Descrete jFET versus IC op-amp quality

[W1DAN]

Hi folks!

I am newly registered here and have looked around a bit, but not comprehensively, so I apologize if this has been covered already.

Since about 1980 or so I have occasionally built audio devices with op-amps. These include a radio station style audio console, a multiband audio compressor/limiter, an NAB preamp, and especially phono preamps. These used the NE5534, TL074, '84, and the AD797, usually with feedback style EQ, but sometimes flat where I EQ in a computer.

Last summer I found at a yard sale a Borbely EB585 (Old Colony KT-2) home preamp ($10.00!) that includes a discrete RIAA preamp (see Audio Amateur magazine 04/85 and 01/86). This preamp uses a single set of 2SK147/2SJ72 jFET's in the front end and passive EQ.

I finally fired this unit up a couple of months ago and was very surprised at the clarity of the sound over my past IC op-amp units. The Borbely preamp is not the quietest, bus the instruments now sound like instruments and the sound stage is accurate.

So I am now on a trek to learn why technically the discrete preamp sounds better than a well-made IC unit. This is with the idea of specs like open loop bandwidth, slew rate and distortion that are comparable with the discrete jFET unit.

I have read Erno's newer design articles and am wondering if I should chase after the idea of breadboarding his last RIAA design, or just get some of the newer op-amps. I see a few newer op-amps out there and wonder if these can match or exceed the performance of the Borbely unit. I have also enjoyed a few designs posted here-some excellent work!

Thank you for your thoughts!

Dan

 

[oon_the_kid]

Hi folks!



So I am now on a trek to learn why technically the discrete preamp sounds better than a well-made IC unit. This is with the idea of specs like open loop bandwidth, slew rate and distortion that are comparable with the discrete jFET unit.

I have read Erno's newer design articles and am wondering if I should chase after the idea of breadboarding his last RIAA design, or just get some of the newer op-amps. I see a few newer op-amps out there and wonder if these can match or exceed the performance of the Borbely unit. I have also enjoyed a few designs posted here-some excellent work!

Thank you for your thoughts!

Dan

You can try hanging arounf the Pass part of this forum under amplifiers. Nelson Pass also has a dislike for Opamps, and his preamps and I/V converters are all discrete. If you go to his websites, he will discuss a bit more on his approach .

Oon

 

[jcx]

You can try hanging arounf the Pass part of this forum under amplifiers. Nelson Pass also has a dislike for Opamps, and his preamps and I/V converters are all discrete. If you go to his websites, he will discuss a bit more on his approach .

Oon

which kind of misses the point if the poster is really looking for balanced opinion on the relative merits of op amps isn't it?

mc phono pre requires discrete or xfmr front end to get acceptable noise performance

for mm cartridges some of the lower noise jfet op amps can be used

at line levels op amp noise isn't an issue, many applications can use op amp based circuits with measurable errors held to below the noise floor

many will still be using strawman examples from the 1st few generations of op amps from the 70's to "prove" the superiority descrete circuits in any audio application

technolgy has considerably advanced since then, if you want a contrasting expert opinion you might also seek out posts by Scott Wurcer for one


another intellectual line to pursue is perceptual psychology and psychoacoustics which suggest much "just listen" opinion is naive and likely overwelmed by other considerations than actual circuit properties

 

[Conrad Hoffman]

I'm always curious about the why of things. The first thing I'd do is compare the RIAA response. It would be unusual if the two units matched closely. There's also the noise issue. The presence of broadband noise changes how we perceive response. I assume I don't even have to mention gain, but if one unit tends to get set higher than the other, it will affect the preference. Everybody says louder always sounds better, but I don't know if that's entirely true. Then there are things like output impedance and how it drives cables. And on and on.

 

[W1DAN]

Hi:

Thanks for the replies. I'll look at Nelson's section.

I should define this a little better. I am not worried about noise and assume the RIAA EQ differences are minimal.

I am looking for things such as distortion, Class of operation such as A and B. Is there a valid technical reason I should be working with discrete JFETS in an RIAA preamp if there is an excellent IC op-amp out there that can match the performance of discrete transistors?

Working with an IC is easy, discretes require more effort.

Thanks,
Dan

 

[jcx]

http://www.fortunecity.com/rivendell/xentar/1179/projects/aaphono/aaphono.html

is one example that was designed by someone I would characterize as having "subjectivist" leanings with extreme tweaking tendencies


all op amp stages except the output typically operate Class A - the output may be ccs biased for Class A operation up to some load, Vswing requirement
I have used composite op amp circuit which operate the output push-pull Class A up to 400 mA - multiple op amps used in composite architectures for a "single function" can improve greatly over single op amp circuits
“unmeasurable” low distortion is possible with composite op amp circuits

 

[AndrewT]

have a look at Salas RIAA
many are building this.

It effectively uses two discrete amplification stages in place of two opamps.

 

[jacco vermeulen]

Working with an IC is easy.

Says who ?

 

[EUVL]

If you want to try your hands on a discrete JFET RIAA, I would suggest this (I did) :

http://www.diyaudio.com/forums/analogue-source/54658-all-jfet-open-loop-riaa-pre-amp.html

You can replace 2SK389/2SJ109 with 2SK170/2SJ74 matched pairs.

http://www.diyaudio.com/forums/solid-state/135154-next-best-thing-2sk389-2sj109.html


Patrick

 

[buzzforb]

May help.

http://www.firstwatt.com/pdf/art_diy_opamp.pdf

 

[EUVL]

> http://www.firstwatt.com/pdf/art_diy_opamp.pdf

Been there :

http://www.diyaudio.com/forums/pass-labs/120445-pass-discrete-opamp-dip-8-package.html

implemented for a phono :

http://www.diyaudio.com/forums/pass-labs/120445-pass-discrete-opamp-dip-8-package.html#post1488703
http://www.diyaudio.com/forums/pass-labs/120445-pass-discrete-opamp-dip-8-package-2.html#post1835752

 

[buzzforb]

Exellent EUVL. I was just providing possible reading material for OP.

 

[Conrad Hoffman]

I should define this a little better. I am not worried about noise and assume the RIAA EQ differences are minimal.

I am looking for things such as distortion, Class of operation such as A and B. Is there a valid technical reason I should be working with discrete JFETS in an RIAA preamp if there is an excellent IC op-amp out there that can match the performance of discrete transistors?

If you don't carefully examine the basics, which have far more to do with the sound than the other stuff, and rule them out, you're likely to go 'round in circles like so many others. Treat the circuit like a black box and fully characterize it, assuming nothing.

There are certainly excellent opamps these days, also excellent composite opamps you can build with JFET front ends, and excellent discrete circuits. IMHO there is no reason you can't get identical measured and subjective performance from any of them.

 

[W1DAN]

Conrad:

Your statements are more of what I am leaning towards exploring. I see many folks making phono preamps with different topologies. Tube, germanium, JFET, bipolar, op-amp, steel needle and cone, etc.

I am spending my time working to fully understand the technical parameters of the various forms of discrete JFET preamps and how these parameters affect the sound I hear. In years past I had done this with the NE5534 and AD797, and now need to study the latest IC op-amps (BTW, to me IC op-amps are easier as once you design for them in their preferred parameters, you just plug the chip in. With transistors, you have more interfacing parameters to consider).

One wonder I have is if a class-B op-amp stage can have such low distortion that it has essentially the same performance as a class A discrete JFET amp.

So I am not asking whether JFET designs exist, or which ones are popular (which I am discovering here), but rather should I invest substantial time and effort building a JFET preamp if any model op-amp today, or combo is equivalent.

Thanks everyone and keep the good ideas flowing!

Dan

 

[Conrad Hoffman]

To complicate things further, you can drag an opamp output into class A operation very easily. I like the ease of opamps and can't imagine designing anything discrete that's better than the National LME opamps. Not that somebody can't, just not me! I don't use MC cartridges, but that's the only place where discrete makes any sense to me, and it may not even make sense there. JFETs solve noise problems with sources having higher impedance. For low impedance sources heavily biased (and maybe paralleled) BJTs seem like the way to go, yet people still (sometimes) use JFETs. Go figure.

 

[john curl]

It is almost impossible to make a class B design as low in distortion as a class A design, OPEN LOOP. And this is the 'key', and not believed in by most of the skeptics here. However, we, who follow this knowledge, have become the more well known designers by making successful audio products.

 

[Bonsai]

Come on John, you are confusing the issue , and may I respectfully suggest, abusing your 'guru' power on this one..

Quite a few posters here have already remarked that it's very easy to bias an opamp output stage into class A. The class A/class B thing wrt opamps vs discrete designs is a 100% non-issue ever
since Walt Jung showed us how to do it years ago.

Build a discrete circuit for the fun of it and if you believe it sounds better (a double blind test is the only way to scientifically verify this BTW) and build an opamp based design if you want the best measured performance. My personal opinion (although I have not DB'd it I have to admit): they may sound different, but I doubt you could say the opamp version would be inferior.

 

[W1DAN]

Connrad:

This is what I am trying to determine. Will an IC such as the LME series or whatever be able to come close to a class-A discrete JFET for MM or even MC RIAA preamps. My experience is currently limited to the LM394 feeding an op-amp, the NE5534's capabilities and the AD797.

So I am reading articles to fully understand Borbely's designs (my early EB585 unit can be improved with a quieter supply, paralleling transisitors and cascoding the input stage-this proves I am learning something!). Erno's later articles explain the advantages of cascoding and a few other advances since he had started making kits. One of his main points was to make a clean amplifier before adding feedback. To me this makes sense, and if one looks at IC op-amps with their open loop gain and distortion, much seems to be cleaned up with FB.

Nelson Pass in his DIY web page has an interesting article that I read last night and has been talked about on this board:

http://www.passdiy.com/pdf/distortion_feedback.pdf

Here, Nelson explains that FB reduces the total percentage of distortion, but the remaining distortion becomes more complex. He states that IM products can be very complex and be the cause of masking detail.

One of the first impressions I had of listening to Erno's early EB-585 preamp was the high end seemed to be down in level. To me this may be the lack of IM mixing products. The high end is very clean and a singer's "esses" can actually be pretty sharp to the ear. I have not experienced this with my op-amp circuits.

John:

I figured one could not make the exact same distortion levels with class B as class A. But today, can one make it so close so that the difference is not heard? Yes I have seen folks biasing IC's to class A. I need to read about this option more. Thanks for your input.


Bonsai:

Have you had a chance to directly compare a discrete JFET preamp with a well-designed class-A biased IC circuit? Would be interesting to hear anyone's impressions of such a comparison. It is always hard to correlate numbers into subjective qualities.

Thanks all and happy Father's day!
Dan

 

[john curl]

Dan, not in my experience.

 

[Conrad Hoffman]

Hey, the LM394! I used the preamp design out of the National book using an LM394 going in the comp leads of an LM318 for many years. Worked OK, had very low noise, but I don't think it was as good as just using an LME49710 (I rarely use duals). My guess is it had poor open loop performance because of the 318 and did well only via feedback. The LME49710 has excellent open loop performance compared to opamps of the past. I don't think you can get an LM394 anymore. It was a true paralleling of many transistors to get the statistical noise improvement. The subs I've seen are not.

IMHO, the quality and design of the RIAA network is the biggest factor in sound quality, but for small signals, if class A appeals, there's no reason not to do it. It's not like the power dissipation matters the way it does for a power amp. (I was just going through some old articles and came across the Pass class A power amp from Audio magazine many years ago!) I don't think you'll ever reach a definitive answer on opamps vs. discrete, but wondering if discrete class A is a tad better will drive you nuts! Just build it.