John Curl's Blowtorch preamplifier

[PMA]

Definitely. I just have OPA627 in the DC servo. It works very well, and does not degrade sound John should try.

 

[syn08]

Wouldn't the OPA827 be a better choice?

No, it would certainly be an overkill. The new OPA827 is amazing in the signal path, it's the best JFET opamp I have ever tested (as a trade between gain, noise, DC performance, etc...). I've tried OPA827 in YAP with excellent results. But it's not in production yet and can't be so far purchased. You have to apply for a couple of samples from TI.

 

[PMA]

Way a bit slow, unfortunately. Slower than OPA627. But still fast enough for the DC servo.

 

[Nelson Pass]

Do you mean that they just don't STEAL your name? I am missing out on something.

They have certainly tried, but usually without having paid for the design, either.

And there are a couple who have paid for the design but want their own name
on it.

You get all kinds.

 

[Wavebourn]

Back to resistors;
how good AD5293 are sonically?

 

[syn08]

The JC-1 input stage can be found here.

I am dissapointed. Ultra low noise 2SK170 input stage, then 100k/3k83 feedback resistors?

John, what is the measured noise figure of the JC1? I don't dare to tell what my estimate is, based on calculations http://focus.ti.com/lit/ml/sloa082/sloa082.pdf

Only the noise in a 100k resistor at 25C (298K) over the audio frequency range of 20 Hz to 20 kHz is Eth = 5.73 uV or –104.8 dBV. I'm looking at the JC-1 spec and I get completely different numbers:

Signal-to-Noise Ratio, input shorted
> 122 dB, IHF A-weighted, bias set to Low
> 120 dB, IHF A-weighted, bias set to High
> 113 dB, Unweighted, bias set to Low
> 111 dB, Unweighted, bias set to High

There is a 5pF across the feedback resistor that I didn't consider, transforming the white noise in pink-ish 1/f noise (the corner frequency is on top of my head around 300kHz) but still... Please, tell me I'm missing here something!

 

[Wavebourn]

Can't you explain why you calculate for 100K, if it forms a divider with 3K83?

 

[syn08]

Maybe you are specifying the input shorted unweighted equivalent input noise? Then it would make some sense, I get a feedback network contribution of -119.8dB.

 

[syn08]

Can't you explain why you calculate for 100K, if it forms a divider with 3K83?

That would apply if you specify the input shorted unweighted equivalent input noise.

 

[john curl]

You guys worry too much. THIS IS A POWER AMP. Not a phono stage, or even a line stage. I purposely scaled UP the resistor array in order to NOT load the fragile open loop fet follower. Lower distortion vs slightly lower noise. Wavebourn's right, the 100K resistor doesn't add much at all.

 

[KBK]

Hypothetical time, guys. If you were given $1.00 worth of parts (at the manufacturing $ level) on that circuit, where would you spend it? Is that 5pf already a polystyrene? (for example, the $1.00 could be about 5-6 holco/prp resistors at manufacturing cost levels)

 

[john curl]

I give up. Tell me?

 

[hitsware]

Cosmetics. Even @ board level.

 

[john curl]

No, actually I really do wish that I could use polystyrene 5pf caps and OLD Holco resistors. Unfortunately, they are not practical in this case.

 

[PMA]

Syn08, the 3k83 sees low output impedance of the input jfet buffer.

 

[KBK]

I give up. Tell me?

I have my guesses John. But I'm sure you've tried them all.

My first shots, if let's say I had opportunity to change/swap only ONE part, it would be...at first glance.. R36. But then again, I'm not designing them, I'm only finding myself doing this exercise for personal amusement on any design I come across, for the past 20 years. Not designing them puts me behind the 8 ball to a large extent, but the constant mental (and then actual single change) exercise helps out a bit. Except I rarely have the luxury of doing it with a schematic.

 

[john curl]

I wish that I could increase the load, not reduce it.

 

[john curl]

R36 cannot be easily reduced unless I personally make amps.

 

[KBK]

I'm talking strictly the quality of the resistor. To a ultra low ppm unit. If that was to be swapped out, then do R43, R45 to R48 and R35, all to the same brand and wattage as R36. PRP's best, or similar. This should increase the perception of speed, slam, stability in the bass, drive, a slight opening of the highs and softening of the treble/mids (increased warmth) into slightly more delicacy and liquidity. Soundstage and depth (detail in complex passages) should also improve in step with the others. All a function of the matching and the tempcos of the resistors as a set-as an affect on the flow of the circuit. Once again, just my guess. The effects will be subtle, but definitely there. That's seven resistors and now I'm in debt. I've overspent my $1.

 

[Wavebourn]

No, actually I really do wish that I could use polystyrene 5pf caps

I would twist a couple of wires instead.