Douglas Self Line Input compared to Mackie line input

[Newmarket]

I still have some gold plated phono jacks in my back lab , anybody interested?

JR

Depends. I assume we all know that the thickness of the plating affects the high frequency IMD impulse response
Whilst the "carat" of the gold alters the THD coefficient

 

[JohnRoberts]

Depends. I assume we all know that the thickness of the plating affects the high frequency IMD impulse response
Whilst the "carat" of the gold alters the THD coefficient

for TMI about me and gold plated stuff. Life Hacks

Yet another Peavey story so don't click if you don't want to see it.

JR

 

[thor.zmt]

for TMI about me and gold plated stuff.

To gold "plate" base metals is done to prevent tarnishing.

Have you ever seen a 20 Years old nickel plated RCA shell that spent this time in a city by seaside in the tropics? The gold-plated sockets on the the same machine are still nice and shiny.

From an electronics viewpoint, thermo-couples are made from dissimilar metals. Directly plating gold onto base metals with good adherence is not a cheap process, ENIG (electroless nickel immersion gold) is.

The best connector is no connector, ideally high pressure clamping of wire on wire using non-conducting clamping structures that reliably exclude air and can be nitrogen or co2 flushed before sealing.

As this is not practical, my preference is for heavy gold plate over heavy silver cladding on phosphor bronze (to keep springiness) with airtight covers for unused connectors. That too is hardly cheap.

So in practice, most of the time choice is the cheap plastic Cliff XLR stuff which is silver over bronze, so ideally use some KY, Vaseline or silicon oil before first insertion to preserve the silver. I also really like Neutrick Speakon and Power on, with similar caveats.

This may seem obsessive, but don't get me started on the effect of cables with nickel plated BNC's used with an AP2 on being able to measure low HD after a year (in a city by seaside in the tropics, despite 24/7 AC).

Underestimate connector issues at your peril. This is not about cost/expense, but what works.

Thor

 

[Dimitree]

Ok, up to now we have looked at design principles.

Let's instead have a fully elaborated design, such as I might recommend to a client who needed a low noise, low distortion, low cost, high real world CMRR, balanced Input and SE output circuit for say a Mixer, Audio Interface or Active monitor. It is still based on the "double inverted" topology, but further optimised.

View attachment 117928

I started laying out a PCB for a bunch of channels using this schematic.
Which part do you think is the best to split the schematic in two PCBs? I'd need a PCB for the connectors (on the back of the chassis) and one for the controls (GAIN potentiometer) on the front panel.
Even in SMD format, I don't think I could do a clean" layout with a single PCB and a bunch of wires going to the connectors, so I'm wondering if I can split the signal after the two buffers, so they will be placed on the PCB with the connectors, and the two inverting opamps will be placed on the PCB with the GAIN potentiometer. A flat cable would connect the two circuits.
Is that a bad idea?
Another option would be to put everything on the connector PCB and just the GAIN pot on the front panel PCB, but I think the node/connection between the gain potentiometer and the opamp is really sensible

 

[thor.zmt]

I started laying out a PCB for a bunch of channels using this schematic.
Which part do you think is the best to split the schematic in two PCBs? I'd need a PCB for the connectors (on the back of the chassis) and one for the controls (GAIN potentiometer) on the front panel.
Even in SMD format, I don't think I could do a clean" layout with a single PCB and a bunch of wires going to the connectors, so I'm wondering if I can split the signal after the two buffers, so they will be placed on the PCB with the connectors, and the two inverting opamps will be placed on the PCB with the GAIN potentiometer. A flat cable would connect the two circuits.
Is that a bad idea?
Another option would be to put everything on the connector PCB and just the GAIN pot on the front panel PCB, but I think the node/connection between the gain potentiometer and the opamp is really sensible

Yes, splitting off the final stage (including pot and resistors) is fine. Always look at relative impedances.

Op-Amp outputs are in effect zero ohm. So they handle driving wires well (small build out resistor if needed), no signal loss or crosstalk.

Thor

 

[Estomagado]

Ok, a noob question. Please be patient

Long ago, back in page 2 here, @thor.zmt mentioned (and explained) the "double inverted" topology with front buffers added (and that trick to tighten resistor tolerance for better CMRR) as a better implementation for lower noise.

My question is: how does it compare to a standard instrumentation amplifier, and why the "double inverted" is better, noise-wise? The "double inverted" uses 4 op amps, whereas the instrumentation amplifier uses 3 - one less active part hissing away on our signal, or so it seems for my untrained eyes.

For reference:

Instrumentation amplifier:


Source: National Audio Handbook (1977)
https://archive.org/details/NatSemiAudioHdbk1977_201801/page/n51/mode/2up


"Double inverted" design on Thor's post:

 

[merlin]

Long ago, back in page 2 here, @thor.zmt mentioned (and explained) the "double inverted" topology
My question is: how does it compare to a standard instrumentation amplifier, and why the "double inverted" is better, noise-wise? The "double inverted" uses 4 op amps, whereas the instrumentation amplifier uses 3 - one less active part hissing away on our signal, or so it seems for my untrained eyes.

Interesting question. Just eyeballing it I can't decide which would be quieter (figuring the noise of diff amps makes my brain ache).
For that matter, how would they compare to an instrumentation amp where the final stage is a superbal? (Yes I know this doesn't make a lot of design sense, but it would use roughly the same parts as the double inverted so is an interesting comparison). I think this is a job for a circuit simulator!

EDIT: A very quick and dirty sim with the three circuit configured for the same overall gain suggests the double inverted is only slightly noisier than the instrumentation amp. The superbal looks better because I had to put gain up front, so perhaps not a fair comparison -I'll leave that for you to decide. The graph below shows what my sim thinks the output noise spectrum will be. I haven't bothered to weigh this against any other relative advantages each circuit might have, and real world results may deviate from sim models of course.

 

[Matt Syson]

When you are designing a circuit (or borrowing one from someone else) you have to juggle a whole range of different parameters and not just use the 'borrowed' circuit UNLESS you are simply copying and the overall requirements are identical. Resistor values and tolerances are not often discussed snd in some confugurations resistor values are more significant for noise and even distortion than SOME op amps.
The Nat Semi design IS low noise compared to some others IF you want 54 dB of gain which is not variable unless you compromise on other paramdeters. Simply bunging in a much later 'wonder chip' may improve some aspects but for example the 50K feedback resistors introduce a significant amount of noise.

 

[JohnRoberts]

When you are designing a circuit (or borrowing one from someone else) you have to juggle a whole range of different parameters and not just use the 'borrowed' circuit UNLESS you are simply copying and the overall requirements are identical. Resistor values and tolerances are not often discussed snd in some confugurations resistor values are more significant for noise and even distortion than SOME op amps.
The Nat Semi design IS low noise compared to some others IF you want 54 dB of gain which is not variable unless you compromise on other paramdeters. Simply bunging in a much later 'wonder chip' may improve some aspects but for example the 50K feedback resistors introduce a significant amount of noise.

That nat-semi design is from a circa 1970s application manual. Nat semi published some good app notes back in the day. They were promoting a chip (LM387 a variant on their LM381) designed for phono preamp or tape head amps. That part was OK for early 70s but designed for single supply use and only stable for closed loop gains >10x (20dB).

I would try to forget you ever saw that one. in the context of modern ICs.

JR

PS; FWIW I used the LM387 in a DJ mixer kit article I published back in 1978

 

[Matt Syson]

I am sure we could promote the original OC71 transistors as 'the best thing since sliced bread' for Hi Fi use and many would fall for it because they are Germanium transistors.

 

[thor.zmt]

Long ago, back in page 2 here, @thor.zmt mentioned (and explained) the "double inverted" topology with front buffers added (and that trick to tighten resistor tolerance for better CMRR) as a better implementation for lower noise.

My question is: how does it compare to a standard instrumentation amplifier, and why the "double inverted" is better, noise-wise? The "double inverted" uses 4 op amps, whereas the instrumentation amplifier uses 3 - one less active part hissing away on our signal, or so it seems for my untrained eyes.

IF ALL ELSE = ALL ELSE

Instrumentation has ~3dB less noise, IF Op-Amp noise is absolutely dominant and resistor noise is minimised. And SuperBal has less noise than Instrumentation.

As the TL07X has 18nV|/Hz Ein this is given. In 2024 there is zero justification to use such a poor part.
On the other hand, the "mixing amp" in the dual inverted option operates at zero common mode, which may result in lower HD, less EMI sensitivity etc.

Also we do not get many triple Op-Amp's, but Quads and Dual's you now have a spare OPA looking for a job.

The TL07X is a utter tosh part for audio in 2024, well it also was in 1985. Swap them out for OPA1679, which in turn is a symbolised OPA1654 with less testing and possibly lesser binning on DC Offset and some other parameters. In practical use I found OPA1679 = OPA1654.

Ein is 4.5nV|/Hz at 1kHz or 12dB less than TL07X, 3V into 600 Ohm gives -120dB THD & N. Price in 1kU is 40 cent USD for the Quad. Or 10 Cent per OPA.

Also, switch TINA output for noise to SNR, bandwidth 20-20k, reference level 0dBm (0dBu = 0.775V) for dBm/u reading or 1V for dBV or expected operating level for actual SNR @ operating level.

I should add something. My own use of Balanced connections is somewhat unusual. I actually happen to think that most of the time balanced connections are utter tosh and worse than SE, UNLESS REAL TRANSFORMERS ARE INVOLVED AT BOTH ENDS.

So my balanced connections are all modified to look like this (one or the other version):

And usually the connection is switched to SE.

Personally, in 99% conditions in a small studio or domestic setup, there is no advantage to Balanced connections, instead of (barely) hiding earth/ground loops inside the CMRR of a balanced input, simply sort this out properly and there is no need to hide something not there.

I content that the above Circuit, in SE mode or in Balanced is the lowest noise option and usually also lowest distortion option.

The theoretical drawbacks may be relevant in stage gear or something sold to non-technical pee-pull, in our own studio or home they should not be.

I am sure we could promote the original OC71 transistors as 'the best thing since sliced bread' for Hi Fi use and many would fall for it because they are Germanium transistors.

Actually it is NOS 12AX7, Telefunken Smooth Plate Tube, not Geranium Transistors.

Thor

 

[abbey road d enfer]

Personally, in 99% conditions in a small studio or domestic setup, there is no advantage to Balanced connections, instead of (barely) hiding earth/ground loops inside the CMRR of a balanced input,

IMO it's the only valid reason to do it in a semi-controlled environment.

simply sort this out properly

Which is sometimes a daunting task.
When you start to mix consumer equipment you may end up in situations where the only solution is to balance signals. I discount the solution that uses cheat plugs to lift earth.

 

[Dimitree]

So my balanced connections are all modified to look like this (one or the other version):

View attachment 126316

great infos, thanks for sharing.
could you describe the input/output LC networks? what values do you use? Also, I noticed two symbols for ground, AGND and the one used on C13/C4 for example. Are these actually the same net?
What are the factors that make you decide when using V1 or V2 input circuits? You never include input buffers and AC coupling caps?

 

[abbey road d enfer]

Also, I noticed two symbols for ground, AGND and the one used on C13/C4 for example. Are these actually the same net?

It is clearly indicated that one is the Analog ground, i.e. the node to which all other nodes are referenced, in other words Zero Volt, and the other is Chassis ground.
In mains powered systems, Chassis ground is usually connected to Protective Earth, but in other systems, it's the casing that acts as an electrostatic shield.
In most cases, Analog Ground and Chassis ground are connected, but it's not an absolute necessity.
In the case of mains-powered units, it is mandatory for safety reasons, but it is often detrimental to performance.

 

[Estomagado]

I am a real noob
I did not know of the existence of the Superbal.

Thank you, everyone

 

[Matt Syson]

Ted Fletcher (Of Alice, the mixer manufacturing company) used the term 'Superbal' in the very early 1980's and may or may not have had connections via the BBC equipment department at the time. Steve Dove, who I think was associated with Alice at around that time may have been another involved in working on this scheme. Alice was an 'opposing firm' (competitor) to where I worked at that time so for obvious reasons some information was not shared as the fight to get contracts for the new (at that time) broadening of radio broadcasting in the UK grew many companies. Ted also wrote a good article for Studio Sound magazine that explained why 'ground' may not actually be ground but is simply the reference terminal for the interconnection between stages that have gain.
A 'balanced' or 'differential' input is very easy to obtain and only uses one or two more components much of the time, and can improve SYSTEM performance with respect to picking up mains related and other gremlins.

 

[JohnRoberts]

I can't speak for Steve Dove but IIRC he told me that he consulted for Alice on design*** of some broadcast mixers. I knew Steve back in the day as another console designer, a relatively small club.

JR

PS: In fact After I published my "performance limits in console design" article in RE/P magazine (1980), Steve told me at a later trade show that he borrowed my current source summing bus idea and used it inside a broadcast mixer. I never chased down any schematics to confirm but he said that alternate sum amp topology was more resistant to RF rectification.