The first time i watched this video, id built my first mic from a pcb and mouser parts list, but my knowledge base was "put the part on the matching number on the board". I didn't understand the jfet or its bias, and i'd unknowing had already done it by virtue of the trimmer resistor in the kit and the instructions to turn it till i get a x number on a multimeter, which i didnt how use to use either. I watched this video twice, not understanding any of it and doubting if i ever would. I kept going, trying to learn more until, eventually, i was able to design my own condenser mic circuit and make my own pcb. I remembered this video, and i was happy cos i finally new what Karl was talking about. So i watch this every now and then, being able to take it all in, remembering there once was a time that i couldn't!
Great video, thank you for making it. I'm replacing a bad JFET on an old Sony electret condenser mic and will no doubt have to adjust the bias point due to the original FET being unobtanium. It makes sense that the optimal point for headroom and distortion would not be too far apart.
Great video. I build preamps for hydrophones which have very similar characteristics and requirements. If instead of changing R7, you had changed R6 instead, the bias current through the JFET would have been relatively constant and the voltage at the drain as well, and making R6 greater would simply increase negative feedback lowering gain and reducing distortion. I use this in my designs to reduce preamp gain without changing the bias point when I am dealing with large input signals. I'm pretty sure you could do the same here, i.e. trade off gain and distortion without changing the bias point, by allowing adjustment of R6. Changing R7 is messier because the relationship between R5, R7 and C4 is important. Changing this resistor will change the bias current and the voltage at the drain which you've used to better center the signal. Reducing R7 from 1.8K to 750 should significantly increase the bias current and also significantly increase the corner frequency of the high pass filter set up by R7 and C4. Anyway, it was an interesting and instructive video, although it wasn't always clear what the preamp gain was as the adjustments were being made, and it's easy to improve both dynamic range and distortion numbers by decreasing gain. Finally, let me end with a different question. This circuit uses the 2SK170 JFET, which has a pretty good reputation, but after doing a lot of testing and measurements, I've found the BF862 to have lower noise throughout the audio band, and it has low enough capacitance to allow use of two or more in parallel. See Comparison of Low Cost, Low Noise JFETs, Dimitri Danyuk, “Measurements Rate SMT Low-Voltage n-JFETs Under Consistent Conditions”, 19 April 2013. This is to some degree an academic question, since the BF862 recently went end of life, but I'd like to hear your thoughts on this.
My apologies for not responding sooner - you make some excellent points. I had rather assumed that C4 was sufficiently large that I could safely reduce the value of R7 by a fair margin without having any impact within the audio band - definitely something to watch for, though. I also take your point re: trading gain for distortion & headroom - it would have been good to make clear just what tradeoffs were being made in that regard. Interesting concept re: parallel JFETs too, definitely something worth looking into. Cheers, & thanks for your excellent comment.
I'm really late to the party (and who cares anyway), but the best way to set the bias conditions would be by using a separate voltage divider. This fixes the JFET source voltage regardless of the current flowing through it. That being said, this whole circuit is pretty.. bad. It clearly stems from the tube days: self-biased JFET, no global negative feedback, and an output transformer - complete nonsense in a solid state circuit! C2 and R5 are redundant because the mike capsule itself is a capacitor. They're there because the polarising voltage is already low, and connecting the capsule directly to the gate would further reduce it, thus reducing the sensitivity (I don't know the specs of the capsule, but it's a good idea to always make the polarising voltage as high as possible). The voltage amplification is done at the JFET's drain, meaning the signal will be greatly attenuated due to Miller effect, further reducing the sensitivity. I could go on and on forever, but the point is: I'd rather make a new preamp from scratch than use this one.
@@noneofyerbeeswax8194 - The gate bias point is a few microvolts above ground. R5 handles the leakage current from the JFET and must be kept as high as possible to keep it from dominating low frequency noise. Thus the 1G values for R4 and R5. - C2 is required because the microphone capacitance leakage current is much higher than the JFET leakage current at room temperature. Without C2, R5 would have to be reduced which would increase low frequency noise. - The transformer is used to shape the output to provide what many people consider more pleasant vocals. This is purely subjective so it doesn't help to make the obvious claim that the transformer will increase distortion. - A cascode arrangement can be used to increase first stage gain. This provides a small performance improvement at the cost of the cascode transistor and biasing network.
@@ats89117 I do know how analog circuits work, thanks. :) I didn't know the capsules were this leaky, probably because those I've worked with weren't. Maybe it's because they were all used in studio environment, at a constant temperature and low humidity. In adverse conditions, the situation might have been different, but people normally prefer dynamics there anyway. I've had more problems with bias corruption due to FET leakage because even 10pA over 1gOhm will give us 10mV (not uV) above ground. The only real advantage of an output transformer is that it gives you a balanced, low-impedance output, which would require a few extra semiconductors otherwise. The linearity and distortion figures however are horrible. A cascode mostly solves the problem with Miller capacitance, that's why it's used in high frequency applications. Using it at LF is overkill IMO. It requires an extra bias etc.. I prefer a three-stage, closed loop system where the JFET is a transconductance amplifier, a PNP transistor does the actual voltage amplification, and an emitter follower provides a low-impedance output. Negative feedback makes for great linearity and lets you modify the frequency response as you wish (although personally, I prefer it flat).
@@noneofyerbeeswax8194 You're right about the bias voltage. I'm used to using a 20 to 50 Meg bias resistor whereas this application has two parallel 1G resistors. The output transformer is only there because it gives a more tube-like sound. Once again, this is a personal preference, and the manufacturer probably believed it would help sell more units. Negative feedback could certainly provide greater linearity at the cost of reduced gain, but once again, the manufacturer clearly wasn't aiming for maximum linearity but rather minimum noise and a tube-like sound (the result of the high impedance tube and its impedance matching transformer). I'm not an audiofile, but I know quite a few people who prefer the 'warm' transformer sound for vocals. Your mileage may vary...
Interesting. I'll be getting a Behringer C-3 tomorrow and it might come in handy. I'd also buy one of them cheap but nicely built chinesium mikes with an experimental vacuum tube conversion in mind. First time seeing you and gotta admit that you have a lot in common with Paul Carlson as far as talking about electronics and explaining it goes.
Thanks for this video. With all the "mic mod" kits, it's nice to see some theory and measurements. Can you describe how the signal injection being reference to ground gives valid results and why the injection point wouldn't need to be at the capsule + and - points?
Thanks for that - it's a very good question. For a bias adjustment it's not really an issue - we're applying the signal via the 1nF coupling cap to the gate of the FET (so DC offset won't be a problem), and avoiding having to worry about the polarization voltage. Connecting the signal generator across both capsule leads would be applying the polarization voltage to the signal generator, and the low impedance of the signal generator output (say 50Ω) would basically short the polarization voltage to ground, drawing extra current that wouldn't be the case during normal operation, and thus changing voltages elsewhere in the circuit. Where this gets (even) messier is that the other capsule lead actually has some AC signal riding on top of the polarization voltage, as this mic (shamelessly) copies the U87's negative feedback scheme for HF de-emphasis. Brute-forcing a low-impedance signal onto the gate of the FET completely obliterates this effect, so if we were measuring frequency response we'd get a misleading result. Thankfully, for a bias adjustment, we do manage to get away with it.
26:00 It would've been interesting to also probe the output, in order to (also visually) compare / correlate the waveform to the spectrum and the level of the input signal.
Very useful video, thank you so much for it! I read in one of your responses to the comments that you would not recommend to leave the capsule when performing the biasing connected but rather to attach the signal gen through a 68pf cap (as I understand it, you meant a coupling cap). On the other hand, some advice on biasing the fet I've read don't mention connection through a coupling cap. I'm wondering what effect this cap or its absence might have on finding the best biasing point.
Thank you. It's a very good question - I suspect it doesn't make much difference, but I'm not certain. You're already applying signal to a 1nF coupling cap anyway, so from a DC perspective it should make no difference. That said, I'm a little wary of using a signal generator with low output impedance to substitute for a microphone capsule having enormously high impedance - strikes me there's the possibility that things might behave a little differently. As I understand it, the idea behind coupling the sig gen through such a small capacitor is to give an effective output impedance of the order of a few meg-ohms, and thus avoid "pushing things around" in a way that the microphone capsule could never do.
Thank you very much! Great video... really loved it!!! Quick question... for the THD test, where did you connected the test lead for the signal? Kind regards!
Thank you so much. As I recall, there's a 1nF coupling capacitor between the capsule and the gate of the FET, and I connected to the capsule side of that capacitor. Looking back, I probably should have connected the signal through a high value resistor, as directly connecting a low impedance source to such a circuit may have changed its behaviour somewhat.
Hey Karl! Thanks for making this video. One question... I see how you connected the signal generator to the mic, but how did you then output the signal to the oscilloscope?
Hi, for the scope probe what setting? 1x or 10x..? How did u connect the output from mic to the scope ? Balanced ? Or one leg of the signal only ? Pin 2 hot and pin 1 ground ? Thx
Good question - how do you measure a balanced mic output with a ground referenced instrument? If I recall correctly, I just connected the unbalanced output of my mic preamp (M-Audio DMP-3) to the scope directly, with the scope input set to 1M Ohm impedance. Naturally you're also measuring the behaviour of the preamp at the same time, but that wasn't a problem in this case. Thinking about how you might probe the mic directly, perhaps you could measure with two channels and use the math function to display the difference? You'd want to AC-couple the inputs, as they're at 48v relative to ground because of the phantom power.
@@KarlAdamsAudio hi, I’m new at this, I’m a hobbyist and not too familiar with all the terminology. Would be great if you did another video connected directly to the mic balanced output and show procedure. But isn’t the math function only on digital scopes ? I came across your video while searching about biasing JFET…Because I have original Rode NT1 mic that I recapped and did the Jim Williams (designer of the mic) sibilance mod (changing values of 2 film caps). It sounds great but when I try to use it on louder sources like a tom drum, at 1-2in away, I can hear a little distortion.. I saw your video and thought maybe I need to re-bias the Jfet. I don’t recall the distortion being there prior to the mods. I have a few of these mics and did the mod to all of them. And they all have the same symptoms. Chances are the mods have something to do with it. But I was also going to try putting back the original caps to see if symptoms went away.. Thx
@@Maxxzinno Yes, the math function is a feature of digital scopes. For this sort of work a hand-held 'scopemeter' can be excellent because the input floating, you can measure between any two points without concern for ground. From what I've gleaned, the NT1 circuit is a variant of the transformerless Schoeps circuit, and the mod involves capacitors between the bases of the output transistors and ground. I wouldn't imagine that would disrupt the bias conditions of the JFET. Also note that the JFET in a Schoeps-style circuit acts as a phase-splitter, we derive a signal from both source and drain sides in order to create our balanced output signal. In that case we can't go changing the resistance in the source leg without messing up the balance, so the method for bias adjustment is somewhat different to the one I used.
@@KarlAdamsAudio ok so if it’s not the JFET biasing, It’s possible the NT1 is just too hot for toms at 1-2in.. I will make a video with a sound clip and share the link on this thread…if I can get your opinion on what you think it can be… would be great. Thx Ps; I also have another NT1 that has way more background hiss than the others.. it makes it unusable… any thoughts on that ?
@@KarlAdamsAudio hi so I figured out that the mods I did to the mic is what was causing the distortion. The mod involved changing 2 470pF film caps to 1500 or 2200pF. I read about the mod from the designer of the mic, Jim Williams, on a forum…he had suggested this mod as the only viable mod to better the mic. What it does is lower the sibilance… but I guess it lowers the headroom on the lower frequencies also… I guess for vocal purposes it wouldn’t matter. But because I was using it on drums(toms) it mattered. Basically I reinstalled the original 470pF caps and the distortion is gone. Now the thing is I had messed with the trim pot trying to get it not to distort, so now I wonder… I have another NT1 untouched, can I measure the value of the resistance and just set it to the same on the mic I touched. Or does it really need to be Re-biased now with test equipment…? As for the method for biasing a fet in schoeps mic like NT1, would you have any info on the method ? Thx…
So I'm presuming the zener is used to keep the Drain-to-Source voltage on the FET below 25V. To be honest I'm not totally sure it's going to be necessary. It all depends on how much the quiescent current draw of the emitter-follower stage pulls down the supply voltage to the FET. I think the idea of using a zener to protect the 2n3819 from excessive voltage is particularly applicable to circuit mods that remove the emitter follower stage and drive a higher-ratio transformer directly. I'll look into it further and let you know.
I just checked some voltages on that mic - the Drain-Source voltage across the FET was a bit variable, but eventually stabilized at 7.6V (I don't think I ever saw more than 12V there at any stage). It looks like a popular location for adding a zener diode to this circuit is across C12 - where I measured some some 20.5V. Starting to look like you'd be safe to substitute a 2n3819 without worrying about adding a zener.
Hey Karl, Im trying to bias a microphone a friend built thats a U87 clone. Where would I insert the Freq Generator? And where would I connect the Oscilloscope Probe? And do you disconnect the capsule when you do the biasing?
As I recall, I just connected to signal gen directly to the capsule side of the 1nF coupling cap that leads to the gate of the FET, leaving the capsule connected. That's not really right, though - so maybe don't do it that way I did. Better to get a capacitor of approximately the same value as the capsule (say 68pF or so) - desolder the capsule wire and connect the sig-gen via your 68pF coupling cap to the point where you just removed the capsule wire. In my case I just connected the scope to the output of my mic preamp, but equally you could pick it up from the driven lead of the output transformer instead.
I just built 2 U87 clones, and would like to custom Bias the FETs per each mic, but don't have a distortion analyzer. On the various build threads they say to Bias the FET before the capsule is wired. I have an older Heathkit single trace scope, which should work fine. Need to borrow a signal generator though. Anyway, so far this is the most comprehensive video I've seen on biasing a Mic FET. I'm a little intimidated to try it on my own!
@@Wizardofgosz I wonder if it's feasible to simply use a computer audio interface for all the task - i mean it can output audio frequency signals, it has a mic preamp with 48V phantom power source, it can capture the mic signal, and soundcard oscilloscope software exists. You can drive a stereo output pair where one channel goes into the mic capsule input where you output 1KHz sine reference signal, while on the other channel you generate sine signal of same amplitude but maybe 50-100Hz and connect that to multimeter in AC mode to get approximate absolute values, and the output volume control will control both at the same time - with some tracking error, but i bet you can just disregard it. It seems you could even just use an old MP3 player or PC onboard soundcard to generate signals, considering you have rest of test equipment. I got a 15 year old TASCAM interface which is junk, but it was only a few $ and that's all i could afford, it works, and it'll have to do, because there is no way i can get access to professional test equipment - even just a better Behringer HD interface is a distant dream at this point. I am interested in the topic and will want to build a microphone circuit and adjust it myself, but haven't done yet. Is there something i'm missing?
It would seem the manufactures neither optimized the circuit for either headroom or distortion! Would you in fact ever get as much as 1volt from the capsule in normal use?
Large diaphragm condenser capsule sensitivities are usually around 20mV/Pa, so a signal level of 1V would imply around 50Pa or approx 128dB SPL - pretty damn loud, but you'd easily see that sort of level from a close-miked trumpet, for example. The issue with optimization is the device-to-device variation in the FETs - they're not particularly tightly binned for Idss as supplied, and testing and selecting them prior to assembly (and/or choosing source resistors to match each FET's characteristics) is an very expensive proposition.
Hello Can you make a video on how to fix a noise buzz on small condenser microphones. Its very usual to have this problems on studio recordings. Thanks
Oddly enough, you never specified what resistor value you used between R8 and C6 for the J508 constant current diode replacement at R10. Personally, I think that would probably be the most important part of this mod because it lowers distortion. Care to specify ?
Hi. i have a usb microphone, the JFET is soldered directly to the mic capsule . The orignal JFET is 2sk660, I want to replace it with 2n5457. How can I bias the FET?
I found a advice from 10 years ago but the picture is gone, can I add some resistor with the FET like schematic at 3:33 ? Can you help me please?
Hard to know what to recommend without knowing more about the circuit - the typical transformerless Schoeps-inspired circuit has identical resistors in the source and drain legs of the FET, so changing the source resistor isn't the approach to take. Also, I'm wondering if your mic has an electret capsule, given the gate of the FET is soldered to the capsule without a capacitor to block DC.
Doug Ford's EEVBlog video on electret microphone design is an excellent resource: czcams.com/video/UhG83WS51q8/video.html I note that the 2sk660 incorporates the high value resistor from gate to source (it's a device specifically intended for this application) while the 2n5457 will need an external resistor (of around 1G). Also of note is the IDSS range of the 2n5457 is much higher (i.e. the lowest value for 2n5457 is double the highest value for 2sk660), which may be a challenge when selecting an appropriate resistor value between +ve supply and FET drain.
Thank you,It's very helpful tutorial! Could you please tell to which junction on the PCB you connect the oscilloscope lead, and you mentioned that you connected a resistor instead of R9, what value. I am trying to do the same thing to the V67G that I own. Thanks,
Thank you. The modification was to add an extra 510k resistor in addition to R9 so that together they form a voltage divider - as shown here: www.dropbox.com/s/p8psqqicpsgm0fi/MXLV67mod1.jpg?dl=0 I installed mine on the back of the board (not exactly neatly) like so: www.dropbox.com/s/nw60kyzoo6encdf/MXLV67mod2.jpg?dl=0 As I recall I connected the scope to the output of the preamp, rather than directly to the board - although I'm sure you could take the signal from (say) the transformer side of C8 instead. Cheers & Best of luck with modifying your mic.
Hi Karl, I did most the changes you recommended and the mic sounds a lot better and more sensitive. Though I could not locate a seller for J508. Any replacement I can use instead? Thank you so much and sorry to bother you.
Good question - there's the 1N5307, but if anything that's even harder to find. At a pinch I've also used a selected J202 with gate shorted to source (but you need to test a few of them till you find one that gives around 2.4mA in that configuration). I originally bought my stock of J508 from Future Electronics, but I see they've got none in stock (and the minimum order quantity is 4,000 - eek!). There are some on eBay if you don't mind paying the asking price, which is a bit steep. I'd be happy to post you a couple from my 'stash' (I've still got a bag of them somewhere) also.
Hello, I actually have one of 1N5307, though when I used it did not seem to work very good. Possibly I did not hook it up correctly ( polarity configuration issue) or maybe because I changed the JFET to PF5102 which is working nicely. I live in U.S my email srifai1@hotmail.com. Not sure if you can post one of J508 to me to try unless you think I don't need it. I will pay you. It has been very fun project! Really appreciate your generosity and help.
so, did you do the video?? if so could you provide the link??... I really liked this video, yet the final sketch with added info is clearly a must have, I am planing in building my own mic and it may be that if I buy the same resistors and other components that you used, I could skip the need of a multimeter and I could skip the need for those big machines that you were using and that for what I need could be considered total overkill...
Doing electronics without a multimeter is... naive, to say the least, i'm afraid. "Those big machines" can be relatively easily replaced with an inexpensive audio interface and free software, by the way.
hum nope!... there is nothing naive in the statement... if the guy who is making the video is already using a multimeter... why do you need to shadow at home the very same steps that were already seen on the video, that took the author such a time to properly create!... by the way I just tried to do a smaller version of the amplifier and it was a pain... the authors are tossing these videos out as if omitting info was a cool thing or as if it was some sort of standard... check my channel so you can see what I mean... the problem is a big one... check my video so you can see it...
Because components have *tolerances* - JFETs especially have pretty loose ones, compared to other components. Certain information is "ommited" because it's sort of taken as granted that a multimeter is among the bare minimum tooling you need, to be able to do anything useful / meaningful in electronics. If this was labeled as an electronics *tutorial* (which it's not), THEN i would understand your dissatisfaction. Suppose your (first) circuit doesn't work (and there's a chance it won't). Without any test equipment, how can you even hope to troubleshoot it? Plain guess-work? Well, maybe except for the "shotgun" approach, where you just replace all the components (or just one at a time)...
Hi Tomxico here from Portugal. My studio condenser mic works just fine but sometimes then the signal is very low or even vanish away. Is there a bad capacitor or something ? It's not about fantom Power cos I have oneother condenser mic and works just fine. Can someone help me ? Please I am locked a home due to the virus. Heath to you all. Thanks
I suspect you may have an issue with humidity and/or dust on the capsule. If it provides a leakage path for the polarization voltage, the mic will go quiet (or even silent) for several seconds until the polarization voltage builds up again. If this is the issue, it will get a lot worse when you breathe on the capsule. You can try leaving the mic in a warm environment to evaporate off excess moisture & that may help. Actually cleaning the capsule is quite tricky due to how easily it can be damaged, so I'd be wary of attempting that.
@@KarlAdamsAudio Thank you for your prompt response. You are right. A minute or 2 and the mic gets low signal input and just vanishes. Then after a minute or 2 come to normal db sound. but if working for a full hour or 2 gets normal all the time. Bad capacitor ? Or just the humidity in the capsule ? I will try to clean the capsule. Thank you for your advise.
That does sound to be consistent with a dust + humidity issue. With a K67 style capsule, you have that ring of unsputtered mylar around the outside of the diaphragm - if you get too much dust on there, then condensation from the air will form a conductive path that will discharge your polarization voltage. Even without excessive dust, storing the mic in a cold and/or humid environment can cause similar issues. In extended use, the mic may get warm enough for this to no longer be an issue (this is especially true of tube mics, that have their own built-in heat source). If you have a dusty capsule, then breathing on it (close & open-mouthed, like you're trying to fog up a mirror) should cause it to cut out almost immediately. Cleaning a capsule is labour-intensive, as I understand the technique is to use a fine camel-hair artists brush and distilled water, and (using a bright light & magnification) remove individual dust specks with the tip of the brush.
Probably not - it's not safe to assume that the resistor named 'R7' performs the same role in the circuit. A suitable candidate would be a resistor in the Source leg of the FET, bit it's not necessarily the one called 'R7'.
Are you trying to substitute a condenser capsule for an electret? As that's not going to work due to the absence of the polarization voltage that condenser capsules need in order to function.
@@KarlAdamsAudio thanks for the info bud ..i really just need help to make an electret mic work with a big capsule ive been studing mics trying to find it out here the bias i see on the mic i hav does have a fet resistor but when i swapped the capsule it didnt work i tryed it on another mic and it worked beautiful i think the small electret mics would be a good shell to convert to true condenser i just need help designing a circuit to run piggy back and give it enough power to use a k67 capsule theres very limited.. info online ive been looking any helps appreciated
The first time i watched this video, id built my first mic from a pcb and mouser parts list, but my knowledge base was "put the part on the matching number on the board". I didn't understand the jfet or its bias, and i'd unknowing had already done it by virtue of the trimmer resistor in the kit and the instructions to turn it till i get a x number on a multimeter, which i didnt how use to use either. I watched this video twice, not understanding any of it and doubting if i ever would. I kept going, trying to learn more until, eventually, i was able to design my own condenser mic circuit and make my own pcb. I remembered this video, and i was happy cos i finally new what Karl was talking about. So i watch this every now and then, being able to take it all in, remembering there once was a time that i couldn't!
Great video, thank you for making it. I'm replacing a bad JFET on an old Sony electret condenser mic and will no doubt have to adjust the bias point due to the original FET being unobtanium. It makes sense that the optimal point for headroom and distortion would not be too far apart.
Great video. Found it very interesting as I used to be Service Manager for the UK agents for Sennheiser and Nagra.
Your guest spot on EEVBlog brought me here, loved your segment and decided to subscribe and watch through your video collection. :)
Thank you so much. (and thanks to Dave too, of course).
Same here. Great videos.
I enjoy your presentation style, Karl. Thanks for the vid!
Thank you so much for your support - it is greatly appreciated.
I think this deserves a new updated video streamlined video
Great video. I build preamps for hydrophones which have very similar characteristics and requirements. If instead of changing R7, you had changed R6 instead, the bias current through the JFET would have been relatively constant and the voltage at the drain as well, and making R6 greater would simply increase negative feedback lowering gain and reducing distortion. I use this in my designs to reduce preamp gain without changing the bias point when I am dealing with large input signals. I'm pretty sure you could do the same here, i.e. trade off gain and distortion without changing the bias point, by allowing adjustment of R6.
Changing R7 is messier because the relationship between R5, R7 and C4 is important. Changing this resistor will change the bias current and the voltage at the drain which you've used to better center the signal. Reducing R7 from 1.8K to 750 should significantly increase the bias current and also significantly increase the corner frequency of the high pass filter set up by R7 and C4. Anyway, it was an interesting and instructive video, although it wasn't always clear what the preamp gain was as the adjustments were being made, and it's easy to improve both dynamic range and distortion numbers by decreasing gain.
Finally, let me end with a different question. This circuit uses the 2SK170 JFET, which has a pretty good reputation, but after doing a lot of testing and measurements, I've found the BF862 to have lower noise throughout the audio band, and it has low enough capacitance to allow use of two or more in parallel. See Comparison of Low Cost, Low Noise JFETs,
Dimitri Danyuk, “Measurements Rate SMT Low-Voltage n-JFETs Under Consistent Conditions”, 19 April 2013. This is to some degree an academic question, since the BF862 recently went end of life, but I'd like to hear your thoughts on this.
My apologies for not responding sooner - you make some excellent points. I had rather assumed that C4 was sufficiently large that I could safely reduce the value of R7 by a fair margin without having any impact within the audio band - definitely something to watch for, though. I also take your point re: trading gain for distortion & headroom - it would have been good to make clear just what tradeoffs were being made in that regard. Interesting concept re: parallel JFETs too, definitely something worth looking into. Cheers, & thanks for your excellent comment.
I'm really late to the party (and who cares anyway), but the best way to set the bias conditions would be by using a separate voltage divider. This fixes the JFET source voltage regardless of the current flowing through it.
That being said, this whole circuit is pretty.. bad. It clearly stems from the tube days: self-biased JFET, no global negative feedback, and an output transformer - complete nonsense in a solid state circuit! C2 and R5 are redundant because the mike capsule itself is a capacitor. They're there because the polarising voltage is already low, and connecting the capsule directly to the gate would further reduce it, thus reducing the sensitivity (I don't know the specs of the capsule, but it's a good idea to always make the polarising voltage as high as possible). The voltage amplification is done at the JFET's drain, meaning the signal will be greatly attenuated due to Miller effect, further reducing the sensitivity.
I could go on and on forever, but the point is: I'd rather make a new preamp from scratch than use this one.
@@noneofyerbeeswax8194 - The gate bias point is a few microvolts above ground. R5 handles the leakage current from the JFET and must be kept as high as possible to keep it from dominating low frequency noise. Thus the 1G values for R4 and R5.
- C2 is required because the microphone capacitance leakage current is much higher than the JFET leakage current at room temperature. Without C2, R5 would have to be reduced which would increase low frequency noise.
- The transformer is used to shape the output to provide what many people consider more pleasant vocals. This is purely subjective so it doesn't help to make the obvious claim that the transformer will increase distortion.
- A cascode arrangement can be used to increase first stage gain. This provides a small performance improvement at the cost of the cascode transistor and biasing network.
@@ats89117 I do know how analog circuits work, thanks. :) I didn't know the capsules were this leaky, probably because those I've worked with weren't. Maybe it's because they were all used in studio environment, at a constant temperature and low humidity. In adverse conditions, the situation might have been different, but people normally prefer dynamics there anyway.
I've had more problems with bias corruption due to FET leakage because even 10pA over 1gOhm will give us 10mV (not uV) above ground.
The only real advantage of an output transformer is that it gives you a balanced, low-impedance output, which would require a few extra semiconductors otherwise. The linearity and distortion figures however are horrible.
A cascode mostly solves the problem with Miller capacitance, that's why it's used in high frequency applications. Using it at LF is overkill IMO. It requires an extra bias etc.. I prefer a three-stage, closed loop system where the JFET is a transconductance amplifier, a PNP transistor does the actual voltage amplification, and an emitter follower provides a low-impedance output. Negative feedback makes for great linearity and lets you modify the frequency response as you wish (although personally, I prefer it flat).
@@noneofyerbeeswax8194 You're right about the bias voltage. I'm used to using a 20 to 50 Meg bias resistor whereas this application has two parallel 1G resistors. The output transformer is only there because it gives a more tube-like sound. Once again, this is a personal preference, and the manufacturer probably believed it would help sell more units. Negative feedback could certainly provide greater linearity at the cost of reduced gain, but once again, the manufacturer clearly wasn't aiming for maximum linearity but rather minimum noise and a tube-like sound (the result of the high impedance tube and its impedance matching transformer). I'm not an audiofile, but I know quite a few people who prefer the 'warm' transformer sound for vocals. Your mileage may vary...
Interesting. I'll be getting a Behringer C-3 tomorrow and it might come in handy. I'd also buy one of them cheap but nicely built chinesium mikes with an experimental vacuum tube conversion in mind.
First time seeing you and gotta admit that you have a lot in common with Paul Carlson as far as talking about electronics and explaining it goes.
Thanks for the video. It would have been helpful to note the two source resistance values, max output resistance vs lowest distortion value
wow very informative sir
Thanks for this video. With all the "mic mod" kits, it's nice to see some theory and measurements.
Can you describe how the signal injection being reference to ground gives valid results and why the injection point wouldn't need to be at the capsule + and - points?
Thanks for that - it's a very good question. For a bias adjustment it's not really an issue - we're applying the signal via the 1nF coupling cap to the gate of the FET (so DC offset won't be a problem), and avoiding having to worry about the polarization voltage. Connecting the signal generator across both capsule leads would be applying the polarization voltage to the signal generator, and the low impedance of the signal generator output (say 50Ω) would basically short the polarization voltage to ground, drawing extra current that wouldn't be the case during normal operation, and thus changing voltages elsewhere in the circuit.
Where this gets (even) messier is that the other capsule lead actually has some AC signal riding on top of the polarization voltage, as this mic (shamelessly) copies the U87's negative feedback scheme for HF de-emphasis. Brute-forcing a low-impedance signal onto the gate of the FET completely obliterates this effect, so if we were measuring frequency response we'd get a misleading result. Thankfully, for a bias adjustment, we do manage to get away with it.
Thank you
26:00 It would've been interesting to also probe the output, in order to (also visually) compare / correlate the waveform to the spectrum and the level of the input signal.
26:24 Nevermind... :D
Very useful video, thank you so much for it! I read in one of your responses to the comments that you would not recommend to leave the capsule when performing the biasing connected but rather to attach the signal gen through a 68pf cap (as I understand it, you meant a coupling cap). On the other hand, some advice on biasing the fet I've read don't mention connection through a coupling cap. I'm wondering what effect this cap or its absence might have on finding the best biasing point.
Thank you. It's a very good question - I suspect it doesn't make much difference, but I'm not certain. You're already applying signal to a 1nF coupling cap anyway, so from a DC perspective it should make no difference. That said, I'm a little wary of using a signal generator with low output impedance to substitute for a microphone capsule having enormously high impedance - strikes me there's the possibility that things might behave a little differently. As I understand it, the idea behind coupling the sig gen through such a small capacitor is to give an effective output impedance of the order of a few meg-ohms, and thus avoid "pushing things around" in a way that the microphone capsule could never do.
Thank you very much! Great video... really loved it!!!
Quick question... for the THD test, where did you connected the test lead for the signal?
Kind regards!
Thank you so much. As I recall, there's a 1nF coupling capacitor between the capsule and the gate of the FET, and I connected to the capsule side of that capacitor. Looking back, I probably should have connected the signal through a high value resistor, as directly connecting a low impedance source to such a circuit may have changed its behaviour somewhat.
It is questionable whether the actual circuit uses a 2SK170, as it has been long obsolete.
Hey Karl! Thanks for making this video. One question... I see how you connected the signal generator to the mic, but how did you then output the signal to the oscilloscope?
G'day & Thank You - as I recall, I had the scope connected to the line output of the mic preamp (M-Audio DPM3).
Hi, for the scope probe what setting? 1x or 10x..? How did u connect the output from mic to the scope ? Balanced ? Or one leg of the signal only ? Pin 2 hot and pin 1 ground ? Thx
Good question - how do you measure a balanced mic output with a ground referenced instrument? If I recall correctly, I just connected the unbalanced output of my mic preamp (M-Audio DMP-3) to the scope directly, with the scope input set to 1M Ohm impedance. Naturally you're also measuring the behaviour of the preamp at the same time, but that wasn't a problem in this case. Thinking about how you might probe the mic directly, perhaps you could measure with two channels and use the math function to display the difference? You'd want to AC-couple the inputs, as they're at 48v relative to ground because of the phantom power.
@@KarlAdamsAudio hi, I’m new at this, I’m a hobbyist and not too familiar with all the terminology. Would be great if you did another video connected directly to the mic balanced output and show procedure. But isn’t the math function only on digital scopes ?
I came across your video while searching about biasing JFET…Because I have original Rode NT1 mic that I recapped and did the Jim Williams (designer of the mic) sibilance mod (changing values of 2 film caps). It sounds great but when I try to use it on louder sources like a tom drum, at 1-2in away, I can hear a little distortion.. I saw your video and thought maybe I need to re-bias the Jfet.
I don’t recall the distortion being there prior to the mods.
I have a few of these mics and did the mod to all of them. And they all have the same symptoms. Chances are the mods have something to do with it.
But I was also going to try putting back the original caps to see if symptoms went away..
Thx
@@Maxxzinno Yes, the math function is a feature of digital scopes. For this sort of work a hand-held 'scopemeter' can be excellent because the input floating, you can measure between any two points without concern for ground. From what I've gleaned, the NT1 circuit is a variant of the transformerless Schoeps circuit, and the mod involves capacitors between the bases of the output transistors and ground. I wouldn't imagine that would disrupt the bias conditions of the JFET. Also note that the JFET in a Schoeps-style circuit acts as a phase-splitter, we derive a signal from both source and drain sides in order to create our balanced output signal. In that case we can't go changing the resistance in the source leg without messing up the balance, so the method for bias adjustment is somewhat different to the one I used.
@@KarlAdamsAudio ok so if it’s not the JFET biasing, It’s possible the NT1 is just too hot for toms at 1-2in.. I will make a video with a sound clip and share the link on this thread…if I can get your opinion on what you think it can be… would be great. Thx
Ps; I also have another NT1 that has way more background hiss than the others.. it makes it unusable… any thoughts on that ?
@@KarlAdamsAudio hi so I figured out that the mods I did to the mic is what was causing the distortion. The mod involved changing 2 470pF film caps to 1500 or 2200pF. I read about the mod from the designer of the mic, Jim Williams, on a forum…he had suggested this mod as the only viable mod to better the mic. What it does is lower the sibilance… but I guess it lowers the headroom on the lower frequencies also… I guess for vocal purposes it wouldn’t matter. But because I was using it on drums(toms) it mattered. Basically I reinstalled the original 470pF caps and the distortion is gone. Now the thing is I had messed with the trim pot trying to get it not to distort, so now I wonder… I have another NT1 untouched, can I measure the value of the resistance and just set it to the same on the mic I touched. Or does it really need to be Re-biased now with test equipment…? As for the method for biasing a fet in schoeps mic like NT1, would you have any info on the method ? Thx…
If you were to replace the input fet with a 2n3819 where would the zener diode go in the circuit?
So I'm presuming the zener is used to keep the Drain-to-Source voltage on the FET below 25V. To be honest I'm not totally sure it's going to be necessary. It all depends on how much the quiescent current draw of the emitter-follower stage pulls down the supply voltage to the FET. I think the idea of using a zener to protect the 2n3819 from excessive voltage is particularly applicable to circuit mods that remove the emitter follower stage and drive a higher-ratio transformer directly. I'll look into it further and let you know.
@@KarlAdamsAudio Thank you!
I just checked some voltages on that mic - the Drain-Source voltage across the FET was a bit variable, but eventually stabilized at 7.6V (I don't think I ever saw more than 12V there at any stage). It looks like a popular location for adding a zener diode to this circuit is across C12 - where I measured some some 20.5V. Starting to look like you'd be safe to substitute a 2n3819 without worrying about adding a zener.
@@KarlAdamsAudio Wow, thank you so much for your help!
Hey Karl, Im trying to bias a microphone a friend built thats a U87 clone. Where would I insert the Freq Generator? And where would I connect the Oscilloscope Probe? And do you disconnect the capsule when you do the biasing?
As I recall, I just connected to signal gen directly to the capsule side of the 1nF coupling cap that leads to the gate of the FET, leaving the capsule connected. That's not really right, though - so maybe don't do it that way I did. Better to get a capacitor of approximately the same value as the capsule (say 68pF or so) - desolder the capsule wire and connect the sig-gen via your 68pF coupling cap to the point where you just removed the capsule wire. In my case I just connected the scope to the output of my mic preamp, but equally you could pick it up from the driven lead of the output transformer instead.
I just built 2 U87 clones, and would like to custom Bias the FETs per each mic, but don't have a distortion analyzer.
On the various build threads they say to Bias the FET before the capsule is wired. I have an older Heathkit single trace scope, which should work fine. Need to borrow a signal generator though.
Anyway, so far this is the most comprehensive video I've seen on biasing a Mic FET.
I'm a little intimidated to try it on my own!
@@Wizardofgosz I wonder if it's feasible to simply use a computer audio interface for all the task - i mean it can output audio frequency signals, it has a mic preamp with 48V phantom power source, it can capture the mic signal, and soundcard oscilloscope software exists. You can drive a stereo output pair where one channel goes into the mic capsule input where you output 1KHz sine reference signal, while on the other channel you generate sine signal of same amplitude but maybe 50-100Hz and connect that to multimeter in AC mode to get approximate absolute values, and the output volume control will control both at the same time - with some tracking error, but i bet you can just disregard it.
It seems you could even just use an old MP3 player or PC onboard soundcard to generate signals, considering you have rest of test equipment.
I got a 15 year old TASCAM interface which is junk, but it was only a few $ and that's all i could afford, it works, and it'll have to do, because there is no way i can get access to professional test equipment - even just a better Behringer HD interface is a distant dream at this point. I am interested in the topic and will want to build a microphone circuit and adjust it myself, but haven't done yet. Is there something i'm missing?
@@SianaGearz I should think this is possible.
It would seem the manufactures neither optimized the circuit for either headroom or distortion! Would you in fact ever get as much as 1volt from the capsule in normal use?
Large diaphragm condenser capsule sensitivities are usually around 20mV/Pa, so a signal level of 1V would imply around 50Pa or approx 128dB SPL - pretty damn loud, but you'd easily see that sort of level from a close-miked trumpet, for example. The issue with optimization is the device-to-device variation in the FETs - they're not particularly tightly binned for Idss as supplied, and testing and selecting them prior to assembly (and/or choosing source resistors to match each FET's characteristics) is an very expensive proposition.
Thank you for your prompt and kind reply, I have subscribed.
Hello Can you make a video on how to fix a noise buzz on small condenser microphones. Its very usual to have this problems on studio recordings. Thanks
Oddly enough, you never specified what resistor value you used between R8 and C6 for the J508 constant current diode replacement at R10. Personally, I think that would probably be the most important part of this mod because it lowers distortion. Care to specify ?
Indeed, I did rather gloss over that point, didn't I? I have opened up the mic & confirmed that the added resistor was 510kΩ.
What other transistors can be used instead of the j508? Any diodes.?
How many volts should the tone generator be maxxed out sending tone?
Good question - from memory it's something in the order of a volt or two before the head amplifier clips.
Hi. i have a usb microphone, the JFET is soldered directly to the mic capsule . The orignal JFET is 2sk660, I want to replace it with 2n5457. How can I bias the FET?
I found a advice from 10 years ago but the picture is gone, can I add some resistor with the FET like schematic at 3:33 ?
Can you help me please?
Hard to know what to recommend without knowing more about the circuit - the typical transformerless Schoeps-inspired circuit has identical resistors in the source and drain legs of the FET, so changing the source resistor isn't the approach to take. Also, I'm wondering if your mic has an electret capsule, given the gate of the FET is soldered to the capsule without a capacitor to block DC.
Doug Ford's EEVBlog video on electret microphone design is an excellent resource:
czcams.com/video/UhG83WS51q8/video.html
I note that the 2sk660 incorporates the high value resistor from gate to source (it's a device specifically intended for this application) while the 2n5457 will need an external resistor (of around 1G). Also of note is the IDSS range of the 2n5457 is much higher (i.e. the lowest value for 2n5457 is double the highest value for 2sk660), which may be a challenge when selecting an appropriate resistor value between +ve supply and FET drain.
Thank you,It's very helpful tutorial! Could you please tell to which junction on the PCB you connect the oscilloscope lead, and you mentioned that you connected a resistor instead of R9, what value. I am trying to do the same thing to the V67G that I own. Thanks,
Thank you. The modification was to add an extra 510k resistor in addition to R9 so that together they form a voltage divider - as shown here:
www.dropbox.com/s/p8psqqicpsgm0fi/MXLV67mod1.jpg?dl=0
I installed mine on the back of the board (not exactly neatly) like so:
www.dropbox.com/s/nw60kyzoo6encdf/MXLV67mod2.jpg?dl=0
As I recall I connected the scope to the output of the preamp, rather than directly to the board - although I'm sure you could take the signal from (say) the transformer side of C8 instead.
Cheers & Best of luck with modifying your mic.
Thank you,
Hi Karl, I did most the changes you recommended and the mic sounds a lot better and more sensitive. Though I could not locate a seller for J508. Any replacement I can use instead? Thank you so much and sorry to bother you.
Good question - there's the 1N5307, but if anything that's even harder to find. At a pinch I've also used a selected J202 with gate shorted to source (but you need to test a few of them till you find one that gives around 2.4mA in that configuration). I originally bought my stock of J508 from Future Electronics, but I see they've got none in stock (and the minimum order quantity is 4,000 - eek!). There are some on eBay if you don't mind paying the asking price, which is a bit steep. I'd be happy to post you a couple from my 'stash' (I've still got a bag of them somewhere) also.
Hello, I actually have one of 1N5307, though when I used it did not seem to work very good. Possibly I did not hook it up correctly ( polarity configuration issue) or maybe because I changed the JFET to PF5102 which is working nicely. I live in U.S my email srifai1@hotmail.com. Not sure if you can post one of J508 to me to try unless you think I don't need it. I will pay you. It has been very fun project! Really appreciate your generosity and help.
beautiful video. can you also add a sketch of your connections
Thank you - I agree that a sketch would have made things clearer, perhaps I could touch on that in a later video. Cheers.
so, did you do the video?? if so could you provide the link??... I really liked this video, yet the final sketch with added info is clearly a must have, I am planing in building my own mic and it may be that if I buy the same resistors and other components that you used, I could skip the need of a multimeter and I could skip the need for those big machines that you were using and that for what I need could be considered total overkill...
Doing electronics without a multimeter is... naive, to say the least, i'm afraid. "Those big machines" can be relatively easily replaced with an inexpensive audio interface and free software, by the way.
hum nope!... there is nothing naive in the statement... if the guy who is making the video is already using a multimeter... why do you need to shadow at home the very same steps that were already seen on the video, that took the author such a time to properly create!... by the way I just tried to do a smaller version of the amplifier and it was a pain... the authors are tossing these videos out as if omitting info was a cool thing or as if it was some sort of standard... check my channel so you can see what I mean... the problem is a big one... check my video so you can see it...
Because components have *tolerances* - JFETs especially have pretty loose ones, compared to other components.
Certain information is "ommited" because it's sort of taken as granted that a multimeter is among the bare minimum tooling you need, to be able to do anything useful / meaningful in electronics.
If this was labeled as an electronics *tutorial* (which it's not), THEN i would understand your dissatisfaction.
Suppose your (first) circuit doesn't work (and there's a chance it won't). Without any test equipment, how can you even hope to troubleshoot it? Plain guess-work? Well, maybe except for the "shotgun" approach, where you just replace all the components (or just one at a time)...
Hi Tomxico here from Portugal. My studio condenser mic works just fine but sometimes then the signal is very low or even vanish away. Is there a bad capacitor or something ? It's not about fantom Power cos I have oneother condenser mic and works just fine. Can someone help me ? Please I am locked a home due to the virus. Heath to you all.
Thanks
I suspect you may have an issue with humidity and/or dust on the capsule. If it provides a leakage path for the polarization voltage, the mic will go quiet (or even silent) for several seconds until the polarization voltage builds up again. If this is the issue, it will get a lot worse when you breathe on the capsule. You can try leaving the mic in a warm environment to evaporate off excess moisture & that may help. Actually cleaning the capsule is quite tricky due to how easily it can be damaged, so I'd be wary of attempting that.
@@KarlAdamsAudio Thank you for your prompt response. You are right. A minute or 2 and the mic gets low signal input and just vanishes. Then after a minute or 2 come to normal db sound. but if working for a full hour or 2 gets normal all the time. Bad capacitor ? Or just the humidity in the capsule ? I will try to clean the capsule.
Thank you for your advise.
That does sound to be consistent with a dust + humidity issue. With a K67 style capsule, you have that ring of unsputtered mylar around the outside of the diaphragm - if you get too much dust on there, then condensation from the air will form a conductive path that will discharge your polarization voltage. Even without excessive dust, storing the mic in a cold and/or humid environment can cause similar issues. In extended use, the mic may get warm enough for this to no longer be an issue (this is especially true of tube mics, that have their own built-in heat source). If you have a dusty capsule, then breathing on it (close & open-mouthed, like you're trying to fog up a mirror) should cause it to cut out almost immediately. Cleaning a capsule is labour-intensive, as I understand the technique is to use a fine camel-hair artists brush and distilled water, and (using a bright light & magnification) remove individual dust specks with the tip of the brush.
not can i change r7 7 on any mic and replace it with that adjustable resistor for the bias?
because i have r7 on my mic but a little closer to the input of the mic
im trying to install a big capsule on an electret mic but it seems not to have enough power
Probably not - it's not safe to assume that the resistor named 'R7' performs the same role in the circuit. A suitable candidate would be a resistor in the Source leg of the FET, bit it's not necessarily the one called 'R7'.
Are you trying to substitute a condenser capsule for an electret? As that's not going to work due to the absence of the polarization voltage that condenser capsules need in order to function.
@@KarlAdamsAudio thanks for the info bud ..i really just need help to make an electret mic work with a big capsule ive been studing mics trying to find it out here the bias i see on the mic i hav does have a fet resistor but when i swapped the capsule it didnt work i tryed it on another mic and it worked beautiful i think the small electret mics would be a good shell to convert to true condenser i just need help designing a circuit to run piggy back and give it enough power to use a k67 capsule theres very limited.. info online ive been looking any helps appreciated