MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

Op-amps for a Precision AC/DC Mixer
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page 1, 2  Next [all]
Author Op-amps for a Precision AC/DC Mixer
Sugarfree
Hey guys,

looking for a perfect op-amp for a custom mixer module (in 5U format).

My objectives:
- top of the line pro-audio quality
- mixing of both AC and DC signals
- transparent sound
- big headroom
- low noise, low distortion
- temperature-stable for DC operation
- precise handling of DC signals
- cost doesn't matter at this point

what would you recommend?

Analog Devices OP285GP - this one is used in MOTM830
TI LME49720
TI Burr-Brown OPA627
TI Burr-Brown OPA827
TI Burr-Brown OPA1612
TI Burr-Brown OPA1622

any other candidates?
loki
When you say "top of the line pro audio quality" do you understand the implications? Recording studio equipment implies 600 ohm source and output impedances and the ability to handle signal levels of 20 V RMS or more.

Is this a synthesizer module? Then then implied range of signals and impedances is different.

Is the 1/f noise corner frequency important?

I suggest you list the characteristics in order of importance and recognize there is some compromise between AC and DC performance. At the bottom line which is more important?

The op-amps you have listed are all high performance in their individual ways. Some are more suited to the audio side of the task while others are more suited for the DC side. There is quite a range of prices and I suggest you eliminate the most expensive ones first.
Sugarfree
I meant "pro audio quality" as it applies to a modular synthesizer. I'm thinking about a mixer module in MU format. I've read that many modular mixers are made with TL072 and that it's in fact neither great for audio nor for DC.

So my question is, which op-amp provides a high quality balance between audio and DC applications. Again, I don't want to include cost as a factor.

To answer your question, both AC and DC performance within a modular synthesizer environment would be equally important.
adam
how about this one?

http://www.mouser.co.uk/ProductDetail/Texas-Instruments/LME49720MAX-NO PB/?qs=7lkVKPoqpbbSdCvbWPyjHw==
loki
Write a product data sheet for your mixer. Specify its characteristics in the audio domain and the DC signal domain. Then pick an op-amp that will allow you to meet those specifications.

If you want to brag to your customer about the expensive op-amps you used, use the OPA627. They are about $22 in unit quantities. You could build a mixer every bit as good for your purposes with the LME49720 that adam recommended.
Synthiq
Sugarfree wrote:
I've read that many modular mixers are made with TL072 and that it's in fact neither great for audio nor for DC.

It seems like not just mixers are using TL07X amplifiers but most modules are using them. If this is the case, the input signals to your mixer are already compromised regarding noise and distortion so I'm just wondering if these fancy amplifiers will make any difference in practice.

If you have followed Graham Hinton's crusade to educate us about the importance of a low resistive ground in the system, you know many systems fail in this respect and the noise on the 0V bus often exceed the noise of an individual amplifier and may very well be the limiting factor. And if you want to take advantage of the low noise in the suggested amplifiers, all resistors around the amplifier must be well below 1kohm or they will dominate the noise so all input signals must first be buffered if you want to keep input resistance to the standard 100kohm.

Essentially all signals will at some point pass through a VCA and so far I have not seen any VCA that has a distortion even close to the distortion of even a TL07X. For instance the 2164 specifies 0.02% THD but only at 0dB gain. At -20dB it increases to 0.15% while a TL07X specifies 0.004% at unity gain.

As for headroom, this will be limited by the supply voltage. The TL07X and most of the suggested amplifiers can go 1.5V from the rails and only one goes all the way to the rails and has a 0.9dB advantage. If that is important or not, I don't know.

I can agree that 18uV/C offset drift for the TL07X isn't the best I have seen and may affect dc performance if temperature varies a lot.
mskala
I suggest using a TL072 and setting fire to a $20 bill.
ricko
I have used LME49710 in an active crossover, and they really are amazing. 0.000003% THD in the right usage. Less noise than a resistor, supposedly. And unity gain stable, which is a good step for a jelly bean.

But you may find that the place to put ultra low distortion circuits is at the mixer rather than the individual input buffers: if you just have a single oscillator to each input, distortion will be harmonic, while if you have multiple tones you will get intermodulation distortion.

My advice? Build it with the designed opamps, then substitute other unity gain-srabke opamps one by one, so the you catch as soon as a problem occurs. And start at the output end of your typical patches and go backwards in you upgrades. VCAs before filters, filters before VCOs, etc.
slow_riot
Your post indicates to me that you are not experienced enough to implement your objectives.

Almost all of those criteria come from overall design, of which opamp choice likely only comprises 5% .

No opamp is perfect, even the "everything and the kitchen sink" opamp the AD797 has huge problems with stability, and the low input impedance makes it useless for 75% of synthesizer circuits. Opamp choice comes down to "most acceptable compromise".

Doug Self is quite a good read on low noise design.
Synthiq
ricko wrote:
I have used LME49710 in an active crossover, and they really are amazing. 0.000003% THD in the right usage.

What are the distortion specs for your speaker elements?
CLee
There are folks here who have had a hand in designing some very high-end audio gear. Maybe they'll chime in, but my personal feeling is that a mixer in a modular synth isn't a place that requires a particularly high spec op amp. Don't throw a 741 in there but you don't need a real tweeky one either. The signals you're mixing are 5v p/p, so signal-to-noise isn't that demanding. If the slew rate can handle 10v p/p audio you should be OK. The reason mixing consoles have higher spec'd parts have more to do with lower signal levels, the number of signals being summed, and the long signal chain.

Places I'd be more concerned with op amp choice would be filters and places with higher gain, such as the current to voltage converter in an OTA circuit like a VCA. OTA's knock the signal level way down at their inputs.
Graham Hinton
Sugarfree wrote:
what would you recommend?


If you are going to have a 100k input and 1k output it is pointless even trying to choose. You will never obtain the benefits.

loki wrote:
Recording studio equipment implies 600 ohm source and output impedances and the ability to handle signal levels of 20 V RMS or more.


In the 1950s maybe... 600 ohms is a telephone line standard for maximum power transfer, modern (1970+) professional equipment is designed for maximum voltage transfer. It would also be designed to be able to drive, say, four 600 ohm loads in parallel just in case someone multed to some vintage gear on a jackfield, but the output would be 50 ohms max and the input would be a "bridging" load of around 15k to 25k, i.e. not enough to effect a 600 ohm line with a few sitting across (bridging) it.

As for 20V rms or +28dBu, the equipment that can take that level usually attenuates by 6db on the input and has to have its output complimentary to achieve the same level. This does not improve the overall snr.

Sugarfree wrote:
I've read that many modular mixers are made with TL072 and that it's in fact neither great for audio nor for DC.


That is more a function of the components around them and the whole design. Most synthesizer audio paths use resistor values at least 10 times greater than they should for low noise.

Synthiq wrote:
If you have followed Graham Hinton's crusade to educate us about the importance of a low resistive ground in the system, you know many systems fail in this respect and the noise on the 0V bus often exceed the noise of an individual amplifier and may very well be the limiting factor.


I haven't even started to go on about low noise design. 0V noise, important as it is, does not equate to signal noise although it does contribute. The noise in the signal path is mainly the worse device in the chain and that can easily be above -60dBu.

Quote:

And if you want to take advantage of the low noise in the suggested amplifiers, all resistors around the amplifier must be well below 1kohm or they will dominate the noise


I think you've dropped a nought there. 10k is commonly used around 5534s and as low as 4k7.

Quote:
so all input signals must first be buffered if you want to keep input resistance to the standard 100kohm.


Why would anybody want to keep that? Yes, I know, because they have a 1k outputs. OK then, why does anybody really want those?

Quote:
For instance the 2164 specifies 0.02% THD but only at 0dB gain. At -20dB it increases to 0.15% while a TL07X specifies 0.004% at unity gain.


Whoooa! You are quoting figures without any reference to signal level. Distortion measurements at normally made at maximum level before clipping. Lower levels will of course have a higher TDN+N because the noise floor stays the same.

Ricko wrote:
0.000003% THD in the right usage.


When I was working for SSL I was discussing a problem with some other engineers and someone (who shall remain nameless) walked in, rubbed his chin and said "hmmmm, sounds like 0.0025% distortion to me". He became a standard joke, everybody went around imitating him and then bursting out laughing. That was 30 years ago, maybe they still do. Oh, and actually the reason we had a problem was a 0.3dB level drop.

Quote:
Less noise than a resistor, supposedly.


A resistor's noise (which is not distortion) is dependent on its value, did you have any particular one in mind?
Synthiq
Graham Hinton wrote:

Synthiq wrote:


And if you want to take advantage of the low noise in the suggested amplifiers, all resistors around the amplifier must be well below 1kohm or they will dominate the noise


I think you've dropped a nought there. 10k is commonly used around 5534s and as low as 4k7.

Several of the suggested amplifiers have noise in the range from 4nV/sqrt(Hz) down to 1.1nV/sqrt(Hz) which is the thermal noise in resistors from 1kohm down to 75ohm. If you go above these values the resistor noise will dominate. The noise figure for the 5534 with a 5kohm source resistance is specified as 0.9dB so this shows the noise will be dominated by the 5kohm resistor, not the amplifier.

Graham Hinton wrote:

Quote:

For instance the 2164 specifies 0.02% THD but only at 0dB gain. At -20dB it increases to 0.15% while a TL07X specifies 0.004% at unity gain.


Whoooa! You are quoting figures without any reference to signal level. Distortion measurements at normally made at maximum level before clipping. Lower levels will of course have a higher TDN+N because the noise floor stays the same

The THD for the TL07X was specified at +/-15V supply and 6Vrms input signal and the gain of 1 and load resistance higher than 2kohm. If the amplitude is reduced to a more typical 5Vpeak, the THD would be reduced if it behaves as most other amplifiers, but as an engineer I will not speculate what the exact value would be but am confident the 0.004% (0.003% at 1kHz where the 2164 was measured) is a worst case. Since the 2164 has a current input, it is difficult to directly compare the two amplifiers but with 30kohm input resistor the THD was specified with a 0dBu signal so with a 5Vpeak signal the THD will be higher and probably closer to 0.1% but is hard to say since the THD versus input voltage plot in CoolAudio's datasheet has an x-axis from 20Vrms to 20kVrms which is obviously wrong. So I was most likely conservative when I stated the THD in the 2164 was 5 times higher than the TL07X, but I'm happy to change my view if anyone can present more reliable measurements that contradicts my conclusions so far.
555x555
Lots of true things spoken in this thread. However, if you're learning, a mixer is a pretty good thing to get your design hands wet, IMO. To the OP: I'd say the summary of this is (1) don't worry about the op-amp, get a TL07x and move forward. You'll learn something. (2) if that's not satisfying to you, hit the books, and in learning about the design of mixers and all the ways noise and distortion can enter a system, you'll also learn what to look for in an op-amp. I second the Douglas Self recommendation for further reading. The reason a TL07x or TL08x is used so widely is usually bc it behaves pretty reasonably in all sorts of conditions, it's not exorbitantly priced, and it's actually pretty OK quality for a lot of applications. Very high quality mixers, although they use a lot of 5532s, often also have a TL07x here and there when that's more appropriate. There's nothing non-pro about them. The other reasons for using them you definitely want here: while you're trying out your diy and design skills, you probably won't have to worry very much about the op-amp's non-ideal behavior. This means easier troubleshooting if/when something goes wrong.
Troubleshooter
Better put your money in the stuff around the opamp. Start with the good old 5532.
Sugarfree
thanks for suggesting 5532.

Perhaps, I should have explained, that it's not going to be a DIY project. I want to design a feature set and interface, and then hire someone with expertise to design the circuit. I asked about op-amp recommendation to better understand their impact on the overall performance.
I fundamentally disagree with "the input signals to your mixer are already compromised" approach. Quality of the source signal (which often becomes a creative choice), is really independent of the purpose of a mixer. This for me means transparent, distortion-free summing of both AC and DC.
The mixer I have in mind will include: four-quadrant mixing, voltage control, panning, aux I/O and more, so it will be quite capable of doing many things at once.
mskala
Sugarfree wrote:
Perhaps, I should have explained, that it's not going to be a DIY project. I want to design a feature set and interface, and then hire someone with expertise to design the circuit.


Then maybe let them design the circuit? People at the skill level you're looking for usually aren't eager to be told which choices to make on points where they see it as their job to decide.
slow_riot
mskala wrote:


Then maybe let them design the circuit? People at the skill level you're looking for usually aren't eager to be told which choices to make on points where they see it as their job to decide.


This happens regularly in the audiophile industry, an engineering company will be given a product spec that includes use of a specific opamp so that when a customer or reviewer opens it up they can sleep well at night knowing that none of the huge profit margin was spent on something as ordinary as a TL07x.
Sugarfree
I still like to know what is inside smile
plus it always helps to communicate the needs.

There's a huge, tangible difference between what is acceptable and cost-effective vs. high-end pro audio though. I'm really not interested in a Mackie mixer, what I want is a Neve.
GuitarBuilder
I've had good results with the OPA1692.
CLee
Sugarfree wrote:
I still like to know what is inside smile
plus it always helps to communicate the needs.

There's a huge, tangible difference between what is acceptable and cost-effective vs. high-end pro audio though. I'm really not interested in a Mackie mixer, what I want is a Neve.


Having recorded on a Neve, the last thing they are is silent and distortion free. They sound great, but not transparent :-)
Sugarfree
CLee wrote:
Sugarfree wrote:
I still like to know what is inside smile
plus it always helps to communicate the needs.

There's a huge, tangible difference between what is acceptable and cost-effective vs. high-end pro audio though. I'm really not interested in a Mackie mixer, what I want is a Neve.


Having recorded on a Neve, the last thing they are is silent and distortion free. They sound great, but not transparent :-)


yes, it's true... well sort of. Depending on what you consider to be a transparent mix. Is it DAW? Mixing on an analog Neve produces the most natural and pleasing sound to my ears. Even comparing to an analog SSL, a digital Neve or System 5.
slow_riot
If I recall Neve were very heavy passive component users, transformers on every input, little inductors on outputs and maybe even board to board connections.

It's definitely a different style compared to a high signal purity approach, and it wouldn't combine well with the need for DC precision. Some kind of soft limiting which could be overdriven might be nice.

It would definitely be interesting to see higher levels of engineering with regards to some aspects of synthesizer design from a wider range of sources. For one thing you would need to change the input and output impedances like Graham recommends. Unbalanced I/O can create conflicts as well if you are trying to patch between 2 systems.
loki
Rupert Neve lives in Wimberly Texas less than 50 miles from where I am. I was looking at his web site this morning. Google him and take a look at his 5088 mixing console. Custom transformers on all the inputs and outputs, power from a 90 Volt supply.

Sugarfree wrote:
I still like to know what is inside smile
plus it always helps to communicate the needs.

There's a huge, tangible difference between what is acceptable and cost-effective vs. high-end pro audio though. I'm really not interested in a Mackie mixer, what I want is a Neve.


I suggest again that you write a product spec. Until you define in engineering specs the level of performance you want, you are just waving your hands. I have most of a decade of experience as an analog applications engineer and I've worked at a couple of pro audio companies whose names you would recognize. You are going to have to upgrade your skills just to write a spec.

Don't waste money on parts that have levels of performance that far exceed your requirements. If you want Neve level of performance you need to design the entire synthesizer to that level of quality.
CLee
Sugarfree wrote:
CLee wrote:
Sugarfree wrote:
I still like to know what is inside smile
plus it always helps to communicate the needs.

There's a huge, tangible difference between what is acceptable and cost-effective vs. high-end pro audio though. I'm really not interested in a Mackie mixer, what I want is a Neve.


Having recorded on a Neve, the last thing they are is silent and distortion free. They sound great, but not transparent :-)


yes, it's true... well sort of. Depending on what you consider to be a transparent mix. Is it DAW? Mixing on an analog Neve produces the most natural and pleasing sound to my ears. Even comparing to an analog SSL, a digital Neve or System 5.


Well I was joking, but by transparent I would mean uncolored. The old wire-with-gain analogy that's used for some mic pre's. The Neve adds something to the sound, something good, but not transparent in that sense. Your OP seemed to be trying for more the wire with gain than a colored warm sound...
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page 1, 2  Next [all]
Page 1 of 2
Powered by phpBB © phpBB Group