John Curl's Blowtorch preamplifier part II

Status
Not open for further replies.
Jump to Latest
#3,443
PMA said:
Crss=CGS/(1-(Vds-Vgs)/PB).5

Crss=Ciss+ CDS/(1-Vgs/PB).5
Click to expand...

One thing I always want to know - did you have to look it up, or did you know it off the top of your head?

I am rather in awe of the people I have met who are walking encyclopedias of this sort of information... I sure as heck can't remember where I put my glasses from one minute to the next, much less formulas for things like this...

_-_-bear
 
#3,446

Yes Stinius if you follow that I think you get the industry accepted answers. Notice the source bootstrapped in measuring the Crss. This is necessary because the channel acts like a low value R shorting somewhat the drain and source. So now the first circuit measures Cgs + Cgd and the second just Cgd. Does anyone know if a GR impedance bridge has a driven shield? I think I can dig one of those up.
 
Last edited:
#3,447

Attachments

  • hands_up_poster_pic.gif
    hands_up_poster_pic.gif
    107.3 KB · Views: 333
#3,448
john curl said:
I do know this: If I directly replaced a 2SK146-J73 pair with a 2SK240-J75 pair, with the SAME Idss, my measured distortion with a 10K input pot, would drop. This is what I did with the JC-80 discussed here on previous pages.
Click to expand...

Are you saying that the -lower- transconductance devices (k240/j75) gave
lower distortion than the higher transconductance devices (k146/j73)?

If this is the case the circuit is sub optimal.
 
#3,449
Terry, don't be childish. The 240's had LESS non-linear input capacitance than the 146's. That is why the distortion with the 240's was lower. Gm makes very little difference in this amplifier, because the series R's in the sources reduce it, in any case.
 
#3,451
dimitri said:
http://www.jstage.jst.go.jp/article/elex/4/16/517/_pdf
Click to expand...

Thank you Dimitri this is a more concise treatment. Same problem here in that these measurenents should agree with the Motorola ones and not the Toshiba ones. Again Ciss is measured vs Vgs not Vgd, drain and source AC shorted.

I think this stuff is important because these properties are at the forefront in determining fine scale performance and getting correct models might help the sims match reality.
 
#3,452
john curl said:
Terry, don't be childish. The 240's had LESS non-linear input capacitance than the 146's. That is why the distortion with the 240's was lower. Gm makes very little difference in this amplifier, because the series R's in the sources reduce it, in any case.
Click to expand...

WRT non linear C versus IP device transfer, of course we agree here,
see my post 3333.

http://www.diyaudio.com/forums/anal...rch-preamplifier-part-ii-334.html#post2036486

WRT 146/240 differences, passive vol control b4 the IP stage is the culprit.

This was also discussed previously.

http://www.diyaudio.com/forums/soli...-blowtorch-preamplifier-1721.html#post1798446

cheers

T
 
#3,453
scott wurcer said:
Yes Stinius if you follow that I think you get the industry accepted answers. Notice the source bootstrapped in measuring the Crss. This is necessary because the channel acts like a low value R shorting somewhat the drain and source. So now the first circuit measures Cgs + Cgd and the second just Cgd. Does anyone know if a GR impedance bridge has a driven shield? I think I can dig one of those up.
Click to expand...

I have both a GR digibridge 1658 and the old fashioned manual version 1650a. The 1650a is two terminal system the 1658 takes jigs and is four terminal. The way I read the ap note the DUT generates the driven shield.

The ap note is not the way I would try to make a measurement that low.
 
#3,454
simon7000 said:
I have both a GR digibridge 1658 and the old fashioned manual version 1650a. The 1650a is two terminal system the 1658 takes jigs and is four terminal. The way I read the ap note the DUT generates the driven shield.

The ap note is not the way I would try to make a measurement that low.
Click to expand...

Upon more careful reading of Dimitri's link I think I can rationalize the Toshiba graphs with those test jigs if Cgs = 24pF and Cgd @ 0 ~ 40pF This is far off of any models that I can find.

What's worse in the LSK270 model Pb = 1.2v and Mj = .59 while most 2SK170 models Pb = .6V or so and Mj is closer to .4. I have seen blind curve fitting in characterization before so I need to run some models vs datasheet curves and then maybe measure some.

No wonder sims don't match reality.
 
#3,455
#3,460
Well, everyone, I thought we were talking about non-linear input capacitance, but it shifted over to quantum mechanics, instead.
In truth, except for measuring Cin changes slightly above and below Idss, we are, as usual, bogged down.
Actually, I like the quantum mechanical cite. Perhaps, I should not really care that it is posted here, except it does lead astray.
 
Last edited:
Status
Not open for further replies.
Top Bottom