Douglas Self Line Input compared to Mackie line input

[Dimitree]

While looking at the schematics for the Mackie Onyx 1220 mixer, I noticed they implemented a line input with +/-20db trim using only two opamps stages.



I'm wondering how this approach compares to the praised Douglas Self line input proposed on his book. The Mackie looks more useful since unity gain should be at center of the pot, and there is more gain range.. but what is the tradeoff compared to the Self approach (below)?

 

[Dimitree]

anyone?

 

[RSRecords]

Free from pot dependence?
Both work really well. The Self should have lower noise with the 4k7 resistors maybe?
Don't really know the reason for the parallel R4, R8 in the self design.

 

[ccaudle]

Don't really know the reason for the parallel R4, R8 in the self design.

That keeps the impedances balanced on the input diff-amp. It could be drawn as a single parallel equivalent, but by drawing as two separate resistors it is (should hopefully be?) obvious that they are matching the impedance of R2 and R7.

what is the tradeoff compared to the Self approach

The first thing which jumps out is that the Mackie circuit always has 10k in series with the second stage op-amp input, which has a noise penalty compared to the Self circuit.

 

[Dimitree]

Thanks! What is the amount of noise penalty with that 10K in series? Also, using a THAT 1200 instead of the first opamp and R1/R2/R8/R9 would be possible?

 

[arthurdb]

The Mackie approach has overall, a better 'gain' control : symmetrical boost and cut, perfectly centered around 0dB, and finer control towards the center of the pot's travel. This circuit is pretty common and can be found in high end stuff, I recall seeing it on at least one Studer console (with a 30dB range IIRC)

Self's topology has the advantage of being more tolerant of high level signals. If you have a signal hot enough to overload IC3A in the Mackie's schematic, you can adjust the gain all you want, you won't prevent the first stage from clipping. Since the gain control is wrapped in the feedback loop of the diff amp in Self's design, lowering the gain will lower the output of that first amplifier. Hence, the volume control can actually help you avoid clipping the first stage.

The tradeoff is that 0dB will be offset by a little (not a big deal, that can be compensated for in later stages), and the boost/cut won't be symmetrical. And you'll get finer control towards the beginning of the pot's travel.

In both designs the range can easily be increased or decreased, within practical limits, I don't think you can realistically get 40dB of range with Self's circuit.

 

[thor.zmt]

While looking at the schematics for the Mackie Onyx 1220 mixer, I noticed they implemented a line input with +/-20db trim using only two opamps stages.

View attachment 117411

Technically speaking, they used a fixed gain balanced input, followed by an active gain/attenuation adjustment stage that looks a like an Equaliser without frequency dependent element.

This circuit creates a noise penalty twice.Especially the gain adjustment circuit in effect first attenuates the signal by around 10dB before boosting the now much more noisy signal back to nominal (0dB gain setting).

I'm wondering how this approach compares to the praised Douglas Self line input proposed on his book.

I am not sure who "praises" Mr. Self and for what exactly.

The Mackie looks more useful since unity gain should be at center of the pot, and there is more gain range. but what is the tradeoff compared to the Self approach (below)?

View attachment 117412

I personally do not consider either design particularly good. The "Mackie" carries a large noise penalty, the "Self" severely limits adjustment range.

Mr Self has a few better options in his book.

If we need gain trim, the "double inverted" circuit is better, in my view.


Something like this in my view beats of the others all around. Using a log pot you get ~ -6dB gain at around centerpoint. Lowering gain lowers noise from an already low baseline.

As all Op-Amp's operate inverting we have a noise penalty, but next to resistor noise it's not that relevant. On the positive side, any common-mode distortion doesn't happen.

Also, if we have another inversion in the full system (say tone control or mix amp etc.) we can flip input polarity and have a very simple SE input, or even more important a high impedance (1M) Instrument input using a nice J-Fet follower ahead of a resistor into mix input.


As long as the jack has a switch that shorts the Jack Tip to ground when not plugged in the impact on noise is nominal. If not, extra switching is needed. A Mic-Input can be added using the classic "British Mic-Pre" frontend (e.g. Elliott Sound Project 66). Using a coarse gain switch in (say) 12dB steps can be added, fine adjustment with gain trim. Voila, simple universal input for a mixer or similar.

On the other hand, if a balanced out is needed (say to feed a AD input), the Birt Circuit is preferable, though gain adjustment becomes more complex.

A High-Z input for this can be made from a J-Fet operates as phase splitter. A Mic-Input can be added again using the classic "British Mic-Pre" frontend (e.g. Elliott Sound Project 66). And we have a very simple and high performance Mic/Line/Inst Input to ADC circuit .

Thor

 

[JohnRoberts]

I am not sure who "praises" Mr. Self and for what exactly.




Thor

Apparently lots of people.... A quick search reveals that he has written 6 books in the subject of audio design. I never read one myself, but 6 books don't get published if a significant number of readers didn't appreciate (buy) them.

JR

 

[k brown]

'Self' promotion?

 

[Dimitree]

thanks for the suggestions,
what do you think about this approach? using the gain stage only if needed, with a switch. And probably replacing the first opamp with a THAT 1240.
Also, I'm wondering if direct output would be better if wired directly on the line input connections

 

[thor.zmt]

Apparently lots of people.... A quick search reveals that he has written 6 books in the subject of audio design. I never read one myself, but 6 books don't get published if a significant number of readers didn't appreciate (buy) them.

I have those books.

They are good compendiums of what has been done before Mr. Self, heavy on objective measurements, moreso than most, something I greatly appreciate and value Mr Self's work for.

Would I commend Mr. Self for the designs he him-self (duh) presented or would I second his rude dismissal of rather clever designs that seem more subjectivist inspired, nope.

But yes, he has always been excellent at Self promotion (duh).

Thor

 

[thor.zmt]

what do you think about this approach? using the gain stage only if needed, with a switch. And probably replacing the first opamp with a THAT 1240.

Not a lot, tbh. Look at the schematic in my post up. It will do objectively and subjectively better.

Also, I'm wondering if direct output would be better if wired directly on the line input connections

You show an SE Direct out.

Most "sound cards" worth having will have balanced inputs.

As remarked, Mr. Birt's rather ingenious design is excellent for that, it's covered in one or more of Mr. Self's books.

For BAL in SE out I think "double inversion" is best, preferred with intentionally inverted polarity out.

I have done these fully discrete, with 1Mohm per polarity input impedance using strictly discrete circuitry based around the so called Sziklai or inverted Darlington building block, pure Class A, over +33dBu undistored input and an Ein of around 1uV at -6dB. Naturally with gain trim.

That is performance way past IC solutions and in SMD quite cost/space efficient fot mass production.

For BAL in BAL out either the Birt circuit or a fully diffential amplifier [FDA - e.g. OPA1632) is the correct choice. If we approach Birt as black box it is an FDA, BTW.

Using an FDA/Birt also gives a valid SE out with full CMRR. so Birt is more versatile.

Thor

 

[gyraf]

I'm wondering if direct output would be better if wired directly on the line input connections

your solution solves potential problems with balanced/unbalanced sources and soundcard(s), I'd keep it like this.

I would add a twin for C16 in series with R16 - just to keep CMRR still-good at loow frequencies for your soundcard/direct output

your "gain" switch will click, and your fader will probably scratch over time: they're seeing the DC input bias current of IC3B. Probably better to DC the no-gain-path with an electrolytic also.

/Jakob E.

 

[Dimitree]

your solution solves potential problems with balanced/unbalanced sources and soundcard(s), I'd keep it like this.

I would add a twin for C16 in series with R16 - just to keep CMRR still-good at loow frequencies for your soundcard/direct output

your "gain" switch will click, and your fader will probably scratch over time: they're seeing the DC input bias current of IC3B. Probably better to DC the no-gain-path with an electrolytic also.

/Jakob E.

thanks for the suggestions,

I added another cap to ground, on the direct output,
I moved C15/R14 after the "gain" switch, but I'm not sure if that is correct.. because C15 has opposite polarity with C12.. Probably I should have left C15 and R14 like before and add a couple like C12/R10 on the no-gain-path?

I'm sorry for the dumb questions, I'm still trying to learn so I'm making a lot of mistakes



what I meant with wiring the Dir Output directly to the Line in was something like this:

I'm trying to understand what is the difference with a (impedance) balanced output like shown before and this simple connection

 

[Dimitree]

Not a lot, tbh. Look at the schematic in my post up. It will do objectively and subjectively better.

thanks, I'm studying this approach as proposed here: https://www.analog.com/media/en/training-seminars/design-handbooks/P2 Ch6_final.pdf
it should be Figure 6-26: "All Inverting" balanced line receiver, if I'm not mistaken.
You are right, now I see this circuit also in the Self book, but he doesn't give it much details about how good it perform, so I never payed too much attention to this.
I'm trying to understand if resistor matching is crucial to outperform a THAT 1240.

 

[thor.zmt]

I'm trying to understand if resistor matching is crucial to outperform a THAT 1240.

Resistor matching affects CMRR ONLY.

Give the potential of impedance mismatches in the wire impedance, when using long cables, it is mostly a datasheet sales spec, than relevant in practice.

Using short cables the potential for wire impedance is minimised, but so is the potential to pick up common mode noise.

Using long cables CMRR likely goes out of the window but the potential to pick up noise is high.

If you reliably need high CMRR that is not depending on the precise balance of wire resistances, you would need to buffer the inputs at very high impedance.

Bill Withlock has a patent which is implemented by THAT as "InGenious" (IIRC). TI also has suitable IC's.

Thor

 

[Zen Zeven]

I'm interested in tweaking the Self input circuit but can't get the mathematics to work. What value does A (attenuation) get in the calculation? Is it 9.9dB (from -3.3 to +6.6dB) or only -3.3dB as that is the attenuated part. What value do I put in to the calculation -3.3 or 9.9? What kind of gain structure can be achieved with this circuit? Could I get something like -7.5 to +15dB or even -10 to +20dB and still have the impedance under control.

Can someone help me get the mathematics usable

Gain=
[R7 x (R2/A)] / [R7 + (R2/A)]
----------------------------------------
R1


in other word


[10000 x (4700/?) ] / [10000 + (4700/?)]
----------------------------------------------
R1



BTW the circuit performs well I have build it a few times as it is

 

[Dimitree]

Resistor matching affects CMRR ONLY.

Give the potential of impedance mismatches in the wire impedance, when using long cables, it is mostly a datasheet sales spec, than relevant in practice.

Using short cables the potential for wire impedance is minimised, but so is the potential to pick up common mode noise.

Using long cables CMRR likely goes out of the window but the potential to pick up noise is high.

If you reliably need high CMRR that is not depending on the precise balance of wire resistances, you would need to buffer the inputs at very high impedance.

Bill Withlock has a patent which is implemented by THAT as "InGenious" (IIRC). TI also has suitable IC's.

Thor

thanks Thor,

so if I understand correctly, we are talking about two different aspects: CMRR and "inherent" circuit noise:
CMRR is improved using precision resistors or ICs like THAT 1240. If you also want high CMRR when there could be impedance mismatch between driver and receiver, than THAT "InGenious" solution is even better.
Aside from CMRR, the circuit proposed by Mackie or by Self that I linked on post #1 are inherently noisier than the "double inverted" configuration, when adjustable gain is needed.

Did I get it right?

 

[johnR]

I am not sure who "praises" Mr. Self and for what exactly.

I'm pretty happy with my Soundcraft Delta mixer, which is full of Doug Self's circuit designs.

 

[arthurdb]

I'm interested in tweaking the Self input circuit but can't get the mathematics to work. What value does A (attenuation) get in the calculation?

I'd reckon A is not in decibels but a ratio. The value of A depends on where you set the pot at and thus could be anything between 0 (max attenuation) and 1 (no attenuation). The resulting gain will also not be in decibels but a ratio. db = 20 * log10(gain)