Thanks, I was looking at the new 2 Ohm 2 mH Grados and it would seem difficult to take complete advantage of its noise floor. Your Benz at 12 Ohms and .4mH puts the trade off far out in frequency.
Last edited:
Yep, the new Grado is quite a strange MM animal. Can't imagine how they got to 2ohm and 2mH, that sounds pretty odd to me.
Power supply
I have decided to do a PCB for the raw power supply (rather than using perf board and wiring). See the attached pictures.
Testing went very well, 0.28-0.3nV/rtHz and 0.008% distortions at all frequencies and output levels up to 40Vpp. I also got only 25uV output hum (input shorted), before even throwing everything in the steel case. This is an excellent level, equivalent to 2.5nV input hum. It's actually a little better than HPS3.1 in the same stage (before case), probably because of a better layout of the input stage. But the key to low hum is the dual mono construction and the ground loop(s) handling, as described on my web site under "wiring".
I have decided to do a PCB for the raw power supply (rather than using perf board and wiring). See the attached pictures.
Testing went very well, 0.28-0.3nV/rtHz and 0.008% distortions at all frequencies and output levels up to 40Vpp. I also got only 25uV output hum (input shorted), before even throwing everything in the steel case. This is an excellent level, equivalent to 2.5nV input hum. It's actually a little better than HPS3.1 in the same stage (before case), probably because of a better layout of the input stage. But the key to low hum is the dual mono construction and the ground loop(s) handling, as described on my web site under "wiring".
Last edited:
Case closed
Before closing the lid:
The new box installed in the phono stack, ready for listening tests.
All I can tell so far: after 108dB of gain at 60Hz, zero hum, ear on grille on my 97dB horns. And also zero hiss.
Before closing the lid:
The new box installed in the phono stack, ready for listening tests.
All I can tell so far: after 108dB of gain at 60Hz, zero hum, ear on grille on my 97dB horns. And also zero hiss.
http://www.synaesthesia.ca the Low Noise Designs part was updated with HPS4.1 schematics, Gerbers, comments, etc... Measurements to follow ASAP (they were mostly posted here, anyway).
Comments welcomed, as usual.
Comments welcomed, as usual.
http://www.synaesthesia.ca the Low Noise Measurements page was updated with the HPS 4.1 measurements.
Currently, continuing the listening tests. So far, I was able to identify two significant differences to HPS 3.1: low level resolution/imaging and the vinyl noise handling.
Compared to HPS 3.1, I like better HPS 4.1 at low listening levels. It seems to have better resolution/imaging. With HPS 3.1 I can't resist turning up the volume 🙂
Pops are, to me, better handled by HPS 3.1. The entire vinyl noise appears more distinctly in HPS 4.1. HPS 3.1 has a more "soothing" effect on the vinyl noise. I guess it's related to the head amp pulse response (less overshooting in HPS 3.1) because this changes with the MC capacitive loading. 47Kohm/220pF is HPS 3.1 is audibly equivalent to 47Kohm/470pF in HPS 4.1. But also switching both to 100ohm input impedance levels the field?!
I'll invite a few friends to capture their opinions as well, then I'll draw the conclusions of this exercise. HPS 5.0 will be a new generation of Peltier cooled devices...
Currently, continuing the listening tests. So far, I was able to identify two significant differences to HPS 3.1: low level resolution/imaging and the vinyl noise handling.
Compared to HPS 3.1, I like better HPS 4.1 at low listening levels. It seems to have better resolution/imaging. With HPS 3.1 I can't resist turning up the volume 🙂
Pops are, to me, better handled by HPS 3.1. The entire vinyl noise appears more distinctly in HPS 4.1. HPS 3.1 has a more "soothing" effect on the vinyl noise. I guess it's related to the head amp pulse response (less overshooting in HPS 3.1) because this changes with the MC capacitive loading. 47Kohm/220pF is HPS 3.1 is audibly equivalent to 47Kohm/470pF in HPS 4.1. But also switching both to 100ohm input impedance levels the field?!
I'll invite a few friends to capture their opinions as well, then I'll draw the conclusions of this exercise. HPS 5.0 will be a new generation of Peltier cooled devices...
Last edited:
exiting ! You start to listen ! can´t wait to see the "cool" one. I look if i can find some gallium transistors.
HPS 5.0 is going to use BF862 in SMD. Probably the best low noise JFET available today, so sorry a P-channel equivalent doesn't exist...
You have to compare the equivalent paralleled BF862's to that FET. The gm/C is the important figure of merit.
Ok, then it is for high frequency work only.
Interfet has other interesting Fets for audio too.
Anyway, i thought that source could be interesting.
Interfet has other interesting Fets for audio too.
Anyway, i thought that source could be interesting.
Yup, last time I've checked it was $50 a pop, in quantities.
BF862, I got 200 pcs for half of that. All you would save with Interfet is SMD soldering points, paralleled BF862 for the same transconductance are, performance wise, significantly better.
FYI, the technical content of the Interfet site could use some oversight.
Like Linear Systems the data sheets are skeletons, they are clearly relying on the few folks with years of experience out there who know to look for exotic JFETS.
Like Linear Systems the data sheets are skeletons, they are clearly relying on the few folks with years of experience out there who know to look for exotic JFETS.
The point was to enable everyone to DIY something of real performance. BF862's typically equal 2SK170's on noise at 1/3 the C's. Yes there is a risk as the noise is not guaranteed, but they can be had for as little as .05$. A front end that requires $50 FETs (invariably only available as part of a group buy) is another matter entirely.
On a technical point the Interfet C's are specified at a reverse bias of 4V. Depending on junction grading this can be only 25-33% of the extrapolated zero bias capacitance (the SPICE parameter). Still 200mS at 5mA is impressive in a single package.
On a technical point the Interfet C's are specified at a reverse bias of 4V. Depending on junction grading this can be only 25-33% of the extrapolated zero bias capacitance (the SPICE parameter). Still 200mS at 5mA is impressive in a single package.
People at Revox (i know the design team very well) told me about the BF682. They got great results so you are in good company. As far as i can see from the data sheet they have between 25 and 40mS so some sorting may be needed. Idss varies a lot too, between 10 and 25mA.
- Status
- Not open for further replies.