Need help with ROG: fetzer valve (gain stage) + buffer

[Alexius II]

Hello everyone, my first post here on SS guitar! 

I would like some opinions about a thing I'm trying to do.

I have a big tube amp (rack), but it's quite bulky.
Using it every time I have an idea and would like to record it... well, too unpractical
(takes too much time to plug everything, to warm up tubes, to set the microphone etc.)

I decided that for these quick ideas a "direct" recording would be the way to go (guitar into pc + some amp simulation software)
Since my soundcard does not have a high impedance input, I have to make one
(input on soundcard is unbalanced line-level)

So, here is the idea: I would like a high impedance input + some voltage gain. Since I like the sound of the "Fetzer Valve", this part is covered. I would also like a low impedance output, so I need a buffer.

I did these schematics in DIY Layout Creator, so you can see exactly what I'm doing:


Then, my idea was to simply join them together like this:


And now: the questions:
1. Would this work? (why / why not)
2. How can I improve this circuit? I've got Fetzer valve figured out, but I have no experience/understanding of buffers.
3. Do I really need the in/out capacitors in the buffer? How to they work (what do they do)?
4. Do I really need the 1M gate-source resistor in the buffer? Does it add any noise?
5. I like the 2N5457 in the fetzer valve, but would the MPF102 be better for the buffer?


Looking forward for your help... and thank you in advance

[mensur]

Hello Alexius,
Welcome to the forum 
First of all you don't need so much components in your preamp.
I suggests to build DC coupled preamp,but here's the answers to your questions:
1.Yes, it would work,
2.By building DC coupled preamp, which also has voltage+current(buffer) gain.
3.Yes you do, capacitors have infinite resistance on DC voltage, hence they stop DC voltage to your pot(input buffer cap),and output(out buffer cap),also they shape frequencies.
4.In that kind of biasing, which is called voltage divider bias(or AC coupled follower), you polarize the gate to half of supply, which means that if you have +/- 10VAC signal, it will pass unaffected(unclipped),without it, if you have +/-2VAC signal it will clip the buffer.
5.Generally, more current, better the buffer, so none of them are good as buffers, but you can try 2SK170.

[rowdy_riemer]

I wouldn't use the Fetzer valve topology for a buffer. A simple op-amp buffer would work great. A MPF102 source follower would be fine two.

[Alexius II]

First, thank you very much for your answers! 

Ok, I've read some articles to learn what is this DC coupling you mentioned.
I think I get the basic idea... but since I learn as I go, more help will be needed 
(I study something completely different, so this is just my free-time "hobby")

So here goes (and correct me if I'm wrong):

Is this ok, or must the voltage divider (1M + 1M) of the buffer also be removed?
(there must be no clipping in the preamp!)


And now, the new set of questions:

1. First, you suggested direct coupled preamp. Why is that? Is the sound better?
 (cost-wise there is no reason for leaving something out)

2. You said that less components are needed? Which ones would you omit and why? Did you mean those DC blocking capacitors and the pot?

3. If I understand correctly, now I can't just put a "volume" pot where it was. How would I lower the signal (voltage)? At the end of signal line? I would like to have some sort of "volume" control, because I'm affraid with some guitars (with active pickups) the output will be enough to clip my soundcard input. On the other side - with ten times lower output vintage pickups I may want to have all the gain I can.

4. And the final one: I only have these transistors available: j201, MPF102 and 2N5457. Which would make the best buffer? I found this page: "jFet bias calculator". I can measure both Vp and Idss of my transistors, but I do not understand the importance of Vrs or Vgs. 

[mensur]

Omit 1M + 1M voltage divider, cause you've already have DC voltage on the gate of a buffer.
1.You don't mess up the frequencies, simpler the deign, greater the tone.
2.Yes, I meant those DC blocking capacitors, pot and the divider.Cause we want to make it simpler.
3.After the 1uF(use wima MKS4 63V poly cap),put 10K log pot.If you use active pickups, you don't need buffer, hence you already have low output resistance,and decent voltage gain.
4.Measure the current of each FET, you will do that, if you tie gate and source together, and conect - of 9V battery to them, and drain of the FET tie to the mA meter and +9V.

(I study something completely different, so this is just my free-time "hobby")

Tell me about that, I finished Faculty of sport and physical education, on the department of sport management, and higher degree of tennis trainee

[Alexius II]

Ok, the voltage divider is gone.



I understand if the audio signal passes through a cap, it has to be a good one.
For that 1uF I will pick something nice, polyester or possibly polypropilene.
It doesn't really have to be polarized, right? (like in the previous schematic)

The only thing that worries me now is possibility of clipping in the buffer.
How would one calculate that?

I read that high power (passive) humbuckers can produce up to 1 volt (peak).
On the Fetzer Valve website the calculator says I will get a gain of 5 (or 14dB).
So if I understand correctly, the buffer has to "withstand" a swing of +/- 5 volts?
Is this possible without the 1M/1M voltage divider?

Ok, I will go and test my FETs for current.
Does it have to be with battery, or can I use something else?
I have one small power supply I used for testing some guitar pedals (9v transformer, caps, LM7809 regulator).

Thanks again, man... you are very helpful
Faculty of sport you say? I'm at Faculty of civil engineering 

[mensur]

The buffer will not clip cause you've done the same thing with DC coupling as with voltage divider.
The best thing for you is to take 2N5486 FET's, for voltage gain, as so as buffer.
Fet are just like the triodes(but only enhanced one's)
I will give you the example:
Lets say that our 2N5486 have Vgs off -3V, and 12mA of current(you can find that info on the datasheet, transfer characteristics),
We want that maximum current consumption be 10mA(we get that from Ohm's law: 18V/10mA=1800R or 1.8K, that is our Rd)
We want linear operation,so we chose mid point of VGS(off): -3 / 2 = -1.5V,and Id form the trans.characteristics is about 3.5mA, so Rs will be -1.5V / 3.5mA= 428R or, 430R.
We get voltage gain from this formula:
Av = gm x Rd (bypassed Rs, for Cs use 10uF el.cap, corner freq will be 37Hz),
gm will be gm= gm0 x ( 1 - (VGS / VP ))
gm0 will be: gm0= 2xIDSS / VP
So,
Idss=12mA
Vp(or Vgsoff)=-3V
gm0= 2xIDSS / VP=2*12mA/-3V=0.008mS
gm= gm0 x ( 1 - (VGS / VP ))=0.004mS
Av=0.004mS * 1.8K = 7.2x or 17.15dB's

[Alexius II]

Wow, you've given me a lot of theoretical knowledge. I now have to slowly digest it

I think now I understand the DC coupling effect. And for your example - that is VERY helpful. I always felt that one could learn a lot from this kind of examples, even without serious theoretical background.

Ok, so if I understand your calculations are for the "gain stage" of this preamp? (replacement for "fetzer valve"?)
I went through everything again and learned how to modify parameters for other transistors, using Idss, Vp and other relations from datasheet. This should not be a problem any more.

The only thing left that I don't understand is, how to determine the Rs of the buffer. The most common values I found in other designs were between 3.3k and 10k. Can this be calculated to something more optimal? How to do it? What should be the Vgs (or Vrs)?

Thanks again for all the help! 

[mensur]

No problem bro,
Yea, the calcs are only for gain stage only.
The value of Rs on the buffer the determine voltage gain of the buffer, as so the output impedance of the buffer, but that is negligible.It also determines quiescent current of the FET,just as s regular Rs.
In our case the Vgs is higher than the Vgs(off), which is about 10V.
The gain of the buffer is:Av= (gm x Rs) / (1 + (gm X Rs))
Output impedance is:Zo= (Rs x (1/gm)) / (Rs + (1/gm))
Put 10K and you will be fine.

[Alexius II]

Great, so this corresponds to what I already knew: gain is almost unity and output impedance around few hundred ohms (for buffer).

I went and measured a few of my 2N5457 and MPF102 transistors.
Most of 2N5457 were around Vp:-1.5V, Idss:2.5mA
and MPF102 were quite different: Vp:-4.5V, Idss:13.5mA

So I suspect MPF102 is better for my application (more current)?
One problem though, I couldn't find any graph of Id/Vgs relation for it (datasheet doesn't have it).

Is there a way to calculate/draw it myself?
I found this equation on wiki (for linear region):

...but I dont know what is Vds. I suspect this is hard to do without measuring..?

I need to have something like this, right?
(example for Vp:-4V and Idss:14mA)


Is it ok to just approximate? 

[J M Fahey]

VERY short answer:
Use
Rs:4k7
Rd: 10K with 9V battery or 15/22K with 18V battery
Do not use a capacitor in parallel with Rs, because you do not need too much gain here, you want to adapt levels and impedance, this is not a full-blown preamp. 2x to 3x will do.
As a buffer use an NPN bipolar transistor, with the same passive components. any general purpose audio will do (BC546, etc.)
Use a regular 1uF electrolytic, you will hear *no* difference whatsoever, especially going into a PC soundboard.
Of course Mensur's calculations are correct, but I fear you find this complicated and give up, unnecessarily.
Good luck to both of you.
(And thanks for posting in English and not in Ciryllic, he he)

[Alexius II]

He he, that IS a very short answer 

So you ment with 18V supply I should use Rs:15k and Rd::22k?
(I didn't fully understand that sentance, sorry)

For which transistor would this be?

About that capacitor: as far I understand the Cs (in parallel with Rs) has an effect on increasing gain... and also acts as some kind of frequency filter (which I don't understand). If I can get around 2x or 3x gain without it, as you recommended, that is probably fine by me.

BC546 is something easily available, so I will take it into account. Using bipolar NPN transistor would further lower the output impedance, right? Some other benefits besides that?

About that output cap: I have a couple 1uF/63V polyester foil caps, that I can use. Not realy WIMA, but no doubt better than any electrolytic I think. My soundcard is nothing special (M-Audio 1010lt) and realy, I doubt I would hear the difference.

Of course Mensur's calculations are correct, but I fear you find this complicated and give up, unnecessarily.

I do indeed, but I am also curious and like to learn new (complicated) things... so I won't give up that easily  I feel these little projects are all about learning new stuff, otherways I would have bought a cheap (active) DI box or something similar


Hehe, English is a language but Ciryllic is a writing... (which I can't read/write)
Though, we could communicate just fine in some of our languages of the Balkans, but I doubt anyone else would be interested in reading that 
This way anyone can learn from this posts

[mensur]

Use MPF102's for both, voltage gain and as buffer.
Use 680R for RS(voltage gain) without bypass cap, BTW. Cs reduce internal resistance of the channel, hence more gain.As for Rd use 2.2K.
These are resistance values foe MPF102 FET (with -4.5V, 13.5mA Idss characteristics).
From my cacs. it will bring you to -2.12V of Vg(humbacker sweetspot) and 3.46mA Id.
Gain will be around 3 to 4x.As for buffer use same character. FET with 2.7K Rs, everything else remains the same.I don't like bipolars as buffers, you don't need much current there.

[J M Fahey]

I visited Yugoslavia over 20 years ago, and still have some money bills (¿Dinars?) written in 4 languages, one of them used Cyrillic characters.
I also visited Bulgaria in the same trip.
Yes, I know it's an Alphabet, I used to speak a little Russian although it's *very* rusty now.
I usually buy many Fets, measure and clasify them, for later use.
The ones with Vp around 3V are labelled "General Purpose"; the (few) ones around 2V are for high gain, critical stages; and those around 4V are used as switches and other lowly tasks.
If you have only 2 or 3 you are stuck with them.
What I meant before was: use 4K7 always as Rs; and Rd=10K to 22K depending on +B (supply voltage) so as to have it biased to *about* 1/2  +B; I find the use of trimmers unnecessary and somewhat amateur; definitely *no* production amp will have one trimmer per FET, no way !!!

[Alexius II]

I was a kid and bit too young to remember anything from the old Yugoslavia, but they say that were the good times 

Anyway, thank you both. I went through some online theory pages and found a few other equations that helped me understand the few missing relations. I think I get it now... I hope


I will now go again through the calculations for my transistor:

These are the equations I used for the gain stage:

Vgs = -Id*Rs
Vds = Vdd-Id*(Rd+Rs)
Id = Idss*(1-(Vgs/Vp))^2 (the characteristic curve is drawn from this equation)
gm0 = 2*IDSS/VP
gm = gm0*(1-(VGS/VP))
Av = (gm*Rd)/(1+gm*Rs) (if there is no Cs)

My transistor:
MPF102
Vp = -4.5V
Idss = 13.5mA

I choose Vgs to be 1/2 of Vp, so
Vgs: 0.5*(-4.5V) = -2.25V

Id = 13.5mA*(1-(2.25V/4.5V))^2 = 13.5mA * 1/4 = 3.375mA
Rs = Vgs/Id = 2.25V/3.375mA = 666ohm (closest is 680R)

gm0 = 2*13.5mA/4.5V = 0.006mS
gm = 0.006mS*(1-(2.25V/4.5V)) = 0.003mS

Then I choose Rd. I understand that with larger Rd come larger voltage gain and output impedance + lower current.

Rd = 2.2k  (18V/2200R = 0.0081A = 8mA)
Av = (0.003mS*2200R)/(1+0.003mS*680R) = 2.17 (= 6.7dB)


Ok, I hope everything is OK this far.

Then comes the buffer (aka "source follower" or "common drain amplifier") 
Only the Rs is necessary to bias it. I've read some theory about it, but I'm not sure. The way I understood it, you have to choose Rs in a way, that at the source you have half the voltage (Vcc). If this is correct, I then used the identical MPF102 transistor with calculated Id = 3.375mA (for Vgs = 1/2 Vp) inserted in the equation:

1/2 Vcc = 9V = 3.357mA * Rs
Rs = 9V / 0.003357A = 2680R = 2.7k

Is this ok, or am I making things up?