Lol, a lot of anti-capacitor audiophool mumbo jumbo as usual.
They are. No measureble distortion precisely because the signals are so small! The rational for using the cap is to keep DC out of the MC coil.
Electrolytics do dry out, its true, but it's rare to find any piece of audio equipment that doesn't have any electrolytics in it. Replace one, replace 'em all.
This is an old wives tale. Modern electrolytics do not require a polarizing voltage for longevity. Millions of them are used this way in equipment worldwide.
You get a dull thump, nothing unusual. Bang pretty much anything in an MC head amp and you get a dull thump- there's a lot of gain there. (Valves are far more microphonic than electrolyctics).
Not sure how you arrive at this figure with nelgigible voltage across the cap... Leakage is not likely to be more than a couple of microamps.
True, but the power supply is not shown and coudl well provide a soft start.
Still, it's quieter than a JFET, even a 2SK369.
Last edited:
LOL, a lot of design-by-AP-with fingers-in-ears mumbo jumbo as usual.
This simply tells us you have never measured the extra noises from an electrolytic in this position. Go ahead, measure, and prove me wrong.
It is quite capable of being equalled by K369s simply by paralleling multiple devices, in the same way - and without resorting to such consumer-level compromises.
Anyway Merlin, we still have not seen your own design for a MC amplifier. I believe we may assume that it does not exist.
It would take rather more than than three JFETs to beat three BJTs!
I would be inclined to use an AD797 simply for ease of use and relaxed power supply requirements (I dislike 'tweaky' circuits). This can match a single K369 for noise (just).
By the way, your input JFET appears to be zero-biased. Does this not cause problems on larger positive peaks?
Last edited:
Hmmm. My 70-ies top-of-the-line, very well regarded SONY TA-E88 has 1000uF in series there....
Still sounds top-of-the-line, too.
Jan
Wouldn't that take several hundred millivolts?
Agreed that handwaving and assertion about caps is no substitute for data, either measurements, ears-only listening, or (preferably) both. Modern electrolytics are far better than caps of 30 years ago, and a loopback test on any modern sound card will confirm the vanishingly low distortion when used for coupling.
OK, here are the results of a quick and dirty test.
Measured on an AP1 at 1kHz, bandwidth 10-30kHz, 100k load. JFETs are three different ones I plucked out of my drawer at random.
Distortion was powerfully dominated by the second harmonic in the J309, but 2nd and 3rd were about equal in the J111. There were no significant changes with frequency.
The rise at low levels is the noise floor. Looks like gate-current clipping sets in around 6Vrms, which is 84mVpeak at the gate of the FET. Obviously this is enough for MC use or even MM use if you don't mind your dust clicks being distorted
An externally hosted image should be here but it was not working when we last tested it.
Vgs(off) for the three JFETs measure:
J309 = 1.68V
J112 = 3.57V
J111 = 6.67V
Last edited:
The PASS B1 and the DCB1 do +ve biasing on the gate on every alternate half cycle of the waveform.
Distortion numbers are much lower than 0.1%
Sound quality reviews are almost universally good, or better.
Last edited:
.. Obviously this is enough for MC use or even MM use if you don't mind your dust clicks being distorted
Distortion of clicks is indeed something to consider as most records have these artifacts.
The blue curve is best I presume.. because there's less distortion at the lowest level.
Or is there another way to look at the results?
I wouldn't read too much into the distortion at this point, I was just curious to see at what level the distortion shot up due to forward gate current.
The absolute distortion levels could be misleading because the three FETs produce very different amounts of gain (about x7 for the J309, less than unity for the other two!), so the true distortions may in fact be similar at similar output levels.
I would like to try out Rod's folded cascode (or a version of it), but he hasn't posted the biasing circuit for the BJT base. @Rod?
Set a current source for 7mA and the gain becomes gm x Rout (2K5 here).
An externally hosted image should be here but it was not working when we last tested it.
But how do you suppose Rod supplies the base of the BJT? I'm guessing a simple resistor divider, with a bypass cap referenced to the supply voltage (not to ground)?
I wouldn't use a CCS in place of the 286 as this would spoil the noise performance. I also dislike batteries because they eventually go flat (BTW, it would require one ~2V battery from the supply to the base, NOT a 10V battery from base to ground. The base supply must be reference to the supply voltage not ground, otherwise you get negative PSRR!) Actually I'm beginning to doubt Rod's claim about the stellar PSRR of the folded cascode; I think an ordinary cascode probably works out the same if you reference the base voltage properly.Well, I happen to own a low voltage shunt regulator (from 1.5 up to 10V).
Ask Salas for details
You could use two 4,8V lithium units.
btw leave the 286 ohm out and put the CCS directly on the semis.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.