Douglas Self Line Input

[mitsos]

Someone ever had a look at the Rolls RM64 input stage ?

Not really clear what they're trying to do..... Design doesn't do anything very well (IMO)

JR

thanks, I didn't want to be a dick and say I hate rolls, but I really don't like them, mainly after looking inside a couple bellari preamp.  Seems they were trying to save every penny in that design, and Abbey's comment about "lesser parts count" makes me think maybe that's all they try to do: use less parts.

Sorry to be negative, I don't like to talk bad about anyone, but I am tired of friends asking me to "upgrade" some piece of equipment they bought after reading online (GS mainly, but elsewhere too) that they could swap a chip and make it sound like an API/Neve/whatever. 

Anyway, I personally will never own anything with the name rolls on it. Unless it's followed by "royce". And then maybe.

sorry for the rant, it's late and I had a couple drinks.

 

[mitsos]

It might be interesting to see how it behaves with modern uber opamps, I'd still put it on the bench and ring it out before trusting that schematic.

JR

to follow up with this, I got all excited but in the end, I ended up scrapping this circuit. Some opamps would oscillate (OPA2134 IIRC), others would cause havoc down the line in the channel strip (pops in the EQ in/out switch, for one thing).  There are electros before and after, so there shouldn't have been any DC but in the end I went a different route.

I did a simple, one amp diff input followed by a noninverting opamp for a bit of gain when needed.  The difference in sound was phenomenal.  A comparison between this, the Doug self and the original console circuit was not even close. You could "feel" the crack of the snare drum.  I'm using LM49860 opamps my new fave, they are phenomenal.  Just have to be careful routing the wires as I did get some motorboating at first. 

 

[pucho812]

Interesting....  you got a schematic?

 

[mitsos]

top secret...  8)  ;D

just kidding.. it's really simple, didn't sketch it out.  Just a differential amp at about -6dB gain, then a low gain API 312-type gain stage.  Each of these use one half the LM49860.  Two caps before the diff amp, one after, and the usual 2 in the "API" stage. This is going into a high impedance so the output cap can be smaller than usual.

http://upload.wikimedia.org/wikipedia/commons/thumb/a/a2/Op-Amp_Differential_Amplifier.svg/300px-Op-Amp_Differential_Amplifier.svg.png

I'm doing a bunch more channels tonight and tomorrow. Lots of resistors to match.  :'(

 

[pucho812]

Interesting.

 

[mitsos]

here is a schematic of how I did it.  Seems to be working OK, but ymmv, use at your own risk, don't do this at home, etc, all the other warnings apply.  You can vary the values to get different gain settings, as well as vary cap values if needed? I'm not 100% sure of the input cap values, I think 22uF should be sufficient. I'm using bipolar caps.  The 330uF cap is what I had on hand, I didn't calculate it here.

I also did not use C4, mainly because I was unsure of the value, my gut told me to use the 100K to calculate the HPF frequency, but I wasn't sure so I left it out and it seemed to work fine.  F response was pretty much flat from 20-20K. 

As is, the input stage is set to about -6dB gain, while the second stage goes from about 6-17, for a total of 0-11dB gain. 

 

[abbey road d enfer]

The lack of a differential after the input stage means common mode noise does not get cancelled.

The cross-path between U1B's output and U1A's negative input IS the differential.

 

[abbey road d enfer]

here is a schematic of how I did it. 

It's too bad you got deterred by the difficulties of implementation, in particular oscillations. The original schemo is more a block diagram than a real application. The fact that one opamp is in the NFB path of the other challenges the phase margin. BW reduction is mandatory, with the application of small caps across the FB resistors.
The so-called "Doug Self input stage" works very nicely, as thousands of Soundcraft mixers testify.

 

[JohnRoberts]

The lack of a differential after the input stage means common mode noise does not get cancelled.

The cross-path between U1B's output and U1A's negative input IS the differential.

Yup, R111..  Thanks, I missed that from my quick look...  The differential version of CM signal has one more opamp path delay so may degrade CMR at very HF, but LPF in the input should reduce very HF inputs. There is a passive one pole LPF in each input. I guess the matching of these two LPF also impact HF CMRR, but I clearly missed the first order CM cancellation. mea culpa  :'(

JR 

 

[mitsos]

here is a schematic of how I did it. 

It's too bad you got deterred by the difficulties of implementation, in particular oscillations. The original schemo is more a block diagram than a real application. The fact that one opamp is in the NFB path of the other challenges the phase margin. BW reduction is mandatory, with the application of small caps across the FB resistors.
The so-called "Doug Self input stage" works very nicely, as thousands of Soundcraft mixers testify.

Yes, I feel the same way... Had a bunch of channels already done that way and went back and redid them because the board has to be put to work this weekend. Could have tried the feedback caps I suppose, but got very depressed by the whole thing that I lost a nights sleep!  In the end the more conventional implementation works better, to my ears anyway.  But again, comparing a non-fully-working doug self input is not 100% fair.  The feedback caps would go across both  4K7 resistors? I may be cutting a console in half "soon" (to make 2 small consoles out of 1)  so I may try this again, although I'd rather have both mic and line inputs on one of those. 

Do you know of any particular soundcraft boards that use that circuit? The few schems I have here are all different..

thanks!

 

[abbey road d enfer]

Do you know of any particular soundcraft boards that use that circuit?

The 6000 (amazingly, there's not a cap in view!). I'm sure there are others but I haven't kept all schemos.

 

[mitsos]

thanks, I'll look for the schemo.

 

[mitsos]

Turns out this guy has a bit too much gain, I'll change those input transistors to 20K-24K or so to drop another 6dB on the input, maybe change the second stage gain as well to give a greater range.  But all in all it's working fine. 

I finally got the console I mentioned above, it's a TAC scorpion 24-16-2 that I want to cut in half and make two 12 channel mixers with 4-8 groups and some extra line inputs for mixing (like the new API the Box thingy). Only one will have a master of course, but that's OK.  The cAPI ACA-Bo will splice into the group channels perfectly, they are all inverting amp based mixers anyway, so that's the plan for masters/summing.

Thing is, I'd like to have 12 mic/line channels based on the above schemo. I am thinking of putting a SPDT switch between the two stages, pole going to the second stage, so that I can switch between a mic and line input for the other part.  Easiest way would be to have a transformer handle the mic in, but that could get costly down here... SO... any suggestions for a solid state mic input that not a THAT chip?  Here is a schemo of the proposed trafo based solution..
thanks!

 

[madswitcher]

Hi all,

Has anyone actually looked at Doug Self's book on this subject - "Small Signal Amplifier Design"? 

There is a chapter devoted to line in amps and how to turn the concept into a practical circuit - its on page 359 for those who have a copy of the book. 

Cheers

Mike

 

[PRR]

> has a bit too much gain

You seem to be -dropping- a good line level all the way down to mike level. Then gaining it back up again.

That's just doubling your hiss (two opamps). Not exact double, but makes you ponder.

Since you have a good low-hiss mike-amp *with* a balancing transformer, you can take a line input a "good way".... through a pad and transformer. No opamp needed or wanted. CMRR is very good, RFI rejection very good, and that "line source" can be mis-wired on another leg of a 208V delta power system and nothing will smoke. And certainly across the 2V-3VAC typical ground difference between stage and sound-booth.

 

[mitsos]

Actually, yes, that's a bit too much drop. Probably will be more like 6dB (12K/6K or 20K/10K resistors in the diff amp).  Would you still prefer the pad? The "line only" circuit further up sounds pretty good that's why I thought I'd stick with it. Plus I have a bunch of dual opamps...

 

[abbey road d enfer]

> has a bit too much gain

You seem to be -dropping- a good line level all the way down to mike level. Then gaining it back up again.

Since the internal operating level is -2 dBu, that makes sense. Min gain of the mic pre is about 6dB, so it is quite reasonable to drop a nominal +4 to -8.

 

[mitsos]

btw, when I said it has too much gain, I was speaking of the first version of this circuit (12K/6K resistors) at -6dB input amp, then to a +6-17dB or so non inverting. So you'd have unity +10 dB.  Which in tests, was very hot, so much so the fader at minimum wouldn't fully cut the sound. That's why I thought to change it to -12dB. 

For reference, the original line (U) pad on these things uses 10K series Rs and (I think, the schemo is not very clear) 820R shunt, for 27dB of attenuation. Or I could go 15-16dB to allow more room for the EQ.  Not really set in stone yet, as is the case of the opamp. That will probably be set more like a preamp, 6-40dB or so, for about 50-60 mic gain total.