| Home |
| Search |
| Today's Posts |
|
#1
|
|||
|
|||
Balanced In the 2nd
Hi
Did not get any response on rec.audio.tech, so I post it here... For a small 4-input-mixer, I tried to design a active balanced input. It does include bootstrapping for high common-mode input impedance and a DC-servo. As I'm not an expert in this field, may you have a look at it? http://n.ethz.ch/student/groners/RAP...ed_Line_In.pdf A few specific questions: * which opamps would you use for the instr.-amp and the bootstrapping? Should operate on +-22V. Thought about OP275 and LT1128. * is there a suitable opamp for the DC-servo, working with +-22V? * is there an easy way to make the RFI-filter more effective (without lowering the bandwith)? Would like a 2nd order Bessel, but that gives large inductors. * if I wanted to trim CMRR (I'll use 0.1% resistors), I would trim R16 and C8, or what? Thanks for any input! Samuel |
|
#2
|
|||
|
|||
|
"Samuel Groner" wrote ...
For a small 4-input-mixer, I tried to design a active balanced input. It does include bootstrapping for high common-mode input impedance and a DC-servo. As I'm not an expert in this field, may you have a look at it? http://n.ethz.ch/student/groners/RAP...ed_Line_In.pdf If I am reading your schematic correctly, (the connection dots are miniscule) any "DC-servo" paths appear to be AC-coupled. How does that work? And why multiple paths? |
|
#3
|
|||
|
|||
|
"Samuel Groner" wrote ...
For a small 4-input-mixer, I tried to design a active balanced input. It does include bootstrapping for high common-mode input impedance and a DC-servo. As I'm not an expert in this field, may you have a look at it? http://n.ethz.ch/student/groners/RAP...ed_Line_In.pdf If I am reading your schematic correctly, (the connection dots are miniscule) any "DC-servo" paths appear to be AC-coupled. How does that work? And why multiple paths? |
|
#4
|
|||
|
|||
|
"Samuel Groner" wrote in message om... Hi Did not get any response on rec.audio.tech, so I post it here... For a small 4-input-mixer, I tried to design a active balanced input. It does include bootstrapping for high common-mode input impedance and a DC-servo. As I'm not an expert in this field, may you have a look at it? http://n.ethz.ch/student/groners/RAP...ed_Line_In.pdf You might try a common-mode inductor for RFI-proofing rather than an RC network; it's probably more effective, and won't mess up your common-mode rejection at lower frequencies. A few specific questions: * which opamps would you use for the instr.-amp and the bootstrapping? Should operate on +-22V. Thought about OP275 and LT1128. Another good possibility is the OPA2604. * is there a suitable opamp for the DC-servo, working with +-22V? The OP07 is rated at +-22V. For safety, probably you shouild really run it at +-21V. That goes for the others, too, except for the OPA2604, which will work at +-24V. * is there an easy way to make the RFI-filter more effective (without lowering the bandwith)? Would like a 2nd order Bessel, but that gives large inductors. See above. A common-mode inductor might allow you to eliminate the bootstrapping circuit, too, saving space and money. Peace, Paul |
|
#5
|
|||
|
|||
|
"Samuel Groner" wrote in message om... Hi Did not get any response on rec.audio.tech, so I post it here... For a small 4-input-mixer, I tried to design a active balanced input. It does include bootstrapping for high common-mode input impedance and a DC-servo. As I'm not an expert in this field, may you have a look at it? http://n.ethz.ch/student/groners/RAP...ed_Line_In.pdf You might try a common-mode inductor for RFI-proofing rather than an RC network; it's probably more effective, and won't mess up your common-mode rejection at lower frequencies. A few specific questions: * which opamps would you use for the instr.-amp and the bootstrapping? Should operate on +-22V. Thought about OP275 and LT1128. Another good possibility is the OPA2604. * is there a suitable opamp for the DC-servo, working with +-22V? The OP07 is rated at +-22V. For safety, probably you shouild really run it at +-21V. That goes for the others, too, except for the OPA2604, which will work at +-24V. * is there an easy way to make the RFI-filter more effective (without lowering the bandwith)? Would like a 2nd order Bessel, but that gives large inductors. See above. A common-mode inductor might allow you to eliminate the bootstrapping circuit, too, saving space and money. Peace, Paul |
|
#6
|
|||
|
|||
|
On 6 Jun 2004 09:19:41 -0700, in rec.audio.pro you wrote:
Hi Did not get any response on rec.audio.tech, so I post it here... For a small 4-input-mixer, I tried to design a active balanced input. It does include bootstrapping for high common-mode input impedance and a DC-servo. As I'm not an expert in this field, may you have a look at it? http://n.ethz.ch/student/groners/RAP...ed_Line_In.pdf A few specific questions: * which opamps would you use for the instr.-amp and the bootstrapping? Should operate on +-22V. Thought about OP275 and LT1128. Why do you need bootstrapping, what source Z's are you dealing with? * is there a suitable opamp for the DC-servo, working with +-22V? * is there an easy way to make the RFI-filter more effective (without lowering the bandwith)? Would like a 2nd order Bessel, but that gives large inductors. try the Wurth Electronics CM chokes, say 60uH part# 74279452, remove the 500R, and leave the 1nF's. as most RF interference will be common mode I * if I wanted to trim CMRR (I'll use 0.1% resistors), I would trim R16 and C8, or what? Ive never been a great fan of DC servos, and since you already have input caps, (Im assuming that you use the servo to get rid of the o/p cap), why not just use a good quality o/p cap. put the circuit into Spice, and see what the sub sonic stability is like Again I would increase the f/b R's R8,R10, a bit, say 3k3, reducing the loading on the ic Thanks for any input! Samuel my .02 euro's martin Serious error. All shortcuts have disappeared. Screen. Mind. Both are blank. |
|
#7
|
|||
|
|||
|
On 6 Jun 2004 09:19:41 -0700, in rec.audio.pro you wrote:
Hi Did not get any response on rec.audio.tech, so I post it here... For a small 4-input-mixer, I tried to design a active balanced input. It does include bootstrapping for high common-mode input impedance and a DC-servo. As I'm not an expert in this field, may you have a look at it? http://n.ethz.ch/student/groners/RAP...ed_Line_In.pdf A few specific questions: * which opamps would you use for the instr.-amp and the bootstrapping? Should operate on +-22V. Thought about OP275 and LT1128. Why do you need bootstrapping, what source Z's are you dealing with? * is there a suitable opamp for the DC-servo, working with +-22V? * is there an easy way to make the RFI-filter more effective (without lowering the bandwith)? Would like a 2nd order Bessel, but that gives large inductors. try the Wurth Electronics CM chokes, say 60uH part# 74279452, remove the 500R, and leave the 1nF's. as most RF interference will be common mode I * if I wanted to trim CMRR (I'll use 0.1% resistors), I would trim R16 and C8, or what? Ive never been a great fan of DC servos, and since you already have input caps, (Im assuming that you use the servo to get rid of the o/p cap), why not just use a good quality o/p cap. put the circuit into Spice, and see what the sub sonic stability is like Again I would increase the f/b R's R8,R10, a bit, say 3k3, reducing the loading on the ic Thanks for any input! Samuel my .02 euro's martin Serious error. All shortcuts have disappeared. Screen. Mind. Both are blank. |
|
#8
|
|||
|
|||
|
If I am reading your schematic correctly,
(the connection dots are miniscule) any "DC-servo" paths appear to be AC-coupled. Do I mix up the terminology? The circuit is AC-coupled. How do you call the opamp with the lowpassfilter that eliminates the offset at the output of the diff-amp? Not DC-servo? And why multiple paths? You mean R3/R4 and R5/R6? Thought that R3/R4 would help sources with some leakage-current. Samuel |
|
#9
|
|||
|
|||
|
If I am reading your schematic correctly,
(the connection dots are miniscule) any "DC-servo" paths appear to be AC-coupled. Do I mix up the terminology? The circuit is AC-coupled. How do you call the opamp with the lowpassfilter that eliminates the offset at the output of the diff-amp? Not DC-servo? And why multiple paths? You mean R3/R4 and R5/R6? Thought that R3/R4 would help sources with some leakage-current. Samuel |
|
#10
|
|||
|
|||
|
try the Wurth Electronics CM chokes, say 60uH part# 74279452, remove
the 500R, and leave the 1nF's. as most RF interference will be common mode OK, the CM choke sounds good. Have to check where I get a good (and small) one. Ive never been a great fan of DC servos, Why not? and since you already have input caps, (Im assuming that you use the servo to get rid of the o/p cap), why not just use a good quality o/p cap. Is there a chance that a good elco (say the Roederstein non-polarized) will last for 20 years to keep the pot silent? Just trimming the diff-amp will not last forever, I guess? Thanks Samuel |
|
#11
|
|||
|
|||
|
try the Wurth Electronics CM chokes, say 60uH part# 74279452, remove
the 500R, and leave the 1nF's. as most RF interference will be common mode OK, the CM choke sounds good. Have to check where I get a good (and small) one. Ive never been a great fan of DC servos, Why not? and since you already have input caps, (Im assuming that you use the servo to get rid of the o/p cap), why not just use a good quality o/p cap. Is there a chance that a good elco (say the Roederstein non-polarized) will last for 20 years to keep the pot silent? Just trimming the diff-amp will not last forever, I guess? Thanks Samuel |
|
#12
|
|||
|
|||
|
Samuel Groner wrote:
try the Wurth Electronics CM chokes, say 60uH part# 74279452, remove the 500R, and leave the 1nF's. as most RF interference will be common mode OK, the CM choke sounds good. Have to check where I get a good (and small) one. You can salvage one from a scrapped PC power supply. Or you can buy a Corcom filter which contains a common mode choke and some capacitors, from Buerklin. Ive never been a great fan of DC servos, Why not? I don't know why he doesn't like them, but they have always worried me because there are so many things to go wrong. It's more stuff in there, just to avoid AC coupling. If you have the space, good film capacitors are not all _that_ expensive. and since you already have input caps, (Im assuming that you use the servo to get rid of the o/p cap), why not just use a good quality o/p cap. Is there a chance that a good elco (say the Roederstein non-polarized) will last for 20 years to keep the pot silent? Just trimming the diff-amp will not last forever, I guess? What's wrong with a 20 uF mylar? --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
|
#13
|
|||
|
|||
|
Samuel Groner wrote:
try the Wurth Electronics CM chokes, say 60uH part# 74279452, remove the 500R, and leave the 1nF's. as most RF interference will be common mode OK, the CM choke sounds good. Have to check where I get a good (and small) one. You can salvage one from a scrapped PC power supply. Or you can buy a Corcom filter which contains a common mode choke and some capacitors, from Buerklin. Ive never been a great fan of DC servos, Why not? I don't know why he doesn't like them, but they have always worried me because there are so many things to go wrong. It's more stuff in there, just to avoid AC coupling. If you have the space, good film capacitors are not all _that_ expensive. and since you already have input caps, (Im assuming that you use the servo to get rid of the o/p cap), why not just use a good quality o/p cap. Is there a chance that a good elco (say the Roederstein non-polarized) will last for 20 years to keep the pot silent? Just trimming the diff-amp will not last forever, I guess? What's wrong with a 20 uF mylar? --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
|
#14
|
|||
|
|||
|
On 7 Jun 2004 10:40:38 -0400, in rec.audio.pro you wrote:
Samuel Groner wrote: try the Wurth Electronics CM chokes, say 60uH part# 74279452, remove the 500R, and leave the 1nF's. as most RF interference will be common mode OK, the CM choke sounds good. Have to check where I get a good (and small) one. You can salvage one from a scrapped PC power supply. Or you can buy a Corcom filter which contains a common mode choke and some capacitors, from Buerklin. Ive never been a great fan of DC servos, Why not? I don't know why he doesn't like them, but they have always worried me because there are so many things to go wrong. KISS..... and if its self powered, less battery time It's more stuff in there, just to avoid AC coupling. If you have the space, good film capacitors are not all _that_ expensive. Me and the late Baz Porter, always seemed to have trouble with VLF hunting on the servo mechanism on wide gain range amps , albeit at a very low level, and I did some sims a year or so ago, in LTspice, yep they show up, sometimes in the millihertz region. and since you already have input caps, (Im assuming that you use the servo to get rid of the o/p cap), why not just use a good quality o/p cap. Is there a chance that a good elco (say the Roederstein non-polarized) will last for 20 years to keep the pot silent? Just trimming the diff-amp will not last forever, I guess? What's wrong with a 20 uF mylar? --scott martin Serious error. All shortcuts have disappeared. Screen. Mind. Both are blank. |
|
#15
|
|||
|
|||
|
On 7 Jun 2004 10:40:38 -0400, in rec.audio.pro you wrote:
Samuel Groner wrote: try the Wurth Electronics CM chokes, say 60uH part# 74279452, remove the 500R, and leave the 1nF's. as most RF interference will be common mode OK, the CM choke sounds good. Have to check where I get a good (and small) one. You can salvage one from a scrapped PC power supply. Or you can buy a Corcom filter which contains a common mode choke and some capacitors, from Buerklin. Ive never been a great fan of DC servos, Why not? I don't know why he doesn't like them, but they have always worried me because there are so many things to go wrong. KISS..... and if its self powered, less battery time It's more stuff in there, just to avoid AC coupling. If you have the space, good film capacitors are not all _that_ expensive. Me and the late Baz Porter, always seemed to have trouble with VLF hunting on the servo mechanism on wide gain range amps , albeit at a very low level, and I did some sims a year or so ago, in LTspice, yep they show up, sometimes in the millihertz region. and since you already have input caps, (Im assuming that you use the servo to get rid of the o/p cap), why not just use a good quality o/p cap. Is there a chance that a good elco (say the Roederstein non-polarized) will last for 20 years to keep the pot silent? Just trimming the diff-amp will not last forever, I guess? What's wrong with a 20 uF mylar? --scott martin Serious error. All shortcuts have disappeared. Screen. Mind. Both are blank. |
|
#16
|
|||
|
|||
|
What's wrong with a 20 uF mylar?
Do not have the space in this design and when loaded with 1k, it will give nice phase shift at LF. BTW, R7 should come before C3/C4, to give better LF response. Samuel |
|
#17
|
|||
|
|||
|
What's wrong with a 20 uF mylar?
Do not have the space in this design and when loaded with 1k, it will give nice phase shift at LF. BTW, R7 should come before C3/C4, to give better LF response. Samuel |
|
#18
|
|||
|
|||
|
"Samuel Groner" wrote in message m... What's wrong with a 20 uF mylar? Do not have the space in this design and when loaded with 1k, it will give nice phase shift at LF. Something not yet made clear: Is this intended as a microphone input, or line-level? I've been assuming the latter. Peace, Paul |
|
#19
|
|||
|
|||
|
Something not yet made clear: Is this intended as a microphone input, or
line-level? I've been assuming the latter. Line, with -6 dB or +4 dB gain. Samuel |
|
#20
|
|||
|
|||
|
"Samuel Groner" wrote in message om... Something not yet made clear: Is this intended as a microphone input, or line-level? I've been assuming the latter. Line, with -6 dB or +4 dB gain. Could you post the link to your schematic again? Thanks! Peace, Paul |
|
#21
|
|||
|
|||
|
On Tue, 08 Jun 2004 17:28:35 GMT, "Paul Stamler"
wrote: "Samuel Groner" wrote in message . com... Something not yet made clear: Is this intended as a microphone input, or line-level? I've been assuming the latter. Line, with -6 dB or +4 dB gain. Could you post the link to your schematic again? Thanks! Did the earlier post already roll off your news server? Here's the link: http://n.ethz.ch/student/groners/RAP...ed_Line_In.pdf Peace, Paul ----- http://mindspring.com/~benbradley |
|
#24
|
|||
|
|||
|
and it looks like you want to nake this thing
'perfect.' This is not too far from truth. ;-) Samuel |
|
#25
|
|||
|
|||
|
I don't know why he doesn't like them, but they have always worried me
because there are so many things to go wrong. KISS..... and if its self powered, less battery time It's more stuff in there, just to avoid AC coupling. If you have the space, good film capacitors are not all _that_ expensive. Me and the late Baz Porter, always seemed to have trouble with VLF hunting on the servo mechanism on wide gain range amps , albeit at a very low level, and I did some sims a year or so ago, in LTspice, yep they show up, sometimes in the millihertz region. Any idea where it comes from? Do not see a second pole which would be required to get 180 degree phase shift for a classic oscillation. Samuel |
|
#26
|
|||
|
|||
|
Well, how about a caveat? Positive feedback and half megohm
impedances can be a tricky combination. Are you talking about the DC-servo? Is there another way to implement it? Samuel |
|
#27
|
|||
|
|||
|
|
|
#28
|
|||
|
|||
|
No, I mean the common-mode bootstrap. Just makes me nervous.
You know the old saying "Oscillators don't; amplifiers do." So would you lower the value of the resistors? Removing the bootstrapping would degrade CMRR. Samuel |
|
#29
|
|||
|
|||
|
Yes, but before getting to that point, you can get all 0.1 percent
resistors for all four units you're going to make (perhaps some extras - you'll probably get price breaks with 100 qty of each value), measure then with a high-precision meter (such as a Fluke or HP DMM or RLC meter that displays five or more digits) (as you measure each one, write the value on a small piece of paper and tacky-glue it to the paper), then sort by value. Pick the pairs that are nearest each others' value and you'll have matched pairs to perhaps 0.01 percent. Do the same with capacitors (else CMRR will drop around their rollof frequencies), so that C1 == C2, C3 == C4, C5 == C6, and C7 == C8 to as accurate as you can measure and sort them. You may want to keep track of where each exact measured value of each resistor and capacitor goes for one amp, so you can put the same values into a simulation and see how closely it maches the actual unit. Measure CMRR and then see if you think it needs trimming. IMO perfectly matched resistors/caps do _not_ guarantee best CMRR - they result in opamp-limited CMRR. If we allow slight unbalances, we even get better than the opamp-CMRR. I had the pleasure to measure an input with 120 dB CMRR from 20 Hz to 40 kHz - it used a 5534, an OP275 and 2 trimmers. Or is this just BS what I'm writing here? I do not claim that we need 120 dB, but it won't hurt... Samuel |
|
#30
|
|||
|
|||
|
"Samuel Groner" wrote in message om... IMO perfectly matched resistors/caps do _not_ guarantee best CMRR - they result in opamp-limited CMRR. If we allow slight unbalances, we even get better than the opamp-CMRR. I had the pleasure to measure an input with 120 dB CMRR from 20 Hz to 40 kHz - it used a 5534, an OP275 and 2 trimmers. Or is this just BS what I'm writing here? I do not claim that we need 120 dB, but it won't hurt... It won't hurt on its own merits, but if the technique that gets you there causes stability problems or other issues, then you have to decide if the benefit is worth the trade-off. Not to mention the extra costs in money and board real estate. Remember the old saying: "An optimist thinks the glass is half full. A pessimist thinks the glass is half empty. And an engineer thinks the glass is twice as big as it needs to be in order to hold the specified amount of water." Peace, Paul |
|
#31
|
|||
|
|||
|
Hello Samuel,
A colleague of mine sent me a copy of your post. You, and members of the group, should be aware that the circuit you reference is covered by my US patent 5,568,561 and other issued and pending foreign patents. I have exclusively licensed the patent to THAT Corp, www.thatcorp.com, and they manufacture the InGenius(R) integrated circuit. Manufacture of this circuit for sale may create legal liability. Many of the parts in your circuit are unnecessary and will actually impair performance. I would be happy to advise you if your intent is a circuit for your own use ONLY. Bill Whitlock, president, Jensen Transformers, Inc. (from my home office) |
|
#32
|
|||
|
|||
|
A colleague of mine sent me a copy of your post. You, and members of the
group, should be aware that the circuit you reference is covered by my US patent 5,568,561 and other issued and pending foreign patents. I have exclusively licensed the patent to THAT Corp, www.thatcorp.com, and they manufacture the InGenius(R) integrated circuit. Manufacture of this circuit for sale may create legal liability. Many of the parts in your circuit are unnecessary and will actually impair performance. I would be happy to advise you if your intent is a circuit for your own use ONLY. Nothing else was intended... Which parts are unnecessary? The DC-servo probably, as the opamp will have low offset in this config. The first bootstrapped resistors are overkill as well. Samuel |
|
#33
|
|||
|
|||
|
The schematic shown on the datasheet
http://www.thatcorp.com/datashts/1200data.pdf has very high performance. It typically produces 90 dB CMRR at 60 Hz with a perfectly balanced driving source, and this degrades to only 70 dB with a source that's unbalanced by 600 ohms (this can NOT be said of ordinary diff-amps like the SSM2141). Using the recently adopted test standard IEC 60268-3, where 10-ohm imbalances are inserted into the driving source to simulate real-world equipment, the CMRR of the THAT parts will remain essentially unchanged. Values of 10 k-ohms for all internal resistors will work well. What's more important to the bootstrap performance is the the gain of the "+1" stages be extremely close to 1.0000. This requires very high open-loop gain and high gain-bandwidth in the op-amps. The implementation of my circuit by THAT Corporation does an excellent job in this regard. Production of the chip has been delayed by ESD (electrostatic discharge) protection issues, but we expect an announcement of full production availability at the AES convention in October. My original paper on this circuit is available at the Jensen website http://www.jensen-transformers.com under "white papers." Bill Whitlock "Samuel Groner" wrote in message om... A colleague of mine sent me a copy of your post. You, and members of the group, should be aware that the circuit you reference is covered by my US patent 5,568,561 and other issued and pending foreign patents. I have exclusively licensed the patent to THAT Corp, www.thatcorp.com, and they manufacture the InGenius(R) integrated circuit. Manufacture of this circuit for sale may create legal liability. Many of the parts in your circuit are unnecessary and will actually impair performance. I would be happy to advise you if your intent is a circuit for your own use ONLY. Nothing else was intended... Which parts are unnecessary? The DC-servo probably, as the opamp will have low offset in this config. The first bootstrapped resistors are overkill as well. Samuel |
|
#34
|
|||
|
|||
|
The schematic shown on the datasheet
http://www.thatcorp.com/datashts/1200data.pdf has very high performance. It typically produces 90 dB CMRR at 60 Hz with a perfectly balanced driving source, and this degrades to only 70 dB with a source that's unbalanced by 600 ohms (this can NOT be said of ordinary diff-amps like the SSM2141). Using the recently adopted test standard IEC 60268-3, where 10-ohm imbalances are inserted into the driving source to simulate real-world equipment, the CMRR of the THAT parts will remain essentially unchanged. Values of 10 k-ohms for all internal resistors will work well. What's more important to the bootstrap performance is the the gain of the "+1" stages be extremely close to 1.0000. This requires very high open-loop gain and high gain-bandwidth in the op-amps. The implementation of my circuit by THAT Corporation does an excellent job in this regard. Production of the chip has been delayed by ESD (electrostatic discharge) protection issues, but we expect an announcement of full production availability at the AES convention in October. My original paper on this circuit is available at the Jensen website http://www.jensen-transformers.com under "white papers." Bill Whitlock "Samuel Groner" wrote in message om... A colleague of mine sent me a copy of your post. You, and members of the group, should be aware that the circuit you reference is covered by my US patent 5,568,561 and other issued and pending foreign patents. I have exclusively licensed the patent to THAT Corp, www.thatcorp.com, and they manufacture the InGenius(R) integrated circuit. Manufacture of this circuit for sale may create legal liability. Many of the parts in your circuit are unnecessary and will actually impair performance. I would be happy to advise you if your intent is a circuit for your own use ONLY. Nothing else was intended... Which parts are unnecessary? The DC-servo probably, as the opamp will have low offset in this config. The first bootstrapped resistors are overkill as well. Samuel |
|
#35
|
|||
|
|||
|
Thanks for all the comments! They are very helpful indeed. I'll solder
a prototyp and see what happens. Samuel |
| Reply |
| Thread Tools | |
| Display Modes | |
|
|
Similar Threads
|
||||
| Thread | Forum | |||
| ESX Q1204 Amp Balanced Cable Connections? | Car Audio | |||
| balanced or unbalanced patch cables? | Pro Audio | |||
| FS: SOUNDSTREAM CLOSEOUTS AND MORE!! | Car Audio | |||
| balanced power | Pro Audio | |||
| balanced power | Pro Audio | |||
Linear Mode
