KM84 circuit question

I am curious, when you set the operating point of the fet (via resistor from source to ground) in the KM84 circuit, are you loosing something by increasing that resistor? I guess what I am asking is, is there a real benefit to finding the fet that will operate correctly with a 3k9 resistor v.s. increasing the resistor to fix the operating point?


Thanks

Increasing the source resistor will reduce the current through the FET and reduce the load driving capability as well. This is a problem if you are driving a transformer directly. If you add a second stage, a source follower of 2 to 2ma you don't have to worry about the reduction in current of the first stage. I have published a circuit like that some time ago.
http://www.groupdiy.com/FORUM/viewtopic.php?t=5875&highlight=mic+amplifier+idea&sid=e6c71572ac83abbb9686cbe6faa5833f

I have seen 2.2K to 10K in the older u87s I have worked on.

3.9k is not set but it might be the best compromise value for the simple fet gain stage: however you will note on the u87 schematic the drain is speced at 10V. The idea is to change the Source R to get the drain at 10v. The current in the drain leg will be the same as the current in the source leg.

Now the source R sets the input level to the FET it should not go beyond the voltage drop across the source R. I noted the PDF "microphones" show the km84u87 type fet circuit fragments with a 3.9K. I did some thinking and also noted the u47 has a-1.1V bias the U67 a -1.6V to -1.8 V bias etc..... I think the IRT/neumann whoever that designed the circuit picked bias points for a few reasons.

The 3.9k with the drain at 10V has about -.9something as the bias for the fet.

Also note the AT4060 thread about how hot a UNITY gain circuit can be with enough capsule voltage for capsule sensitivity. Schoeps are less than 1 or +3db depending on a jumper looking at a schematic. I think the TLM103 is a gain of one.

The above microphones(not km84 circuit) are <= x1 and are noted for the hot output level I reasoned I want about 1 volt or more input headroom to the FET circuit. Again if the Source R is a higher value for the drain to be at 10V the fet can not turn on as much in respect to source voltage side of the signal swing. So I think 3.9K source R and 10V drain V might be about the best compromise for input headromm and output signal swing.

Now I have worked on a stock early 70's u87 that had a factory installed 10K source R I would have liked to compare it to another with a lower value to try to tell if the 2nd harmonic was stronger in the 10K one at high signal levels. If anyone remember PRR and myself posted that maybe the at4060 circuit was designed for a "sound"

bias point bias point............................................... harmonics harmonics...............................

thanks guys.. I'll chew this all over and probably ask more questions.... I know how to build the circuit, but now I am trying to really understand what's going on. Thanks for the help.

I was hoping someone would question what I wrote. Does anyone see any mistakes in my reasoning?

The stock KM84and U87 http://recordist.com/ampex/schematics/neumann/u87sch.jpg
have gain controlling networks
(so do the KM54,3 if I understand the schematics http://members.tripod.com/~coreyeng/km54a.PDF ) so maybe with the stock circuits it does not matter as much.

Question should the charge voltage modulation via the .47uf cap from the drain in the u87 be called a form of feedforward.

I am working on a km84 circuit right now. I am actually attempting to bias the FET on a breadboard. I am applying phantom power into R9/R10, and I have stuffed R6 (56K), R7(10K), and R4(47k). Drain connects to R4, 1G from gate to ground, and 3.9k from source to ground. I measure 1.5V from source to ground on 3 FET's so far, but drain only measures 1.6V. Changing the source resistor to 2.2k gives source 0.8V, and drain 0.9V. Either I am measuring the drain at the wrong point, or I don't have something connected correctly. I have 20V before R4. How do I get a good Drain measurement?

Thanks

Drain resistor should be 4k7.


EDIT: This is incorrect, 47k is the correct value (see below).

Insomniaclown

what schematic are you using?

something to read
http://www.datasheetcatalog.org/datasheet/vishay/70595.pdf
Also what JFETs are you using?  What is the IDSS spread and other parameter range?

I test then select JFETs for the circuit the JFET will be used in

You need to use ohms law to figure out what Id the KM84 circuit JFET operates at, you need a higher IDSS at the circuit D to S voltage than Id

I got 3.9K to work in KM84 like circuits by testing JFETs from a bag of 1000 JFETs to find some that work at the values of a circuit fragment in the PDF "Microphones".  I did this to check if maybe 3.9K is a hint.

I have learned more over the years about JFET microphone circuits so my older and newer posts should be questioned.

Thanks Magnetosound. I thought that might be the case, but wasn't sure.

Gus- Thanks for all the info! I have some reading to do. I have read through the pdf Microphones, but mainly looking at capsule polarization. I am using the 2N3819 for the FET.

Schematic attached.

Just checked my BOM for my km84s and the relative threads, 47k for drain. When I was trying to select a 2n3819, initially I tried a stripboard test jig, the voltages didn't correspond to what I expected (I used a bench power supply) but i was still able to sort the fets into best to worst order. I went ahead with the circuit, temporarily wired a trimmer for the source R. I kept going back to that thread where Gus said "come on people, drop in a trimmer and adjust for 10v.." or something to that extent.

check here for Bias method
http://recforums.prosoundweb.com/index.php/t/9790/0/

goodluck
Matt

Here are some measurements with a couple of different source R's. The KM84 schematic I posted states the current draw at 0.4mA as well as some other measurements. I am wondering if I should also stuff the capsule polarization part of the circuit as there is a feedback capacitor there that connects to the drain leg. Might allow me to get the measurement I seek.

R4-47K
Source R=2.2K
Current- 0.394ma
Voltage across source resistor- 0.86
Phantom in- 45.9
Voltage before R7- 23.7
Voltage before R4- 19.6
Voltage at drain- 0.9
Voltage drop across FET- 1V

Source R=3.9K
Current before 56K- 0.388mA
V across SR- 1.53
Phantom in- 45.9
V@R7- 24
V@R4- 20
V@ Drain- 1.606

Source R-3.9K
R4-4.7k
Current- 0.595mA
V across SR- 2.33V
V@R7- 11.58
V@R4- 5.33
V@Drain- 2.5

Tested 7 2n3819 with negligible differences between them. According to the schematic I have, most of the measurements are very close. I tried the 2.2K source R just to see. I think if I add a couple hundred ohms to that, I can get the 0.95V @ the source. I was thinking of running up to the electronics store to grab a couple more FET's to test. The 10 I have are from digikey. Whenever I have ordered packs of 10, transistors tend to be consistent in measurements. However, packs from the electronics store vary much more. Going to figure out this IDSS.

The 47k seems to be the right value for R4. I remember reading about 2 errors on 2 different schematics for the km84. One was R6 being noted as 5.6K when it should be 56K (I think that was you Gus), and another where R4 was listed as 4.7K when it should be 47K, which is this schematic http://www.sdiy.org/oid/mics/Neumann_KM83-84-85.GIF.

I took a look at that link MatthisD, but I don't really have a way to measure distortion. It's a bit over my head to tell you the truth. After testing some FET's from the store, I was just going to build up the circuit, and see how it sounds.

21.5VDC -10VDC =11.5VDC

11.5VDC/47K =about .24ma

Don't worry about the source voltage adjust the source resistor value for about 10VDC at the drain to start.

Read the link MatthisD posted

Insomniaclown said:

I took a look at that link MatthisD, but I don't really have a way to measure distortion. It's a bit over my head to tell you the truth. After testing some FET's from the store, I was just going to build up the circuit, and see how it sounds.

Click to expand...

Hi again Ryan, I spent too many weeks in the forum scratching me head about drain and source voltages. It just made much more sense once I built the entire circuit and stuck a fet in with the 10k trimmer (I actually used 25k). Then replaced with a 7.8K resistor in the case of the selected 2n3819.
I didn't measure the distortion as such, I applied a 1K signal into the gate capacitor and used a scope on the drain, adjusted the trimmer until I got maximum gain, kept going until it began to clip, I went back and forth a few times between max gain and first sign of clipping and settled for somewhere in between. This could be a crude method but I got the Mic fired up with healthy gain at least. I'm now more worried about putting the dremel tool to the capsule venting!!
Matt

Insomniaclown said:

The 47k seems to be the right value for R4. I remember reading about 2 errors on 2 different schematics for the km84. One was R6 being noted as 5.6K when it should be 56K (I think that was you Gus), and another where R4 was listed as 4.7K when it should be 47K, which is this schematic http://www.sdiy.org/oid/mics/Neumann_KM83-84-85.GIF.

Click to expand...

Sorry, my bad - that is the schematic I was looking at! I happened to have the mic folder open when I read your OP and posted a quick reply. When I saw the proper schematic I remembered the 47k from the U87 and knew I'd got it wrong. 

(The 10k(ish) trimmer works well, but best remember to put about 1k in series so you don't short the source to ground by accident.)

Ok. Thanks very much for the info guys. I'll get that trimmer in place and adjust for 10V at the drain. I guess I was too focused on the source voltage.

One issue for me is I don't own a scope. It would make things much easier if I did, but I may be able to wrangle something up using my DAW and some DC blocking caps. I have been meaning to get a simple tone generator.

Re-read the link MatthisD posted. Makes more sense on the second time around. So I'll set a trimmer for 10V at the drain, and then recheck once I have the mic constructed for drift, as well as listening for noise, distortion and such. Not going to be scientific, but should be fun!

Edit: Wow! Holy finicky batman! I started off with a trimmer, but after awhile, it was becoming troublesome, so I just made source R 10K, and measured drain voltages for each FET. Now they are all over the place ranging from 4V to 18V. Then I grabbed one of the 18V FET's, and got the drain to 10.22V using a 6.8K source resistor. So it seems that source R must be within a certain range to even get the FET to work at all, which explains every FET in my bag measuring the same at 3.9K. I also discovered that decreasing source R causes the drain to drop off sharply, which means the range is very small for adjustment. Now I understand why you have such an issue finding FET's that bias with a source R of 3.9k Gus. Very interesting.

I am going to go through the rest that showed a voltage over 10V @10k to see how many I can get to bias to 10V drain. Then I'll save all my pairings for the future.

Hi Ryan,
You can try Klaus Heine's method (it really works well): replace Rs with 10k trimmer, connect AC voltmeter to drain, apply 1kHz tone to gate and find maximum output level.
You can use soundcard (DAW) for this task, but remember about impendance matching. I'd measure THD and other parameters after the microphone's output transformer (ratio 7:1).
Another hint : to get same gain as mikes have, you can increase feedback capacitor (4pF) to 5.1pF.

Thanks Bezen4uk for the explanation. That makes more sense. Might give that a try today if I find time. Think I am going to do some work on the microphone body today. I have a nice piece of brass to chop up. Thanks for the tip on the feed back capacitor as well! Much appreciated.

Edit: Just thought of 2 other things after I posted. The schematic I posted shows the input cap value as 470pF. I was planning on making this 1000pF as this is more common in LDC mics. I have some 470pF caps, so I might try it out and see what happens. I am going to venture a guess that the frequency response will be more limited with a 470pF cap?

I am using a tab T49 (10:1) transformer for the output, and an RK 47 microphone capsule in this mic. I know the 10:1 will shave off a couple of db's, but that's OK with me as long as I can get the FET working to it's potential. I am going to try this mic with the RK 47 capsule first, and then possibly try out a 797 audio k67 capsule that I have. Would love to hear some opinions on a brighter vs flatter capsule with this circuit. I always love talking gear!  ;D

Edit #2- My best FET measures 10.5V drain and 2.1V source with source R 6.81k. Not perfect, but a good starting point. I think I'll get s few more (a million or so) to see if I can get closer. Now, where is that hacksaw.

Insomniaclown said:

Edit: Just thought of 2 other things after I posted. The schematic I posted shows the input cap value as 470pF. I was planning on making this 1000pF as this is more common in LDC mics. I have some 470pF caps, so I might try it out and see what happens. I am going to venture a guess that the frequency response will be more limited with a 470pF cap?

Edit #2- My best FET measures 10.5V drain and 2.1V source with source R 6.81k. Not perfect, but a good starting point. I think I'll get s few more (a million or so) to see if I can get closer. Now, where is that hacksaw.

Click to expand...

..haha, I have a 470pf as specified, in one and a 1000pf polystyrene in the other. No huge differences between the two but i've got ceramic feedback caps which I will change to polystyrene and also the lone 470pf to 1000pf. The feedback capacitor value is a tricky one, I would like to record anything from loud overheads in a small room to softly-strummed acoustic guitar.

Similar to yourself; 10.5v drain 1.45v source from a variety of 30 or less 2n3819

You're not using a trimmer for the source resistor?

Hey MatthisD. Well, I started using a trimmer, but then my big hands made it difficult to get it dialed in, so I started just swapping in resistors until the drain was what I wanted, then checked the source. It worked. Although, I do plan on grabbing another 10 to 20 2n3819's on monday, so I'll try the trimmer again with those. I'll just make sure to have less coffee. I guess it would be a good idea to just use a trimmer in the circuit as well for fine tuning.

Thanks for the info on the input and feedback capacitors. I think I'll be using this bad boy mostly for vocals and acoustic guitar, but I know I'll be putting it in front of a cranked amp.

Ohms law

You can add resistors together in parallel for a non stock value

About trimmers. You can take a 10k and add a 4.7K fixed resistor in series  For 4.7k to 14.7K
Or you could use a 5K trimmer and a 4.7K for 4.7K to 9.7K and have finer control of the trimmer

A simple setup for trimming a simple gain stage like the 84 uses could be a few switchable fixed resistors (1k,2.2k,3.3k,4.7k, 6.8k) selecting one at time with a rotary switch, in series with maybe two switched potentiometers 5k OR 10k SPDT

Remove the trim circuit and measure it, use two or more resistors in parallel to make a non standard value if the trim circuit measures at a non standard value.

To do DIY stuff in electronics OHMs law and power equations are useful

I googled "ohm's law resistors in parallel" here is one link.
http://www.physics.fsu.edu/users/ng/courses/phy2054c/Labs/Expt02/Expt-02.htm

10k in parallel with(||) 18k about 6.4k
10k || 15k about 6K
...