+1 to Syn
I note the LSK489 that Bob Cordell used in his Vinyltrak and Scott has used in some designs is getting hard to source already.
I note the LSK489 that Bob Cordell used in his Vinyltrak and Scott has used in some designs is getting hard to source already.
Chris: Given vinyl noise is 10-15dB above preamp noise and vinyl distortion is 100-1000 times that of the preamp I fret it not. I use batteries and Jan's silentswitcher as its quick and removes a whole parcel of potential problems. I would claim no audible benefits just piece of mind 🙂.
I'd guess that our perspectives are pretty similar. But in my dotage I'm finding better and better excuses to play with vacuum valves. That I can still see them and still solder at their scale is entirely and completely coincidental.
Everything interesting in phonographs happens where the diamond meets the PVC, but I will never knock piece of mind. Second only to perception focusing chemistry, it's the most important part of audio reproduction.
Much thanks, as always,
Chris
Nearly everything, but there are some other interesting forgotten backwaters to re-explore.
The best solution to the potential power supply noise problems is to provide a decent PSRR in the gain stage(s). Gain stages like in the Blowtorch, Vendetta SCP-2, etc... with virtually zero PSRR require more unobtanium to build low noise power supply buffers. Not that low noise bipolar transistors would not be as good...
In general, it is much difficult to build a very low noise power supply (while keeping the other parameters unaffected) than providing a decent PSRR in the gain stage(s), although the latter is not free in terms of complexity. The best discrete Jung regulator I was able to build was around 5nV/rtHz. The best TI and LT/Analog ultra low noise IC regulators are significantly less noisy, and the whole performance make any discrete regulators simply moot.
I measured a batch of 25 bf862 and another of 25 bsp135 with about 15 different resistor values for autobias and the results were very close, almost identical in both separate batches.Just out from the factory...no hand matched"...It took me two full days to do it but in the end the result was clear.
TI
TPS7A47xx positive
TPS7A33 negative
Slightly worse noise compared to the LT/Analog regulators.
TPS7A47xx positive
TPS7A33 negative
Slightly worse noise compared to the LT/Analog regulators.
My impression is that the LSK/LSJ supply is kept to a minimum, to drive/keep the prices up.
When you look at the noise levels, make sure you scroll to the graph with the large Cset (22uF). A number of the curves are with 4.7uF.
The benefits of going single ended wrt noise have been discussed in quite a few places on this forum. You have confirmed what I understood - and what I read years ago in a nat semi app note where they turned one of the input transistors in the LTP off to get lower noise (it was for an RIAA or NAB tape head preamp IIRC in the Audio and Radio Handbook). Going complementary only adds complexity and especially so when the devices are not as tightly matched as those from a single type number.
As to the distortion, -110 dB below the output (which is at c. -30 dBV) seems ok to me (disclaimer: simulation) when the LP distortion is orders of magnitude higher. Ditto noise.
Bonsai, it was probably National's LM387 Low Noise Preamplifier IC, datasheet attached. They omitted half of the LTP for better noise performance.
This is also discussed pretty thoroughly in the book you are holding in the photo within your "avatar" here on diyAudio, it appears in the upper left of each posting you create.
_
Yes. Let me name just one... the equality of the pos and neg voltage numbers. It is strange that great care to make each supply be exactly the same when it doesnt really matter much IMO. What does matter is they both stay the same abs value change under dynamic conditions.
It doesnt matter if one is +15 and he other is -14. (except clipping which I am ignoring as not normal).
What does matter is they stay the same. If one side goes down, that the other side goes down same amount. ETC.
Master-Slave supplies will do this. Also, my preferred is the shunt push-pull output. Fast in sourcing and sinking.
One of the best places to improve its supply is in the power amp... esp its OPS. These plus and minus supplies are unregulated and exhibit a lot of diff between the rails under dynamic loading.
Another solution I introduced several decades ago and again more recently as 10-15 years ago (In Monster Products Power Amps) is starting to gain traction is shown in this simple SPICE model of back in the day when i worked at LLNL.
View attachment Bridged C Reg.pdf
View attachment Bridge C Reg-2.pdf
View attachment Bridged C Reg 3.pdf
Here I switch in a low z load (2 Ohm) on one side of the PS and look at the other side.
Notice what happens when this proof of concept/idea has a large enough Cc bridged across the PS rails.
THx-RNMarsh
Last edited:
Whoa, that was completely inaccurate. I shouldn't rely on fading memory, sorry for the error.
It was the LM381 (ends in 1) which offered the half-LTP option for superior noise. You get that by shorting pin 2 to pin 3. Correct datasheet attached below.
_
Last edited:
Stereo Power Amplifier SE-R1 | Reference Class R1 Series | Technics US
Interesting techniques or marketing blurb? 🙂
Interesting techniques or marketing blurb? 🙂
You could say the same about InterFet, and then there is the tiny University associated fab in Utah (IIRC) that makes FET's for nuclear instrumentation, no samples $250 each in die form.
At this point I don't think there is any large application for low noise JFET's, microphones and other large volume applications want low charge uncertainty which rules out the big input C.
- Status
- Not open for further replies.