FET DI: Impedance converter...

[3nity]

I'm looking to design a discrete DI but i'm still lost on some things..
How do i know the impedance out the DI??

I'm still confused on how to know the Impedance of the FETs?

Sugestions...?
Thanks.

 

[Svart]

the input impedance of the FET is typically very high, like 5Mohm+ and commonly higher than 10Mohm these days.  This is usually in the datasheet somewhere.

The output impedance is a value of drive ability usually given by figuring out how much power you lose through your devices at some frequency. If you have an ideal powersupply with the ability to source a lot of current then MOSFETs can have a very low output impedance because they typically have much less than 1 ohm of resistance for DC and relatively the same for AC impedance, also in the datasheet.  Since the part will have some degree of resistance/impedance then when you add a load into the equation you are now sinking currents and your resistance/impedance element can now be measured in voltage drop.  At some point if you drop your load low enough your FET will start to become increasingly resistive in relation to the rest of the circuit and you will see a curve in the voltage plot.  This is usually where you would see someone say "this can drive xxx load".  Do this with a frequency range instead of DC and you'll have impedance.

Somewhere along the line, someone used the term impedance to describe drive ability.  I don't really care for that description because it's not totally accurate.  I like "drive capability" or "current sourcing ability" much better as you are dealing with being able to source currents over time.  Impedances can change while the device is working and they can change over frequency so measuring at a point in time or at a single frequency can give you different results than other times/frequencies.  Sorry, this is the RF talking..

 

[Winston OBoogie]

I'm looking to design a discrete DI but i'm still lost on some things..
How do i know the impedance out the DI??

I'm still confused on how to know the Impedance of the FETs?

Sugestions...?
Thanks.

As Svart said, the intrinsic input Z is high.  I'd use a J-Fet for a DI over a MOSTET myself but the same applies.  You'll want to put a gate resistor to audio common on it though and a good value for a guitar might be 2M.  Less is OK for other sources.

If you're just going to use a single J-Fet and self bias with a resistor, then output Z is low and you can do the sums pretty simply:

Z out = rsRs/rs+Rs

rs = 1/gm

Rs = Source/bias resistor

Since Rs is going to be quite large compared to rs, output Z will be 1/gm give or take.

You could also use another J-Fet (or whatever) as a constant current sink instead of a resistor.  Lots of info online regarding J-Fets and current sources/sinks.

Even though J-Fets are more immune than other devices, you should probably put a little R.F. filter on the input if you're coming into it straight from the outside world.  (I do believe Svart's a good guy to ask about R.F.  )

Actually, I take what I said back about not using a MOSFET.  If your input voltage is quite small (guitar level), about the simplest MOSFET follower would be a Zetex ZVP3310 with drain to audio common and a resistor from source to the voltage rail.  It'll self bias with about 4 volts-ish on the source so calculate your current based on the supply drop.   It's a simple Nelson Pass circuit so do some searching...

 

[clintrubber]

FWIW, note that input-capacitance can 'spoil' high MegaOhms-figures pretty quickly. For example, at 10 kHz, 20 pF of ||-capacitance brings the magnitude of the total impedance already below 800 kOhm.

In other words, going for say 10M input-resistance but neglecting parasitics buildup will give you a significantly lower resulting impedance.

Note though that a 100M || 1 pF input with a 10cm cable might be technically interesting but might sound artificially or too bright. Some attenuation of highs might be what you want, but then again, you could always attenuate latter as well.


Finally, do you want to make a DI that's the only load on the signal-source or do you want to make a 'sniffer' ? If it's the latter then I'd go for the highest Ohms & least pFs, otherwise for pleasing sound.

 

[Gus]

Mosfets have two issues
Device capacitance.  Depends on the size of the device.  Could help, could hurt(miller cap)
Need to protect the gate the insulator is often only good for 20VDC before it breaks down.

Jfet input R is almost what ever you make it before the gate drain leakage becomes an issue.  Microphones often use 1gig

Don't overlook a bootstrapped BJT input.  Something I designed and allowed diystompboxes to use as a beginner project.  I should have named it something else look at what happens with different Hfe transistors at different gain settings.  Made for a guitar to tube amp boost clean at low gain and distorts as the gain is turned up.
http://www.diystompboxes.com/smfforum/index.php?board=6.0

 

[Wonderlandaudio]

You may use the n-chanel fet, PNP transistor tandem as found in National Semiconductor Application Note 32.
PM me in case you cant find the pdf. It definitely works great

 

[skipwave]

So much good stuff here:

http://www.groupdiy.com/index.php?topic=27526

The schem JPK posted is the next DI design I'm going to try.

 

[ppa]

just developed

Pier Paolo

 

[ppa]

with bootstrap to avoid any sink current in the jfet drain

Pier Paolo

 

[3nity]

Ohh yeah i'm liking the simple and elegant design of the JPK fet DI..
I'm gonna try it very soon!!
thanks.

 

[ppa]

obviously these my designs are derived from the rafafredd fet DI , like all saw in this topic, I wanted to simply and/or modify them, for example changing the sink current used in  rafafredd version with two resistor using a bootstrap. This is the super simplified version.................... obviously derived from rafafredd fet DI.........if someone has not yet seen this. This is a no academic circuit  (but only simulated! not builded) , infact it does not magnetize the out transformer...........

Pier Paolo

 

[clintrubber]

Hi,

Looks like R11 has not the intended value, like shown (3k3) the input-Z is very low for a DI.

If you want to go for CCS-less I'd go for AS004.pdf @ the Jensen-site, works fine.
http://www.jensentransformers.com/as/as004.pdf

Bye,

 Peter

 

[ppa]

Hi,

Looks like R11 has not the intended value, like shown (3k3) the input-Z is very low for a DI.

thanks, this is a mistake , I wanted 3.3M instead 3.3k (I've done this schematic in a few minutes).

 

[ppa]

If you want to go for CCS-less I'd go for AS004.pdf @ the Jensen-site, works fine.
http://www.jensentransformers.com/as/as004.pdf

I've now seen this schematic , but at this point the rafafredd DI should be modifyed to insert it in API 312 clone preamps or other preamps using directly the main supply voltages that often are +-16v and +-24V.

 

[skipwave]

I've now seen this schematic , but at this point the JPK  DI should be modifyed to insert it in API 312 clone preamps or other preamps using directly the main supply voltages that often are +-16v and +-24V.

Use just the +24 rail or +16, or two 9V batteries in series.

 

[Gus]

http://twin-x.com/groupdiy/displayimage.php?pos=-3274

A MPSA18 9VDC powered DI PI idea fragment.  Bootstrapped input, outputs from C and E about 6.6 ma idle current (3VDC/470 ohms + 9VDC/55K).

Have not built it.  Posted as something a little different from what is in the thread so far

 

[3nity]

I like the simplicity of the Gus deisgn!
However i went for the JPK design before and i'm about to testing it.
board layout its done ...however i did a mistake i did figured out a litle late board its etched i use the footprint for BF245 instead of a 2sK170!!!  ???

 

[3nity]

I did some listening trought the JPK fet di using my epiphone les paul...and it sounds good..
Altough i did compared to the JLM fet DI and the JLM sounded a litle more bright!

The JPK what i love its his price...board + parts costed me around 5$

I will do a layout for something different like the National app. notes and i'll report back!
thanks.

 

[ppa]

if the power supply go on when the original rafafredd DI is connected to mic input, the input transformer will be magnetized by DI's output on transient. I've modified this DI also for this.

Pier Paolo

 

[3nity]

i'm looking to run the JPK di on 48V.
Ohm laws tells me i should change some resistors..
I should change R11 from 1K to 3K2 and R6 to...320??