No regulator for best sound?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I've posted several times on a late-1980s Magnavox (Philips) CD player. Due to it' simple, accessible layout, it's an easy "breadboard" project.

The stock unit's PSU contains only one main voltage regulator: an LM7805. Its 5V gets routed to the SAA7210, TDA154x, and a few other ICs.

What's surprising is that the analog section (I/V and LPF opamps, etc.) are fed unregulated power! E.g., +/- 15v. Indeed, there is over 50mV of AC ripple on the +/- 15v opamp power pins. Compared to less than 2mV on the +5v DAC.

As part of my experiments with this unit, I am no longer using the stock I/V and LPF opamps ...Inside this CD player (see photo) I am currently feeding the DAC to a discrete I/V output stage. The discrete I/V is designed to work on 12-18vdc (supposed to sound best at 18vdc).

So I fed the discrete I/V the same unregulated +15v that orig fed the opamps. The sound quality is very good.

Because the discrete I/V can run on 12v, I decided to regulate the stock unregulated +15v with an LM317 circuit (the LM317 was config'd to output 12v). While the ripple reduced several orders in magnitude, the sound quality worsened: bass became puny and dynamics became very weak.

Returning to stock unregulated +15v, performance returned to orig.

Any guesses as to the why the sound worsened despite improve ripple performance? E.g., lower voltage (power) for I/V, etc.

Note: This CD player's stock transformer can only offer a max of 15vdc. So that's the max voltage I can afford for any project power.

Below: Stock PCB of Magnavox CD 2000
3715982035_87bbb15be9.jpg


Below: Same Magnavox CD 2000 as above, but using Rudolf Broertjes' I/V
(Note: the I/V is simply powered by the opamp socket 15v power pin)
An externally hosted image should be here but it was not working when we last tested it.
 
Last edited:
Administrator
Joined 2004
Paid Member
Hi hollowman,
I rather suspect that it was the lower B+. Too bad Philips made such cheap ... stuff. I serviced those machines and am well aware of how many corners were cut to save a little $$.

For all the effort you are putting into that machine, consider installing a new transformer that supplies higher voltage and regulate that down to the same levels (minus the ripple). If the regulators get hot, consider using an inexpensive switching regulator for the digital chips (the 5V supply).

Basically, eliminating the ripple and noise can only lower the noise floor and distortion from outside effects. There is no way running your analogue sections off a supply with ripple and noise will ever be better than a nice quiet power supply.

-Chris
 
hollowman said:
Because the discrete I/V can run on 12v, I decided to regulate the stock unregulated +15v with an LM317 circuit (the LM317 was config'd to output 12v). While the ripple reduced several orders in magnitude, the sound quality worsened: bass became puny and dynamics became very weak.
What measurements did you do to determine which sound was closest to the original sound heard in the studio? All you have expressed is a preference, which tells us not very much.
 
...... There is no way running your analogue sections off a supply with ripple and noise will ever be better than a nice quiet power supply.

-Chris
While that's generally true, there could be exceptions?
Like maybe the lower voltage of the regulated supply is having an impact?
Or the regulators are not working well?

It seems to be known that the I/V does not works as well on lower voltges. Why is that?
It sounds like time to get in there with a scope.
 
Time to experiment further ..

While that's generally true, there could be exceptions?
Like maybe the lower voltage of the regulated supply is having an impact?
Or the regulators are not working well?

It seems to be known that the I/V does not works as well on lower voltges. Why is that?
It sounds like time to get in there with a scope.
Lots of possibilities, I realize! I'm betting on lower voltage.
May have to break out one of my small Amveco xformers to experiment.
BTW: I used a Tangentsoft TREAD regulator (it's based on LM317).
 
Administrator
Joined 2004
Paid Member
Hi Chris,
While that's generally true, there could be exceptions?
No, not with signal circuits. For a power amplifier, some see an improvement, but this wastes power. The best compromise is to regulate the voltage amp stage and allow the current amplifier stage to run off unregulated power. The voltage gain of the output stage is commonly just less than one, so noise will not be amplified. Amplifiers that are set up this way tend to have wider signal to noise ratio, and also seem to be open and airy sounding even when the bass is pounding away. The regulator insulated them from all / most power supply noise and variations. My Marantz 300DC is designed like this for example.
Like maybe the lower voltage of the regulated supply is having an impact?
Or the regulators are not working well?
Could be either or both. With an oscilloscope, you could check the regulated voltage for signs of drop-outs.
It seems to be known that the I/V does not works as well on lower voltges. Why is that?
Could be you don't have enough voltage compliance (range to maximum peak output voltage). It also could be the circuit that picks up the signal from the I-V converter.
It sounds like time to get in there with a scope.
You're blind otherwise. I'd recommend a 20 MHz bandwidth as an absolute minimum. Buy the best you can afford, used is also a valid choice. I'm using a $20,000 oscilloscope (Agilent 54642D) that I could never afford. I saw one for about $1,000 and jumped on it. So whatever you end up with, as long as it's a good one, is a wise purchase and multiplies the dollars you spend. I had been looking for a few years.
"Chance favours the prepared mind" :)

-Chris
 
.....
Could be either or both. With an oscilloscope, you could check the regulated voltage for signs of drop-outs.
....
You're blind otherwise. I'd recommend a 20 MHz bandwidth as an absolute minimum. Buy the best you can afford, used is also a valid choice. I'm using a $20,000 oscilloscope (Agilent 54642D) that I could never afford. I saw one for about $1,000 and jumped on it. So whatever you end up with, as long as it's a good one, is a wise purchase and multiplies the dollars you spend. I had been looking for a few years.
"Chance favours the prepared mind" :)

-Chris
Definitely worth getting a scope if anybody is at all serious about analogue electronics.
I'd agree anything below 20MHz bandwidth is of limited use.
I would also say that a minimum of two channels will be useful quite often.
You can often get something that works on ebay for surprisingly little cash.
Just get something you can see working before buying it.
You can always trade up to a nicer instrument later.
 
As the CD player is obviously built to a budget and has ripple on the voltage rails why not increase the capacitance after the rectifiers?
A simple substitution with some decent higher value low ESR caps should help.
I don't think I will bother with modding CD player outputs again, after building an external DAC which outperformed my CD player by a large margin, and allowed me to connect other digital sources such as the TV!
 
I decided to regulate the stock unregulated +15v with an LM317 circuit (the LM317 was config'd to output 12v). While the ripple reduced several orders in magnitude, the sound quality worsened: bass became puny and dynamics became very weak.



Goodness. This could have been exciting news in 1985. Dunno what is your excise for learning it this late :confused: Perhaps frequenting the scope-heads forum a bit too much :p

These 3-pin monstrosities simply destroy the sound. Want an audible upgrade upon raw dc? Look at a nice shunt. Just mho of course.

Others think that properly built passive filtering is unmatchable with active solutions but at the cost of lots more iron and capacitance.
 
Didn't break out my 'scope ... yet ... but I may in a few days....

Anyway, exclusively for the I/V, I'm now using a totally-external PSU based on the tracking pre-regulator tweaked by Martin Clark. The dedicated xformer is a small Amveco toroidal.
Most important, the new reg. is outputting 18.5v with super-clean <1.0mV ripple (read with my Fluke 87 DMM on 'AC')

The sound?
Definite improvement in most aspects [compared to un-regulated DC]. Most aspects except -- perhaps -- slam and dynamics. Hard to tell.

The suggested shunt may be worth a shot.
 
I know folk hate those 3-pin regs but before giving up - Try adding a good quality cap to the power rail close to the audio circuit. A Nichicon MUSE cap. You want this nice cap to be large enough to shunt bass frequencies so that no signal current has to be handled by the reg. Depending on the circuit, 470uF might be good.
 
Administrator
Joined 2004
Paid Member
Hi Bigun,
Good suggestion as the power supply has to be designed as a system. You have to make darned sure that the load side discharges more quickly than the unregulated side or you stand the chance of destroying the regulator when you turn it off. You can arrange diodes to prevent this. Look in any datasheet for the 3-pin regulator chips from the old days. They took pains to show and explain this.

All,
Shunt regulators are no better than series regulators if they are designed well. You can certainly create a bad shunt regulator!
As the CD player is obviously built to a budget and has ripple on the voltage rails why not increase the capacitance after the rectifiers?
Simple. High capacitance causes other problems in the higher frequency area. It is easy to eliminate a 120Hz ripple. It is very difficult to eliminate high order harmonics and possible bursts of HF garbage caused by inductance, high capacitance with low ESR.

Series resistance actually is extremely beneficial to you. It limits the charging current peaks and thus, the higher energy noise you don't want. Look at any performance graphs for regulators and they are all going to show decreasing performance with frequency rising. That means they can block low frequency noise a lot better than higher frequency noise. The same holds true for discrete voltage regulators and most shunt regulators. Exactly like an audio amplifier. The urge to install larger capacitors is due to the current fad <sic>. If you take the time to think about these things more, you can see where the problems are.

-Chris
 
I have found regulators take their toll no matter how good they are.

They offer convenience (like CDs) and they save space (those things again) but they do not sound anywhere near as good as brute force lclclc filter.

The brute force approach is wasteful of space and electricity but with preamps and DACs this is not insurmountable.

Requires a little experimentation to get the voltage you need but after a few you start to intuit what will be needed and can arrive at your required voltage without too much trouble. The bleed resistor is always needed for fine tuning along with keeping the current through the chokes at the level required.

I recently replaced the shunt reg in the SALAS phono amp. I thought this one had to be good enough and could not be bettered. But it was trounced by the brute force filter in every way that means something to me.

I do realize some people like the sound of regulators just like many have grown to accept that ported loudspeaker bass is what bass sounds like. Each of us has to find our own way ...
 
I do realize some people like the sound of regulators just like many have grown to accept that ported loudspeaker bass is what bass sounds like. Each of us has to find our own way ...

Very well said. I guess i am among those who have grown into accepting regs.

The brute force applied to raw dc is still essential: no regulator can do miracles with crappy rectifiers and caps. Chokes are also compulsory if ultimate quality is sought.

It all probably boils down to where your musical preference lies: tone and unrestriced dynamics vs soundstage, bass control and good prat.

Very interesting you mention the Salas phono. At the beginning i thought it showed some real promise but i never managed to fully relax with it. Multiple coupling cap changes, multiple loading resistor changes. Every time it seemed like i got it just right...
 
Administrator
Joined 2004
Paid Member
Hi Rick,
With all the different regulators out there, and each has different performance, I have a little trouble with what you are saying. It sounds like a strong case of expectation bias. The regulator Salas designed has less noise than various battery types. Certainly the noise from most electronics will be greater than that regulator (and many others). Certainly in the case of a DAC, you have two power supplies (one digital, one analogue) and the board layout alone can have a much greater effect than the regulator used.

Certainly some regulators are better than others, but with such varied characteristics you would have some that were better than a poorly regulated "brute force" supply. In fact, electronic regulators are not just easier to design compared to what you are using. In equipment where cost is not a major concern, you will find the best power supplies available for each purpose. Absolutely zero of these use anything even close to what you are suggesting. I have redesigned some of these power supplies in customer equipment and ended up with a much quieter power rail which was noticeable to hear with unaided listening. In 100% of the cases, customers had a strong preference for what I had done. This process included pulling out excessively large capacitors and installing "little ones" in their place.

You may be right in comparison with some of the IC regulators, but not to the majority. Note that in using even the regulator Salas designed, if you don't connect the common to the correct point in the circuit, you can destroy its performance. When you come up with what you are saying, it's time to look at how you installed them. No way can you make the type of claim you just have.

There are a couple ways you can test how good a power supply is. Take a large polypropylene capacitor and hook on end to the B+, the other from a 1 Meg resistor to your common point. Take the junction of the capacitor and resistor and connect it to the input of an amplifier. Once your power supply is running the intended load, increase the volume of the amplifier and listen to it. You can try this with every power supply including your approach. A better way is to do the same thing, but replace the amplifier with a spectrum analyser. Then you will have actual numbers you can use to compare them.

If you compare a bunch of regulators, don't touch the volume control once you have set it. Just disconnect the speakers until your connections have been made. You might cause the protection on your amplifier to cycle, but that's okay since there isn't any load on the amp while you are changing connections. Basically, the more noise there is on the supply line, the less good the power supply or regulator is.

I look forward to hearing your results.

Best, Chris
 
From my point of view, the object of the game is for the output of the amplifier to be a faithful amplified version of the input.
That can be influenced by the power supplies. Two main things going on here, firstly line related signals, i.e. harmonics of the mains frequency and switching effects with a mains PRF. Secondly, load related signals, i.e. audio signals and harmonics etc thereof put onto the rails by the amp. Surely both things are worth measuring before looking for qualitative psycho-accoustic influence on the stereo illusion or user happiness?

Another way of looking at regulators vs LC or RC or 'more C' is that we are talking about active vs passive filters?
The fundamental stumbling block with a purely passive approach is that the supply frequency (let alone its harmonics) is within the signal bandwidth. Passive filters are essentially, on the whole bi-directional devices. That means any passive filter which blocks mains related noise from flowing to the amps is also impeding the amp's ability to draw from the supply at audio frequencies. Which tends to mean that the only way you'll keep the supplies remotely clean is to have a very low supply impedance compared to the load impedance. Which isn't trivial across 5Hz to 50kHz. Although obviously it's easier for a pre-amp working into a high impedance than a power amp.
So, it seems to me there is often going to be signal on the rails. So maybe the question should be more about what is going to be on the rails and how that will affect various audio circuits?
Regulators will also put some audio related signals on the rails. They have finite BW, so a big pulse of audio will pull the rail and then the regulator may overshoot as it corrects.
Maybe if we don't like the way things sound with a certain regulator or decoupling, we need to be re-analysing PSRR if we can't measure anything gross on the rails?

This is a different way of thinking to what I'm used to outside of audio, where mostly regulators supply 'clean DC' and signals are mostly higher frequency than regulator BW, so signals are mostly kept in their box by decoupling capacitors.
 
That means any passive filter which blocks mains related noise from flowing to the amps is also impeding the amp's ability to draw from the supply at audio frequencies.


Amps are not supposed to draw audio frequencies from the mains. I am not convinced this is what really happens, but at least, by intention audio frequency current draw should exclusively be taken care of by the caps. For one, caps offer by far lower impedance then the reflected through the transformer impedance of the mains. And they offer pretty constant availability of current, unlike the mains.
 
Administrator
Joined 2004
Paid Member
Hi analog_sa,
In most circuits, especially simple ones, whatever is on the power supply will end up in the output signal, attenuated a little but still very much there. With op amps, at least they have high PSRR in most cases.

As for a limited view, ouch! That was insulting a bit.

By definition, a power supply that can deliver power to the load without modulation of the power it is delivering is exactly what it should be doing, not a limited view at all. It should not add anything to the load or subtract anything, perfection in other words. We can't reach this goal, but the better the supply is, the closer it will be to this ideal.

Given that you view my definition as "limited", why not offer your own view for examination? I honestly don't think yours could be any different than mine unless it is flawed. Go for it.

As far as amplifiers go, they do draw from their main filter caps - the ones closest to the amp and load. They draw at audio frequencies plus high harmonics as each side draws one polarity for the most part. What comes from the AC wall current are simply spikes of current that are higher with increasing load from the amplifier. I could see these spikes following the low bass line of the music, so you might get modulated pulses.

-Chris
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.