Active MS encoding (was: An inverting amp..)

[Viitalahde]

Here's the deal..

My next project is an active M/S encoder/decoder for the mastering rig. Since there's a real chance I'll be hitting the decoder pretty hard, I want a lot of headroom. +/-24v rails, at least. I also want to build it discrete, just for the fun. (Yep, after building a lot of things for other people's consoles with IC's, I just don't get the kick out of building with them!

I've been wanting to try the Gainbloak for a long time, but since I'm going to need L-R & M-S, I need to invert somewhere.

The Gainbloak doesn't appear to invert very well.. So what do you people suggest?

 

[BradAvenson]

I don't see why you can't use a current feedback opamp as an inverter.

http://www.national.com/an/AN/AN-597.pdf

I would try it.

 

[Viitalahde]

That's what I understood from Tamas's comments from the big thread.. I might have misunderstood it, maybe he meant that the Gainbloak would make a crappy balancing receiver?

Anyhow, I plan not to use any differential amps in the encoder/decoder. M= buffered L&R summed through precision resistors and a 6dB amp, S= buffered L and inverted R, resistors and a 6dB amp.

EDIT: Those application notes seem helpful, thanks! :thumb:

 

[BradAvenson]

http://www.intersil.com/data/an/an9420.pdf
http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1154,C1009,C1146,D4234
http://focus.ti.com/lit/an/sboa081/sboa081.pdf

There are a number of articles on current feedback opamps since they seem to be fairly fashionable for the IC makers these days. There should be lots of useful information out there, just check the sites of any vendor that makes CF opamps.

 

[Samuel Groner]

As Brad has pointed out, using CF opamps in inverting mode is basically possible. They aren't optimum with respect to noise at low gains though. The inverting input current noise is releatively high, and you can't lower the feedback impedance for stability reasons. It ain't horrible, but good to know nonetheless.

Samuel

 

[Viitalahde]

Here's more thinking of the M/S encoder/decoder. I've gave more thought to the use of a differential amp, and might actually go that route afterall.

I'm now thinking of two possible variations:

A) No input buffer, L/R or M/S signals go straight the basic differential amplifier (=L-R or M-S) and to a series feedback (noninverting) amplifier, 6dB of gain (L+R or M+S). 20k precision resistors or so.

This amplifier would be an OPA604 on +/-24v rails, and for the fun & testing, biased to class A with a JFET current source. This biased IC would in turn go to a FETBloak buffer.

PROS: The OPA604 would most likely give better CMRR figures than any discrete op-amp.

CONS: No input buffer (I like to reduce the total number of amp stages in the chain whenever I can), so the CMRR is completely dependant on source impedances. There is a possibility that the encoder's source comes throug a passive attenuators, which by 1% chance might be set up for different L/R balance to compensate for balance errors in the received mixes. The OPA604 is going to get HOT. Will require a big heatsink.


B) FETBloak input buffers to same type of differential & series feedback amps as previously. This time it would be a Forssell 1st generation JFET amp. No output buffer.

PROS: Source impedances no longer an issue, more fun with discrete parts..

CONS: Some pretty good matching is needed to make the JFET amp operate properly in the differential mode.

 

[Samuel Groner]

Can you draw a schematic of both versions?

Samuel

 

[radiance]

Why not make the MS decoder part with two transformers?
As described in this thread...

http://www.groupdiy.com/index.php?topic=257575&highlight=#257575

 

[Viitalahde]

I don't want any more transformers in the chain - I just can hear most of them. An active one will be better in my application.

I'll sketch up a schematic some time today.

 

[Viitalahde]

So, here are a couple of quick sketches, hope they're not too unclear.

Option A.. The OPA604 current source inspiration comes from this page: http://tangentsoft.net/audio/opamp-bias.html

Option B.. I'm beginning to think myself this is better. The Forssell op-amp could probably be made close enough, and the preceeding impedances are not an issue anymore.

Here's the Forssell op-amp: http://forsselltech.com/downloads/schematics/Class A JFet Opamp.PDF

 

[Samuel Groner]

I'd go with B. You might want to reduce the impedance of the feedback network of the second stage for lower noise and perhaps trim one or the other resistor for good CMRR (didn't think about which one).

Samuel

 

[Viitalahde]

Yep, B seems good now as I drew it.

You're probably right about the impedances of the 2nd stage.. The FETBloak has plenty of drive. Good tip about the trimming resistor too. I'd trim at the inverting input.

Today I had a Focusrite Red6 on the bench, and seeing a couple of precision resistor modules (those SIL-looking things) I thought I could try those out in these.

 

[Marik]

[quote author="Viitalahde"]I don't want any more transformers in the chain - I just can hear most of them. [/quote]

Hey Jaakko, understand that, besides they are costly.
Why then not a resistive matrix A la Neumann?
Simple, elegant, neutral sounding, cheap, and works just as good.

 

[Viitalahde]

[quote author="Marik"]Why then not a resistive matrix A la Neumann?[/quote]

'Coz I'm completely out of balance. :razz:

If I ran my stuff balanced, I'd do that.. The only time I've missed it!

 

[chrissugar]

[quote author="Viitalahde"]Today I had a Focusrite Red6 on the bench, and seeing a couple of precision resistor modules (those SIL-looking things) I thought I could try those out in these.[/quote]

Those precision resistor modules are also used in all the balanced input receivers in the Cranesong Avocet and the STC8. Also discrete opamps. They work and sound excellent.
I also had this thought to use a similar combination. :grin:

chrissugar

 

[Marik]

[quote author="Viitalahde"][quote author="Marik"]Why then not a resistive matrix A la Neumann?[/quote]

'Coz I'm completely out of balance. :razz:

If I ran my stuff balanced, I'd do that.. The only time I've missed it![/quote]

Single inverter will do.

 

[Viitalahde]

I'm not sure a single inverter will do. Doing that might result in a impedance mismatch again.

The Analag Compressor I just finished has an output Z of some ~250 ohms if I remember correctly. Going through a single inverter would result in having one Z at 250 ohms and the other below 10 ohms.

...

Just ordered some 2SK170BL's from Banzai: http://www.banzaieffects.com/

0,89€ for a piece, best price I've seen anywhere. They also had a weird special, 100 pieces of 510ohm resistors for 0,83€. :razz: 500 resistors for me please. It's a good shop, I've ordered some stuff before.

Now I only need some 2SJ74BL's.. I think my local supplier has GR's only. I'd probably find the Q6 for the last voltage amplifier of the Forssell amp from there, but not so sure about the output amplifier's Q7..

 

[chrissugar]

You can buy 2SJ74 from B&D Enterprises: www.bdent.com

2SJ74
http://www.bdent.com/search/search.jsp?contains=2sj74
They also have the 2SK170 at a very good price:
http://www.bdent.com/search/search.jsp?contains=2sk170
If you buy more than 100 of each the prices are lower than the one on the +50 column. A couple of years ago I bought some hundreds of each and they give me a very good price. :grin:

Buy the BL version only and select for Ids.
The 2SJ74GR (from your local supplier) is not usable in Fred's opamp. You need at the output the fets (Q7 and it's pair) to have Ids at least 10mA, and for the voltage amp something in the 6-8mA region.
You will really love these opamps, I built tons of them and they sound great. :grin:

 

[bcarso]

You may want to screen the SJ74's for breakdown voltage, as the preamp causes the written spec to be routinely exceeded. I discussed this with Fred and he says he's never seen an actual problem in use. As JR says, YMMV.