Active ribbon-mic

[abbey road d enfer]

Do you typically choose the surface mount or metal can packages?

The original PCB I made was with the TH version, because I'm not good with SMD's, but the price of the SMD version is nearly half, so I layed out a version that accepts both, but I never prototyped it.

Any noise/performance differences?

I know of one person who did a PCB compatible with both versions; he didn't notice any difference in performance.

 

[turnerk6]

I'm trying to simulate the input impedance but with the 10nF on the input to ground, I'm seeing the impedance go from 22k to near zero once it gets higher than 10khz. Is that just a weird part of the spice simulation? If I want to have switchable input impedance, the only thing needing changed is the 22k resistor at the input, correct?

 

[abbey road d enfer]

I'm trying to simulate the input impedance but with the 10nF on the input to ground, I'm seeing the impedance go from 22k to near zero once it gets higher than 10khz.

At 10kHz, the impedance of each leg is about 1.6 kiloohm, for a total of 3.2k. At 20kHz the total is 1.6k, which is within the rules for a 150 ohms microphone. Not "near zero".

Is that just a weird part of the spice simulation?

Probably finger fart.

If I want to have switchable input impedance, the only thing needing changed is the 22k resistor at the input, correct?

If you want a significantly higer Z, you should also decrease the input capacitors.

 

[turnerk6]

Okay, my simulation was showing around 700 ohms when I really zoomed in so I must be simulating it incorrectly. I'm charting VoltIN/CurrentIN.

 

[turnerk6]

Hey abbey road d enfer, could you take a look at this and make sure I'm thinking about this right? I'm thinking about being able to switch input impedance of a 30dB boost (600, 1.4k, and 6.8k). I'll attach my simulation with the 600 ohm resistor on the inputs. For the 1.4k and 6.8k, I would just replace the 600 with 1.4k and 6.8k?

 

[abbey road d enfer]

For some reason I can't open this new file. Try re-posting.

 

[turnerk6]

Try that one. The electrolytics may change from uA to some other weird prefix during the .txt conversion process. That one is the 600 ohm impedance version 30dB. For the 1.4k, I would replace the 600R with 1.4k and likewise for the 6.8k impedance mode.

 

[abbey road d enfer]

The electrolytics may change from uA to some other weird prefix during the .txt conversion process.

Yes it happens sometimes. Just a benign nuisance.

That one is the 600 ohm impedance version 30dB.

Impedance should be measured with a .ac command. With a .tran command, you have a "divide by zero" that happens every half-cycle.
Beware that your IN+ and IN- labels are not connected.
You should use the SI prefixes: instead of Maximum Timestep ".000001" you should type "1u". It is more readable and less prone to mistakes.

For the 1.4k, I would replace the 600R with 1.4k and likewise for the 6.8k impedance mode.

If you use the circuit in unbalanced mode, the input impedance at LF and MF is governed by the input resistor.
If you use it balanced, the actual input Z is twice.

 

[KZhack14]

Hello everyone, stumbled upon this thread as I was looking to modify some booster circuits for different dynamic capsules as a side interest.

I recently bought a DM1-styled inline booster (not mentioning the brand) from a local market just to take it apart and see the circuit inside. Apart from the 10uF Capacitors are there to block DC I haven't traced the circuit out, but I can;t quite understand why the K170 FETs are oriented in reverse of each other. There seem to be a 1M Ohm resistor or something in between + and - input (correct me if im wrong) to do some kind of impedance matching? It also looks like the LED indicator is soldered to the ground.

From first impression this circuit looked like someone may have taken inspiration from the circuit designs mentioned in this thread and simply added a little bit of extra stuff.

My plan is to repurpose/rebuild this circuit with 1-2stages of high pass filter pre-boost and RF filter at the end of the circuit to clean up the signal and perhaps swap out the caps for better power block/filter.

 

[abbey road d enfer]

I can;t quite understand why the K170 FETs are oriented in reverse of each other.

They are used as a differential pair, so tehy are connected symmetricaly. Positioning them facin each other just makes layout easier.

There seem to be a 1M Ohm resistor or something in between + and - input (correct me if im wrong) to do some kind of impedance matching?

Not impedance matching. Just providing the necessary DC path for correct operation.

It also looks like the LED indicator is soldered to the ground.

It may be. You need to trace the circuit.

From first impression this circuit looked like someone may have taken inspiration from the circuit designs mentioned in this thread and simply added a little bit of extra stuff.

It is probably their take on the original FetHead, which is known to have issues with excessive voltage across the FET's, which results in premature faileur and excessive noise. Maybe thay have solved this problem with some of the additional components.

 

[KZhack14]

They are used as a differential pair, so tehy are connected symmetricaly. Positioning them facin each other just makes layout easier.

Not impedance matching. Just providing the necessary DC path for correct operation.

It may be. You need to trace the circuit.

It is probably their take on the original FetHead, which is known to have issues with excessive voltage across the FET's, which results in premature faileur and excessive noise. Maybe thay have solved this problem with some of the additional components.

I see. Thanks for the opinions. I'll attempt to trace it & draw the schematic out so I can get more insight soon.

 

[KZhack14]

Hello everyone, stumbled upon this thread as I was looking to modify some booster circuits for different dynamic capsules as a side interest.

I recently bought a DM1-styled inline booster (not mentioning the brand) from a local market just to take it apart and see the circuit inside. Apart from the 10uF Capacitors are there to block DC I haven't traced the circuit out, but I can;t quite understand why the K170 FETs are oriented in reverse of each other. There seem to be a 1M Ohm resistor or something in between + and - input (correct me if im wrong) to do some kind of impedance matching? It also looks like the LED indicator is soldered to the ground.

From first impression this circuit looked like someone may have taken inspiration from the circuit designs mentioned in this thread and simply added a little bit of extra stuff.

My plan is to repurpose/rebuild this circuit with 1-2stages of high pass filter pre-boost and RF filter at the end of the circuit to clean up the signal and perhaps swap out the caps for better power block/filter.

Ok. I've attempted to do a visual trace after a long time not practicing this. Excuse the messy drawing. A couple of things stumped me regarding the circuit:
1. First glance this seems like a pseudo balanced circuit. And the R1 resistor in parallel to FET source pins to ground had me confused to its supposed function. Also center pin out should be labeled pin 3 (-) not pin 2.
2. Either my eyes are deceiving me or the colors on R3 & R4 are not accurate as the color code should give 30k & 300k ohms respectively yet I'm measuring only 15K Ohm for both resistors.
3. Is R2 functioning something like a HPF before any gain happens? Also thought that R7 & R6 may be functioning like a voltage divider to send the correct voltage to power the LED.

I've measured the preamp with OSM if anyone's interested. Seems like a pretty linear boost with roughly 20dB self-noise when reference channel is silent. (Measured this with pink noise going out into my interface at 0db output & 0db gain)
Any thoughts on this circuit are welcome, I've a feeling I may have traced certain parts wrongly even though I triple checked.

 

[abbey road d enfer]

1. First glance this seems like a pseudo balanced circuit.

Not pseudo, it's symetrical enough to qualify.

And the R1 resistor in parallel to FET source pins to ground had me confused to its supposed function.

Nothing unusual here. Just like a vacuum tube uses a cathode resistor, FET's use source resistors.

Also center pin out should be labeled pin 3 (-) not pin 2.

Correct.

2. Either my eyes are deceiving me or the colors on R3 & R4 are not accurate as the color code should give 30k & 300k ohms respectively yet I'm measuring only 15K Ohm for both resistors.

R3 &R4 are supposed to be identical. However their value - as well as R5, R6 & R7 - seem to be pretty high. I'm not sure the FET's operate in their best performance zone.

3. Is R2 functioning something like a HPF before any gain happens?

No. It's here just to define the DC resistance of the input.

Also thought that R7 & R6 may be functioning like a voltage divider to send the correct voltage to power the LED.

If my amended schemo is correct - which I'm far from certain - these resistors are part of the chain that powers the FET's. The LED is powered by R8.

Any thoughts on this circuit are welcome, I've a feeling I may have traced certain parts wrongly even though I triple checked.

What about this revised schemo? Seems more correct to me, although the LED and its resistor break the beautiful symmetry.

 

[KZhack14]

Thanks Abbey. Your correction was most helpful. I'll find sometime within the week to do some multimeter probes to verify the connections and update the schemo

 

[KZhack14]

Alright. I got the schematic probed. Pin 2 is indeed connected to the + of 10uF cap. So I'm going to attempt a guess at how the circuit works:

1. Although output is balanced, it seems like the designer took an asymetrical approach as to how the phantom power is routed. If I'm guessing, the FET's are getting about 10-12V & 1-2mA current after going through all the resistors from pin 2. Pin 3 is supplying most of the power required by the LED. If the specs on the FETs are accurate, then the power given is operating way under its safety limits.

2. The circuit seems to be made to amplify an unbalanced signal (ie the +ve side of input) as the -ve is completely grounded. Im uncertain if this has an undesirable effect on SNR/CMRR of the signal.

Before I attempt to add any additional filters to the circuit, I'd like to first attempt improvements to the SNR/Noise floor of the circuit, probably by replacing the grounded caps & move around some resistors to make the circuit symetrical. This is what I came up with:



Can anyone kindly point out any flaws with my modified circuit if any, and maybe suggest more efficient designs other than the above schematic.

 

[abbey road d enfer]

1. Although output is balanced, it seems like the designer took an asymetrical approach as to how the phantom power is routed.

That's right and is definitely a flaw.

If I'm guessing, the FET's are getting about 10-12V & 1-2mA current after going through all the resistors from pin 2.

Actually they are operating at less tha 0.5uA, which is probably the main reason for the poor noise performance, in conjunction with the 750r source resistor.

2. The circuit seems to be made to amplify an unbalanced signal (ie the +ve side of input) as the -ve is completely grounded.

Corerct. I had not noticed it. It does not prevent the thing to operate. Seems to me like it's the same PCB as the guitar version, with its 1 Megohm input resistance. It sure works for a mic, but is not optimal.

Im uncertain if this has an undesirable effect on SNR/CMRR of the signal.

No. The LED and the way power is derived takes good care of ruining CMRR.

Before I attempt to add any additional filters to the circuit, I'd like to first attempt improvements to the SNR/Noise floor of the circuit, probably by replacing the grounded caps

The main improvement will be decreasing the source resistor to a much lower value, which would result in significantly increasing the operating current. 2SK170 wants to operate at 1mA minimum, 2-3 even better.. I would start with 68 ohms. The problem then is that you'll also need to decrease the 10k resistors.

& move around some resistors to make the circuit symetrical.

You'll need to add gate resistors to each FET. I typically use 22k, but the value is not critical.

This is what I came up with:

Pls find my suggestions.
From the start, the topology is not optimal.
It solves the two main issues of the original FetHead, but adds two other issues, poor noise performance and poor headroom.

 

[KZhack14]

Correct. I had not noticed it. It does not prevent the thing to operate. Seems to me like it's the same PCB as the guitar version, with its 1 Megohm input resistance. It sure works for a mic, but is not optimal.

No wonder. I'm suspecting that because where I got this from this booster has 2 variants: the other one taking 1/4" input. If so, can this circuit effectively be rehoused as an "Active High-Z instrument DI" or by cutting the phantom power rails from pin 2 &3, replace power source from a separate DC input, this circuit can be rehoused into something like a FET Booster pedal?

No. The LED and the way power is derived takes good care of ruining CMRR.

The main improvement will be decreasing the source resistor to a much lower value, which would result in significantly increasing the operating current. 2SK170 wants to operate at 1mA minimum, 2-3 even better.. I would start with 68 ohms. The problem then is that you'll also need to decrease the 10k resistors.

So are you saying the LED with its resistor is contributing to the noise factor as it can't be effectively shunted to ground? So now I understand the FET's need more current to operate as I was reading up their spec sheet.

You'll need to add gate resistors to each FET. I typically use 22k, but the value is not critical.

So if I'm understanding this based on some of the common source Amp schematic I found, Rg is basically placed in parallel from V+ (which is connected to FET gate) to V-. Is my 220k resistors doing the same thing? It sounds like 22k is the common mid point for best noise to performance ratio?

Again thank you Abbey so much for taking your time to enlighten me on the circuit. I'll be getting my hands on the FETs to start breadboarding this soon.

 

[abbey road d enfer]

No wonder. I'm suspecting that because where I got this from this booster has 2 variants: the other one taking 1/4" input.

That's what I suspected.

If so, can this circuit effectively be rehoused as an "Active High-Z instrument DI" or by cutting the phantom power rails from pin 2 &3, replace power source from a separate DC input, this circuit can be rehoused into something like a FET Booster pedal?

It's a possibility. Now, I'm not sure the circuit would be good with only 9V... It would take some additional mods.
Actually, a different topology should be used for that. For example, an unbalanced circuit with impedance-balanced output.

So are you saying the LED with its resistor is contributing to the noise factor as it can't be effectively shunted to ground?

No. I say it's not good for CMRR. If the cable length is small and/or EMI/RFI is low, it shouldn't be an issue.

So if I'm understanding this based on some of the common source Amp schematic I found, Rg is basically placed in parallel from V+ (which is connected to FET gate) to V-.

Do you mean in+ and in- ?

Is my 220k resistors doing the same thing? It sounds like 22k is the common mid point for best noise to performance ratio?

There is almost no difference between 22k and 220k in this position. The advantage of 220k (or even more) is that it increases the input resistance. However, due to the large capacitance of the FET, the AC impedance is not so good.

 

[KZhack14]

Do you mean in+ and in- ?

Yes yes, on the (300k) Rg side (pic copied from EE Stackexchange forum)


Anyway, based on what I can gather from you so far, I'll try the breadboard with a pot for my 220k resistors to see if that affects anything. And I'll probably add a 10-22nF cap to shunt XLR pin2 & 3 out for RF/EMI rejection.

 

[KZhack14]

Ok after a month or so I finally got my hands on some 2SK170BL from China. See the attached pics for the completed mod. On first round of measurement the noise floor did not change much, but I did see a 3dB increase gain running some pink noise.

However after a short while when I repeat the measurement test, I'm hearing some sputtering kind of noise. Could this be a result of mismatched FETs? (For context I broke the G leg of the right transistor TWICE, and finally settled on the closest FET I have on hand. Original left transistor hFE measures about 442-445, right replacement transistor measures somewhere around 436-437)

Also a question regarding DC isolation came to my mind while doing this. Since the polarized caps are there to block DC, can I effectively replace them with an iso-transformer for the same function? I'm thinking of taking either the schematic from Alctron MA-1/Cloudlifter with cascoded FET or the FEThead circuit and mod it to something like this:



Btw, can anyone tell me the differences between the current versions of 2SK170BL and 2SK170GR (I found these variants on the platform I shop)?