bcarso
Well-known member
In exchange for other problems you could bootstrap out protection diode capacitance. Seems to me Pease mentions this being used in some of the National CMOS opamps. Or maybe I dreamt it.
Mic Preamp Schematic Collection
- Thread starter Samuel Groner
- Start date
Help Support GroupDIY Audio Forum:
[quote author="bcarso"]In exchange for other problems you could bootstrap out protection diode capacitance. Seems to me Pease mentions this being used in some of the National CMOS opamps. Or maybe I dreamt it.[/quote]
I once bootstrapped the shielded cable on a mic pre gain pot inside a console with an almost 2' long run, to improve HF CMRR and stability. Only maybe 30pF a foot of capacitance but not a great place for it in that circuit.
I suspect the small geometry diodes are pretty similar and have never experienced a problem from them in that application, but concede I never looked for one there. Even a high impedance microphone is still only 600 ohms.
JR
I once bootstrapped the shielded cable on a mic pre gain pot inside a console with an almost 2' long run, to improve HF CMRR and stability. Only maybe 30pF a foot of capacitance but not a great place for it in that circuit.
I suspect the small geometry diodes are pretty similar and have never experienced a problem from them in that application, but concede I never looked for one there. Even a high impedance microphone is still only 600 ohms.
JR
Samuel Groner
Well-known member
This protection thing is more difficult than I expected. I already half-way decided to change to 1N400x as I realised that the leakage of this diode--and perhaps even that of the 1N914B--would decrease the DC precision of the frontend of design B. So I guess we should use a FDH300 here... :?
Samuel
Samuel
G
Guest
Guest
[quote author="JohnRoberts"]
I've seen clamps make with 2 transistors connected with bases floating on both and connected back to back (Cs to Es) with one end to ground and the other to the input line. This would clamp when the be junction zenered + the collector diode approx. +/- 7.2 V or so. Another approach is zener inside diode bridge connection. A few diodes in series will reduce capacitance somewhat.
[/quote]
Sure, using both junctions seems to be better, but I don't like a floating base, probably some high enough resistors between bases and emitters would help.
I've seen clamps make with 2 transistors connected with bases floating on both and connected back to back (Cs to Es) with one end to ground and the other to the input line. This would clamp when the be junction zenered + the collector diode approx. +/- 7.2 V or so. Another approach is zener inside diode bridge connection. A few diodes in series will reduce capacitance somewhat.
[/quote]
Sure, using both junctions seems to be better, but I don't like a floating base, probably some high enough resistors between bases and emitters would help.
jdbakker
Well-known member
You don't care about bias current per se as much as you do about differences in bias current, right ? From what I've seen modern diodes from the same batch/reel are pretty well matched (just like modern BJTs). Remains a possible temperature difference; that is easily cured by mounting the diodes close together, with a blob of heat sink goo / thermally enhanced potting compound for good measure. It's not like the diodes have any self-heating under normal use, so as long as you can eliminate external temperature gradients you should be fine.
It may be that over-current events (as caused by shorting the input to ground) affect the diode's reverse current permanently.
So why not use a DC servo instead of those OPA627s ? That's what I'm planning to do, with a trimmed OPA134.
JDB.
[what's good enough for Deane Jensen is good enough for me]
It may be that over-current events (as caused by shorting the input to ground) affect the diode's reverse current permanently.
So why not use a DC servo instead of those OPA627s ? That's what I'm planning to do, with a trimmed OPA134.
JDB.
[what's good enough for Deane Jensen is good enough for me]
[quote author="mediatechnology"]Sam et al;
Hebert and Thomas of THAT Corporation published an AES Paper on the peak current requirements of mic preamp protection diodes. They measured 2-3A peaks IIRC with 47 uF input caps.
I also recently corresponded with Walt Jung who did some of the original ADI SSM2017 protection work. 1N4148s will pop here and 1N400X-series most likely will too.
I've questioned a lot of engineers about this gain switch/pot thing and the consensus so far has been a precise stepped switch with a post-pre gain trim pot. Some prefered having non-linear dB steps with more precision towards the top.
I haven't found a rev-log worth a damn. The 10K RV4s that Digi-Key have are good but still a little touchy. Can't readily get 2.5% (taper) rev log.
There are some charts here on the last tab of the 1510/1512 gain calculator xls I did recently: www.ka-electronics.com/THAT1510/1510_1512_Gain_rev1.xls
Sam: I've sent you a PM with a pdf. Nice work.[/quote]
I'm not sure how much of a "problem" this is. Over a couple decades I've been responsible for plenty of phantom powered mic preamps with no major issues related to clamp diodes. IIRC the majority of my channels were protected by small geometry diodes (914/4148). I believe these can transiently handle 1A and the failure mechanism is heat so the fault must persist more than transiently. FWIW the majority of my mic channels used 22 UF caps which would be only 1/2 the potential charge of a 47 to dissipate. Also where the clamps are located relative to several ohms of build out resistance can also make a difference in robustness.
I am not calling anybody liars. The results are what they are, but I suspect they are performing pretty extreme tests with hard, persistent shunts to ground. By all means if using very large front end capacitors this may be a valid design consideration.
In my dotage I'm becoming more a fan of the "best capacitor is no capacitor" school of design, or as BCARSO wrote "there's no capacitor like no capacitor". I was attracted to this site by the thread about DC coupled mic preamps as I have been scratching on such a project half heartedly for several years. I have two or three general design directions and may publish something here (of course I also need to find the time to enter schematics and finish details, not currently a high priority). I bought some high voltage caps a few years ago to accomodate the unusual rail voltages needed, but they're still sitting untouched back in my lab. I got distracted pretty full time on my current project (I should be writing code instead of this right now).
I fear there is only limited commercial value in obscure improved mic preamps. There’s a difference between actually making something better and being able to convince people to spend hard earned money for one, especially when the improvements are subtle (and not gold plated, etc).
JR
PS: For background I managed Peavey's mixer engineering group for several years so was responsible for many such mic channels. With the number of channels sold under my watch if there was a serious failure problem I would have been in the middle of it. Biggest issue I recall was other companies wireless mic receivers being smoked by phantom power. Not my fault mon, if you’re plugging into a mic input, there’s a very good chance you’ll see phantom voltage sometime in your product life.
Hebert and Thomas of THAT Corporation published an AES Paper on the peak current requirements of mic preamp protection diodes. They measured 2-3A peaks IIRC with 47 uF input caps.
I also recently corresponded with Walt Jung who did some of the original ADI SSM2017 protection work. 1N4148s will pop here and 1N400X-series most likely will too.
I've questioned a lot of engineers about this gain switch/pot thing and the consensus so far has been a precise stepped switch with a post-pre gain trim pot. Some prefered having non-linear dB steps with more precision towards the top.
I haven't found a rev-log worth a damn. The 10K RV4s that Digi-Key have are good but still a little touchy. Can't readily get 2.5% (taper) rev log.
There are some charts here on the last tab of the 1510/1512 gain calculator xls I did recently: www.ka-electronics.com/THAT1510/1510_1512_Gain_rev1.xls
Sam: I've sent you a PM with a pdf. Nice work.[/quote]
I'm not sure how much of a "problem" this is. Over a couple decades I've been responsible for plenty of phantom powered mic preamps with no major issues related to clamp diodes. IIRC the majority of my channels were protected by small geometry diodes (914/4148). I believe these can transiently handle 1A and the failure mechanism is heat so the fault must persist more than transiently. FWIW the majority of my mic channels used 22 UF caps which would be only 1/2 the potential charge of a 47 to dissipate. Also where the clamps are located relative to several ohms of build out resistance can also make a difference in robustness.
I am not calling anybody liars. The results are what they are, but I suspect they are performing pretty extreme tests with hard, persistent shunts to ground. By all means if using very large front end capacitors this may be a valid design consideration.
In my dotage I'm becoming more a fan of the "best capacitor is no capacitor" school of design, or as BCARSO wrote "there's no capacitor like no capacitor". I was attracted to this site by the thread about DC coupled mic preamps as I have been scratching on such a project half heartedly for several years. I have two or three general design directions and may publish something here (of course I also need to find the time to enter schematics and finish details, not currently a high priority). I bought some high voltage caps a few years ago to accomodate the unusual rail voltages needed, but they're still sitting untouched back in my lab. I got distracted pretty full time on my current project (I should be writing code instead of this right now).
I fear there is only limited commercial value in obscure improved mic preamps. There’s a difference between actually making something better and being able to convince people to spend hard earned money for one, especially when the improvements are subtle (and not gold plated, etc).
JR
PS: For background I managed Peavey's mixer engineering group for several years so was responsible for many such mic channels. With the number of channels sold under my watch if there was a serious failure problem I would have been in the middle of it. Biggest issue I recall was other companies wireless mic receivers being smoked by phantom power. Not my fault mon, if you’re plugging into a mic input, there’s a very good chance you’ll see phantom voltage sometime in your product life.
G
Guest
Guest
[quote author="JohnRoberts"]
Not my fault mon, if you’re plugging into a mic input, there’s a very good chance you’ll see phantom voltage sometime in your product life.[/quote]
Especially during a live performance when everybody are under stress.
[quote author="mediatechnology"]
I've questioned a lot of engineers about this gain switch/pot thing and the consensus so far has been a precise stepped switch with a post-pre gain trim pot. [/quote]
I am both developer and end user and second that.
Not my fault mon, if you’re plugging into a mic input, there’s a very good chance you’ll see phantom voltage sometime in your product life.[/quote]
Especially during a live performance when everybody are under stress.
[quote author="mediatechnology"]
I've questioned a lot of engineers about this gain switch/pot thing and the consensus so far has been a precise stepped switch with a post-pre gain trim pot. [/quote]
I am both developer and end user and second that.
jdbakker
Well-known member
[quote author="mediatechnology"]Hebert and Thomas of THAT Corporation published an AES Paper on the peak current requirements of mic preamp protection diodes. They measured 2-3A peaks IIRC with 47 uF input caps.[/quote]
Did they use current limiting resistors, like they show in the THAT1510/1512 data sheet ? Did they use a CMC or other inductors in the input signal path (which may help to flatten peaks) ?
[quote author="mediatechnology"]I also recently corresponded with Walt Jung who did some of the original ADI SSM2017 protection work. 1N4148s will pop here and 1N400X-series most likely will too.[/quote]
(emphasis mine)
With all due respect to Walt Jung, I find that hard to believe. 1N400x diodes are rated for 30A 8.33ms half-sinewave surges, and I've often seen them cope with such. Picture the turn-on current in a 1N400x in a bridge rectifier between a 24VAC transformer and a 1000+uF capacitor; there is no way that a 47uF cap holds anywhere near that amount of energy.
[quote author="mediatechnology"]I've questioned a lot of engineers about this gain switch/pot thing and the consensus so far has been a precise stepped switch with a post-pre gain trim pot. Some prefered having non-linear dB steps with more precision towards the top.[/quote]
Just to be clear: do you mean they want smaller steps in the high-gain range ? If so, any idea why ?
Thanks,
JDB.
Did they use current limiting resistors, like they show in the THAT1510/1512 data sheet ? Did they use a CMC or other inductors in the input signal path (which may help to flatten peaks) ?
[quote author="mediatechnology"]I also recently corresponded with Walt Jung who did some of the original ADI SSM2017 protection work. 1N4148s will pop here and 1N400X-series most likely will too.[/quote]
(emphasis mine)
With all due respect to Walt Jung, I find that hard to believe. 1N400x diodes are rated for 30A 8.33ms half-sinewave surges, and I've often seen them cope with such. Picture the turn-on current in a 1N400x in a bridge rectifier between a 24VAC transformer and a 1000+uF capacitor; there is no way that a 47uF cap holds anywhere near that amount of energy.
[quote author="mediatechnology"]I've questioned a lot of engineers about this gain switch/pot thing and the consensus so far has been a precise stepped switch with a post-pre gain trim pot. Some prefered having non-linear dB steps with more precision towards the top.[/quote]
Just to be clear: do you mean they want smaller steps in the high-gain range ? If so, any idea why ?
Thanks,
JDB.
[quote author="mediatechnology"]I've questioned a lot of engineers about this gain switch/pot thing and the consensus so far has been a precise stepped switch with a post-pre gain trim pot. Some prefered having non-linear dB steps with more precision towards the top.[/quote]
I believe I may have addressed this already in another thread, but when I was doing this for production designs we would tool special taper pots that were not just the usual reverse audio logarithmic curve, but required multiple (3 IIRC) screening operations overlaid to get a good adjustability from hop off down in the low ohm region with likewise relatively high resistance for adjustability in the low gain end. The extra steps made the part cost more but this was a considerable perceived value issue with customers especially non technical types who don't want to hear an explantion for why it's hard to do. IIRC we made a few customs in the 25k total resistance range with good adjustability at loe ohms end.
Users don't usually give very good specific feedback when asked but what they seem to really like is good output from a SM-58 with gain pot at 12 o'clock, no twitchiness around max gain, and ability to handle hot dynmaic mics. :grin:
Switches are very nice for precision and tracking between channels. Also use of quality resistors on a good low contact resistance switch my be preferable to typical potentiometer resistive substrate, but that is another subtle or esoteric benefit that may look better on paper than measure or express sonically.
For mass production it's basically a cost issue. I vaguely recall ALPS developing a rotary volume control years ago that actually consisted of stepped resistances. My recollection fails me regarding the actual technology used to fab the Rs (probably screened). Nowadays with 01x02 SMT resistors I guess a true discrete stepped switch could be made in the footprint of a small pot, but won't hold my breath waiting for that to happen.
JR
I believe I may have addressed this already in another thread, but when I was doing this for production designs we would tool special taper pots that were not just the usual reverse audio logarithmic curve, but required multiple (3 IIRC) screening operations overlaid to get a good adjustability from hop off down in the low ohm region with likewise relatively high resistance for adjustability in the low gain end. The extra steps made the part cost more but this was a considerable perceived value issue with customers especially non technical types who don't want to hear an explantion for why it's hard to do. IIRC we made a few customs in the 25k total resistance range with good adjustability at loe ohms end.
Users don't usually give very good specific feedback when asked but what they seem to really like is good output from a SM-58 with gain pot at 12 o'clock, no twitchiness around max gain, and ability to handle hot dynmaic mics. :grin:
Switches are very nice for precision and tracking between channels. Also use of quality resistors on a good low contact resistance switch my be preferable to typical potentiometer resistive substrate, but that is another subtle or esoteric benefit that may look better on paper than measure or express sonically.
For mass production it's basically a cost issue. I vaguely recall ALPS developing a rotary volume control years ago that actually consisted of stepped resistances. My recollection fails me regarding the actual technology used to fab the Rs (probably screened). Nowadays with 01x02 SMT resistors I guess a true discrete stepped switch could be made in the footprint of a small pot, but won't hold my breath waiting for that to happen.
JR
[quote author="mediatechnology"]jdb:
JohnR:
But other consoles required me to change input devices a lot due to noise degradation. First you'd change the LM394 and sometimes that would nail it, other times you'd have to change the protection devices and the LM394. And sometimes replace the caps because they got leaky for completely different reasons. The Harrisons were one of the worst. Seems like the fragility of the input device has a lot to do with it. Didn't somebody use power transistors in their mic preamp for low noise performance? Doubt they would have a problem with stored-charge punch-through as well as some discretes. (Isn't the LM394 a bunch of small-geometry devices in parallel making it more fragile?)
I'm going with the no cap is best approach at least in active phantom inputs.
But having said that, Sam's designs here as well as other gain blocks could have their rails flown as well. The no-capacitor thingy I'm working on could be adapted to a lot of active mic pre circuits including these.[/quote]
The input noise failure mode was describe in the old M&F "Low Noise Design" text as a consequence of low noise devices being allowed to zener (b-e junction reverse biased approx 7V). It's pretty common to see small diode clamps across low noise b-e junctions to prevent that from occurring. Perhaps this isn't widely known but it seems pretty widely practiced. Two of the three designs that started this thread have them, and the MAT02 array has the diodes built in. Note: the LM394 metal can schematic shows clamp diodes but not the plastic package? Not sure if that's true.
In general the use of lots of parallel structures is no more or less robust. I recall a series of power devices (Ring Emitter Transistors) that were essentially a bunch of devices in parallel.
Yup, no magic about flying the pre up to phantom, but still several ways to scratch that itch... I'm leaning toward just flying the input pair, but that's only one of several I'm kicking around.
I would be surprised if somebody hasn't already flown a pre and A/D up there then passed the digital output via optical... Sounds obvious for hi performance digital desks... but some of those folks lack imagination (or just don't share our distaste for capacitors). :roll:
JR
JohnR:
But other consoles required me to change input devices a lot due to noise degradation. First you'd change the LM394 and sometimes that would nail it, other times you'd have to change the protection devices and the LM394. And sometimes replace the caps because they got leaky for completely different reasons. The Harrisons were one of the worst. Seems like the fragility of the input device has a lot to do with it. Didn't somebody use power transistors in their mic preamp for low noise performance? Doubt they would have a problem with stored-charge punch-through as well as some discretes. (Isn't the LM394 a bunch of small-geometry devices in parallel making it more fragile?)
I'm going with the no cap is best approach at least in active phantom inputs.
But having said that, Sam's designs here as well as other gain blocks could have their rails flown as well. The no-capacitor thingy I'm working on could be adapted to a lot of active mic pre circuits including these.[/quote]
The input noise failure mode was describe in the old M&F "Low Noise Design" text as a consequence of low noise devices being allowed to zener (b-e junction reverse biased approx 7V). It's pretty common to see small diode clamps across low noise b-e junctions to prevent that from occurring. Perhaps this isn't widely known but it seems pretty widely practiced. Two of the three designs that started this thread have them, and the MAT02 array has the diodes built in. Note: the LM394 metal can schematic shows clamp diodes but not the plastic package? Not sure if that's true.
In general the use of lots of parallel structures is no more or less robust. I recall a series of power devices (Ring Emitter Transistors) that were essentially a bunch of devices in parallel.
Yup, no magic about flying the pre up to phantom, but still several ways to scratch that itch... I'm leaning toward just flying the input pair, but that's only one of several I'm kicking around.
I would be surprised if somebody hasn't already flown a pre and A/D up there then passed the digital output via optical... Sounds obvious for hi performance digital desks... but some of those folks lack imagination (or just don't share our distaste for capacitors). :roll:
JR
mikep
Well-known member
[quote author="mediatechnology"]And Danish Audio Connect DACT can do them as well in custom with SMT Rs.[/quote]
I recently re-resistored some DACT smd 23-pos switches for a custom taper. not much fun but it worked out great.
I recently re-resistored some DACT smd 23-pos switches for a custom taper. not much fun but it worked out great.
jdbakker
Well-known member
[quote author="JohnRoberts"]I would be surprised if somebody hasn't already flown a pre and A/D up there then passed the digital output via optical...[/quote]
I've looked at doing just that. The main stumbling block (as far as I could see) is low-jitter master clock distribution. The best I could think of was transformer coupling of the 24.576MHz MCLK into a low-noise limiting differential amplifier, but even that will easily double the jitter of a good low-noise clock source over impedance-matched DC coupling.
I think Wayne's setup with the servo and the LDRs is very interesting, but I'm still working on a brute-force no-cap mic pre using high voltage DOAs.
JDB
[...after I'm done with Mk1 of the HD recorder]
I've looked at doing just that. The main stumbling block (as far as I could see) is low-jitter master clock distribution. The best I could think of was transformer coupling of the 24.576MHz MCLK into a low-noise limiting differential amplifier, but even that will easily double the jitter of a good low-noise clock source over impedance-matched DC coupling.
I think Wayne's setup with the servo and the LDRs is very interesting, but I'm still working on a brute-force no-cap mic pre using high voltage DOAs.
JDB
[...after I'm done with Mk1 of the HD recorder]
[quote author="jdbakker"][quote author="JohnRoberts"]I would be surprised if somebody hasn't already flown a pre and A/D up there then passed the digital output via optical...[/quote]
I've looked at doing just that. The main stumbling block (as far as I could see) is low-jitter master clock distribution. The best I could think of was transformer coupling of the 24.576MHz MCLK into a low-noise limiting differential amplifier, but even that will easily double the jitter of a good low-noise clock source over impedance-matched DC coupling.
I think Wayne's setup with the servo and the LDRs is very interesting, but I'm still working on a brute-force no-cap mic pre using high voltage DOAs.
JDB
[...after I'm done with Mk1 of the HD recorder][/quote]
While in general digital audio should be able to be passed asynchronously it seems a digital clock could be cap coupled over a wide voltage range without too much difficulty. You might loose sync while rails are ramping around but the last codec I worked with had a PLL that would just hunt and recover (I think... I never actually tried this with one).
Passing non square wave digital results which may have some changing DC content could probably be managed with long time constants. I believe there are also digital transfer schemes that don’t have any changing dc content.
One of my various scratchings on the subject is a discrete 3 input differential amp to level shift down from 48V to 0V and convert differential to SE at the same time. I'm almost tempted to finish that approach just to prove out that oddball diff amp but my latest thinking is to float up only the parts than need to be up there so the bulk of the circuitry would remain down in +/- 15V world...
It seems there may be more different ways to accomplish this task than customers to buy one..
JR
I've looked at doing just that. The main stumbling block (as far as I could see) is low-jitter master clock distribution. The best I could think of was transformer coupling of the 24.576MHz MCLK into a low-noise limiting differential amplifier, but even that will easily double the jitter of a good low-noise clock source over impedance-matched DC coupling.
I think Wayne's setup with the servo and the LDRs is very interesting, but I'm still working on a brute-force no-cap mic pre using high voltage DOAs.
JDB
[...after I'm done with Mk1 of the HD recorder][/quote]
While in general digital audio should be able to be passed asynchronously it seems a digital clock could be cap coupled over a wide voltage range without too much difficulty. You might loose sync while rails are ramping around but the last codec I worked with had a PLL that would just hunt and recover (I think... I never actually tried this with one).
Passing non square wave digital results which may have some changing DC content could probably be managed with long time constants. I believe there are also digital transfer schemes that don’t have any changing dc content.
One of my various scratchings on the subject is a discrete 3 input differential amp to level shift down from 48V to 0V and convert differential to SE at the same time. I'm almost tempted to finish that approach just to prove out that oddball diff amp but my latest thinking is to float up only the parts than need to be up there so the bulk of the circuitry would remain down in +/- 15V world...
It seems there may be more different ways to accomplish this task than customers to buy one..
JR
jdbakker
Well-known member
[quote author="JohnRoberts"]While in general digital audio should be able to be passed asynchronously it seems a digital clock could be cap coupled over a wide voltage range without too much difficulty. You might loose sync while rails are ramping around but the last codec I worked with had a PLL that would just hunt and recover (I think... I never actually tried this with one).[/quote]
This approach works well, if you/your customers are content with 30..100ps clock jitter (and many are, judging by a lot of equipment out there). I'm aiming for 2..3ps jitter at the ADC MCLK input.
[So why do I think I need such tight jitter control ? To keep the impact of jitter-induced sample point uncertainty below the converter's THD+N floor. I did the math some weeks back, but I can't find my notes ATM. If you like I can search some more in the morning]
[quote author="JohnRoberts"]Passing non square wave digital results which may have some changing DC content could probably be managed with long time constants. I believe there are also digital transfer schemes that don’t have any changing dc content.[/quote]
Plenty don't, such as Manchester encoding. The simple way would be to hook an AES transmitter to the floating ADC and cap/transformer-couple the signal out.
Expanding on this, you could give each floating converter its own free-running low noise crystal clock, and use an ASRC (asynchronous sample rate converter) like the TI SRC4392 on the backplane to re-sync all bitstreams to one global master clock. This will keep jitter almost as low as the direct-coupled case, at the cost of possible nasty intermodulation by the EMI generated by all those clocks.
[quote author="JohnRoberts"]One of my various scratchings on the subject is a discrete 3 input differential amp to level shift down from 48V to 0V and convert differential to SE at the same time.[/quote]
Sounds interesting ! Have you any schematics/rough sketches ?
My plan (read: partially-finished breadboard) consists of a few DOAs designed to run from +60/-15 supplies, dropped into a standard differential Cohen mic pre (like Samuel's Design A, with all '5532s replaced by DOAs), with a few floating servos to preserve DC balance. Plan B would involve a diff-in/diff-out DOA, but I'm not sure if imperfect matching between each output's compensation network might not make an oscillator out of such a beast.
JDB.
This approach works well, if you/your customers are content with 30..100ps clock jitter (and many are, judging by a lot of equipment out there). I'm aiming for 2..3ps jitter at the ADC MCLK input.
[So why do I think I need such tight jitter control ? To keep the impact of jitter-induced sample point uncertainty below the converter's THD+N floor. I did the math some weeks back, but I can't find my notes ATM. If you like I can search some more in the morning]
[quote author="JohnRoberts"]Passing non square wave digital results which may have some changing DC content could probably be managed with long time constants. I believe there are also digital transfer schemes that don’t have any changing dc content.[/quote]
Plenty don't, such as Manchester encoding. The simple way would be to hook an AES transmitter to the floating ADC and cap/transformer-couple the signal out.
Expanding on this, you could give each floating converter its own free-running low noise crystal clock, and use an ASRC (asynchronous sample rate converter) like the TI SRC4392 on the backplane to re-sync all bitstreams to one global master clock. This will keep jitter almost as low as the direct-coupled case, at the cost of possible nasty intermodulation by the EMI generated by all those clocks.
[quote author="JohnRoberts"]One of my various scratchings on the subject is a discrete 3 input differential amp to level shift down from 48V to 0V and convert differential to SE at the same time.[/quote]
Sounds interesting ! Have you any schematics/rough sketches ?
My plan (read: partially-finished breadboard) consists of a few DOAs designed to run from +60/-15 supplies, dropped into a standard differential Cohen mic pre (like Samuel's Design A, with all '5532s replaced by DOAs), with a few floating servos to preserve DC balance. Plan B would involve a diff-in/diff-out DOA, but I'm not sure if imperfect matching between each output's compensation network might not make an oscillator out of such a beast.
JDB.
[quote author="jdbakker"][quote author="JohnRoberts"]While in general digital audio should be able to be passed asynchronously it seems a digital clock could be cap coupled over a wide voltage range without too much difficulty. You might loose sync while rails are ramping around but the last codec I worked with had a PLL that would just hunt and recover (I think... I never actually tried this with one).[/quote]
This approach works well, if you/your customers are content with 30..100ps clock jitter (and many are, judging by a lot of equipment out there). I'm aiming for 2..3ps jitter at the ADC MCLK input.
[So why do I think I need such tight jitter control ? To keep the impact of jitter-induced sample point uncertainty below the converter's THD+N floor. I did the math some weeks back, but I can't find my notes ATM. If you like I can search some more in the morning]
[/quote]
Does cap coupling the clock cause actual jitter (an uncertainty from one clock to the next) or some fixed timing error due to C, perhaps interacting with device input C? This is beyond my understanding of things digital I guess but I don't see how a constant timing error from a passive component causes noise, variable timing errors sure.
[quote author="jdbakker"]
[quote author="JohnRoberts"]One of my various scratchings on the subject is a discrete 3 input differential amp to level shift down from 48V to 0V and convert differential to SE at the same time.[/quote]
Sounds interesting ! Have you any schematics/rough sketches ?
[/quote]
Now that I think about it (this was a few years ago) it doesn't do the SE conversion but level shifts two differential signals centered at some dc voltage from 0 to 48V and references them down to two outputs at nominal 0V all DC coupled while ignoring common mode content.
Imagine a discrete opamp with 3 input transistors instead of 2 in the long tail pair, perhaps I should call it a long tail trio? :grin: two inputs look like conventional - inputs driving two output followers negative, The third input acts like a common + input for both outputs. A little odd but uses less parts in an all discrete design (I think this is in my version 2 or so that has first stage outputs centered up at phantom voltage.
The third input is fed by equal value resistors from the two first gain stage outputs. The differential signal cancels out so what remains is CM and DC level. A third resistor from the junction of these two resistors connected to the ground reference works like the simple 2 input diff amp for referencing between dc voltage levels. To wit - input 1 R input and R feedback, - input 2 R input and R feedback, + input common fed by 2R from each gain stage with R to ground reference.
[quote author="jdbakker"]
My plan (read: partially-finished breadboard) consists of a few DOAs designed to run from +60/-15 supplies, dropped into a standard differential Cohen mic pre (like Samuel's Design A, with all '5532s replaced by DOAs), with a few floating servos to preserve DC balance. Plan B would involve a diff-in/diff-out DOA, but I'm not sure if imperfect matching between each output's compensation network might not make an oscillator out of such a beast.
JDB.[/quote]
I'm still not sure why Cohen gets credit for that topology. The much earlier version that I first saw (Transamp/Buff) doesn't pass DC offsets to the output so may be preferable for your app. (Cap couple between low noise input transistors collector and opamp - input. You also need to provide DC feedback path around opamp). Since this capacitor is inside the overall negative feedback loop it is not an audio quality concern.
JR
PS: When I was momentarily thinking of possibly selling this I worked on an all discrete approach as being more "sellable" to a public that needs some hooks to become true believers. To just make it work well I plan to use opamps with a few discretes as appropriate.
This approach works well, if you/your customers are content with 30..100ps clock jitter (and many are, judging by a lot of equipment out there). I'm aiming for 2..3ps jitter at the ADC MCLK input.
[So why do I think I need such tight jitter control ? To keep the impact of jitter-induced sample point uncertainty below the converter's THD+N floor. I did the math some weeks back, but I can't find my notes ATM. If you like I can search some more in the morning]
[/quote]
Does cap coupling the clock cause actual jitter (an uncertainty from one clock to the next) or some fixed timing error due to C, perhaps interacting with device input C? This is beyond my understanding of things digital I guess but I don't see how a constant timing error from a passive component causes noise, variable timing errors sure.
[quote author="jdbakker"]
[quote author="JohnRoberts"]One of my various scratchings on the subject is a discrete 3 input differential amp to level shift down from 48V to 0V and convert differential to SE at the same time.[/quote]
Sounds interesting ! Have you any schematics/rough sketches ?
[/quote]
Now that I think about it (this was a few years ago) it doesn't do the SE conversion but level shifts two differential signals centered at some dc voltage from 0 to 48V and references them down to two outputs at nominal 0V all DC coupled while ignoring common mode content.
Imagine a discrete opamp with 3 input transistors instead of 2 in the long tail pair, perhaps I should call it a long tail trio? :grin: two inputs look like conventional - inputs driving two output followers negative, The third input acts like a common + input for both outputs. A little odd but uses less parts in an all discrete design (I think this is in my version 2 or so that has first stage outputs centered up at phantom voltage.
The third input is fed by equal value resistors from the two first gain stage outputs. The differential signal cancels out so what remains is CM and DC level. A third resistor from the junction of these two resistors connected to the ground reference works like the simple 2 input diff amp for referencing between dc voltage levels. To wit - input 1 R input and R feedback, - input 2 R input and R feedback, + input common fed by 2R from each gain stage with R to ground reference.
[quote author="jdbakker"]
My plan (read: partially-finished breadboard) consists of a few DOAs designed to run from +60/-15 supplies, dropped into a standard differential Cohen mic pre (like Samuel's Design A, with all '5532s replaced by DOAs), with a few floating servos to preserve DC balance. Plan B would involve a diff-in/diff-out DOA, but I'm not sure if imperfect matching between each output's compensation network might not make an oscillator out of such a beast.
JDB.[/quote]
I'm still not sure why Cohen gets credit for that topology. The much earlier version that I first saw (Transamp/Buff) doesn't pass DC offsets to the output so may be preferable for your app. (Cap couple between low noise input transistors collector and opamp - input. You also need to provide DC feedback path around opamp). Since this capacitor is inside the overall negative feedback loop it is not an audio quality concern.
JR
PS: When I was momentarily thinking of possibly selling this I worked on an all discrete approach as being more "sellable" to a public that needs some hooks to become true believers. To just make it work well I plan to use opamps with a few discretes as appropriate.
Samuel Groner
Well-known member
Basically: if I can skip a servo, I'll do. I've thought quite a bit about adding servos to this topology and I didn't come up with a simple yet convincing solution. The problem is the injection point--the obvious place after the input coupling capacitor is somewhat unhappy as the coupling capacitor will interact with the servo action, potentially producing some low-frequency wobbling. Using the inverting input will degrade CMRR (though I skipped the detailed analysis for a quantitative estimate). Then there's the way of providing a pair of servo-controlled current sources to the inverting input. That's cool because at the same time you can eliminate the common mode offset, improving headroom--but it is relatively complex and I'm unsure how to estimate the noise contribution of the current sources (but there might be a chance that there is another design in preparation for this schematic collection using this approach...). Then again we could use the collector resistors of the input pair (or the noninverting input of the according opamp), but this might increase distortion a bit due to the resulting collector current imbalance (surely a subtle effect though).
But perhaps I'm missing something? If you've got another solution, please let us know!
If you are interested, I've a presumably nice 80 V DOA in the pipline. For sure I would be curious to see your fully differential opamp!
Likely there are almost as many opinions on this as are engineers. Personally (being 90% user and only 10% developer) I don't have much or perhaps even any use for trims, that's why there aren't any.
Samuel
Samuel Groner
Well-known member
I edited the first post to include design D and E.
Samuel
Samuel
mikep
Well-known member
[quote author="jdbakker"]
I've looked at doing just that. The main stumbling block (as far as I could see) is low-jitter master clock distribution. [/quote]
neumann tried to tackle this issue with their digital microphone, the D-01 (IIRC). I remember reading a paper about it a few years ago. AES-42 is the standard they came up with, might be worth a look.
mike
I've looked at doing just that. The main stumbling block (as far as I could see) is low-jitter master clock distribution. [/quote]
neumann tried to tackle this issue with their digital microphone, the D-01 (IIRC). I remember reading a paper about it a few years ago. AES-42 is the standard they came up with, might be worth a look.
mike
jdbakker
Well-known member
[quote author="Samuel Groner"]I've thought quite a bit about adding servos to this topology and I didn't come up with a simple yet convincing solution. The problem is the injection point--the obvious place after the input coupling capacitor is somewhat unhappy as the coupling capacitor will interact with the servo action, potentially producing some low-frequency wobbling.[/quote]
I was planning to use the same injection point that you use on your Design B. It might help to make the injection symmetrical by using an inverting amp to couple into the node connecting R9 and R10; the nice thing about that is that (with sufficient decoupling) the AC impedance of both input legs is even better matched, improving CMRR. I have to test and see how that works out.
The interaction with the input coupling caps is fixable by making the time constant of the servo sufficiently larger than that of Rin*Cin. A factor of 10 is usually enough, in my experience. This may mean that the mic pre needs to settle for half a minute after turning P48 on or off, but I don't care too much (customers might feel differently).
[quote author="Samuel Groner"][quote author="I"]My plan consists of a few DOAs designed to run from +60/-15 supplies, dropped into a standard differential Cohen mic pre, with a few floating servos to preserve DC balance. Plan B would involve a diff-in/diff-out DOA, but I'm not sure if imperfect matching between each output's compensation network might not make an oscillator out of such a beast.[/quote]
If you are interested, I've a presumably nice 80 V DOA in the pipline.[/quote]
Yes, please ! Mine is rather inspired by your DOA as it is, and I'm looking to design a pre not a DOA.
[quote author="Samuel Groner"]For sure I would be curious to see your fully differential opamp![/quote]
It is still very much a work in progress. For a rough idea, take the SGA-SOA, add a collector resistor to Q2, duplicate the gain and output stages (Q4-Q7) to get a negative output, and add a regular op-amp to set the common mode voltage on the output. Will post schematics if I ever get anything that's even marginally stable.
[quote author="JohnRoberts"]Does cap coupling the clock cause actual jitter (an uncertainty from one clock to the next) or some fixed timing error due to C, perhaps interacting with device input C? This is beyond my understanding of things digital I guess but I don't see how a constant timing error from a passive component causes noise, variable timing errors sure.[/quote]
There are two different issues here. First:
[quote author="JohnRoberts"] [...] the last codec I worked with had a PLL that would just hunt and recover (I think... I never actually tried this with one).[/quote]
[quote author="I"]This approach works well, if you/your customers are content with 30..100ps clock jitter[/quote]
This 30..100ps clock jitter is mostly due to the use of PLLs. I know of no integrated audio PLL with typical clock jitter specs much better than 100ps. This is partly because many of them use on-chip ring oscillators or integrated LC VCOs, which have much lower Q than a crystal, and partly because it is hard (read:expensive) to build a phase comparator with a noise figure that is comparable to that of a good XO. If you know of any commercial PLLs which achieve much less than 30ps jitter on a 24.576MHz clock do let me know.
The other thing is:
[quote author="I"]The main stumbling block (as far as I could see) is low-jitter master clock distribution. The best I could think of was transformer coupling of the 24.576MHz MCLK into a low-noise limiting differential amplifier, but even that will easily double the jitter of a good low-noise clock source over impedance-matched DC coupling.
[/quote]
Assuming a good low-noise MCLK has 2.5ps jitter, sinusoid signal shape, 24.576MHz @ 3Vpp (LVCMOS), and the ADC clock input has its threshold exactly at mid-voltage. The slew rate at the switching point is
dV/dt = (2*pi*f*A)
= (2*pi*24.576M*(3V/2))
= 231MV/s = 231uV/ps
A little under 600uV (231*2.5) wideband noise is enough to double the 2.5ps jitter! With the supply servo it's likely that converter and backplane ground will have more noise between them than that, so single-ended capacitor coupling is out. Differential coupling is possible, but most semiconductor differential to single ended converters have a few ps jitter, too. So it's not the capacitors per se that are introducing the jitter, it's the whole clock chain. Some can be gained by using a clipped sine as that speeds up the edge rate, but this introduces harmonics and makes the system sensitive to group delay. RF transformers work best, but even then you'd likely need some narrowband filters on the receiving end to minimize the jitter.
[quote author="JohnRoberts"][quote author="I"][...] dropped into a standard differential Cohen mic pre [...][/quote]
I'm still not sure why Cohen gets credit for that topology.[/quote]
I get the feeling that most people here know of the Cohen design (or the Green pre), so for me it's more a matter of using a commonly understood term than giving credit.
Do you mean like the Valley People mic pre (as discussed in this thread) ? I'll look into it, but I'm not too happy about the large cap/resistor values. I understand that the feedback should cancel most of the badness, but I'm still trying to see how a no-cap pre will sound (just like I want to know what an ADC sounds like when it's fed directly by a transformer).
JDB.
I was planning to use the same injection point that you use on your Design B. It might help to make the injection symmetrical by using an inverting amp to couple into the node connecting R9 and R10; the nice thing about that is that (with sufficient decoupling) the AC impedance of both input legs is even better matched, improving CMRR. I have to test and see how that works out.
The interaction with the input coupling caps is fixable by making the time constant of the servo sufficiently larger than that of Rin*Cin. A factor of 10 is usually enough, in my experience. This may mean that the mic pre needs to settle for half a minute after turning P48 on or off, but I don't care too much (customers might feel differently).
[quote author="Samuel Groner"][quote author="I"]My plan consists of a few DOAs designed to run from +60/-15 supplies, dropped into a standard differential Cohen mic pre, with a few floating servos to preserve DC balance. Plan B would involve a diff-in/diff-out DOA, but I'm not sure if imperfect matching between each output's compensation network might not make an oscillator out of such a beast.[/quote]
If you are interested, I've a presumably nice 80 V DOA in the pipline.[/quote]
Yes, please ! Mine is rather inspired by your DOA as it is, and I'm looking to design a pre not a DOA.
[quote author="Samuel Groner"]For sure I would be curious to see your fully differential opamp![/quote]
It is still very much a work in progress. For a rough idea, take the SGA-SOA, add a collector resistor to Q2, duplicate the gain and output stages (Q4-Q7) to get a negative output, and add a regular op-amp to set the common mode voltage on the output. Will post schematics if I ever get anything that's even marginally stable.
[quote author="JohnRoberts"]Does cap coupling the clock cause actual jitter (an uncertainty from one clock to the next) or some fixed timing error due to C, perhaps interacting with device input C? This is beyond my understanding of things digital I guess but I don't see how a constant timing error from a passive component causes noise, variable timing errors sure.[/quote]
There are two different issues here. First:
[quote author="JohnRoberts"] [...] the last codec I worked with had a PLL that would just hunt and recover (I think... I never actually tried this with one).[/quote]
[quote author="I"]This approach works well, if you/your customers are content with 30..100ps clock jitter[/quote]
This 30..100ps clock jitter is mostly due to the use of PLLs. I know of no integrated audio PLL with typical clock jitter specs much better than 100ps. This is partly because many of them use on-chip ring oscillators or integrated LC VCOs, which have much lower Q than a crystal, and partly because it is hard (read:expensive) to build a phase comparator with a noise figure that is comparable to that of a good XO. If you know of any commercial PLLs which achieve much less than 30ps jitter on a 24.576MHz clock do let me know.
The other thing is:
[quote author="I"]The main stumbling block (as far as I could see) is low-jitter master clock distribution. The best I could think of was transformer coupling of the 24.576MHz MCLK into a low-noise limiting differential amplifier, but even that will easily double the jitter of a good low-noise clock source over impedance-matched DC coupling.
[/quote]
Assuming a good low-noise MCLK has 2.5ps jitter, sinusoid signal shape, 24.576MHz @ 3Vpp (LVCMOS), and the ADC clock input has its threshold exactly at mid-voltage. The slew rate at the switching point is
dV/dt = (2*pi*f*A)
= (2*pi*24.576M*(3V/2))
= 231MV/s = 231uV/ps
A little under 600uV (231*2.5) wideband noise is enough to double the 2.5ps jitter! With the supply servo it's likely that converter and backplane ground will have more noise between them than that, so single-ended capacitor coupling is out. Differential coupling is possible, but most semiconductor differential to single ended converters have a few ps jitter, too. So it's not the capacitors per se that are introducing the jitter, it's the whole clock chain. Some can be gained by using a clipped sine as that speeds up the edge rate, but this introduces harmonics and makes the system sensitive to group delay. RF transformers work best, but even then you'd likely need some narrowband filters on the receiving end to minimize the jitter.
[quote author="JohnRoberts"][quote author="I"][...] dropped into a standard differential Cohen mic pre [...][/quote]
I'm still not sure why Cohen gets credit for that topology.[/quote]
I get the feeling that most people here know of the Cohen design (or the Green pre), so for me it's more a matter of using a commonly understood term than giving credit.
Do you mean like the Valley People mic pre (as discussed in this thread) ? I'll look into it, but I'm not too happy about the large cap/resistor values. I understand that the feedback should cancel most of the badness, but I'm still trying to see how a no-cap pre will sound (just like I want to know what an ADC sounds like when it's fed directly by a transformer).
JDB.
ruffrecords
Well-known member
[quote author="Wavebourn"]So, what NPN beasts may be called today low noise devices for say 600 Ohm input?[/quote]
I am not sure why everyone gets so interested in very low noise transistors. Most modern types, whether PNP or NPN can give a mic preamp noise figure of between 1 and 3dB. You will probably need a PNP to get 1dB but 2dB is achieveable with NPN. This means the output noise is at most 3dB worse than the input noise at max gain. Unless you have a really low output mic or want to record something that is very quiet then the extra dB or two is not relevant. At that sort of gain, the self noise of most mics will exceed the mic pre noise anyway.
Plus, it is easy to design a mic pre with low noise figure at high gain. What is less easy is designing one that has a good noise figure at all gains. Manufacturers tend to quote NF at highest gain because that is dominated by input noise. Designers often think little about output noise which starts to dominate once the gain is reduced. For example suppose you have a pre with a -130dBu EIN. With a gain of 60dB the output noise (due to inout noise) is -70dBu. If the output noise is -80dB it won't significantly affect the total noise. But if you reduce the gain by 20dB, the noise at the output due to the input noise is -90dBu. The output noise at -80dBu now dominates and your wonderful input noise figure is wasted. A really good mic pre will be hard pushed to achieve better than -90dBu output noise so even with one of these output noise will dominate for gains less than 40dB and IME a gain of 40dB is more typical in real recording situations.
Ian
I am not sure why everyone gets so interested in very low noise transistors. Most modern types, whether PNP or NPN can give a mic preamp noise figure of between 1 and 3dB. You will probably need a PNP to get 1dB but 2dB is achieveable with NPN. This means the output noise is at most 3dB worse than the input noise at max gain. Unless you have a really low output mic or want to record something that is very quiet then the extra dB or two is not relevant. At that sort of gain, the self noise of most mics will exceed the mic pre noise anyway.
Plus, it is easy to design a mic pre with low noise figure at high gain. What is less easy is designing one that has a good noise figure at all gains. Manufacturers tend to quote NF at highest gain because that is dominated by input noise. Designers often think little about output noise which starts to dominate once the gain is reduced. For example suppose you have a pre with a -130dBu EIN. With a gain of 60dB the output noise (due to inout noise) is -70dBu. If the output noise is -80dB it won't significantly affect the total noise. But if you reduce the gain by 20dB, the noise at the output due to the input noise is -90dBu. The output noise at -80dBu now dominates and your wonderful input noise figure is wasted. A really good mic pre will be hard pushed to achieve better than -90dBu output noise so even with one of these output noise will dominate for gains less than 40dB and IME a gain of 40dB is more typical in real recording situations.
Ian
