| redrabbit |
There are single-ended versions...
TLOxx datasheets show us... an input opamp, and an opamp at each output.....
(hey, why is that first opamp inverted?)
...or there's the input opamp, with a single buffer and distributed via 'build-out' resistors....
http://www.diyaudio.com/forums/show...7930#post197930
(thanks/credit to PRR)
Balanced versions of these.......
The first schematic is similar to the Audio Metrics DA16:
http://www.radiosystems.com/Manuals/da8-16manual.pdf
...and in the Sigma Electronics ADA-106, or some Leitch models, that buffer the differential signal, then employ build-out resistors at the outputs. (no schem available.)
The first balanced version converts the balanced input(s) to single ended....then distribute.....then re-balance the outputs.
Is there a way to keep the entire process balanced all the way through?
Is there a reason it is not done this way?
Is it because it is too difficult to keep the differential signal matched (cmr?) , especially when adding a level control?
As you might guess, I'd like to build a high quality balanced DA.
If there were to be the "ultimate" overkill DA, how would it be done ?
But in real-world "do-able" DIY terms (using IC opamps)....how ?
=RR= |
|
|
| Keith Taylor |
Redrabbit, the first op amp is inverting because the person that drew it made a mistake and transposed the inputs. Inverting should be on top and non inverting below. As shown, the amplifier has positive feedback and would thus make a better oscillator than an amplifier!
As for maintaining balanced throughout, implementing gain controls would be a bit more difficult and I don't believe you gain anything performance wise. If you are seriously interested in a balanced system have a look at www.thatcorp.com. The "outsmarts" line driver and "ingeneous" line receiver look set to become the standard for proffessional use.
Keith |
|
|
| sek |
The usual suspects among audio professionals come from Analog Devices and Texas Instruments, although THAT outperforms them (for a considerably higher price).
My experience also is that going balanced all the way through a device is not beneficial unless you go overboard with capacitor and resistor selection, potentiometer accuracy, board layout and wiring. Having a balanced input receiver and a balanced output transmitter is all that is required in order to interface complex electronic devices to each other.
OTOH, going symmetric all the way can be benefical in digital interfacing systems, where i.e. only the input stage before an ADC and the post filter stage after a DAC are analogue opamp circuitry! In these special cases, the circuit complexity is usually low enough to justify higher quality symmetric component selection, as the parts count is usually reasonably low.
If you're talking about symmetric transducer preamps, passive line stages, Pass amps and transformer driven electro- or magnetostats, that's another story. But in that case an opamp interface would be out of question anyways, I suppose... ;)
Cheers,
Sebastian. |
|
|
| lineup |
| quote: | Originally posted by redrabbit
Balanced versions of these.......
The first schematic is similar to the Audio Metrics DA16:
http://www.radiosystems.com/Manuals/da8-16manual.pdf
As you might guess, I'd like to build a high quality balanced DA.
If there were to be the "ultimate" overkill DA,
how would it be done ?
But in real-world "do-able" DIY terms (using IC opamps)....how ?
. |
Hello.
To make a good balanced driver, Output .. is not as easy, as we may think.
Same goes for making a good balanced Input stage.
Certainly not just take
- one non-inverted and
- one inverted Opamp, for output
Yeah, okay :) we can do it, if we do not care to much about true Precision and low distortion.
In fact, I would prefer 99 times out of 100 to use
un-balanced signal like normal
before a badly tracking, not very well matched
balanced signal input
There are such really high quality IC-s for professional Audio use.
By Analog Devices.
They are easy to use, and are TRULY matched to 'the last drop' !!!! :cool:
Also, they do not cost more, than we can afford to buy them.
It should not be too difficult to find a dealer close to you.
Building my own homemade Balanced driver + Receiver + Amplifier
is something I would NOT want to do.
If I wanted this and really needed, I would go for some dedicated IC, like the ones by Analog Devices.
Audio Regards :)
lineup
====================
Recommended: :cool: :cool: :)
|
|
|
| sek |
And just for the sake of completeness, there's also the DRV134 transmitter and INA134 receiver from Texas Instruments, as TI claims them to be an improvement over the AD devices.
Those four ICs (in general) are also used in many projects here on diyaudio.com, just do a search for the part numbers. :)
The THAT ICs would still be better and - in the sense of lineup's explanation - also be among the parts a DIYer could afford for the occasional project. They are just a tad little more difficult to source for many of us. :(
Cheers,
Sebastian.
PS @lineup: "point of view" ;) |
|
|
| lineup |
| quote: | Originally posted by sek
And just for the sake of completeness, there's also the DRV134 transmitter and INA134 receiver from Texas Instruments, as TI claims them to be an improvement over the AD devices.
Those four ICs (in general) are also used in many projects here on diyaudio.com, just do a search for the part numbers. :)
The THAT ICs would still be better and - in the sense of lineup's explanation - also be among the parts a DIYer could afford for the occasional project. They are just a tad little more difficult to source for many of us. :(
Cheers,
Sebastian.
PS @lineup: "point of view" ;) |
Thanks, sek, Sebastian.
I will remember the good Texas alternatives.
Just if anyone asks again.
Personally, I have never had the need to go balanced, ever.
Regards
lineup :) still, today, a reasonably Balanced man :D |
|
|
| Keith Taylor |
Redrabbit, I had a couple of thoughts after posting above. The problem with a fully symetrical (balanced) system is that if you disturb the balance of one of the outputs you are likely to do the same to all the others for any conceivable topology that comes to mind. Also note that the THAT CORP line driver chips and possibly others, need to be fed from a low source impedance (less than 500 ohms) This tends to put the kibosh on a normal gain control pot between the two stages. I find it hard to imagine why you would want gain controls on a distribution amplifier when its main function is to take a signal and split it N ways, preferably unaltered.
Keith |
|
|
| redrabbit |
Thanks for the replies everyone.
I have used the balancing chips before (INA, SSM, THAT) . Strange though, the balanced drivers are always in more demand.......digikey is often out of them.
The reason for my quest to build a DA is.... firstly: to learn something..... secondly: to customize it to my needs..... thirdly, to pet my ego. (NNITO)
I have plenty of receivers SSM2143, INA2137.....but I'm out of the THAT 1646 drivers at the moment.
I do enjoy the challenge of building perfectly symmetrical 3 chip instrumentation amps, or line drivers, I won't enjoy building 12 of them. :dead:
You're right Keith, THAT1646 datasheet says: | quote: | devices must
be driven from a low-impedance source, preferably
directly from opamp outputs, to maintain the
specified performance. |
So another buffer opamp before each might be necessary.
QUESTIONS.....
If I went with these chips and I wanted gain (level) adjustment, I assume I'd wire the receivers , for +2 gain (as stated in the datasheets)...and add a (trim) pot before the balancing output chip ?
Also, what about the DA method I described in my OP....about the circuit that drives one balanced, current boosted output, then uses build-out resistors ?
I assume this consists of one receiver, and one driver with a power buffer on it's output (discrete buffer, of packaged -BUF634) with a single gain control (??)
http://i5.photobucket.com/albums/y1...ot/DSCN3358.jpg
http://i5.photobucket.com/albums/y1...ot/DSCN3359.jpg
Perhaps I'll do up a rough schematic of either version.
=RR= |
|
|
| Keith Taylor |
By +2 gain you presumably mean x2 gain which is 6bB. If you were to cascade this stage with a balanced driver which also has 6 dB gain (single ended to differential) you will end up with 12 dB gain overall. I would be inclined to use the line receivers with -6dB gain to give a system gain of unity.
A buffer with build out resistors is another approach to achieving a splitting function. This is a common method used in VDA's (video distribution amps). Each output has a 75 ohm source impedance defining resistor to give minimal reflections when used with 75 0hm coax cable. The low driving source impedance in the amplifier results in good isolation between outputs. In audio we do not concern ourselves with reflections so the build out resistors will be given a value for different reasons including, do we want the system to tolerate shorted outputs? how many?. How long, and therefore what capacitance are the cables going to have? Apart from any of the foregoing the resistors will be needed to preserve the stability of the driving amp when loaded with cable capacitance. For a given cable capacitance per metre the source impedance has a big effect on high frequency losses. There is a Rane Corp tutorial on this subject.
Keith |
|
|
| AndrewT |
Hi,
when terminated coax is used to feed the buffered audio signal to the next stage, what impedance does the buffer see at it's output?
Is it the 75r series resistor at the buffer + 75r load resistor to ground at the far end of the 75ohm coax?
At audio frequencies does the cable have no effect on this total 150r load? |
|
|
| jacco vermeulen |
| You can still go expensive on the AD stuff. |
|
|
| Keith Taylor |
AndrewT, you are correct in that for a single load on the buffer stage described the amp would be seeing 150 ohms. You need to generate a signal of twice the voltage you want as half of it is lost in the source impedance defining resistor and the other half arrives at the far end of the cable. The source resistor is called a back terminating resistor I think. This type of system results in a high "return loss" both for forward and reflected signals. This type of system with matched source/cable/load impedances has very limited application in audio outside of telecommunications.
In the real world of audio, reflections due to what ever cause are going to arrive back at the source much sooner than the period of the highest frequency signals and thus can be ignored. This reality is why source impedances tend to be low and load impedances high and the cable characteristic impedance does not matter. It goes by the name of "voltage matching" and has been in use for 30 odd years.
Keith |
|
|
| AndrewT |
Thanks Keith,
but does the buffer see only the 150r as a load, or does it see 150r//cable capacitance, or something else?
I'm thinking about very long twisted pairs (CAT5) travelling throughout the house. Well, very long in terms of domestic interconnects. |
|
|
| Nordic |
From today's TI newsletter...
DRV601
"DirectPath™ Stereo Line Driver with Variable Input Gain: The DRV601 is a cost-effective, integrated stereo audio line driver designed to drive 600Ù or high-line impedances. An integrated charge pump enables operation from a single external supply voltage and the device allows input gain to be set by external resistors.
SNR: 109dB
2VRMS/Ch output voltage into 600Ù at 3.3-V supply
Applications: Set-top boxes, CD/DVD players, DVD receivers, HTIB, PDP/LCD TVs
DRV601 Price (1ku): $0.75" |
|
|
| Keith Taylor |
Andrew, you must have been at your computer when I posted! Yes all cables will have capacitance which will be in parallel with the load. It is not the value of the capacitance alone that counts but the value in relation to the source impedance of the generator (the device driving the cable). With these two elements we have a low pass filter. This is explained in the Rane notes mentioned above.
Keith |
|
|
| AndrewT |
that gives me F-3dB of 193kHz for 110ohm CAT5 that are about 50m long and 150pF/m.
I can live with that, but obviously 75ohm or 50ohm would allow higher bandwidth.
I had a feeling that terminated cables didn't behave like that. |
|
|
| redrabbit |
| quote: | Originally posted by Keith Taylor
Maybe it's another incarnation of you? Cheers
Keith |
Yes, that's my other ego.
I went over there because I knew they (mediatechnology, etc) have a lot of experience with the "THAT" audio products.
Anyone wishing to see my block diagram/schematic should visit the link Keith gave above.
I've already built it, so far, so good. Very simple. Testing it this week.
=RR= |
|
|
| mediatechnology |
| quote: | | Redrabbit, the first op amp is inverting because the person that drew it made a mistake and transposed the inputs |
Correct. Based on the typography it looks like it came from TI. They probably ought to fix that one. |
|
|
| redrabbit |
| quote: | Originally posted by mediatechnology
Correct. Based on the typography it looks like it came from TI. They probably ought to fix that one. |
(sarcastic giggle)...As far back as 1999 (or longer??) this mistake has been present. The current TL08x datasheet still wrong.
http://focus.ti.com/lit/ds/symlink/tl081.pdf
But the TL07x has it correct (shows the same circuit application).
Datasheets too often have mistakes. I don't think they pass thru the lab for approval.
Can you find a mistake in this Analog Devices schematic?
=RR= |
|
|
| Keith Taylor |
Hello again Redrabbit, that stage is certainly novel and seems to be an unusual configuration of a differential output stage employing cross coupled feedback loops. At first glance you tend to think uh oh, the load is being driven with in phase signals instead of antiphase. Closer examination reveals that the lower stage takes its feedback signal from the upper stage output, which is inverted. Similar remarks apply to the "inner" feedback loops. Nothing jumps out at me as being wrong; just different. These stages are notorious for their general instability and require extremely close matching of resistor values. This is best done by laser trimming on the chip as in the single chip products by AD, Burr Brown, THAT etc.
I think you are likely to get some good advice on Wayne's forum. I became his 100th member the other day!
Cheers Keith |
|
|
| redrabbit |
Thanks Keith for some insight on the above OP471 circuit.
The pin numbers are wrong though, 12 & 13 are reversed.
=RR= |
|
|
| mediatechnology |
I think that there's a second error in the ADI drawing in addition to the pin numbers of the top op amp.
The + and - symbols are also wrong on the bottom op amp. The pin numbers are correct, it just shows the inverting an non-inverting symbols wrong.
If you compare it to the SSM2142 CCOS circuit, drawn correctly below, you'll see it:
Image Courtesy of Analog Devices, SSM2142 Data Sheet
Here's the original Thomas Hay 1980 AES figure:
Image Courtesy AES, Preprint 1723
BTW Thanks Keith. |
|
|
| redrabbit |
I tried building this 2 years ago, and it never worked.....at a time when I knew less than now (which is will be more, in a few years, as long as they don't cut the power grid. :xmastree: )
I think Wayne is correct......one sign is that the inner 10k feedback resistors would/should be on the negative input side of the lower opamp...???
I planned to re-do this at some point, because I have several ceramic OP471's.
(or, the OP470 , a low-noise version)
=RR= |
|
|
| mediatechnology |
FWIW the Thomas Hay cite is closer to the OP471 circuit (despite the error) than the SSM2142. The OP471 and Hay inputs use a non-inverting follower to buffer the input. The FB resistors reference to ground. Hay probably serves as a better model to fix the ADI error than the 2142 does.
It would seem, based on casual observation only, that the output balance in the corrected OP471 cite and the Hay version would not have the sensitivity to input source resistance that the SSM2142 does. Source impedance doesn't become part of the gain equation and balance like it does with the 2142. |
|
|
|
