Regulated Power Supplies

[Haselsh1]

I shall soon be looking for a new power amplifier and they may well be:

Croft Series 7
Croft Series 7 R

As I understand it, the R has a fully regulated power supply whilst the other does not.

How important is a regulated power supply and why does it double the cost of the amplifier...?

[dowser]

Is it a regulated supply in the Croft hybrid series? In Glenn's earlier models you got more and better supplies as you move up the chain. The actual stage designs from Glenn's early stuff I have seen are pretty straight forward, the complexity is in the power supplies - adding more increases transformer, rectifier, smoothing components - which tend to be the big/heavy/expensive parts. But...I have not actually looked at Glenn's Series 7 stuff - would love to, but out of my price range :)

Richard

[Firebottle]

How important is a regulated power supply and why does it double the cost of the amplifier...?

A very good question Shaun. I can only see regulation of the valve input stage plus an additional heatsink and a bigger mains transformer for the FET output stage :scratch:

Regulation of the valve stage will give improved linearity but there are better ways of doing it IMHO.
I'm about to determine the cost of my AXOLOTL power amplifier now the circuit is finalised.

:)

[lurcher]

Regulation of the valve stage will give improved linearity but there are better ways of doing it IMHO.

There are so many more things that it can do than that, I would suggest that linearity is the least of them.

A friend of mine who designs measurement instrumentation once said "people would understand if they relabelled the circuit and where it said power supply, replace that with INPUT".

[Firebottle]

A very good point Nick, I like your friends interpretation :)

It's not lost on me but also depends on the PSRR of the circuit of course.

[StanleyB]

When it comes to the output stage, a clean and noise free power supply is 100x better than a regulated one. Most audio power amp designers make their output stage capable of handling a 5 to 10% variations in the power supply. The output of the amp is not dependent on the power supply if the power rail is above the safe margin before the onset of clipping under the most difficult load condition, but on the feedback loop gain of the circuit.

[Arkless Electronics]

There are so many more things that it can do than that, I would suggest that linearity is the least of them.

A friend of mine who designs measurement instrumentation once said "people would understand if they relabelled the circuit and where it said power supply, replace that with INPUT".

Spot on!

[Arkless Electronics]

I shall soon be looking for a new power amplifier and they may well be:

Croft Series 7
Croft Series 7 R

As I understand it, the R has a fully regulated power supply whilst the other does not.

How important is a regulated power supply and why does it double the cost of the amplifier...?

This is exactly the sort of fundamental and very important area that audiophiles should be chewing the fat over.... but as it's technical it will probably get little attention.... unlike if it was about cables.... but don't get me started on that one!

I can't find much info relating to the Croft 7R but IMO a regulated supply (or several, depending on the application) can give a vast improvement to the sound quality. Some, as can be seen from replies above, have their own take on this.
The type and implementation of a regulated supply is crucial though and the design of a really good one is often more difficult than that of the amplifier itself!
Valved regulated supplies of the type Croft uses in pre-amps (I would guess the same would be used in the 7R but can't find any info) are not worth bothering with IMO as they are total crap and not much better than no regulated supply! Why? because although it is possible to design a really good regulated supply with valves it would be very expensive indeed to do the job properly. The regulator would be maybe 10 x as complicated and expensive as the pre amp circuitry itself to do the job properly. Solid state lends itself to this application much more easily and cheaply and, at the end of the day, even if money was no object a valved regulator could not really match the performance of a solid state one, which would be a fraction of the price.

In effect a regulator is an amplifier but put to another use, hence yes they add greatly to cost.

Sorry if the above is sweeping statements and generalisations but to address this technically would mean only the likes of Firebottle, Lurcher etc would be likely to join in the debate....

[Haselsh1]

Jez, thank you so much for taking the time to type a response to my query. I would like a Croft Series 7 R and fully understand that a regulated power supply must be an 'upgrade' over a non regulated supply. I think that in the Croft power amps the power supply is almost totally solid state with only the signal path having valves. This all leaves me in quite a dilemma but fortunately I have a couple of months to think about it.

[lurcher]

Spot on!

One of the friends colleagues also said when asked how he would design a audio amp: "first I would design a low noise wideband DC coupled amplifier, then connect a low noise voltage reference to the input of that, then use the output of that to power another similar amplifier that actually drove the loudspeakers".

Over the years I find my designs become carefully crafted power supplies with simple output stages added after them. As Jez said, not all regulators are created equally, and I find different types are better suited to different tasks (who would have guessed it).

Warning. If anyone finds me at a show or bake off with a drink in my hand, don't ask me about power supplies unless you want a long conversation involving arm waving and looking for a notepad to sketch on.

[Arkless Electronics]

One of the friends colleagues also said when asked how he would design a audio amp: "first I would design a low noise wideband DC coupled amplifier, then connect a low noise voltage reference to the input of that, then use the output of that to power another similar amplifier that actually drove the loudspeakers".

Over the years I find my designs become carefully crafted power supplies with simple output stages added after them. As Jez said, not all regulators are created equally, and I find different types are better suited to different tasks (who would have guessed it).

Warning. If anyone finds me at a show or bake off with a drink in my hand, don't ask me about power supplies unless you want a long conversation involving arm waving and looking for a notepad to sketch on.

It looks like we are reading from the same hymn book on this Nick!

[Light Dependant Resistor]

Regarding regulated supplies, history would show devices for regulation
are very numerous. In the same way oranges and boysenberries are fruit,
there are very common, and not so common ways of going about this.

Good electronics design usually boils down to dividing everything into
smaller and smaller segments, until errors for the most part are
eliminated, but a good saying I recall Ben Duncan saying, as one
door opens many below it, are flung open wide too. So discover one
small portion that fixes a issue and many more await discovery
improving further for those inclined to not give up and look.

So it is with regulated supplies,

But why regulation ? It is all about controlling loads, an example of a large and
complex load is a loudspeaker, sadly nature is not kind to us when
copying ability to recreate actual sound audio accurately – requiring every nuance
of engineering skill.

The most common speaker is a unique combination of magnetism
and a coil, involving a paper or plastic cones, and a crossover network.
Involving capacitance, inductance, resistance and all of
the impedance products those three combining provide, it is a difficult
load. It's success is in its ability to reproduce low, mid and
high frequencies with reasonable accuracy at reasonable
cost, the last part has very much guaranteed its continued success.

The interested reader might want to branch off and
also explore the zobel network, it is provided on
amplifiers not for the ability to assist loudspeaker
driving, rather simply to assist driving the speaker lead.
attached to the amplifier. Another zobel at the speaker
end of that lead and a third related to the speakers voice coil
should be provided but usually are not. See post 17
http://www.diyaudio.com/forums/multi-way/190300-best-place-put-zobel-network-2.html

The less common speakers are planar types like electrostatics
and there are good reasons for these as they can and do
reproduce mid and high frequencies with greater accuracy.
The best example being the Quad ESL57, a lovely subject in itself.

Regulation of power supplies is somewhat essential for good audio,
although valve amplifier get away without too much because their
voltage rails are very high typically 300 or more volts and they
invariably importantly, (the exception being OTL - output transformer-less designs )
have a known type of load they are driving namely a transformer
that for the most part is an inductor. So direct comparison with
transistor amplifiers is for the most part invalid, as 99.99999%
of solid state amplifiers do not have output transformers
the exception that I know of being the Quad 50e

So valve amplifiers have it fairly easy, they just have to drive a transformer
that then drives the loudspeaker . But they can benefit from
earlier stages being regulated, and the typical device is the LM317
or LM350 or LM338. Their design parameters being a maths formula expressed as
Vref x R2/R1 +1 , the LM317 involves a resistive divider
which can be good in some circuits but also a disaster in others

Many on the forum would use and appreciate the Quad 303 amplifier
that has a regulated power supply, it is one of few commercially
available examples of where the DC rails of the complete amplifier
are regulated, and few would argue its still good/ excellent audio ability.

Let us take a look at what it did. Instead of regulating the positive
rail it regulated the earthy end of the positive rail. It took what would
normally be just on 80 volts and tightly regulated what was called a
class B amplifier with transistor triples to 67 volts, and being
single ended required capacitors on its output as exactly half the
regulated voltage 33.5v was present on the capacitor prior to
coupling audio.

But lets go further back, it would not have been possible
without a simple device called a zener diode https://en.wikipedia.org/wiki/Zener_diode

So zeners are a major part of almost all voltage regulation circuits,
as are resistive dividers. But there are other devices also with
internal reference voltages that are very useful like transistors.
and led's. A comprehensive study of some devices that can be used as
references and the noise they exhibit is here:
http://www.diyaudio.com/forums/parts/35821-some-noise-measurements-leds-zener-diodes-7.html#post417008

Almost all amplifiers today fall into two categories either those
that use sophisticated regulation of the front end of the amplifier, and
allow the higher voltage and current section hopefully controlling the
loudspeaker to be unregulated this combination is usually very good for
audio.

Or those that use very simple regulation, relying and hoping on the
output stage not combining wrongly with earlier stages, once again the
resistor to drop current and zener diode is very popular but not
necessarily nearly good enough.

The very best amplifiers will attempt full regulation, and back to the earlier
point of division, will then further divide sections into
smaller sections with individual regulation. The sum of
all those regulated stages, assisting the ability to
properly drive a load.

But what about current regulation ? Current regulation is very valid
for all amplifiers particularly in their earlier stages, and is
IMO inexcuseible it it is not provided in amplifiers given the simplicity
of a good Vref/r design, vs what you usually get which is a V/R
or the next best but not a lot better a resistor zener based supply.

Cheers / Chris

[Sovereign]

Well written Chris, thanks for making it easy to understand

[Macca]

So what differences would we expect to hear in the sound between an amplifier with adequate regulation and one that has had special attention paid to it?

[lurcher]

So valve amplifiers have it fairly easy, they just have to drive a transformer that then drives the loudspeaker

Hmm, you do know how a transformer works dont you? Any and all impedance variation with frequency that the solid state amp has to drive is transformed up in the same way the output voltage is transformed down, so the valve amp has exactly the same issues as a solid state (or OTL) one.

[Light Dependant Resistor]

Hmm, you do know how a transformer works dont you? Any and all impedance variation with frequency that the solid state amp has to drive is transformed up in the same way the output voltage is transformed down, so the valve amp has exactly the same issues as a solid state (or OTL) one.

The transformer is usually designed to assist impedance matching in valve amps which it does, as seen by provision of selectable secondary taps for different loads.
as seen in this Marantz 8B

Cheers / Chris

[Light Dependant Resistor]

So what differences would we expect to hear in the sound between an amplifier with adequate regulation and one that has had special attention paid to it?

Hi Martin
More of what we know and appreciate in music is usually revealed, you start to hear depth, better stereo separation, clarity.


Cheers / Chris

[lurcher]

The transformer is usually designed to assist impedance matching in valve amps which it does, as seen by provision of selectable secondary taps for different loads.
as seen in this Marantz 8B

Cheers / Chris

Yes, but the main reason for that is so the nominal load the speaker presents to the amplifier is transformed into the specified load that the output valve(s) sees. It does nothing to help with the leading and lagging load that a typical speaker presents (the zobel helps a little).

[Light Dependant Resistor]

Yes, but the main reason for that is so the nominal load the speaker presents to the amplifier is transformed into the specified load that the output valve(s) sees. It does nothing to help with the leading and lagging load that a typical speaker presents (the zobel helps a little).

Hi Nick.
Yes there are certainly challenges in designing valve amp transformers, to get sufficient bandwidth through them.

The zobel will assist the amplifier drive a speaker lead, not the speaker. There are fundamental differences between solid state and valve amps
that readers might not know namely that a valve amp power increases with rising speaker impedance, whereas solid state amps do the opposite decreasing power as impedance rises.

It would be nice for speaker manufacturers to specialise in designing with valve amps in mind, great past examples might include
the LS 3/5A with 11-15 ohms, and to a degree the ESL57 but very low impedance at high frequencies and impedance curve
that was awkward to say the least made it fussy.

Cheers / Chris

[Arkless Electronics]

I'm afraid Chris that IMO most of your explanation is spurious, irrelevant and in many places simply incorrect. I agree with Nick. In your last post you also say that the output power of a valve amp increases with rising speaker impedance. It doesn't. As Nick says a transformer will transform a specific load impedance into a specific impedance seen by the valves. Valve output stages give their highest power and simultaneously lowest distortion when they are correctly matched to the load (usually about 6 - 8K for EL84's for example), hence the taps for different speaker impedance on most valve amps.

[Light Dependant Resistor]

[QUOTE=Arkless Electronics;684765 In your last post you also say that the output power of a valve amp increases with rising speaker impedance. It doesn't./QUOTE]

It is explained further here: http://education.lenardaudio.com/en/14_valve_amps_7.html
As is the purpose and placement of a Zobel http://education.lenardaudio.com/en/12_amps_3.html
and here: http://web.archive.org/web/20091027010126/http://uk.geocities.com/cyrilb2@btinternet.com/downloads_5.html

Cheers / Chris

[lurcher]

Sorry, but I only had a quick read of the first link and there at least two things I noticed that were outright wrong, plus other that were only slightly wrong. The two gems I saw were:

""The the output Impedance of a valve amp (Tetrode configuration) without negative feedback, is Infinite, Damping factor is zero."

Even though a pentode has a high Ra, its far from infinity, So the output impedance will be that reflected in the output transformer turns ratio. Not that there are many tetrode/pentode amps out there that don’t use feedback of some sort. For example, the Ra of a el84 is 38k, so with a typical 6k transformer, that would have an output impedance of 50R, far from infinite.

"Regardless of how much Negative feedback is applied it is not possible for a valve amp to achieve a Zero output Impedance (100% damping factor) similar to a solid-state amp."

Look at the literature, there was a fad in the late 50's where amps had controls that allowed variable output impedance (as you mention), but they could be set to have a +ve output impedance, a zero output impedance or a -ve output impedance. Crowhurst wrote several papers on the subject.

[Arkless Electronics]

http://i1303.photobucket.com/albums/ag156/Jez1235/New%20Picture_zpsmqiecn3o.png (http://s1303.photobucket.com/user/Jez1235/media/New%20Picture_zpsmqiecn3o.png.html)


Maximum power output is when Rl = 2XRp (for push pull) about 7200 Ohms for the 6L6 valve in the graph. This is the reflected impedance of the speaker load hence the need for output taps for 4, 8, 16 Ohms etc in order to optimise that reflected impedance for different speaker loads.

[Light Dependant Resistor]

Valve amps are enabled with much greater voltage swing, when loads typified by a actual speaker
with higher impedance are connected.

Where provided always use the transformer tap matching your speaker

It is further explored and answered as I have already provided
here: http://sound.westhost.com/valves/valve-trans2.html

It is further discussed here with similar conclusions see post 22
http://www.diyaudio.com/forums/tubes-valves/279415-output-transformer-saturation-impedance-load-curve-3.html

and here http://www.hans-egebo.dk/Tutorial/amplifiers.htm

Cheers / Chris

[lurcher]

Valve amps are enabled with much greater voltage swing, when loads typified by a actual speaker with higher impedance are connected.

Yes, of course, but you were talking about power, not voltage.

[Light Dependant Resistor]

Yes, of course, but you were talking about power, not voltage.

power watts = E x I given they also are current devices, the balance of those two combining
as a amplifier leans toward them exhibiting higher power output into speaker loads with higher impedance
which is the opposite of what solid state amplifiers do.

[Arkless Electronics]

Valve amps are enabled with much greater voltage swing, when loads typified by a actual speaker
with higher impedance are connected.

Where provided always use the transformer tap matching your speaker

It is further explored and answered as I have already provided
here: http://sound.westhost.com/valves/valve-trans2.html

It is further discussed here with similar conclusions see post 22
http://www.diyaudio.com/forums/tubes-valves/279415-output-transformer-saturation-impedance-load-curve-3.html

and here http://www.hans-egebo.dk/Tutorial/amplifiers.htm

Cheers / Chris

This is an interesting point which probably has much to do with "valve sound" in spite of being relatively ignored.... Yes a valve amp will provide a bigger voltage swing into a higher impedance, and although it's counter intuitive this does not mean it gives more power, just that it gives more power relative to a voltage source, therefore a valve amplifiers poor output regulation can work in both directions ie up as well as down as far as influencing a speaker's frequency response. I believe there's been a "crossed line" in as much as I interpreted you as claiming "valve amps give increasing output power into higher load impedance". Period. The graph is correct and maximum power output occurs when Rl = 2 X Rp (for push pull) is correct as I said. Maximum power transfer theorem states that maximum power is delivered from source to load when generator impedance = load impedance and this remains true, hence a load line has to be chosen for a particular set of conditions which combines max power with min distortion. It's actually far more critical to get the optimum match with pentodes than with triodes.....
I hadn't really thought about the implications of this for "some speakers seem suited to valve amps only", as many say, so thanks for stirring the little grey cells there Chris! :) This is a matter that could provoke major debate amongst the valve V transistor cognoscenti. It's a fault in amps that show this behaviour, in that it is a deviation from a perfect voltage source and therefore makes an amp unduly coloured into speakers without a constant impedance. Negative feedback will HUGELY effect this and make a valve amp increasingly like a transistor amp as feedback is increased. If enough feedback is applied then the ultimate result will be that although the output impedance is constant the clipping point will change with the varying impedance of the speaker load in two directions... Discuss ;)

[lurcher]

power watts = E x I given they also are current devices, the balance of those two combining
as a amplifier leans toward them exhibiting higher power output into speaker loads with higher impedance
which is the opposite of what solid state amplifiers do.

One last go.

Yes,

P = E * I

And Ohms Law

I = E / R

Combining

P = E^2 / R

So if the resistance (for example doubles) the voltage will have to do more than just increase to produce more power, it will have to increase 1.4 times.

But we are far from talking about regulated power supplies and this is a pointless conversation.

[CageyH]

So, are regulated power supplies any good then? ;)

[Marco]

Yeah, let's rein-in the 'technical tennis', chaps, and get back to helping Shaun with his enquiry, conducting the dialogue in layman's terms! ;)

Marco.

[Firebottle]

Yes, any sort of added regulation is going to have benefits.

[Arkless Electronics]

Back on topic then.... Yes, as Alan says regulation is a good thing. How good can vary from jaw dropping improvements to hardly noticeable depending on where and how in the circuitry it's used and on the sensitivity of a particular topology to voltage quality. The quantity and quality of the regulators required to do the job will vary according to how many different voltage rails there are in a unit that can benefit from regulation and of course how much money is thrown at the problem... Often a really good regulator can suffice for several stages and/or both channels.
Oh and for completeness here I'll say that there are designers who are anti regulation but they're in the minority.....

[Frazeur1]

If one can get an audition between the Croft "R" versions and non "R" in context of your own system, then maybe you will know if it really matters or not. Then it will be game, set, and match. No more tennis. Maybe easier said than done though....

[anthonyTD]

Glad to see the thread back on topic, :)
Although its good to re-discuss interesting subjects like valve amps versus Transistor amps, and the diffrent power transfer characteristics etc from time to time, as it does indeed give the old grey matter some serious stimulation, however, its very difficult sometimes to not get into circular discussions, and end up chasing each others tails, A few of us here know this all too well! :eek: :)
A...

[Jimbo]

The only advantage will be if you can hear it and it's what your looking for.All the technical guff is then only theoretical subjectivity.

[Light Dependant Resistor]

This is an interesting point which probably has much to do with "valve sound" in spite of being relatively ignored.... Yes a valve amp will provide a bigger voltage swing into a higher impedance, and although it's counter intuitive this does not mean it gives more power, just that it gives more power relative to a voltage source, therefore a valve amplifiers poor output regulation can work in both directions ie up as well as down as far as influencing a speaker's frequency response. I believe there's been a "crossed line" in as much as I interpreted you as claiming "valve amps give increasing output power into higher load impedance". Period. The graph is correct and maximum power output occurs when Rl = 2 X Rp (for push pull) is correct as I said. Maximum power transfer theorem states that maximum power is delivered from source to load when generator impedance = load impedance and this remains true, hence a load line has to be chosen for a particular set of conditions which combines max power with min distortion. It's actually far more critical to get the optimum match with pentodes than with triodes.....
I hadn't really thought about the implications of this for "some speakers seem suited to valve amps only", as many say, so thanks for stirring the little grey cells there Chris! :) This is a matter that could provoke major debate amongst the valve V transistor cognoscenti. It's a fault in amps that show this behaviour, in that it is a deviation from a perfect voltage source and therefore makes an amp unduly coloured into speakers without a constant impedance. Negative feedback will HUGELY effect this and make a valve amp increasingly like a transistor amp as feedback is increased. If enough feedback is applied then the ultimate result will be that although the output impedance is constant the clipping point will change with the varying impedance of the speaker load in two directions... Discuss ;)

Some good constructive comments, thanks.:) Yes valve systems should review the option of speaker systems with slightly
higher impedance. A good test would be the LS3/5a with 11-15 ohm specification. You are exactly right with combining
maximum power with minimum distortion, valves being capable of very good distortion figures and higher order harmonics
being very attractive for designing with. The efforts of firms like Quad, Leak, McIntosh, Dynaco forming a excellent base
upon which to learn.

Although regulation of heater supplies and DC rails is often used, it would be interesting to try a pair of LM317
or LM338 arranged as a AC regulator in a valve amp , a circuit that is not often used as far as I know

Cheers / Chris

[StanleyB]

The LM317 might not be very good at those high voltages.

[Light Dependant Resistor]

The LM317 might not be very good at those high voltages.

On the contrary, the LM317 is capable to several hundred volts, rather it relies upon the
input to output differential voltage not ever being exceeded that voltage being 37v, a HV version
increases that to 57v. A sensible approach would be designing with a 5-10v differential

Cheers / Chris

[Light Dependant Resistor]

Which asks testing of a pair of LM317 handling duty as a AC regulator
what are it's limitations / benefits ? given input to output differential of
37v of one device.

There is a peak to peak specification given in application notes
which first presents itself as a limitation for high voltage AC use
exceeding the differential , but a pair of devices may well be different.
as well may be choice of R2/R1 values

Certainly for AC Heater supplies the AC regulating 317 circuit is very
appealing, not to confine the first suggestion to be dismissed.

Cheers / Chris

[StanleyB]

Unless you use a protection diode between the input and output pins of the LM317, and also use a high value electrolytic cap at the output the LM317 can go faulty at mains switch on. Low value caps can read like a low value resistance on switch on, which would draw a large current through the LM317. It is then likely to give up the ghost.
But the flip side of using a big output capacitor is that could cause the input to be more than 100V higher compared to the output at switch on. This too can take the LM317 down on switch on.
Then we have the variable resistor to contend with that you would use to set the required voltage. Those things hate heat over a length of time and tend to start acting up when they have to handle high voltages. In a valve amp heat is a common thing.

[Light Dependant Resistor]

Unless you use a protection diode between the input and output pins of the LM317, and also use a high value electrolytic cap at the output the LM317 can go faulty at mains switch on. Low value caps can read like a low value resistance on switch on, which would draw a large current through the LM317. It is then likely to give up the ghost.
But the flip side of using a big output capacitor is that could cause the input to be more than 100V higher compared to the output at switch on. This too can take the LM317 down on switch on.
Then we have the variable resistor to contend with that you would use to set the required voltage. Those things hate heat over a length of time and tend to start acting up when they have to handle high voltages. In a valve amp heat is a common thing.

There is usually need for 2 protection diodes with DC circuitry using a LM317
the additional one used if the adjustment pin has capacitance bypassing,Most
circuits include up to 10uf here as up to 80db of ripple rejection can be achieved,
likewise to observe input /output capacitance values and voltage ratings.

However the LM317 AC regulation circuit is different, using 2 devices
it is shown at Figure 52 of the data sheet http://www.ti.com/lit/ds/symlink/lm117.pdf
it has instant appeal for heater supplies, but research needed for its application
in higher voltage stages, due to limitations with input to output differential.

As regards slow turn which is advisable there is a wealth of information at diy audio
usually involving thermistors.

As regards pots yes they can be unreliable, better to use fixed resistance
values. vout = Vref x r2/r1 + 1 , not exceeding differential and to use a heatsink

My experience with slightly higher voltage use with LM317 and LM337
in this case using pass transistors was in a Quad 405 supply
the differential being about 7 volts. That circuit continues to work well
with some 12 years of continual use.

Cheers / Chris

[anthonyTD]

Hi Chris,
As you say,
They do actualy work fine with high voltage applied, as Stan and others may have explained, as long as you use a zener across the in and out to protect the maximum voltage accross the device, i have used them in the past to good affect, for eg, at the end of capacitor mulitiplying circuits,where the use of large, high value capacitors was not posible.
There is usually need for 2 protection diodes with DC circuitry using a LM317
the additional one used if the adjustment pin has capacitance bypassing,Most
circuits include up to 10uf here as up to 80db of ripple rejection can be achieved,
likewise to observe input /output capacitance values and voltage ratings.

However the LM317 AC regulation circuit is different, using 2 devices
it is shown at Figure 52 of the data sheet http://www.ti.com/lit/ds/symlink/lm117.pdf
it has instant appeal for heater supplies, but research needed for its application
in higher voltage stages, due to limitations with input to output differential.

As regards slow turn which is advisable there is a wealth of information at diy audio
usually involving thermistors.

As regards pots yes they can be unreliable, better to use fixed resistance
values. vout = Vref x r2/r1 + 1 , not exceeding differential and to use a heatsink

My experience with slightly higher voltage use with LM317 and LM337
in this case using pass transistors was in a Quad 405 supply
the differential being about 7 volts. That circuit continues to work well
with some 12 years of continual use.

Cheers / Chris

[Arkless Electronics]

Just "wrap around" the 317 with a transistor and Zener diode to keep input/output differential below 40V ;)

[Arkless Electronics]

An observation here.... REALLY GOOD regulated supplies (state of the art), and hence I don't mean things like LM317, 7815 etc but discrete or partially discrete circuitry, are one of the most difficult things to layout and build properly in the whole pantheon of electronics and have such unseen subtleties to them that I wonder how many far below optimum implementations there are out there and to what extent some of the few anti regulated supply protagonists have based their judgement on listening to poor implementations??? Just a thought....

[anthonyTD]

I dont think anyone here was stating the humble LM317 to be the ultimate in Regulation, as far as low impedance, frequency, and signal to noise Jezz, just stating a few reasons why it can, and has been used to good affect, i have [like others here] designed my own bespoke regulators over the years, and continued to refine them for diffrent applications etc, i am sure you know how to design a decent regulator too, and i would go as far as to say that for me, the most interesting bit is learning how to design around specific issues in each case. As the saying goes, there is nothing new under the sun. ;)
And yes, i think your absolutely right, in sugesting that many a good circuit design has ultimately been let down over-all by bad power supply, and regulation!
I think Nick mentioned earlier on in the thread, that for a power supply and regulation to work properly, in any specific design, it almost always ends up being much more complicated than the circuit its designed to supply, and IMHO, he's absolutely right. :)
An observation here.... REALLY GOOD regulated supplies (state of the art), and hence I don't mean things like LM317, 7815 etc but discrete or partially discrete circuitry, are one of the most difficult things to layout and build properly in the whole pantheon of electronics and have such unseen subtleties to them that I wonder how many far below optimum implementations there are out there and to what extent some of the few anti regulated supply protagonists have based their judgement on listening to poor implementations??? Just a thought....

[Arkless Electronics]

I dont think anyone here was stating the humble LM317 to be the ultimate in Regulation, as far as low impedance, frequency, and signal to noise Jezz, just stating a few reasons why it can, and has been used to good affect, i have [like others here] designed my own bespoke regulators over the years, and continued to refine them for diffrent applications etc, i am sure you know how to design a decent regulator too, and i would go as far as to say that for me, the most interesting bit is learning how to design around specific issues in each case. As the saying goes, there is nothing new under the sun. ;)
And yes, i think your absolutely right, in sugesting that many a good circuit design has ultimately been let down over-all by bad power supply, and regulation!

I think you may have misinterpreted my post Anthony :) Neither I nor, to the best of my knowledge, anyone else is trying to claim great things from 3 terminal regulators... My point is that when regulators are designed for ultimate performance in noise, transient response, bandwidth and low output impedance they become ornery critters that are just waiting to metaphorically bite your hand off.... hence many are probably well below optimum.

A very high performance design I did recently needed a 1.5GHz spectrum analyser to fully sort out!

[anthonyTD]

Aye, i realised that Jezz when i re-read your post,:doh: which is why i added the last bit :D Another key factor is the availabilty of Good components these days, therefore, the cost between a good design, and a great one is neither here nor there.I remember when i was a kid, my Dad would salvage everything he could from scrap TV's, and Radio's etc, even now, he still cant bear to throw anything away! :)
I think you may have misinterpreted my post Anthony :) Neither I nor, to the best of my knowledge, anyone else is trying to claim great things from 3 terminal regulators... My point is that when regulators are designed for ultimate performance in noise, transient response, bandwidth and low output impedance they become ornery critters that are just waiting to metaphorically bite your hand off.... hence many are probably well below optimum.

A very high performance design I did recently needed a 1.5GHz spectrum analyser to fully sort out!

[StanleyB]

I think you may have misinterpreted my post Anthony :) Neither I nor, to the best of my knowledge, anyone else is trying to claim great things from 3 terminal regulators...
There are some extremely good ones out there, but not easy to get hold of. I use two of them in the Caiman DAC, but the paperwork I had to do just to satisfy the manufacturer would make them illogical to use for non-critical applications. A better way would be to use a decent LDO and capacitance multiplier.

[Arkless Electronics]

Aye, i realised that Jezz when i re-read your post,:doh: which is why i added the last bit :D Another key factor is the availabilty of Good components these days, therefore, the cost between a good design, and a great one is neither here nor there.I remember when i was a kid, my Dad would salvage everything he could from scrap TV's, and Radio's etc, even now, he still cant bear to throw anything away! :)

Indeed, the rapid disappearance of many of the best parts for analogue electronics is a major bugbear for me and I'm sure for yourself! Anything that ceases to be used in HUGE quantities for consumer equipment, due to "improvements" in technology, tends to be obsoleted by the manufacturer.

It may be a surprise to many audiophiles but many of the best parts ever made for use in analogue gear, and I'm thinking mainly of discrete transistors and FET's, were made between about 1985 and 2000 ish and are now either unobtainable or available in small quantities at silly prices only.... and no nothing of truly equal performance is always available to replace it. One has to spend far too much time looking for ways around the issue using alternative circuitry etc.

I keep all sorts of old boards, scrap gear etc etc myself as you never know when a part made from solid unobtainium will be needed and I may just have one on an old board... it's rather wasteful of space of course!

Bizarrely, with the rise in popularity in valve amps and the huge market for guitar amps, it's getting easier to find fairly exotic valve paraphernalia such as valve sockets, multi section caps and even decent new valves than parts for some solid state stuff!

Here's a prediction for the future: Power transistors for normal class A/B and class A amps will become hard to get as virtually all consumer applications are taken over by class D switchers..... not for a long while yet though!

What's your take on this?

[Arkless Electronics]

There are some extremely good ones out there, but not easy to get hold of. I use two of them in the Caiman DAC, but the paperwork I had to do just to satisfy the manufacturer would make them illogical to use for non-critical applications. A better way would be to use a decent LDO and capacitance multiplier.

It's all relative of course Stanley :) Even an LM317 can give excellent results in the right application and is of course very cheap. It's performance could even be described as good, depending on what parameters are important to your own application. Yes there are some much improved modern POL regulators available these days (it's easy to forget just how old something like the venerable LM317 is!) but for ultimate performance nothing can match a properly designed and built regulator made with discrete components or based on the latest op amps combined with discrete components.
When you want the lowest noise and greatest bandwidth possible there's no IC regulator that can match the best discrete one ;)

[StanleyB]

300 nV √Hz noise figure from those regulator can't be easily bettered by even an expensive discrete regulator.

[Arkless Electronics]

300 nV √Hz noise figure from those regulator can't be easily bettered by even an expensive discrete regulator.

I'm talking 1nV √Hz!

[Light Dependant Resistor]

I'm talking 1nV √Hz!

There are some very nice regulators now doing just 40 uV to 60uV of noise, such as the LT1963A, and LT3015
but noise is not the only parameter for a good regulated supply, perhaps we can think of some more ?

Anthony TD has mentioned a few times the capacitor multiplier, which for readers unaccustomed to it
uses the current gain of a transistor a term called hfe to magnify the capacitance. The base of the transistor has a
capacitor connected to ground, and provides seemingly larger capacitance.

For audio which in terms of frequency occupies just 20 hz or slightly less, and to celebrate the life work of
David Blackmer lets put an upper figure of 40Khz,( the confining to 20khz has been pushed by CD sampling
and the Nyquist sampling therorem dictating an upper limit well certainly until oversampling came in. )
So getting regulators to provide maximum benefit to audio frequencies is what is needed.

We think of current being pushed toward where it is being used, but not so much about the time it
takes to get there and by that I mean maintaining phase relationship. In the same way protection
diodes dissipate capacitors downstream returning current, it should be noticed if we use a
fast diode like a UF series or a transistor to do this instead, there is great benefit for audio.

So Anthony build a cap multiplier for a positive supply and instead have the NPN emitter
returning to the front input of the regulator, and give it a wee bit of emitter degeneration like 10 ohms.
let us know how it goes ? Having done this many times it is I know excellent for audio use.

Cheers / Chris

[anthonyTD]

Like you, i have kept much "junk", and have stock of some NOS Transistors, op amps, power resistors, etc, etc,from "the golden era" also, i seem to have an almost endless stock of Millitary Spec resistors,[unfortunetly not all common values] you know, the type with Brass end caps etc, i seem to have adopted the mind set of my Dad, that i might need them one day! :eyebrows: As you say, at the moment, with the ever popular valve amp, paticularly Guitar amps still being made and used on a large scale, there is a good supply of high voltage parts, and valve bases etc being produced, however, with the trend of everything in the consumer market becoming ever more efficient, how long this will last is anyones guess. :rolleyes:
Indeed, the rapid disappearance of many of the best parts for analogue electronics is a major bugbear for me and I'm sure for yourself! Anything that ceases to be used in HUGE quantities for consumer equipment, due to "improvements" in technology, tends to be obsoleted by the manufacturer.

It may be a surprise to many audiophiles but many of the best parts ever made for use in analogue gear, and I'm thinking mainly of discrete transistors and FET's, were made between about 1985 and 2000 ish and are now either unobtainable or available in small quantities at silly prices only.... and no nothing of truly equal performance is always available to replace it. One has to spend far too much time looking for ways around the issue using alternative circuitry etc.

I keep all sorts of old boards, scrap gear etc etc myself as you never know when a part made from solid unobtainium will be needed and I may just have one on an old board... it's rather wasteful of space of course!

Bizarrely, with the rise in popularity in valve amps and the huge market for guitar amps, it's getting easier to find fairly exotic valve paraphernalia such as valve sockets, multi section caps and even decent new valves than parts for some solid state stuff!

Here's a prediction for the future: Power transistors for normal class A/B and class A amps will become hard to get as virtually all consumer applications are taken over by class D switchers..... not for a long while yet though!

What's your take on this?

[anthonyTD]

:)
There are some very nice regulators now doing just 40 uV to 60uV of noise, such as the LT1963A, and LT3015
but noise is not the only parameter for a good regulated supply, perhaps we can think of some more ?

Anthony TD has mentioned a few times the capacitor multiplier, which for readers unaccustomed to it
uses the current gain of a transistor a term called hfe to magnify the capacitance. The base of the transistor has a
capacitor connected to ground, and provides seemingly larger capacitance.

For audio which in terms of frequency occupies just 20 hz or slightly less, and to celebrate the life work of
David Blackmer lets put an upper figure of 40Khz,( the confining to 20khz has been pushed by CD sampling
and the Nyquist sampling therorem dictating an upper limit well certainly until oversampling came in. )
So getting regulators to provide maximum benefit to audio frequencies is what is needed.

We think of current being pushed toward where it is being used, but not so much about the time it
takes to get there and by that I mean maintaining phase relationship. In the same way protection
diodes dissipate capacitors downstream returning current, it should be noticed if we use a
fast diode like a UF series or a transistor to do this instead, there is great benefit for audio.

So Anthony build a cap multiplier for a positive supply and instead have the NPN emitter
returning to the front input of the regulator, and give it a wee bit of emitter degeneration like 10 ohms.
let us know how it goes ? Having done this many times it is I know excellent for audio use.

Cheers / Chris

[anthonyTD]

Didnt really have chance to answer this post when i first read it, hence the smiley.
I know exactly where your coming from on this one Chris, having built the first versions of what you describe around 2001.:)

So Anthony build a cap multiplier for a positive supply and instead have the NPN emitter
returning to the front input of the regulator, and give it a wee bit of emitter degeneration like 10 ohms.
let us know how it goes ? Having done this many times it is I know excellent for audio use.

Cheers / Chris