Can anyone tell me what this JFET does in this circuit?

[OneRoomStudio]

...so you could have up to 1.5mV differential worst case voltage across the transformer secondary.

True that the offset would be low, but could still contribute to distortion.

And op-amps run just fine with no load, as long as there is a feedback path. Not sure where the concern about oscillations with no load connected is coming from.

"just fine" might be an overstatement. Here's a good (but brief) TI note on the topic: https://www.ti.com/lit/an/sboa204a/sboa204a.pdf

Since the jack is TRS it would make more sense to connect to tip and ring, not tip and sleeve.

You're right, I mistyped. Tip and Ring.

 

[AUDIO FREQ]

Question:

What are the advantages of electronically balanced outputs compared to transformer balanced? Why did Rupert Neve design his to have both Input and Output transformers? Meanwhile Jensen/Presonus only put Input transformers in things like say, The Eureka?

 

[OneRoomStudio]

Question:

What are the advantages of electronically balanced outputs compared to transformer balanced? Why did Rupert Neve design his to have both Input and Output transformers? Meanwhile Jensen/Presonus only put Input transformers in things like say, The Eureka?

Different times and different technologies. When Neve first developed his circuits in the 1960's, monolithic opamps barely existed, much less were widely available. Transformers were cheaper and more readily available. By the time this Presonus device was designed, IC opamps were much cheaper and more readily available than transformers.
If you're wondering why Neve and others continued to design with transformers, it has a lot to do with nostalgia and aesthetic. In a lot of ways, modern IC opamps are "better," meaning lower distortion, wider bandwidth, better at matching various impedances, etc. But in the last ~30/40 years, people have realized that they like the sound of transformers, imperfections and all. In most cases though, I think it would be hard to hear the difference between a decent 1:1 output transformer and an electronically-balanced IC output. The goal of both is to take an unbalanced signal and balance it, while coloring it as little as possible.

 

[ccaudle]

Here's a good (but brief) TI note on the topic

None of the situations warned about in that TI app note apply to this circuit. That warns against improper input connection, or leaving the output completely unconnected (i.e. no feedback path), neither of which would apply.

What are the advantages of electronically balanced outputs compared to transformer balanced?

Lower cost for equivalent or better distortion performance.

Meanwhile Jensen/Presonus only put Input transformers

Input transformers give some advantages in compatibility with a wide range of signal sources which are difficult to achieve electronically. If you are going to spend money on a transformer the input is the best use.

 

[OneRoomStudio]

None of the situations warned about in that TI app note apply to this circuit. That warns against improper input connection, or leaving the output completely unconnected (i.e. no feedback path), neither of which would apply.

In this situation, they have left R48 connected, which both AC and DC couples the output to an opamp that has no load. At the very least, they should disconnect and ground the left side of R48 to stabilize and disconnect U4A.

 

[ccaudle]

they have left R48 connected, which both AC and DC couples the output to an opamp that has no load

Not a problem. Op-amps have reasonably low output impedance, so there is not much change between load and no load. Usually they are more stable with no load because real world loads have capacitance as well, and capacitance on the output tends to reduce phase margin.

they should disconnect and ground the left side of R48 to stabilize and disconnect U4A

U4A does not have a stability problem, it will just be driving signal into R50, which then goes nowhere because C31 is removed.

 

[OneRoomStudio]

Not a problem. Op-amps have reasonably low output impedance, so there is not much change between load and no load. Usually they are more stable with no load because real world loads have capacitance as well, and capacitance on the output tends to reduce phase margin.



U4A does not have a stability problem, it will just be driving signal into R50, which then goes nowhere because C31 is removed.

While it may not be much of an issue, it seems like a poor practice, and could lead to more noise and unnecessary power consumption. Why not just disconnect it properly? To quote T.I., "Unused op amps in multi-channel devices must be configured properly to avoid possible device degradation, extra power consumption and noise. These unused channels can cause undesired effects on other channels."

 

[Matt Syson]

The original value of 47uF for output caps when combined with a transformer should 'ring' nicely (LF level rise). Not sure what frequency but a good Sowter transformer and 100uF output capacitor gave an interesting rise that I argued about years ago and then as I happened to have said traformer and Audio precision I simply measured the rise (when only 'terminated by 10K as a representative input impedance of whatever the circuit was feeding).
Also in the argument about burning out the transformer coil, this is unlikely as the op amps mentioned have internal current limiting to I think about 30 milliamps. A DC offset will of course increase distortion a bit.
transformers and 'all silicon' input stages both have advantages and disadvantages. I silicon input would struggle to survive a 400 Volt COMMON MODE signal (occasionally found during live gigs with cables going to other areas that are powered from a different source. i was party to a discussion on this 45 years ago where the input caps that would 'isolate' a high voltage (accidental mains application) would be huge (non polarised) so a transformer really was the more obvious choice.

 

[L´Andratté]

If I may ask, why would the input transformer ratio chosen be so relatively low? As far as I know (not very far), to obtain optimum source impedance/noise performance with a FET, like for a tube, the ratio should be as high as practically possible, with regard to other spec. like frequency response...

 

[CJ]

i believe the effects of a dc offset from an opamp into a transformer will depend on the amount of turns the dc is flowing through and the size and material of the core.

remember making an electromagnet in 3rd grade with wire, a battery and a chunk of steel rod? the more turns, the bigger the stuff you could pick up.

so if you are turning an audio transformer into an electro-magnet, it will have to do two jobs instead of one, looking at BH curves with various amounts of DC flux is the way to visually see what is going on.

 

[AUDIO FREQ]

i believe the effects of a dc offset from an opamp into a transformer will depend on the amount of turns the dc is flowing through and the size and material of the core.

remember making an electromagnet in 3rd grade with wire, a battery and a chunk of steel rod? the more turns, the bigger the stuff you could pick up.

so if you are turning an audio transformer into an electro-magnet, it will have to do two jobs instead of one, looking at BH curves with various amounts of DC flux is the way to visually see what is going on.

What is BH curves?

 

[CJ]

https://www.thriftbooks.com/w/trans...rRoCRUwQAvD_BwE#idiq=18892975&edition=4380172

 

[AUDIO FREQ]

https://www.thriftbooks.com/w/trans...rRoCRUwQAvD_BwE#idiq=18892975&edition=4380172

Can't afford it right now, but i'll check the local library (somehow I doubt it).

 

[MisterCMRR]

The choice of input transformer ratio is generally driven by optimizing for minimum equivalent input noise. Generally, the first amplifier stage's noise, both voltage and current, feedback resistor noise, and transformer winding resistance noise play a role. For FET (and vacuum tube) amplifiers, the upper limit to ratio is determined mostly by the extreme difficulty in maintaining HF bandwidth ... it's why Jensen stops at 1:10 to maintain frequency and time domain performance. But sometimes engineers just listen and make the ratio decision based on perceived quality - I suspect that may have been the case for the Presonus. For more insight into the ratio issue, see How to calculate mic input xfmr ratio for tube circuit

Incidentally, I'd like to talk a bit about resistor noise. All resistors, regardless of their construction and materials, make the exactly the same amount of thermal agitation noise based on their resistance. But there is another mechanism for noise in a resistor that does depend on its construction and materials. It's called excess noise, but it only occurs when current flows (when there's voltage across it). This means it's unnecessary to use exotic low-noise resistors where no DC voltage is applied. This includes the resistor across the secondary of a transformer (damping of resonances) and resistor feedback networks around op-amps, etc. But you do need low-noise resistors in locations like biasing resistors for transistors, especially emitter and collector load resistors. This difference between thermal and excess noise is very widely misunderstood.

 

[CJ]

Roderstein.

Bill what is the reason for having partial layers in the 115 K. Mic Input secondary?

 

[MisterCMRR]

The 115KE was a very iterative design, going back and forth between the circuit model of the transformer and it's prototypes. The goal was to manipulate the distributed capacitances (hence resonances and their Q) to achieve a Bessel HF rolloff with no external RC network, just a 150 k resistor. It was quite an achievement. Trying to do that at higher ratios is exponentially more difficult!

 

[CJ]

Did Jensen make the transformer for the ProCo DI box?

 

[MisterCMRR]

As far as I know, no. Generally, makers that use the Jensen DI transformer (12:1 ratio JT-DB-E) will mention it in their advertising. Radial was always our biggest customer for DI transformers, but there were quite a number of others, too.

 

[JohnRoberts]

Incidentally, I'd like to talk a bit about resistor noise. All resistors, regardless of their construction and materials, make the exactly the same amount of thermal agitation noise based on their resistance.

Johnson noise.

But there is another mechanism for noise in a resistor that does depend on its construction and materials. It's called excess noise, but it only occurs when current flows (when there's voltage across it). This means it's unnecessary to use exotic low-noise resistors where no DC voltage is applied. This includes the resistor across the secondary of a transformer (damping of resonances) and resistor feedback networks around op-amps, etc. But you do need low-noise resistors in locations like biasing resistors for transistors, especially emitter and collector load resistors. This difference between thermal and excess noise is very widely misunderstood.

I recall or think I recall reading a white paper from back in the 70s from one resistor brand claiming that their carbon composition resistors expressed lower excess noise than other types of resistors. This is contrary to popular wisdom about resistor media (metal film or foil is considered better). Excess noise voltage is caused by granularity or lack of smoothness inside the conductor. Higher power resistors could be better than low power resistors of the same value.

I tried to calculate the significance of excess noise in typical phantom powered mic preamps where the 6.81k phantom resistors can have ten to twenty volts dropped across them in use. My judgement back then was that excess was not significant compared to the other noise sources involved. But I am not sure I calculated the uV/volt/decade calculus.

I can imagine this being a concern in vacuum tube designs with significant bias voltage across resistors. I am not a tube guy so don't know.

JR

 

[CJ]

Bill what is the purpose of reverse winding a section of an audio transformer?

Is it strictly for noise rejection or does it improve response?

Can it affect inductance?

Can transformer insulation types affect sound?

Thanks!