Accuracy of 2SK170 matching for JFET BOZ?

[Dr.H]

Hi all,

I have about 40 2SK170BL which I've run a simple matching procedure on Idss.

This varies from 700 to 950 measured across a 100R resistor using a 12V battery.

I'd like to build a BALANCED JFET BOZ as a buffer on my LDR pre-amp (as seen here: Jfet BOZ)

Questions:
1. How closely should the Idss be matched? 1%, 5%, 10%?
2. Assuming that I cannot find 4 that are close enough in tolerance, is there a way to compensate accurately for Idss differences when buidling the JFET BOZ buffer?

Thanks, Ryan

[Dr.H]

Consensus looks like 0.1-0.5mA or 0.1%

[AndrewT]

Hi,
selecting by Idss only gives you matched devices at that one current.

If the jFETs are biased to a different current then they must ALSO be matched at the operational bias current.
If the devices have a spread of operating current around the bias current value then even better if they are also matched over that range of currents.
This process of matching is effectively finding Idss pairs and finding those devices that have or have not got matching gm (transconductance).

Idss selection is a very simple and quick test, but is only applicable if the bias current is set to the Idss value.

If you can get devices that are Id +-1% for both Idss and Ibias when fed the same Vgs while thermally coupled, you will get better results than selecting Idss to <0.1% spread.

[Bill Fuss]

I usually match to around 0.02ma or so but that doesn't guarantee great matches in itself. If they are from the same batch that's always better.
Another simple test that can get you much closer matches than just Idss is matching the pinch off voltages too. That will keep you reasonably close in the transconductance numbers.
It takes a second PS and another resistor but it is pretty easy to do.

Whether any of this makes a sonic difference or not is up to you.

Best, Bill

[Dr.H]

Bill, any link to a good jig to measure pinch off voltages?

[Dr.H]

Have a look here:
A closer look at the Fetzer Valve

or here:
http://www.forsselltech.com/media/at.../JFET_Jig1.pdf

[Bill Fuss]

I'll have to try it, it seems right. I generally use 15 volts, others use 10 volts. One of Borbely's jfet articles has a more elabolate testing jig. Google it, I dont have a link.

Best, Bill