Addressing John Curl's concerns on low noise designs

Status
Not open for further replies.
Jump to Latest
#701
Re: Re: HPS measurements

Sigurd Ruschkow said:
Ovidiu - good work!

The noise data you show are at 1kHz if I understand the text correctly,
but it would be more interesting (and easier to compare with say Stereophiles measurements) I think to get a SNR value
with a 20-20k BW and both flat and A weighted values re 0.5mVrms.
Could you maybe use a soundcard to make such measurements?
Click to expand...

I've added some calculated values yesterday - the S/N ratio re: 1mV 60dB gain, unweighted 20-20KHz, is 87dB. A weighted, don't have a final number but it's certainly better than 95dB. Substract 6dB for re: 0.5mV.

I can measure directly the S/N (and will do that ASAP, for the moment I'm busy listening music 🙂 ) but I'm not expecting any significant differences to the calculated values (they actually will be for sure better by some 0.5-1dB), because the corner frequency is relatively low (around 1KHz for HPS 3.1, measured, as mentioned on the web site, before the RIAA correction and 50KHz pole were installed). Comparing with the stereophile data won't be easy, they don't seem to have a consistent method of evaluating the S/N.
 
#703
john curl said:
I would be careful about adding the 50KHz pole. Often, feedback designs always have a pole, but the reasons for a 50K pole can be disproven by actual understanding of the recording process and the natural resonance of MC cartridges.
Click to expand...

That pole is outside the FB loop (therefore independently and very well defined, the closed loop pole is at 1MHz), but otherwise you are right. That pole was and still is an endless point of debate. I've chosed to add it, as in most of the RIAA stages that I've seen lately (including the $29,000 Boulder 2008 phono preamp which, btw, has measured only -90dB crosstalk in midband, compare to -107dB in HPS 3.1 🙂.
 
#704
Re: Re: Re: HPS measurements

Ovidiu - please also have the RIAA curve fully operational, ie do have all the RC components for the RIAA time constants installed.
Then we can all compare numbers with other designs.



Sigurd

syn08 said:


I've added some calculated values yesterday - the S/N ratio re: 1mV 60dB gain, unweighted 20-20KHz, is 87dB. A weighted, don't have a final number but it's certainly better than 95dB. Substract 6dB for re: 0.5mV.

I can measure directly the S/N (and will do that ASAP, for the moment I'm busy listening music 🙂 ) but I'm not expecting any significant differences to the calculated values (they actually will be for sure better by some 0.5-1dB), because the corner frequency is relatively low (around 1KHz for HPS 3.1, measured, as mentioned on the web site, before the RIAA correction and 50KHz pole were installed). Comparing with the stereophile data won't be easy, they don't seem to have a consistent method of evaluating the S/N.
Click to expand...
 
#705
Re: Re: Re: Re: HPS measurements

Sigurd Ruschkow said:
Ovidiu - please also have the RIAA curve fully operational, ie do have all the RC components for the RIAA time constants installed.
Then we can all compare numbers with other designs.



Sigurd
Click to expand...

Huh?! They are installed and all measurements on the web site are with the RIAA network in place. I did some measurements before installing the RIAA network (like the closed loop bandwidth and phase, the noise corner frequency, etc... but these are not shown anywhere (only mentioned).

If you can find a consistent way to compare my measurements with the stereophile hodge-podge, let me know.
 
#709
This is not the problem with the 50KHz pole. Anyone can do it. We had a 60KHz pole in the first JC-2 phono stage. The problem is almost 1dB boost at 2OKHz, that has already been fixed by underdamped 2 pole response of the disc driver amp. Of course, IF you can make your own disc cutter system, then everything is OK.
 
#712
Steve Dunlap said:
What sort of difficulties are you encountering?
Click to expand...

Low input current and zero offset at the same time, without crazy device matching. Theoretically, you can't have both without impacting the noise, so some trade is required. Trying to find an optimum (not even sure if there is one).

One option to avoid this problem is to use a floating input stage (and this will provide zero current through the cartridge) and do the offset compensation in a second stage. But I don't want to use batteries for the floating stage, and I still have to figure out a floating power supply (which is though AC coupled to/through the mains, hence some sort of ideal mains garbage collector).

Looking to find how much current should be allowed through a cartridge, anybody seen any reference about? Is 1uA to much for a MC cartridge? Perhaps 100nA?
 
#714
I would chose Ohm's law to check how much the DC off-centers the assembly. Electrically a cart isn't that shy -- but it depends on the thickness of the coil wires, that vary from cart to cart. Dynavectors are said to have very thin wire, for instance, and a broken coil wire is not that seldom with dynavector carts. I have no numbers, though.

I think it is a major feature of FET-input stages to provide that '0'...

Rüdiger
 
#715
scott wurcer said:
You're probably killing yourself on the DC current in an MC coil. Obsession will always lead you to 0 as the only answer, it just comes with the territory.
Click to expand...

Probably so. To my surprise, the MC current was never measured in Stereophile, it doesn't seem to be a quality criteria.

I would still like to find some references/comparative data about...
 
#716
syn08,
consider this outstanding topology:
 

Attachments

  • prepre_c.gif
    prepre_c.gif
    10 KB · Views: 476
#718
Lumba Ogir said:
syn08,
consider this outstanding topology:
Click to expand...

Thanks, but I don't find anything outstanding in this topology. To add insult to injury, any required manual trimming is a major turn off for me. I'd rather handmatch devices.

Coincidentally, the Hitachi devices (being ion implanted) are incredibly matched out of the tube. Out of 150 2SC2457, 144 have beta between 417 and 423. The problem is matching with the pnp devices, which are 324-331.
 
Status
Not open for further replies.
Top Bottom