matched 2N5952.. how many do i have to buy..??

[JRM]

I have a friend (Maths teacher) that offered himeslf to calculate this as accurately as possible. But on need to know (tell him) what we can consider a match. From the 2N5952 datasheet we can check that Vgs vary from -1.3 to -3.5V. I believe that will be the 95% spread of the Normal function with the central value being -2.4V, so I just need to know how much will we consider a "good match". For instance 1%: 0.02V; 5%: 0.12V; 10%: 0.24V; on the other hand, a 0.01V delta corresponds to 0.41%.

[PRR]

> Maths teacher... need to know (tell him) what we can consider a match.

That's what I thought.

I even did some scribbling about phaser sensitivity (low!) and FET V/R curves (bent, and we may care about the end of the bend).

I was hoping for some empirical data from loose-match or no-match rigs, but I guess the guys have gone fishing.

[JRM]

I've managed to achieve some conclusion regarding how many one should buy if one have a specific desired variance. The only point is that I only know how to do it in the vicinity of the central value (mean) that, in our case is -2,4V. If that's not what we want/prefer, I can't do it (but I might ask).
Taking the normal distribution and assuming that the minimum and maximum values given in the datasheet are the 6-sigma spread we can calculate that the standard deviation is s=0.1833V. Now, fooling around with the erf function et all we achieve for each required precision:
P: 0.5% (=0.012V); N=77
P: 1.0% (=0.024V); N=38
P: 2.0% (=0.048V); N=19
P: 5.0% (=0.12V); N=8
P: 7.6% (=0.1833V); N=6
P: 10% (=0.24V); N=5

The 7.6% isn't because I'm mad is the 1-sigma mark! If one gets 20 it one can be sure to have a 2% matched quartet. It seems quite fine to me but I'm still waiting for the answers of the ones that realy know electronics and phasers...

[jkokura]

that sounds about right! and a 2% match sounds good according to my builds of a phase 45. I'll see about the Phase 90 shortly.


Jacob

[JRM]

2% it's less than 0.05V which sounds nice. But don't forget this is to have JFETs around their mean value of -2.4V. From the Normal distribution curve it's easy to understand that if you want to have a different value, e.g. near -3.0V, you'll have to buy more, as the curve is more steep and has a lower value, meaning that it's much harder to find 4 of a kind (it looks like poker )

[Bullet79]

how do u guys get the reading,..? my dmm get 6.++v

[jkokura]

You can't just read the transistor with your DMM. You need to build RG's Tester circuit. I think I have a link somewhere, but you can go to Geofex and search for fet tester.

Jacob

[Bullet79]

i v built that thing already... but how do i hook the dmm. and what to set on the dmm

[JRM]

It depends on what dmm you have. You'll need one that can read values around 2,4V with an accurancy of at least 0,01V, meaning that will have to have 3 decimal places (0.000), as we neglect the value of the last one.
BUT, we still don't know if -2,4V and 2% match it's ok...

[Bullet79]

i have somthing like this... (cheap one,. but it can read transistor hFE's) will it do.. seriously till this day i still figure out how to use the jfet tester... how to check..?? and where to put the dmm switch DCV..??

[jkokura]

If you go to GEOFex, there is a description and schematic for building a Jfet tester. I have built one and tested many 2N5952's with it. There's lots of info over there, and if you read you should be able to figure it out.

If you want some matched Transistors, there are people around who can help you find some, including myself.

Jacob

[Bullet79]

its true... but the only thing he doesnt mention is where to put my dmm probes...?? and what value is Rset

[jkokura]

I'm sorry, but if you take a look at the document fetmatch, which I'm assuming you are, there's a schematic which includes two lines that go to your DMM.

Try this for a layout: http://www.aronnelson.com/gallery/main.php/v/Richard-Boop-RLBJR65/album81/album75/Jfet_Matcher_6.gif.html

Jacob