Glad you are there.It should be noted I'm possibly the only nutter on this thread with sub-3ohm cartridges
Obviously we really need you.
Hans
Thanks for this aboos.
Could you and/or Hans do a noise comparison of your respective LF solutions?
___________________
There's one last factor that has yet to be looked at. This is PSR.
Could someone sim a frequency response & gain from the floating battery to the output with a selection of cartridges & their matched Duraglit gains.
Then perhaps the same thing with a lesser version like the Leach current mirror and perhaps JC's wonky virtual earth ... matching gains etc of course ..
I'm not sure how LTspice does its 'linear' response. I think you specify 1V source for these 'linear' analysis but I don't know whether this is just a convention in LTspice or if it actually tries to put 1V through the DUT.
Last edited:
Richard,Thanks for this aboos.
Could you and/or Hans do a noise comparison of your respective LF solutions?
___________________
There's one last factor that has yet to be looked at. This is PSR.
Could someone sim a frequency response & gain from the floating battery to the output with a selection of cartridges & their matched Duraglit gains.
Then perhaps the same thing with a lesser version like the Leach current mirror and perhaps JC's wonky virtual earth ... matching gains etc of course ..
I'm not sure how LTspice does its 'linear' response. I think you specify 1V source for these 'linear' analysis but I don't know whether this is just a convention in LTspice or if it actually tries to put 1V through the DUT.
See here
Small-signal model - Wikipedia
That’s how FR simulations are done.
It makes no difference wether 1uV or 1KV is apllied, it only results in a level shift.
So 1V is fine for finding the gain.
Hans
Spice linearizes the circuit around the operating point. That means that
resistances , dc voltages etc are computed and then fixed.
The 1 is just a convenient value to take dBs from. The real AC values
are assumed to be so small that they do not change the operating point.
That may be too optimistic if your circuit oscillates. Stability results have to
be taken with a grain of salt because oscillators are large signal in principle.
I.e. the amplitude grows until it clips somewhere.
resistances , dc voltages etc are computed and then fixed.
The 1 is just a convenient value to take dBs from. The real AC values
are assumed to be so small that they do not change the operating point.
That may be too optimistic if your circuit oscillates. Stability results have to
be taken with a grain of salt because oscillators are large signal in principle.
I.e. the amplitude grows until it clips somewhere.
Richard,
A lot of questions !
Maybe you could reduce your questions to just a few Cart examples.
Hans
Thanks for this aboos.
Could you and/or Hans do a noise comparison of your respective LF solutions?
___________________
There's one last factor that has yet to be looked at. This is PSR.
Could someone sim a frequency response & gain from the floating battery to the output with a selection of cartridges & their matched Duraglit gains.
Then perhaps the same thing with a lesser version like the Leach current mirror and perhaps JC's wonky virtual earth ... matching gains etc of course ..
I'm not sure how LTspice does its 'linear' response. I think you specify 1V source for these 'linear' analysis but I don't know whether this is just a convention in LTspice or if it actually tries to put 1V through the DUT.
I already made the effort to check the PSRR with a 1R, 5R and a 10R cart. It is not as shabby as one would initially think. I got values above 40 dB! (a few dB better with the higher R cartridges)
When looking deeper into this, it is quite obvious. As long as the power supply is floating, the noise from the power source is splitting in half between the both collectors and thus is eliminating as both halfes are highly correlated. As soon as the PS gets split (2x0.75V) with individual non correlated noise sorces and a ground inbetween, PSRR is reduced drastically. That may also be one additional reason why your found achieving good noise performance with floating battery supplies was much easier than with the grounded ones.
With respect to noise between the different approaches to tame the LF bumps, noise is not different at 100 Hz and above. And below there are only little differences. The highest noise contribution is coming from the input while the bumps are created at the output. Hans' approach of using large value capacitors at the bases may have a positive noise influence for the 1/f noise which is not modelled in Spice.
Hans,
there must be a difference between my circuit and yours as I get less than 1 dB reduction at 20 Hz for Rs = 1R and gain of 52.
see attached plot
View attachment Hawking_AB gain50 1R cart.pdf
Andreas,
I suppose you are using the standard 13K resistor giving a collector current of 2.8mA.
However noise can be reduced by 1.5dB with 6mA and even 3dB by using 12mA.
Doing that increases the FR reduction to resp -3dB and -6dB, as mentioned in my above posting.
Also see here Richard Lee's Ultra low Noise MC Head Amp
And be aware that this drop in FR largely depends on the Hfe of the used transistors.
With 5mF, Hfe this is no issue.
Hans
You are right - running at much higher currents requires lower base resistors.
So the exotic very low Rs carts, where every dB of noise reduction counts, require larger values for the base caps to keep frequency response linear. And on top, these caps must be of lowest possible noise not to spoil the advantages gained by higher currents.
So the exotic very low Rs carts, where every dB of noise reduction counts, require larger values for the base caps to keep frequency response linear. And on top, these caps must be of lowest possible noise not to spoil the advantages gained by higher currents.
I really think that a floating MC HA run off a 1.5V battery with 6-12 mA or more current draw is not practical. Sure, you can rig up some charging circuit or have a separate floating supply, but the dynamic range is also limited.
I did some more work on the Newton (+-15V powered 'Duraglit') and got the amplifier noise only down to 242 pico Volts/rt Hz which is the lowest out of all of them so far (10 Ohm Rs - increase Ic for lower Rs) and you have one hell of a high overload capability. The supply is +-7.5V so easy to power off a low noise active regulator where you can get output noise of c. 30 pico Volt/rt Hz
I did some more work on the Newton (+-15V powered 'Duraglit') and got the amplifier noise only down to 242 pico Volts/rt Hz which is the lowest out of all of them so far (10 Ohm Rs - increase Ic for lower Rs) and you have one hell of a high overload capability. The supply is +-7.5V so easy to power off a low noise active regulator where you can get output noise of c. 30 pico Volt/rt Hz
Until I build it and live with it I won't know but still not scared by the dynamic range.
How come 242pV/rtHz if Rs alone is 10ohm?
Hmmm... what "noise active regulator" with 30pV/rtHz are you thinking of? That would be quite outlandish in my book...
Andreas,
Just a question: 2.7mF and 3.9mF Caps are readily available for prices comparable with 100uF Caps, so do you have a problem with them ??
Hans
Bonsai,
With all respect, a moderate sized 1.5V C cell can supply 10.000mAh.
More than enough energy for at least one year listening pleasure for a 12mA Duraglit.��
Hans
P.s why should the dynamic range be limited, I thought we had dealt with that problem
Last edited:
Hans,
not at all.
It was simply trying a different approach and see what can be accomplished.
And the issue with noise is not related to the capacitor value but more on type, brand and manufacturing technologies. So this challange needs to be addressed anyway.
1. The 242 pico Volt/rt Hz is arrived at by subtracting the 10 Ohm Rs noise.
2. 30 pico Volt/rt Hz (Output) regs are shown in the back of the presentation. A big issue with these regs is ESR of the output filter caps - 1000uF and higher.
The +-15 V powered version is all together more practical on a number of fronts in my view. But, I'm not going to decry anyone's efforts on the 1.5V version - just not for me thanks.
1. Ok, so you gave up of posting noise of head amp + Rs
2. Not sure where you got the idea those are regulators, I would call them at best "power supply buffers". By your logic, they have the equivalent noise of the output cap ESR, therefore in a first approximation you would not even need low noise op amps, since the ESR will shunt the noise, anyway. You may want to check this assumption or even better build and measure some, a true regulator or even a power buffer under 1nV/rtHz flat is a very big deal to build. That's before considering the noise corner frequency which could (and usually does) lie right into the audio band around 1KHz. 0.03nV/rtHz is, in all truth, a fantasy.