Hi guys.
I'm going to build an volume control on PGA2311 which will be used with XLR (balanced) input.
Signal will came from typical DAC (non pro stuff).
And here is a question, may I use one PGA2311 per one channel?
PGA2311 is a single-ended chip and I'm not sure is that will work correctly in terms of signal integrity etc.
So let me explain
XLR Input -> LPF -> PGA2311 (hot and cold pins connected to left and right channel input pins).
Or should I use INA13x (or any other opamp) chip to convert balanced singnal to SE -> PGA2311 -> DVR135 to convert it back to balanced again?
First idea is much more simpler and economic but I'm unsure is that will work as it should.
Could you help me with that?
Thank you for your time and effort to help me out
I'm going to build an volume control on PGA2311 which will be used with XLR (balanced) input.
Signal will came from typical DAC (non pro stuff).
And here is a question, may I use one PGA2311 per one channel?
PGA2311 is a single-ended chip and I'm not sure is that will work correctly in terms of signal integrity etc.
So let me explain
XLR Input -> LPF -> PGA2311 (hot and cold pins connected to left and right channel input pins).
Or should I use INA13x (or any other opamp) chip to convert balanced singnal to SE -> PGA2311 -> DVR135 to convert it back to balanced again?
First idea is much more simpler and economic but I'm unsure is that will work as it should.
Could you help me with that?
Thank you for your time and effort to help me out
To keep the benefit of using balanced line, you need to be very careful about gain mismatches between the hot and cold signals.
I can't recall the gain specification for the pga off the top of my head, but would be surprised if it were nearly as good as using a balanced line receiver/ transmitter around the pga.
I presume your goal is noise rejection?
In writing the above I starated asking myself questions like : why a volume control on the middle of a balanced line run?
And: is this person going to get onto hot water with grounding issues? Think about where control for the plan comes from.
I can't recall the gain specification for the pga off the top of my head, but would be surprised if it were nearly as good as using a balanced line receiver/ transmitter around the pga.
I presume your goal is noise rejection?
In writing the above I starated asking myself questions like : why a volume control on the middle of a balanced line run?
And: is this person going to get onto hot water with grounding issues? Think about where control for the plan comes from.
Hi guys.
I'm going to build an volume control on PGA2311 which will be used with XLR (balanced) input.
Signal will came from typical DAC (non pro stuff).
And here is a question, may I use one PGA2311 per one channel?
PGA2311 is a single-ended chip and I'm not sure is that will work correctly in terms of signal integrity etc.
So let me explain
XLR Input -> LPF -> PGA2311 (hot and cold pins connected to left and right channel input pins).
Or should I use INA13x (or any other opamp) chip to convert balanced singnal to SE -> PGA2311 -> DVR135 to convert it back to balanced again?
First idea is much more simpler and economic but I'm unsure is that will work as it should.
Could you help me with that?
Thank you for your time and effort to help me out
I have done exactly the same thing before. It works as it should be, but the sound is not as good as I expected. Probably due to cheap PSU and bad implementation. I used Ebay China contoller. Also THAT chip is better.
+1 for converting to unbalanced and back. Reading Bruno Putzeys' "G-Word" whitepaper is advised.
Make sure your balanced receiver can do less than unity gain, since INAs often don't - common balanced line driver circuitry has 6 dB of gain, so you're generally at least that much hotter than unbalanced, and pro levels could be up to +22 dBu (almost 10 Vrms), while the PGA's input is somewhat limited by its +/-5 V supplies, so you can't run more than 2.5 Vrms into that. Thus depending on where your balanced signal is coming from, the balanced receiver would have to have a gain of -6 dB to -12 dB (maybe -14 dB).
Likewise, it has to be made sure that PGA output level is well-mapped to headphone amp output level. Given that headphones wired up balanced tend to be fullsize models that are not super-sensitive, you'd probably be just fine with enough gain to achieve full output (2.5 µV of noise + 12 dB is 10 µV or 2 dB SPL into something 102 dB / 1 Vrms). That said, if you do need a variable gain control to accomodate more sensitive cans, after the PGA would be a good spot.
Oh, and XLR pin 1 goes straight to chassis.
Make sure your balanced receiver can do less than unity gain, since INAs often don't - common balanced line driver circuitry has 6 dB of gain, so you're generally at least that much hotter than unbalanced, and pro levels could be up to +22 dBu (almost 10 Vrms), while the PGA's input is somewhat limited by its +/-5 V supplies, so you can't run more than 2.5 Vrms into that. Thus depending on where your balanced signal is coming from, the balanced receiver would have to have a gain of -6 dB to -12 dB (maybe -14 dB).
Likewise, it has to be made sure that PGA output level is well-mapped to headphone amp output level. Given that headphones wired up balanced tend to be fullsize models that are not super-sensitive, you'd probably be just fine with enough gain to achieve full output (2.5 µV of noise + 12 dB is 10 µV or 2 dB SPL into something 102 dB / 1 Vrms). That said, if you do need a variable gain control to accomodate more sensitive cans, after the PGA would be a good spot.
Oh, and XLR pin 1 goes straight to chassis.
I'll put in a vote for using a balanced signal path. An advantage of a balanced signal path is that, to a first order, a balanced signal path helps to keep signal currents out of ground, and given a few other techniques, allows signal induced power supply currents to be kept out of the power supples and ground.
The cost is that usually 2x the circuitry is required to provide a balanced signal path, but the benefit is that because signal currents are able to be kept out of ground and the power supplies, PCB layout is greatly simplified.
The typical idea now is that 'balanced' is useful only between separate boxes, and primarily serves to eliminate noise from interface cabling. While that's true, one can consider separate functional blocks within a larger PCB to be 'separate boxes', and so the ground and power supply isolation that one gets from a balanced signal path also helps these circuit subsystems and functional blocks to work properly despite the fact that they share a PCB ground and power supply system. So, while 'balanced' is great between pieces of gear, it's also great within a piece of gear for the same reasons.
In a balanced signaling system, every time a signal is sent to ground through a resistor or other component, there will be an equal and opposite signal also sent to ground. If one arranges the layout so that these two signals to ground can cancel each other within a small piece of PCB foil, then these currents essentially disappear, instead of traveling back to their source, possibly becoming coupled to other circuit nodes due to IR losses in the PCB foil as these currents travel back to their source.
While it's more complicated, there are similar cancellation techniques that can be used to keep transient signal induced currents out of the power supply, and also keep half wave rectified power supply currents out of the ground system, where they can inject significant distortion into a circuit through IR coupling through the PCB ground foil. Those power supply bypass caps can couple half wave rectified signal currents into the ground foil, and as these currents travel back from their source/load, voltages can be generated in the finite impedance PCB foil, and possibly couple them into the signal path.
While the degree of isolation depends upon the degree of signal balance, even basic balanced circuitry that has only modest balancing, for example to 1%, can offer 40dB of reduction of the magnitude of signal currents traveling in ground or the power supplies. With a more closely balanced signal, one can get greater isolation. So, even though the PGA2311 is not precisely balanced, the two channels should match each other to a pretty high degree, and offer some significant advantages.
To summarize, I know that the common parlance is that "balanced is to eliminate cable problems and interference". However, if you dig a little bit deeper, one can take advantage of a symmetric signaling system in order to simplify and minimize the currents traveling within ground as well as the power supplies. I've built some circuits recently that use these techniques, and they perform extremely well, despite their relatively simple and compact PCB layouts.
The cost is that usually 2x the circuitry is required to provide a balanced signal path, but the benefit is that because signal currents are able to be kept out of ground and the power supplies, PCB layout is greatly simplified.
The typical idea now is that 'balanced' is useful only between separate boxes, and primarily serves to eliminate noise from interface cabling. While that's true, one can consider separate functional blocks within a larger PCB to be 'separate boxes', and so the ground and power supply isolation that one gets from a balanced signal path also helps these circuit subsystems and functional blocks to work properly despite the fact that they share a PCB ground and power supply system. So, while 'balanced' is great between pieces of gear, it's also great within a piece of gear for the same reasons.
In a balanced signaling system, every time a signal is sent to ground through a resistor or other component, there will be an equal and opposite signal also sent to ground. If one arranges the layout so that these two signals to ground can cancel each other within a small piece of PCB foil, then these currents essentially disappear, instead of traveling back to their source, possibly becoming coupled to other circuit nodes due to IR losses in the PCB foil as these currents travel back to their source.
While it's more complicated, there are similar cancellation techniques that can be used to keep transient signal induced currents out of the power supply, and also keep half wave rectified power supply currents out of the ground system, where they can inject significant distortion into a circuit through IR coupling through the PCB ground foil. Those power supply bypass caps can couple half wave rectified signal currents into the ground foil, and as these currents travel back from their source/load, voltages can be generated in the finite impedance PCB foil, and possibly couple them into the signal path.
While the degree of isolation depends upon the degree of signal balance, even basic balanced circuitry that has only modest balancing, for example to 1%, can offer 40dB of reduction of the magnitude of signal currents traveling in ground or the power supplies. With a more closely balanced signal, one can get greater isolation. So, even though the PGA2311 is not precisely balanced, the two channels should match each other to a pretty high degree, and offer some significant advantages.
To summarize, I know that the common parlance is that "balanced is to eliminate cable problems and interference". However, if you dig a little bit deeper, one can take advantage of a symmetric signaling system in order to simplify and minimize the currents traveling within ground as well as the power supplies. I've built some circuits recently that use these techniques, and they perform extremely well, despite their relatively simple and compact PCB layouts.
This is a non-issue; balanced signals are defined by the level BETWEEN the two lines and NOT to ground, that's the whole idea.
A PGA is perfect for this application because you can use one half for each phase and you do not have to convert to single ended and back. At the instant you convert to SE you again introduce ground as a reference with all the issues that go with that, and destroys the advantages of balanced connection.
Jan
Indeed, and therefor it is best to use one chip for each channel. The input impedances are not well defined but the two channels on a chip should be reasonably well matched.
Jan
I built a preamp using this technique in 2010. You can get the write-up and schematics here
Ovation SCA-1: A high performance fully balanced line pre-amplifier
The PGA chips are excellent - they do exactly what the spec sheet says they do. Esoteric use them in their top of the line preamps which have consistently gotten very good reviews.
Ovation SCA-1: A high performance fully balanced line pre-amplifier
The PGA chips are excellent - they do exactly what the spec sheet says they do. Esoteric use them in their top of the line preamps which have consistently gotten very good reviews.
What sort of CMRR did you get at various volume settings? I'd be afraid that even relatively good channel tracking would degrade it... I mean, 0.1 dB of mismatch still is ~1%, resistor matching could be a lot better than that. (I recently had two inexpensive Yageo 2.2k 1% MFs from the same tape in hand, they differed by about 0.2 ohms according to my multimeter.)
BTW, I found a minor error on the schematic - one of R47/48 connects to the wrong signal line. I would also recommend a T-type connection to keep common-mode input impedance high (say, 10k+10k with 47k to ground, depending on what the LM4562 is confortable with, similarly for input capacitance, maybe 220p+220p T'd via 47p).
If staying all-balanced is desired, I would use a circuit based on a balanced opamp like OPA1632 in front of the PGA, or something else that gives inherent CMRR.
BTW, I found a minor error on the schematic - one of R47/48 connects to the wrong signal line. I would also recommend a T-type connection to keep common-mode input impedance high (say, 10k+10k with 47k to ground, depending on what the LM4562 is confortable with, similarly for input capacitance, maybe 220p+220p T'd via 47p).
If staying all-balanced is desired, I would use a circuit based on a balanced opamp like OPA1632 in front of the PGA, or something else that gives inherent CMRR.
I was referring to a circuit like Bonsai's (or proposed by the OP), with single-ended buffers feeding two channels of the PGA directly. Common mode to differential mode conversion depends on channel tracking there. I haven't done the math in a while, but asking for 60 dB of CMRR (which would be a decent value for a line-level input though not SOTA) would require matching on the order of 0.1% or 0.01 dB. The part is specified for +/- 0.1 dB, no idea how good it is in real life.
Get why I'm seeing a potential problem there?
Get why I'm seeing a potential problem there?
Yes I see what you mean. Not sure what the consequences would be. If you keep everything balanced, a non-perfect attenuation tracking would mean that the differential voltage (which is what matters in a balanced connection) would be slightly too large or too small compared with the nominal attenuation.
I am hesitating to mention 'ground' but let's talk about the 'center of power supply'. A non-perfect tracking would mean that the signal would move slightly with respect to 'center of supply'. That is a common mode component I agree, but again, not sure what the effect would be on quality however defined.
Does this make sense?
Edit: Not mentioned yet but you should NOT connect the PGA input return pins to ground! Make a 'center point' with two resistors from each XLR signal pin and connect the PGA returns to that, of course.
Jan
I am hesitating to mention 'ground' but let's talk about the 'center of power supply'. A non-perfect tracking would mean that the signal would move slightly with respect to 'center of supply'. That is a common mode component I agree, but again, not sure what the effect would be on quality however defined.
Does this make sense?
Edit: Not mentioned yet but you should NOT connect the PGA input return pins to ground! Make a 'center point' with two resistors from each XLR signal pin and connect the PGA returns to that, of course.
Jan
I think the main thing is gain matching. The input buffers effectively remove any errors due to input loading and ensure the PGA2320 is fed from a low source impedance.
If there are gain errors between the two halves, then yes, there will be a CMRR issues. However, I assumed that on the same chip, the tracking would be good and never experienced any problems. You could of course gain trim or balance these if you like. One way to do this would be to add a pot to give circa +-4% gain adjustment in the feedback loop of the output buffer circuit on one side. Of course this assumes the gains track over the full 7 bit range of the PGA2320.
Practically, the preamp had a very open sound - I really liked it. And the LME49600 buffers make for a truly fantastic headphone amplifier.
My problem was that the boards were enormous - there was no way I could house that thing and have it look decent. I don't know what I was thinking when I laid the boards out - put it down to experience. I re-kindled my C programming skills (which are beginner level nevertheless) so it served a good purpose. The graphics were cool, but far too slow - I would need to use a 16 bit I/F graphic display (I used 8 bit on this preamp) and rewrite the code on a faster processor. The mbed platform is great, but not for intensive graphics.
If there are gain errors between the two halves, then yes, there will be a CMRR issues. However, I assumed that on the same chip, the tracking would be good and never experienced any problems. You could of course gain trim or balance these if you like. One way to do this would be to add a pot to give circa +-4% gain adjustment in the feedback loop of the output buffer circuit on one side. Of course this assumes the gains track over the full 7 bit range of the PGA2320.
Practically, the preamp had a very open sound - I really liked it. And the LME49600 buffers make for a truly fantastic headphone amplifier.
My problem was that the boards were enormous - there was no way I could house that thing and have it look decent. I don't know what I was thinking when I laid the boards out - put it down to experience. I re-kindled my C programming skills (which are beginner level nevertheless) so it served a good purpose. The graphics were cool, but far too slow - I would need to use a 16 bit I/F graphic display (I used 8 bit on this preamp) and rewrite the code on a faster processor. The mbed platform is great, but not for intensive graphics.
Last edited:
I was hoping for that kind of answers
So should I really avoid converting BAL to SE and then SE to BAL?
How about VRMS? Does PGA2311 will be able to accept typical range of balanced signal?
BTW I'm going to build input stage based on good opamp (like OPA1632 or OPA1642) because I want to build good LPF.
Gain missmatch is kinda good on PGA2311.
So should I really avoid converting BAL to SE and then SE to BAL?
How about VRMS? Does PGA2311 will be able to accept typical range of balanced signal?
BTW I'm going to build input stage based on good opamp (like OPA1632 or OPA1642) because I want to build good LPF.
Gain missmatch is kinda good on PGA2311.
See this thread:
http://www.diyaudio.com/forums/analog-line-level/274072-my-pga2310-box.html
I used SSM 2141 and 2142 in and out of the PGA2310.
http://www.diyaudio.com/forums/analog-line-level/274072-my-pga2310-box.html
I used SSM 2141 and 2142 in and out of the PGA2310.
Last edited:
I was hoping for that kind of answers
So should I really avoid converting BAL to SE and then SE to BAL?
How about VRMS? Does PGA2311 will be able to accept typical range of balanced signal?
BTW I'm going to build input stage based on good opamp (like OPA1632 or OPA1642) because I want to build good LPF.
Gain missmatch is kinda good on PGA2311.
I would use the PGA in balanced mode.
I think the problem is wrong approached.
Check how the volume adjustment is done in PGA and you will understand.
The volume it is adjusted in PGA by referring to GND so differential work is for nothing.
Better convert it to unbalanced and use the PGA like in datasheet and if it is needed converted back to balanced.
Check how the volume adjustment is done in PGA and you will understand.
The volume it is adjusted in PGA by referring to GND so differential work is for nothing.
Better convert it to unbalanced and use the PGA like in datasheet and if it is needed converted back to balanced.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.