That is a myth , one of many in Audio.
But of course they are very good , and the current source helps a lot. But they are the worst in respect to power efficiency, and that is the reason I do not use it . The regulator that i use have very low noise and is more efficient.
Efficiency is very low, but this is line preamplifier and I never used shunt regulator yet, and I want to see haw that myth stand against serial regulators.
Yes, that's one way to get the voltage across the 170 larger. I also like the bypassing of the resistors with the 47uF, for lower noise, but be aware that for a fault condition on one of the outputs, say a dead short to common, the bipolar base associated will be abruptly reverse-biased and break down. This typically will result in beta degradation. So a diode connected across the b-e junction to clamp against reverse voltage is a good precaution.
To the comment about inefficiency of shunt regulators: yes, indeed so. In an eight-channel system I designed for a client, the overall power dissipation was constant. At 200W! And it was not a power amp! Someone in the audience of three (it was early in the morning on the last day of an AES convention
) frankly disbelieved this at the time.What is particularly effective for achieving a combination of low susceptibility to RF without a sacrifice of audio frequency performance: local shunt regulators near portions of the circuit that have variable current consumption with signal levels. This permits the use of ground planes where they would otherwise compromise performance. I believe Putzeys has endorsed this approach somewhere, although many have used it for years (some instrumentation I did, not audio, in the 1970s for example).
One more caveat that occurs: again, under fault conditions, the power DMOS parts can see out-of-spec magnitudes of gate-source voltages. Although zeners for protection are somewhat high capacitance, you could bias one "on" and use a small signal diode like a 1N4148 to be normally reverse-biased and only clamp a gate for a good-sized but in-spec overvoltage condition.
bcarso , do you mean using a single current source and then use various shunts near the various parts of the circuit ?. I also like ground planes.
200W and not an amplifie
Actually quite a few current sources. To be precise though, the local shunt reg approach was not followed as much with the 200W behemoth as with some other systems I've done over the years.
And if it is any consolation, from the standpoint of squandering resources, Harman R&D ran out of money, so the 200W system was never completed. It would have cost another 20k to finish, and they would have had to let a secretary go.
I don't know what they wound up using, I think some bog-standard thing from Audio Authority, but since most of the group don't think that one can hear nuances in electronics very well anyway, I'm sure they were satisfied
They are still my friends The system was to be used to compensate for loudspeaker efficiencies, so that double-blind tests using their speaker mover would have each presentation at equal SPL. The line level gains were adjusted based on previous pink noise measurements, and the system faded down and up between step changes in overall gain, with the fader circuitry completely bypassed thereafter. My goal was to do 144dB maximum balanced output signal to zero-signal noise, and it almost managed that (one channel was completed and measured). Vishay S102s were switched with mercury-wetted reeds, with various silicon ahead and behind the attenuators. I would do things a bit differently today
Actually I was not saying the trouth, I am using it in the Orion analog crossovers.
I'm more worried about the shunt ones. If the bipolars momentarily saturate or swing too much the gate-source voltage magnitudes get big. |20V| is usually the momentary maximum, or you puncture the oxide and it's all over. For continuous bias usually |10V| is about as high as recommended.
Of course if no fault condition ever occurs, you'll be fine. But it's very demoralizing when bringing a new board up and some errant condition pops one of the parts
Yes, that's exactly what I'd recommend. It may be that the zeners (without the standby biases) alone are low-enough capacitance, but with the additional diodes you're even closer to adding negligible capacitance.
![shuntreg.LAY].jpg](/community/data/attachments/305/305182-9d1b96756e794da601f3dab5d05b3780.jpg)
This is finall tested shunt regulator version. I built one channel for now, tested it with no load and with a load and it looks stable and no visible riple or high frequency oscilation.
Now I have to assemble second channeland test it with the lineamp(GainWire).
dado
Now I have to assemble second channeland test it with the lineamp(GainWire).
dado
Attachments
Dado it looks very good
How about using those supertex depletion mosfet in place of the jfet's? They have higher early voltage than the 2sk170 at those low Vds. I just do not know if they are noisier.
I will buy some of this deplection mosfets for triyng some ideas.
You can also cascode the main current sources for higher output impedance of the ccs. But your regulator is very good as it is.
How about using those supertex depletion mosfet in place of the jfet's? They have higher early voltage than the 2sk170 at those low Vds. I just do not know if they are noisier.
I will buy some of this deplection mosfets for triyng some ideas.
You can also cascode the main current sources for higher output impedance of the ccs. But your regulator is very good as it is.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.