Hi Mark,
I've used the OPA134x series for input stages in filters and the like, but not the OPA164x which looks to be an improved lineup of similar topology.
Oddly the datasheet for the 164x makes a big deal of saying JFET, the 134x says only FET, although I am sure it was also JFET. Any comment from the IC gurus? Is this just pandering to the audio crowd?
Howie
TI telling that the Opa164x is a JFet amp, probably sets it apart from it’s Opa165x called a Fet amp, which is actually a MosFet amp.
Hans
Hans
Last edited:
The 1652 is indeed a mosfet input opamp. The noise is worse than the 1642.
Both are superb although I’d not use the 1652 for phono apps though as the 1/f is high.
Both are superb although I’d not use the 1652 for phono apps though as the 1/f is high.
@Howie: When I finally finish the backlogged projects I'm going to have a whole heap of THAT stuff in my signal chain. Temped to get some 50m cable reels just cos I can run long interconnects as a test. But I'm doing weird stuff for home audio like side chain EQ for lolz.
Tomchr on here used the THAT1200 in his early iterations of an amplifier he sells boards for and moved away from it to a FET opamp solution as the THAT chips have some power up thump issues with single ended sources. Of course studios always power things up in the right sequence...
For drive side my current illogical interest is using the THAT1200 series in a cross coupled output. Again totally pointless for audio where single ended tends to work well and differential outputs are far less important than being balanced even if they don't.
And I've yet to work out the use cases where the patented bits in the THAT chips actually make a difference. Sure Ed will tell me 🙂
BTW are you going SE to Balanced or putting a driver on a differential signal?
Tomchr on here used the THAT1200 in his early iterations of an amplifier he sells boards for and moved away from it to a FET opamp solution as the THAT chips have some power up thump issues with single ended sources. Of course studios always power things up in the right sequence...
For drive side my current illogical interest is using the THAT1200 series in a cross coupled output. Again totally pointless for audio where single ended tends to work well and differential outputs are far less important than being balanced even if they don't.
And I've yet to work out the use cases where the patented bits in the THAT chips actually make a difference. Sure Ed will tell me 🙂
BTW are you going SE to Balanced or putting a driver on a differential signal?
Bill,
I don’t use differential drivers as mentioned. I do have a friend or two there. Currently waiting for a reel of compressor-limiter chips. Like most folk they are backlogged. Only need 4,000 of their chips for my foreseeable future. Probably more later.
What I find interesting is that so much stuff is made off-shore that my purchases stand out. Basically I can make as many as 50 of one of my products a day. Not exactly screaming high quantity production. I look at it as something to keep me busy now that I am past retirement age.
I also noted in the news one of your crazy fellows claimed to be from the US! Not nice news, not nice results.
I don’t use differential drivers as mentioned. I do have a friend or two there. Currently waiting for a reel of compressor-limiter chips. Like most folk they are backlogged. Only need 4,000 of their chips for my foreseeable future. Probably more later.
What I find interesting is that so much stuff is made off-shore that my purchases stand out. Basically I can make as many as 50 of one of my products a day. Not exactly screaming high quantity production. I look at it as something to keep me busy now that I am past retirement age.
I also noted in the news one of your crazy fellows claimed to be from the US! Not nice news, not nice results.
Last edited:
I just read about that as well, not a deal breaker but sub-optimum in the intended usage here, if I use them I may have to add muting FETs or relays to eliminate that initial offset. These buffers are going to feed crossover inputs which directly feed multi-amped speakers, so thumps are potentially dangerous depending on amplitude.
These buffers will be active differential driven and drive DSP crossovers which have active differential inputs. Given the distance, there will be a bit of capacitance in the intervening cabling, so a high-current output stage is critical. I remember Scott touting the high current and low noise capabilities of ADSL drivers, but the ones I have found are +/-12 V max. Based on my previous experience with the JE990 discrete opamp and despite it's age as a design, I am tempted to integrate them as the output stage, they seem incredibly stable into capacitive loads like long cables...I ran 2 twisted pairs close to 1000 feet moving audio around at UNC (live show remote) driven by a PR&E distribution amp which had a JE990 output stage and it showed very little ringing with a 1 KHz square wave, and flat response past 25 KHz...
I'm sure there is a more competent newer solution, perhaps a discrete current buffer output stage on an OPA164...
Cheers!
Howie
This is the driver you want: TPA6120A2 data sheet, product information and support | TI.com It started as a DSL driver and was redone (new part number?) as an audio driver. If you want super low distortion it is fast enough to go inside a feedback loop with something like an OPA1611. Since they were designed to drive a mile of crappy decaying phone wire they should be pretty sturdy. I still would add an output R/L network. I have used them for audio with very satisfying results.
TI has a number of headphone drivers that are quite good (and low noise).
The thump in the THAT receiver is probably a consequence of the common mode feedback cap charging. Possibly a couple of FET's in the right place to ground an internal stage. Trigger a 555 with the DC offset to control the thump?
Or if you are differential in and balanced out a handful of opamps can do the whole thing, including removing DC if needed. It may not be an issue if these are feeding DSP. I had to so something similar for the buffered outputs for my Shibasoku AM70A to avoid unwonted ground loops.
TI has a number of headphone drivers that are quite good (and low noise).
The thump in the THAT receiver is probably a consequence of the common mode feedback cap charging. Possibly a couple of FET's in the right place to ground an internal stage. Trigger a 555 with the DC offset to control the thump?
Or if you are differential in and balanced out a handful of opamps can do the whole thing, including removing DC if needed. It may not be an issue if these are feeding DSP. I had to so something similar for the buffered outputs for my Shibasoku AM70A to avoid unwonted ground loops.
There is a compendium of balanced line driving on the Pro Audio Design forum Balanced Outputs - Pro Audio Design Forum
Mark,
The typical professional audio power amplifier tends to at full gain produce full power when a real VU meter reads 0 VU. Due to practical considerations (such as virtually no one knows what the voltage should be when the properly installed meter reads O VU) this has become 1.5 VRMS for many amplifiers. However for units sold through music stores where an unsophisticated user may compare them, the louder one for the same input is thought to be more powerful.
Now many users, particularly with electronic crossovers may turn down the “volume” knob. This knob of course does not change the important acoustic parameter of the room size, but actually decreases the amplifier’s sensitivity. It however is almost always just a potentiometer. Thus it also reduces the signal to noise ratio of the audio power amplifier.
A good compromise is to reduce the voltage sensitivity by a factor of 10/1 or 20 dB or a bit less. I should note one major loudspeaker manufacturer uses only 12 volt rails in their loudspeaker processors. Otherwise for that much attenuation the signal processor must drive a bit above 21 volts peak!
So the typical 20 dB drop is a bit much.
The next consideration is how much noise the processor to amplifier cable lets in. There are two sources of noise. The first is crosstalk from say the woofer signal to the high frequency circuit. Raising the send signal level has not surprisingly no effect on this issue. Typical hundred foot cable runs will be much better than -80 dB crosstalk with well balanced send and receive impedances.
The more significant noise source is the EMI from the lighting system. This used to be just a step below horrible, however as the light source has pretty much changed to LED the power levels have dropped. This results in 10 dB less noise or perhaps a tad better. It seems no one has noticed this and changed audio signal levels. Also note that at arena type concerts some folks actually run lighting cables next to audio cables!!!
If you are running active cable drivers adding a capacitance stabilization resistor not only prevents oscillation but also tends to provide a uniform source resistance. It should not surprise you that directly connecting a pair of gain reduced opamps into a not quite balance capacitive load does not produce a source balanced to better than 100 dB. (Headroom)
Of course the build out resistors are never matched that well either. But using 1% parts will get you typically 60 dB or better. That actually improves even more as the load resistance is typical 10,000 ohms or more, so the issue is more capacitive loading of around 30 pF per foot of the typical 100 foot cable.
There really is a reason why my standard is to always use transformers for audio lines that are not in the same rack location. (Amp room etc.)
In an outdoor venue the lowest noise floor is around 35 dBA. The typical signal in such an environment is 80 dB, even allowing an extra 20 dB leaves a signal to noise or distortion ratio of 65 dB. So the improvement in distortion by eliminating transformers is not significant. Quite different than a recording studio environment. Of course when the music is playing loud the crowd noise rises and the transducer distortion masks pretty much any other issues.
The other issue with high volume is hearing damage and health regulations.
The typical professional audio power amplifier tends to at full gain produce full power when a real VU meter reads 0 VU. Due to practical considerations (such as virtually no one knows what the voltage should be when the properly installed meter reads O VU) this has become 1.5 VRMS for many amplifiers. However for units sold through music stores where an unsophisticated user may compare them, the louder one for the same input is thought to be more powerful.
Now many users, particularly with electronic crossovers may turn down the “volume” knob. This knob of course does not change the important acoustic parameter of the room size, but actually decreases the amplifier’s sensitivity. It however is almost always just a potentiometer. Thus it also reduces the signal to noise ratio of the audio power amplifier.
A good compromise is to reduce the voltage sensitivity by a factor of 10/1 or 20 dB or a bit less. I should note one major loudspeaker manufacturer uses only 12 volt rails in their loudspeaker processors. Otherwise for that much attenuation the signal processor must drive a bit above 21 volts peak!
So the typical 20 dB drop is a bit much.
The next consideration is how much noise the processor to amplifier cable lets in. There are two sources of noise. The first is crosstalk from say the woofer signal to the high frequency circuit. Raising the send signal level has not surprisingly no effect on this issue. Typical hundred foot cable runs will be much better than -80 dB crosstalk with well balanced send and receive impedances.
The more significant noise source is the EMI from the lighting system. This used to be just a step below horrible, however as the light source has pretty much changed to LED the power levels have dropped. This results in 10 dB less noise or perhaps a tad better. It seems no one has noticed this and changed audio signal levels. Also note that at arena type concerts some folks actually run lighting cables next to audio cables!!!
If you are running active cable drivers adding a capacitance stabilization resistor not only prevents oscillation but also tends to provide a uniform source resistance. It should not surprise you that directly connecting a pair of gain reduced opamps into a not quite balance capacitive load does not produce a source balanced to better than 100 dB. (Headroom)
Of course the build out resistors are never matched that well either. But using 1% parts will get you typically 60 dB or better. That actually improves even more as the load resistance is typical 10,000 ohms or more, so the issue is more capacitive loading of around 30 pF per foot of the typical 100 foot cable.
There really is a reason why my standard is to always use transformers for audio lines that are not in the same rack location. (Amp room etc.)
In an outdoor venue the lowest noise floor is around 35 dBA. The typical signal in such an environment is 80 dB, even allowing an extra 20 dB leaves a signal to noise or distortion ratio of 65 dB. So the improvement in distortion by eliminating transformers is not significant. Quite different than a recording studio environment. Of course when the music is playing loud the crowd noise rises and the transducer distortion masks pretty much any other issues.
The other issue with high volume is hearing damage and health regulations.
Last edited:
For a line stage its not usually an issue. 100 Ohms would be the correct value to match a differential pair in fact. For headphones (not where I used it) I would use a typical amp output R/L parallel pair type of solution. Current feedback brings challenges as does anything ambitious. A discrete solution would be more involved and probably less sturdy. I did see somewhere around DIYaudio a power amp built from paralleled TPA6120's. i'm sure that would be a handfull to stabilize.
The only issue with good transformers is that they cost more than all the rest of the product combined. Unless with volume you found a more cost competitive source? However given the probability that the power wiring is "not right" and will fry something a transformer is a good thing. The unit should be hipot tested before deployment to be sure.
I use the TPA6120 in one of my products for the headphone amplifier. I have a 10 ohm series resistor.
A series resistor is useful to isolate the opamp from any capacitive load. I wouldn’t drive anything external to the preamp without it. That said, Self shows how to drive a load without a series resistor - it’s a compensation design issue, but you do incur some extra parts.
A series resistor is useful to isolate the opamp from any capacitive load. I wouldn’t drive anything external to the preamp without it. That said, Self shows how to drive a load without a series resistor - it’s a compensation design issue, but you do incur some extra parts.
A trivial part of the system cost! The ones I use are under $10 in 100 piece lots. Listening test show no differences, measurements a small one. The expensive ones don’t pass my hand cranked megger.
Ed I read the Fr is 200Hz to 15kHz on 150-600 Ohm load. What is their load with your application?
PC Mount - Plug-In Printed Circuit Audio Transformers On Triad Magnetics
George
I think it's only the 120X series that have the power up pop issue on pseudo balanced links. The 16XX don't (but then don't fit your requirements).
The bullet proofing required for Pro audio apps is interesting coming from domestic audio where some products make a feature of dying at the first sign of something not as expected and is even considered part of the charm in some circles.
The bullet proofing required for Pro audio apps is interesting coming from domestic audio where some products make a feature of dying at the first sign of something not as expected and is even considered part of the charm in some circles.
George,
That is at full power. As you drop the current level the low frequency response increases. The spec is quite conservative on the high frequency side. As I recall I do have to put an RC snubber across output to prevent rising response. So with a typical 10 K amplifier load they easily do better than 30-20K +/- 1 dB. I don’t have any woofers that really go lower and outdoors atmospheric loss is the high frequency limit.
Last edited:
I used 24 of them in a Whirlwind stage box to make a Press Distribution hub, and they work quite well. I drove it with a Crown DC75 and the system was clean and well isolated, AFAIK that system is still in use. As everyone here knows, just about all audio inputs are bridging loads these days, so there was little LF attenuation passing signal through them. I should still have a few left to test ...somewhere, they are impressive little buggers for the price.
Ed, your idea of using a Megger to test isolation is a particularly good step given the risk when 20 news agencies plug who knows what into your box, and their equipment is powered from a truck parked outside connected to a different 3-phase 208 V panel than the venue...the potential (literally) could be deadly. I should have thought of that back then...
That brings me to another subject: finding parts used in previous projects. Increasingly I have small quantities of parts left over when a run is done and it would be impractical to break them down and put them in some sort of "General Inventory." As a result I standardized on a 12" x 6" x 4" box I just keep all the parts from a single project in. These stack nicely in cabinets and utilize the full depth so I don't have to go digging behind anything. Also the 4 x 6 end panel is the right size for labels made by my thermal mailing label printer. The problem becomes cross-referencing projects to find a specific part. I used to be able to recall ALL of the parts used in every project...
1st world problems, I know...Hey anyone have a B&K 4135 capsule for sale?
Howie
p.s. I am using a pair of tiny Triad xfmrs to pass 50 Hz-200 KHz I&Q signals from a SDR ham rig to a PC for spectral analysis, so HF limit is not an issue. They are not particularly flat depending on how you terminate them, but usable.
Last edited: