages in lieu of going balanced.
The diagram shows the manner of connection for symmetric balanced drive. Because the amplifier only deals with low level signals inside a shielded environment and appears almost as simplistic as it can get, the network can be easily constructed on a small vero-board. As a result it seems ideally suited for the more inexperienced diyer's to construct. In examining the link from the above, some matched pairs of ZTX851 can be purchased (albeit at $25 per matched pair) that could be ideally suited to this network.
https://ka-electronics.com/shop/index.php?route=product/product&path=66&product_id=118
Thanks for the info Hierfi.
It's actually $25 for two matched pairs.
I also have ZTX851 (for when Mouser is out of stock) which can be matched in circuit.
The ZTX951 is even quieter though the improvement with most carts would be de minimis.
It's actually $25 for two matched pairs.
I also have ZTX851 (for when Mouser is out of stock) which can be matched in circuit.
The ZTX951 is even quieter though the improvement with most carts would be de minimis.
There are various ways of skinning this particular cat.
First, low noise bipolar arrays in NPN and PNP are available from THAT https://thatcorp.com/datashts/THAT_300-Series_Datasheet.pdf with single transistor en of 0.75nV/rootHz. Paralleling all four would get down to 0.38nV/rootHz, which is exceptionally quiet.
The ZTX851/STX951 are also exceptionally quiet, with Horowitz&Hill quoting in their table of low noise transistors as having 0.18/0.2nV/rootHz at 10mA Ic. That is an astonishingly low noise for a single device.
The other way of skinning the cat is to use those ZTX transistors in Richard Lee's "Duraglit" head amp. That is two complementary common base transistors. You choose Ic and load resistors to match the characteristics of the cartridge output voltage and resistance. The idea is that the gain is chosen to give 5mV output to feed whatever MM RIAA stage you have. The worst case additional noise is 3dB worse than the noise generated by the cartridge.
I can't think of any reason not to run two of Lee's design in balanced mode, although it will by definition be noisier than the 3dB of the single ended version, although it might be possible to reclaim that by adjusting the current and gain of the pair if used in balanced mode - I have not thought that through though.
First, low noise bipolar arrays in NPN and PNP are available from THAT https://thatcorp.com/datashts/THAT_300-Series_Datasheet.pdf with single transistor en of 0.75nV/rootHz. Paralleling all four would get down to 0.38nV/rootHz, which is exceptionally quiet.
The ZTX851/STX951 are also exceptionally quiet, with Horowitz&Hill quoting in their table of low noise transistors as having 0.18/0.2nV/rootHz at 10mA Ic. That is an astonishingly low noise for a single device.
The other way of skinning the cat is to use those ZTX transistors in Richard Lee's "Duraglit" head amp. That is two complementary common base transistors. You choose Ic and load resistors to match the characteristics of the cartridge output voltage and resistance. The idea is that the gain is chosen to give 5mV output to feed whatever MM RIAA stage you have. The worst case additional noise is 3dB worse than the noise generated by the cartridge.
I can't think of any reason not to run two of Lee's design in balanced mode, although it will by definition be noisier than the 3dB of the single ended version, although it might be possible to reclaim that by adjusting the current and gain of the pair if used in balanced mode - I have not thought that through though.
Hi, thanks for contribution Sawyers,
In my contemplation, post #88, I consider it is worth building dedicated MC, not just MC head to continue with MM preamp. In that case, I think 1st stage should have as much gain as overload allows (e.g. 40 to 60db gain), any comments on that?
Earlier in thread we went through ZTX's, I think it is digested, also touched upon various designs including Duraglit can, very well covered in : https://hifisonix.com/technical/mc-head-amp-circuit-compendium/
In my contemplation, post #88, I consider it is worth building dedicated MC, not just MC head to continue with MM preamp. In that case, I think 1st stage should have as much gain as overload allows (e.g. 40 to 60db gain), any comments on that?
As threads get longer, it becomes difficult to absorb every twist and turn of what is discussed and what is not. So perhaps it is not too surprising that devices get discussed more than once.
I'm aware of the link that Duraglit and endless other configurations are listed and compared in the link you give.
But I know Richard Lee's design from ancient times when we both worked for Wharfedale 33 or more years ago, and count him as a personal friend.
I'm aware of the link that Duraglit and endless other configurations are listed and compared in the link you give.
But I know Richard Lee's design from ancient times when we both worked for Wharfedale 33 or more years ago, and count him as a personal friend.
Hi Sawyers, Sure, I totally understand, sometimes when I see thread with 3000 posts, even subject is interesting I move on. Might be AI will sort bookmarking at some stage, but than again I have no idea where DiyAudio is financed from except (very few) adverts. Looked where to donate 20€ for moderators, but its not an option. As is it is fantastic and I appreciate how it works.
Lucky you with such friends! I'm sole diyer between people I know. Only professional audio industry friend is guy that until recently made software for Devialet, But funny enough he never did circuits or any hardware since school. Pro audio engineer is today turning into something else. I do know many musicians and media production people, they are for me better judges for quality of my hobby than engineers 🙂
Lucky you with such friends! I'm sole diyer between people I know. Only professional audio industry friend is guy that until recently made software for Devialet, But funny enough he never did circuits or any hardware since school. Pro audio engineer is today turning into something else. I do know many musicians and media production people, they are for me better judges for quality of my hobby than engineers 🙂
This is also just for fun:
Nice talks but after few days using all free time looking at screen, my hands are beginning to shake. Urgently I need to pick soldering iron!!!!
Uhhhh, little bit better feel now!
Figured that for this project I will need better scope probes than what I had before, so used piece of silver cable (2 twisted in shield) . This was not audio shop crazy cable, but lab equipment issue. I will solder end of wires to DUT, other end is TRS as this is what my soundcard accepts...
Nice talks but after few days using all free time looking at screen, my hands are beginning to shake. Urgently I need to pick soldering iron!!!!
Uhhhh, little bit better feel now!
Figured that for this project I will need better scope probes than what I had before, so used piece of silver cable (2 twisted in shield) . This was not audio shop crazy cable, but lab equipment issue. I will solder end of wires to DUT, other end is TRS as this is what my soundcard accepts...
Hi, today ZTX851 is 75.5 cents on Mouser for over 10 pcs, you can always diy matching, and have pile of transistors left for other things...
I'm scratching my head, almost sure that yesterday it was about 50 cent per piece....
The ZTX851 matched pairs I sell are epoxied into an aluminum cap nut. Most of the cost is labor.
The mass of the nut improves thermal equilibrium. It reduces servo activity and lowers what would appear to be 1/f noise but isn't 1/f, just LF.
The mass of the nut improves thermal equilibrium. It reduces servo activity and lowers what would appear to be 1/f noise but isn't 1/f, just LF.
This is the servo response (at the servo output not the audio output) of one channel without the transistors glued in open air:
This is the servo response (at the servo output not the audio output) of both channels with the transistors glued and the thermal "hat" installed in open air:
(Some of the noise in the above images are the Velleman scope which is noisy even with the input grounded.)
I got 83db s/n with ksc1845 and HA1457 for mm phono. The noise you get from touching it is a bad chassis ground from your amp to turntable.
Gain is one of the two reasons I am using a transimpedance termination, a connection that the GyroHead supports in that it operates open loop. This maximizes the open loop gain and trends toward neutralizing the collector/base feedback capacitance.
Feedback capacitance is viewed as problematic, particularly when capacitances are high for low noise devices, being listed as 45pF for the ZTX851 at 10Volt (being multiplied by paralleling). For devices operating at lower voltages, as 0.6V in the GyroHead, the capacitance rises dramatically. The conclusion is that the extent of changes in output voltage signal caused by input signal variance moves along a non-linear capacitance transfer function. Transimpedance termination reduces the extent the output voltage swings along the non-linear capacitance of these devices.
It has been reported that the LM394 sounds darker than I believe some SSM devices. Transimpedance termination was employed to neutralize the non-linear bulk capacitance of the LM394 by effectively terminating the collectors in a zero impedance load. The added effect is that bandwidth also increases as a result.
Uhh, it feels all right now!
Spared 1 and half hour today and quickly put head amp as my post#1, courtesy of tubecad.com together. Today I just managed to check few DC points; hefty 50mA current draw for all 4 transistors, DC voltage points ok left and right, seems it works. Tomorrow I will try to measure more.
Reason I put it together is that nobody commented on this, there is no simulation (I mentioned before, SPICE and me are not good friends, this is how I do simulation for simple things, complicated ones I torture friends and volunteers to do).
Than I had all parts (check the manufacturer on SSm2220 , I actually bought these new, you can guess my age then 🙂)
And, I need to do little practice how to actually measure so low R, V,, eN things. So far I only measured power and speakers.
Please note this is practice mini project, not idea for final design...
Managed to make first steps before I need to put kid in bed 🙂
But I did not analyze anything yet. Good thing is amp works from first, nice sines and squares, I will not publish that as its trivial.
Measurement set up, just as this first measurements:
Windows PC with million programs opened, when "serious" measurements are on, Ill clean PC
RME Firefice UCX card connected with USB. The cards itself output impedance is speced at 75ohm, way higher than MC cartridges.... first challenge!
Second challenge is to create few hundred microvolts from digital sound card, but I tried to see what will happen
I did not shield preamp, this is just first check to see if it works:
RTA 0,34 mV digital input 1000Hz
Sweep:
Noise, no input
As said, did not look in details, this is just wild first check, but looks promising to dig into it bit deeper 🙂
Good night all
But I did not analyze anything yet. Good thing is amp works from first, nice sines and squares, I will not publish that as its trivial.
Measurement set up, just as this first measurements:
Windows PC with million programs opened, when "serious" measurements are on, Ill clean PC
RME Firefice UCX card connected with USB. The cards itself output impedance is speced at 75ohm, way higher than MC cartridges.... first challenge!
Second challenge is to create few hundred microvolts from digital sound card, but I tried to see what will happen
I did not shield preamp, this is just first check to see if it works:
RTA 0,34 mV digital input 1000Hz
Sweep:
Noise, no input
As said, did not look in details, this is just wild first check, but looks promising to dig into it bit deeper 🙂
Good night all
I don't see the point in a "balanced" phono stage - given the way a cartridge is constructed. But by all means construct the circuit so that it outputs a balanced circuit - into a balanced preamp ... and then balanced power amp (if you believe 'balanced' is an advantage)!
That's the way to go, IMO! 👍
Nono, I suggest that given:
- the (pre-RIAA) signal coming off the LP is 10x higher in the treble than it is at 1kHz
- and you have to allow for another 10x for transient peaks
... 60dB gain for the 1st gain stage is waaaay too high!
I would suggest 40dB (100x) should the maximum.
Why so low? As the RIAA network has corrected the "HF/LF imbalance" ... a 40dB gain stage is not a problem! 😵
A pickup cartridge is inherently balanced. There are two independent coils, the output of each being the difference in voltage between the pins. So a differential input (not "balanced" - that is the wrong term) is entirely appropriate.
Now my differential input, balanced output phono stage uses passive EQ. There are detractors about that approach, most vocal of which is Doug Self. He maintains that such an approach has a worse SNR and overload margin than just wrapping the RIAA EQ around the input stage. He does not mince his words "Clearly a completely passive RIAA stage is a daft idea" compromising overload margin and noise.
I cannot speak for overload margin, except I d not have an audible problem with it, The measured SNR of Proteus is 80dB A weighted at 1.9V rms output.
The gain structure is the input stage gain is 40dB, passive EQ, then 44dB post EQ. Very different from Self's assumption of at least 50dB input stage gain.
That it is quiet means that at normal loud listening levels, you have put put your ear next to the loudspeaker to hear any hiss.
Now my differential input, balanced output phono stage uses passive EQ. There are detractors about that approach, most vocal of which is Doug Self. He maintains that such an approach has a worse SNR and overload margin than just wrapping the RIAA EQ around the input stage. He does not mince his words "Clearly a completely passive RIAA stage is a daft idea" compromising overload margin and noise.
I cannot speak for overload margin, except I d not have an audible problem with it, The measured SNR of Proteus is 80dB A weighted at 1.9V rms output.
The gain structure is the input stage gain is 40dB, passive EQ, then 44dB post EQ. Very different from Self's assumption of at least 50dB input stage gain.
That it is quiet means that at normal loud listening levels, you have put put your ear next to the loudspeaker to hear any hiss.
Last edited:
When I used to put my (only) "kid" to bed years ago (after some negotiations) I used to tell her she was my favourite daughter. She said "hey... I'm your only daughter". "Yes... but you are still my favourite". Also that I loved her "just as if she were a boy".
Last edited:
Daughter asks her mother; "Mama, do you think I'm too masculine?" Mother: "Noooo, not at all my son"
Hi, Thanks for input
40db for first stage I do agree to, just as minimum, not maximum.
All else I agree with you
It is not a big nono, I use 60db (x1000) gain for decades, it works superb and dead quiet. 340microV +20db (in HF) = 3.4 mV x1000 =3.4V before signal sees RIAA that attenuates it for 40db. And that is 0db record in 20,000 kHz. There is no 0db signal recorded at 20khz, tweeter would melt instantly together with all ears listening to that 🙂. Overload margin need I buy in lower frequencies where it is also not a problem.
40db for first stage I do agree to, just as minimum, not maximum.
All else I agree with you