Perhaps you have forgotten that one of the major functions of a preamplifier is SWITCHING among several input jacks. Although your unit probably does not have a "tape recorder loop" that requires its own switching, your unit certainly does include SWITCHING among the (three??) option boards. You are soon going to discover that the "Star Signal" idea is going to send lots and LOTS of wires to the selector-arm of your switch / relay bank. Where, oh by the way, there is not much in the way of a ground plane.
However, Billy Gibbons reminds me that: I Might Be Mistaken.
However, Billy Gibbons reminds me that: I Might Be Mistaken.
Member
Joined 2009
Paid Member
The sketch is just a concept, a practical implementation would see each input (and other functions) tether to the common Star Signal point through a relay. Controlling these relays allows switching. This doesn't have to be more onerous than other, more traditional topologies.
Whilst I like the novelty, I'll not be adopting such a simplistic approach but will certainly borrow from it.
No plans for a ground plane, I prefer to define the routing of return currents through ground traces for the most part.
Whilst I like the novelty, I'll not be adopting such a simplistic approach but will certainly borrow from it.
No plans for a ground plane, I prefer to define the routing of return currents through ground traces for the most part.
Last edited:
Not a new concept...I think this design dates back to 1801.
https://en.wikipedia.org/wiki/Union_Jack
When I was at college my tutor Peter Platt ( ex RAF ) said something very interesting and way ahead of the trends. " That's the trouble with you Audio types. You always say in the signal path. Everything is in the signal path and is covered by Kirchoff's law ". Perter said " Kirchoff approximately says that the sum of currents entering a junction is equal to the sum of current exiting that junction ". The real version is in two halves and says node and moment in time and in German. That was 1972 and set me up for life when Star Earth's, Bus bars and solid ground plans. Always try the latter if you can not get a star to work. Self hints at the problems.
https://en.wikipedia.org/wiki/Kirchhoff's_circuit_laws
Most problems can not be solved by Kirchoff, Dead Bug gets close. Instead Guassian ( spheres ) solutions are the reality. Crudely put, call the problem a city and solve the roads going in and out as best we can do. An op amp is exactly this problem as we have little access to the city.
https://en.wikipedia.org/wiki/Kirchhoff's_circuit_laws
Most problems can not be solved by Kirchoff, Dead Bug gets close. Instead Guassian ( spheres ) solutions are the reality. Crudely put, call the problem a city and solve the roads going in and out as best we can do. An op amp is exactly this problem as we have little access to the city.
I tried to find a quote online for Gause as extended Kirchoff. My old text book Linear Circuit Analysis Decarlo/Lin P48. " Kirchoff's current law holds for closed curves or surfaces, called Gausian cuves." From my vague memory the surface was thought of as part of a sphere. Doubtless that's another thing.
Member
Joined 2009
Paid Member
Board1 Overview
Piecing it altogether, the first board should be described by the attached scheme.
This will give me a functional pre-amp all on one board. The other boards will be able to tap into the Start-Signal point, Start-Gnd point and pull their own power from the upstream regulator.
Piecing it altogether, the first board should be described by the attached scheme.
- I'll be re-using the dual rail discrete shunt regulator design from the TGM1i thread (proven) and configure it to provide +/-30V rails, about a 6V drop from the raw dc supply coming out of the LM317/337 upstream regulator.
- There will be an off-board shunt volume control sandwiched between two discrete unit-gain no-feedback Diamond Buffers, one of which feeds the output via a dc-protection circuit from the TGM10 thread.
- Feeding the input buffer are three relay-switched options, two line inputs and a phono amplifier circuit.
This will give me a functional pre-amp all on one board. The other boards will be able to tap into the Start-Signal point, Start-Gnd point and pull their own power from the upstream regulator.
Attachments
Last edited:
I often worder if the best shunt regulator is the zener. Often it would do enough.
The FET as a shunt makes sense. These ideas would adapt. TubeCad is a great site. He is not just into tubes.
https://www.tubecad.com/2007/06/blog0109.htm
The FET as a shunt makes sense. These ideas would adapt. TubeCad is a great site. He is not just into tubes.
https://www.tubecad.com/2007/06/blog0109.htm
Member
Joined 2009
Paid Member
I think you're right there Nigel, in fact the regulator I have used is really just an amplified zener diode. A diode alone has the challenge of heat dissipation, adding a transistor makes for a more convenient, should I say 'traditional', means of thermal management. The extra gain does of course make the regulator more accurate on paper at least.
I may modify my scheme to put a buffer and solid state relay (SSR) on each input rather than a combined one.
At the same time, I'm giving some thought as to the differences between a discrete diamond buffer and an op-amp buffer. I can't help thinking that the op-amp would be more transparent unless the discrete buffer is very well designed.
I may modify my scheme to put a buffer and solid state relay (SSR) on each input rather than a combined one.
At the same time, I'm giving some thought as to the differences between a discrete diamond buffer and an op-amp buffer. I can't help thinking that the op-amp would be more transparent unless the discrete buffer is very well designed.
Last edited:
Avoid opamps. An IC buffer with a little gain may be handy. Avoid home brew diamond.
If a FET is used it will be super cheap and super fast in either shunt or series. If a CCS is added to the zener reference it should be a very good regulator. A resistor added to the zener if the FET gate voltage needs to be factored in.
Here is a bit of fun I did with a NE5532 in SE class A. The idea was to drive headphones or a power amp. This idea will drive 330R at 5 Vrms if so. The red 3K9 is a bit of a silly idea that might not be so silly here. It will give a small bias to 2nd harmonic distortion. It allows one to say fully class A although with a high gain transistor like this it would be PP class A of a class AB op amp if the red 3K9 removed. I did go on to make the 220R into a CCS. I wasn't unhappy with this one. I have a hunch NE5532 likes not having to work hard even if it can. Forgive these notes as only my note book. This might give you something special with zero hard work. It would have a headphones gain setting or perhaps a gain of two for normal use. I have no idea what headphones need, my hunch is 1 Vrms for safety. Note a 30R load was used also.
Here is a bit of fun I did with a NE5532 in SE class A. The idea was to drive headphones or a power amp. This idea will drive 330R at 5 Vrms if so. The red 3K9 is a bit of a silly idea that might not be so silly here. It will give a small bias to 2nd harmonic distortion. It allows one to say fully class A although with a high gain transistor like this it would be PP class A of a class AB op amp if the red 3K9 removed. I did go on to make the 220R into a CCS. I wasn't unhappy with this one. I have a hunch NE5532 likes not having to work hard even if it can. Forgive these notes as only my note book. This might give you something special with zero hard work. It would have a headphones gain setting or perhaps a gain of two for normal use. I have no idea what headphones need, my hunch is 1 Vrms for safety. Note a 30R load was used also.
A Zener diode shunt has rather anemic regulation due to its high dynamic impedance, especially if you bias it with a resistor instead of a current source. For example, 36V input and 30V output using an 1N5256 zener and a 1.5K resistor (4mA bias), gives only -30 dB of attenuation. Amplified zeners are better (thanks to their lower dynamic impedance), TL431s are better still, and cascode current source bias is even better. For those unafraid of opamps, you can build a double-shunt regulated reference voltage using two cascaded zeners, then an opamp controlled pass transistor with voltage divider feedback from the output. More than 90dB of attenuation is easily achievable. But now you're designing instead of copying, which may not fit the philosophy of this project.
_
_
Attachments
Last edited:
In real life over and above a RC filter I found an amplified zener quite good. TL431 needs care, the graphs given are really very easy to use. There are very low and very high capacitance options with the in between not usable fo some voltage options. If you try them you will find they really tell the truth. There are some makes that claim dramatically better noise levels ( ON?? ). Here are some tests I did. LM317 with some care is almost as good. The TL431 is with a junk box slow Darlington ( TO3 ). The amplified zener here is not with CCS ( note not really worse than LM7812 although more peaks ). Again just my notes so forgive format. The TL431+ MJ3001 is 4 dB above background. The 7812 would be quiter if the 1.25V LM317. The hiss can be lost in the lower feedback arm due to a noise being bypassed to ground by using a capacitor in paralell with the lower feedback resistor that sets the gain/output voltage.
Member
Joined 2009
Paid Member
For the Shunt regulator -see attached schematic. It shows two zeners in series because of a limitation in the parts library for LTSpice but in reality it is only one diode. The CCS feeding the shunt is bog standard of course. I think the overall design is simple and effective.
Buffers-
in the left corner we have the i.c. buffer from Intersil, the HA-5002. This one isn't the only option, there are others with built-in over-current protection.
in the right corner we have a discrete buffer from RJM (http://phonoclone.com/diy-sapp.html). It is a Diamond Buffer only this incarnation uses a CFP in the output. I'm not sure a CFP is needed nor are all the parallel output devices because we're not driving headphones out of Board1. But the performance of a discrete buffer vs i.c. is what is at stake here.
who's gonna win ?
Buffers-
in the left corner we have the i.c. buffer from Intersil, the HA-5002. This one isn't the only option, there are others with built-in over-current protection.
in the right corner we have a discrete buffer from RJM (http://phonoclone.com/diy-sapp.html). It is a Diamond Buffer only this incarnation uses a CFP in the output. I'm not sure a CFP is needed nor are all the parallel output devices because we're not driving headphones out of Board1. But the performance of a discrete buffer vs i.c. is what is at stake here.
who's gonna win ?
Attachments
The LME49600 caught my eye as the sort of thing...although it is unity gain. http://www.ti.com/lit/ds/symlink/lme49600.pdf
Member
Joined 2009
Paid Member
That was a bit of a bare-bones Spice simulation I used to get a sense of it's sensitivities. I would expect to see some decent caps placed across the zener diodes to shunt as much of their noise away as possible and some decoupling on the power rails would be nice but that should be done close to the relevant circuitry. Additional caps could be used on the CCS for slow start (to ground) if this was deemed useful.
That LME49600 device looks tasty. Once again, it's a Diamond Buffer. Are we sure we can't do better with discrete ? One side-advantage is that the discrete can be designed to run off the full +/-30V rail voltage whereas the i.c. buffers will need their supply pins cascoded with a pair of transistors to drop the voltage down.
That LME49600 device looks tasty. Once again, it's a Diamond Buffer. Are we sure we can't do better with discrete ? One side-advantage is that the discrete can be designed to run off the full +/-30V rail voltage whereas the i.c. buffers will need their supply pins cascoded with a pair of transistors to drop the voltage down.
I have not used any of these buffer ICs so I am going on data sheet info. and hope. I would hope that Intersil and TI have gone to a lot of trouble to match transistors up to get low distortion figures and have gone to a lot of trouble to make low-drift, temperature compensated current sources and so on. I am very wary of op-amps but these buffers "appear" somewhat simpler and ought to be less prone to audio nasties than op-amps and don't "appear" to use loop feedback.
Can you go for less V? How much output swing are you designing for?
Can you go for less V? How much output swing are you designing for?
Member
Joined 2009
Paid Member
I was planning on the voltage to be sufficient for the NAD RIAA amp. I can scale that amp down a bit but I like to see it stay fairly generous - it's not a serious headroom issue but linearity is often improved at the higher voltages. The i.c. guys are limited by their processes however. Dropping volts isn't difficult of course and isn't a roadblock to using an i.c. solution. In fact an i.c. solution makes life easy, just spend $10 and be done with it has quite a lot to recommend it. However, if a discrete buffer could be better then it could be worth the effort ?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.