About design philosophy _ are minimalists right or not ?

[JeffS7444]

My Douglas Self-designed preamplifier, which is chock full of Texas Instruments 5532 opamps. Among it's other design feats, it was designed to use linear, volume control pots, which tend to have more precise left/right tracking than your typical audio-taper pot.

 

[Dimitri]

I would like to understand in practice whether the so-called circuit minimalism is a valuable design approach and those who complicate it are wrong or minimalism has limits that cannot be overcome and complexity is, let's say, a necessary evil.

Can't patent a simple circuit that came straight out of the example in a datasheet. But make it cheap and "good" , attach a story to it and it can can sell.
If you end up adding "enginering pixie dust" to make it "yours", different and even patented especially if you stumble on something that can have a cool name, now the "complicated" has more value. Not necessarily as a product to use and sell, but a design to have for when you sell the company.

The above comment is fictional and has no reference to any persons or other entities of any kind alive or dead in the past, present or future on this or any other planet in this or any other universe known or not yet discovered. (that should cover it )

 

[Blumlein 88]

Another part of the tale behind simple circuits is every component matters. If your circuit is simple, you can afford to use the highest quality components. If your circuit is complex those are cost prohibitive. And that tale is easy for people to think about just that way. In fact all that matters is input vs output. If complex works with cheap components then it can be better or even as good as possible. That isn't an easy concept for those who don't know circuits. If going thru every component leaves a trace on the signal the fewer of those and the higher quality each is the less degradation occurs or so the story goes.

 

[Chr1]

... As in passive preamps presumably?

Perhaps it might be easier for us luddites, if those more knowledgeable could explain what should be avoided when designing quality audio electronics. Presumably anything that adversely affects the measurements within the audible bandwidth?
Thanks!

 

[solderdude]

To be more specific and talking about measurable performance I wonder if very low distortion and noise, the peculiar qualities of good sound, can be achieved also with very simple circuits.
I'm rather confused because often the simplest circuits, if done well, make a very favorable impression on listening.
Are those who complicate things that could be simple wrong ?

Consider that the issue at hand (good sound) is not the complexity of a circuit but rather the fact that people don't really need TOTL technical performance to hear something as 'sounding good'.
With decent enough specs sound is already 'good enough'... take vinyl or analog tape for instance. These are spec'd really poorly opposite digital yet people swear by them because of the simplicity of the format.
So simple can already sound excellent and this plays right in the cards of KISS fans.
When you are a KISS fan and get your hands on something with few parts you are likely to prefer it because it fits your thinking.

Also consider that op-amps may look simple but are usually highly complex and can have parts in them that do not even exist in discrete components.

 

[gino1961]

But why do you want discrete simple line stages? I get the fascination and the implied philosophy. But if what you really want is performance, it is irrelevant.

hi i have to agree with you Probably it is easier to get good performance with opamps
without naming names, years ago a line pre-amp was released which became as famous as its designer, made with discrete components embedded in resin and boxed in modules inserted on a motherboard Some time ago I saw small PCBs pin compatible with a single opamp on them and some passive components sold as an upgrade of the original modules.
Anyway i understand that almost no one is curious about what kind performance can be obtained let's say from a two-transistor circuit when all the parameters are optimized
To be more specific looking at some service manuals of top preamps of mid '70s i found quite popular the Sziklai pair for line and buffer duties

Sziklai pair - Wikipedia

en.wikipedia.org

I tried to simulate a very basic circuit using the topology and i noticed that changing R and C values and Vsupply the distortion was also varying a lot
I remained with the doubt of how low could be the distortion of a very well optimized Sziklai pair Maybe very low indeed and rivalling much more complex preamp circuits
And of course i was using two common bjts of which i had the models Maybe a better selection of bjts could provide even better performance
Then there is the other big problem Power supply noise
If i understand well more complex circuits have usually a better Power Supply noise rejection and this is a very good thing
More basic circuits could need much more complex power supplies to keep noise low
I love batteries But they are too unconvenient

You asked if there are other things than linearity, distortion and low noise. I would add frequency response and beyond that there is nothing in the sound such a unit manages. I've built some simple preamps with 3 transistors. They were good, but so are some op amps. An LM4562 or OPA 891 is probably beyond reproach for an audio line level preamp.

how good were they ? like the opamps ?

Audio Research famous for their vacuum tube preamps, when they finally made a SS model in the 1980's had a few of what they called Analog Gain Modules. Some super great design by them. Many assumed it was a high quality discrete FET design. Nope, they had some good OP amps potted in epoxy to obscure that fact and put covers over that to make it look more substantial than it was. I think they used NE 5532 op amps, but might remember that wrong. I know if the gain modules go bad knowledgable people replace them with LM4562 op amps.

I didn't know that and thank you very much for this interesting information
An attitude that doesn't make such a renowned brand look edifying for sure I guess that the modules were substantially more expensive than the OA ?

 

[gino1961]

Go to the library and get Small-Signal Audio Design by Douglas Self or some of the books by Walter Jung. Both are longstanding Audio Engineering Society members who explain circuit design. There are tube design books too. In design there are always limits and tradeoffs. Today's audio-optimized op amps are a pretty exchangeable gain block with high input impedance, low output impedance, wide power supply range, high gain-bandwidth product, low noise, and low cost. Essentially they are a voltage-controlled current source. They gained a foothold in analog computers, then revolutionized low level audio.

Hi and thank you for the very valuable advice
I have decided for a different approach that is to copy a working schematic and check if it can be improved changing passive and active parts and working parameters
The circuits i am looking at go back to mid 70s and unfortunately are all cap coupled and this could impact the overall performance
Maybe today bjts resistors capacitors can be found that are much better than those of 50 years ago
Just think the pride of beating a good line stage with ten times less parts
Passives are passive But bjts can provide quite different performance depending on the working conditions

 

[roog]

I say No, because like the room where my Stereo is installed, many rooms have similarly troublesome acoustic issues and I find Parametric EQ applied using DSP, (by definition a complex function) to be an effective way to deal with the room issues which in turn improve the sound.

 

[gino1961]

... As in passive preamps presumably?

Perhaps it might be easier for us luddites, if those more knowledgeable could explain what should be avoided when designing quality audio electronics. Presumably anything that adversely affects the measurements within the audible bandwidth?
Thanks!

Hi an active one would allow me to use different power amplifiers without worrying too much about various impedances
There are also power amplifiers with only 10kohm input impedance
Sometimes passives can sound absolutely clean but anemic with little punch

 

[Chr1]

So I understand. And long cable runs are an issue too.
I use one between my AVR and KH310s and it is perfect for the job: transparent and easy for minor tweaks in the volume of the L/R channels when using the AVR as my preamp/source.

 

[gino1961]

Consider that the issue at hand (good sound) is not the complexity of a circuit but rather the fact that people don't really need TOTL technical performance to hear something as 'sounding good'. With decent enough specs sound is already 'good enough'...

Hi perfect and thank you Then if decent specs can be obtained with a low parts count why i see line preamps with dozens and dozens of components ?
design perversion ? or maybe it could be just a marketing strategy To justify an high cost customers must see a lot of stuff inside ?

take vinyl or analog tape for instance. These are spec'd really poorly opposite digital yet people swear by them because of the simplicity of the format.

i have a view on this A digital copy of an analog tape could sound very close to the original at the point that it would be difficult to pick the original in a blind test Digital can be very transparent I read of a demonstration by Millennia with the original feed compared to the same signal passed thorugh many AD and DA stages The differences were minimal I am digital

So simple can already sound excellent and this plays right in the cards of KISS fans.
When you are a KISS fan and get your hands on something with few parts you are likely to prefer it because it fits your thinking.

i dont trust my ears I am 63 and i cannot listen above 14kHz with ears cleaned
My kiss circuit should provide impeccable measurements in terms of noise distortion slew rate and crosstalk at least in line with the better pres around

Also consider that op-amps may look simple but are usually highly complex and can have parts in them that do not even exist in discrete components.

absolutely I have seen they have many bjts and jfets inside And i wonder why are not so popular in high end preamps Strange

 

[Multicore]

I think the less is more ethos can be attractive depending on the context, for example when I, a contrarian, am feeling oppressed by the more is more consumer ethos, which is not exactly uncommon in audio. But I don't think it is a very useful engineering concept. It's too simplistic. A good engineer will not over-complicate a design but that's completely different from seeking to minimize its complexity.

System function and performance goals are primary. Cost and your resources/capabilities are constraints and complexity is what it is. If we let minimum complexity dictate function and performance then we just build zero dB amplifiers.

In this thread we discussed a block diagram hinting at the complexity in a typical powered speaker. With only one quibble we agreed that all the complexities in the design make sense and, properly implemented, would improve the sound of the system. I get that some people are attracted to much simpler systems but improved sound is hard to argue with.

 

[Chr1]

Again, it's all about the measurements surely? And what's even audible.
No hard and fast rules, or generalisations.

Like arguing about optimal cylinders in a motor... To vague a discussion methinks. Horses for courses etc

 

[gino1961]

My Douglas Self-designed preamplifier, which is chock full of Texas Instruments 5532 opamps. Among it's other design feats, it was designed to use linear, volume control pots, which tend to have more precise left/right tracking than your typical audio-taper pot.
View attachment 364396

great and clean layout indeed I am just focusing a suitable "gain block" able to provide a Vgain of 2 or 3 times and low enough Zout
Attenuator is not a problem I could invest in something good if the gain block is providing good performance

 

[Ze Frog]

A lot of it comes down to the designer a their way of implementation and arriving at their end goal.

I've never built a Pre-amp, but speakers I have and it's amazing how many different ways you can design a crossover to achieve a relatively the same result. Sometimes you can take the easy way, other times something more complex work's. Effectively a group of designers would likely all arrive at something different. Minimalism is good, as long as it's not used as a tool by bean counters trying to do something on the cheap or minimal work possible to reach something acceptable rather than great. Thing's can also go the other way of something being so elaborate that it then creates its own set of problems or other variables that then make the job more complex. For instance building a rather OTT crossover you end up with a lot of factoring in of things like DCR in the components used that can complicate things where it's not needed.

 

[solderdude]

Hi perfect and thank you Then if decent specs can be obtained with a low parts count why i see line preamps with dozens and dozens of components ?
design perversion ? or maybe it could be just a marketing strategy To justify an high cost customers must see a lot of stuff inside ?

To be different, to justify prices, because customers like to see that, to improve measured performance.

i dont trust my ears I am 63 and i cannot listen above 14kHz with ears cleaned

We're from the same production year.

My kiss circuit should provide impeccable measurements in terms of noise distortion slew rate and crosstalk at least in line with the better pres around

Noise should be below audible level.
THDistortion should be below 0.01%. Doing any better is technically fine but has no audible consequences.
Slew-rate: when you can reach 50kHz -0.5dB at max output voltage slewrate is more than sufficient. This can be achieved with most circuits, simple or complex and is never really an issue.
Crosstalk... >40dB is good enough. Just play one channel with 40dB attenuation and the other with no attenuation.
Listen with speakers and headphones and see if you can actually hear something on the attenuated side.

absolutely I have seen they have many bjts and jfets inside And i wonder why are not so popular in high end preamps Strange

Because people believe it is better.
Just look at how recordings are made. All that gorgeous recordings probably have passed through many opamps.
Hell... even the vast majority of DACs has opamps in the output ... and then ... with just 1x to 3x gain a pe-amp suddenly isn't good enough for line level signals ?

In my younger years I built 2 pre-amp circuits in one pre-amp with a switch so I can select them.
Both 3x gain.
One with non-biased very old electrolytics (2 in series on the input and 2 in series with the output. It had the well known TL072 that amplified both channels. It also had a simple 7812/7912 regulators and ceramic cap for decoupling.

The other channel was DC coupled, better regulators with compound decoupling in factors of 10 (so 100uF//10uF//1uF//100nF//4.7nF) and used the (in those days as excellent considered) OP27 opamps (one for each channel and separate power supply circuits).
It was for my own 'education' and was made so I could show others what a big leap in sound it made. Switch was on the back.
I never could hear any difference nor could anyone that came to visit.
Of course I always used it in OP27 mode... At least until the day when I had left it on the TL072 section without me knowing. Probably listened to it in that position for years without realizing until I wanted to demo and reached for the switch again.

 

[gino1961]

Can't patent a simple circuit that came straight out of the example in a datasheet. But make it cheap and "good" , attach a story to it and it can can sell.
If you end up adding "enginering pixie dust" to make it "yours", different and even patented especially if you stumble on something that can have a cool name, now the "complicated" has more value. Not necessarily as a product to use and sell, but a design to have for when you sell the company.

The above comment is fictional and has no reference to any persons or other entities of any kind alive or dead in the past, present or future on this or any other planet in this or any other universe known or not yet discovered. (that should cover it )

hi thank you for the helpful advice I will start from something surely working and try to improve on the original
I would try to explain what i mean for KISS
i have a schematic of a line stage with 3 stages input stage-tone control stage-volume-output stage I will keep only the volume and output stage and saving a lot of complexity Tone controls degrade the signal too much They have no place in a preamp that wants to be true to the source signal

 

[JeremyFife]

Consider a pre-amplifier (or a device providing the functionality of one): simple, minimalist design might be good. Adding more components is no longer minimalist, but might be more robust and last longer. Add in much more complex circuitry for DSP and it's better at producing sound.
Minimalist implies as little 'stuff' as possible. Simplicity implies something more. A good solution should be as simple or complex as it needs to be to deliver your requirement

 

[Chr1]

Indeed, but perhaps we're best limiting the discussion to analogue.
Any DSP, or anything digital shouldn't be affecting the sound adversely surely...

 

[Ron Texas]

Many of us don't use preamps. We just plug a DAC with 2V output into a power amp which will reach full power with less than 2V input and don't boost any frequencies in our streamer.