[PhopsonNY]

Hi there,

Does anyone know what combination of Cascoded JFETS and resistor are needed to obtain a 5mA opamp Biasing into Class A.

I cannot seem to find out the exact JFET # to acheive this.

Thanks for all the help.

Cheers

 

[Arzela]

Perhaps one of these
would work:

http://www.cse.ucsc.edu/classes/ee17...ics/2N5486.pdf

The 5485 has a minimum Idss of 4 ma.

 

[morsel]

I would recommend the 5486. The 5485 is a bit low for the job.

 

[puppyslugg]

Of the 2 dozen 2N5486's I have, the lowest Idss was 10ma. Out of 1 doz. 2N5485's, I found 2 with a Idss of 5ma.

 

[tangent]

Quote:

Of the 2 dozen 2N5486's I have, the lowest Idss was 10ma. Out of 1 doz. 2N5485's, I found 2 with a Idss of 5ma.

But realize tht when you cascode two JFETs, you get a lower current draw than either JFETs' IDSS. And, if you're doing your own layout, you can add a "source resistor" to drop the cascode's current draw to arbitrary values. See my Biasing Op-Amps into Class A article for details about all this and pointers to further information.

 

[puppyslugg]

Hi tangent,

I'm confused. (what else is new?!!

)

If I use a 5ma fet for Q1 and say 10ma fet for Q2, with no Rs, what is the amount of bias? I was under the impression, it was Q1 that sets up the bias(?).

 

[tangent]

Quote:

If I use a 5ma fet for Q1 and say 10ma fet for Q2, with no Rs, what is the amount of bias?

Under 5mA. I don't know if there's a formula that lets you predict precisely what happens. Don't take my word for it, though. Try it!

Quote:

I was under the impression, it was Q1 that sets up the bias(?).

Q1 puts a cap on how much current you can get, and in a simple lone-JFET arrangement, Idss = total current through the current source. But when you add Q2, the overall current draw drops significantly. Again, try it.

 

[puppyslugg]

Hi tangent,

OK, I tried it, and you were correct...as usual.

I noticed that the ccs in the Meta is different from the one in the PPA. Is one functionally superior?

 

[morsel]

The extra resistor in the PPA CCS allows for precise current control without FET matching. Check the Vishay-Siliconix paper for more details:

http://www.vishay.com/document/70596/70596.pdf

 

[puppyslugg]

Hi Morsel,

In the case of the ad8610/20, is such precision necessary? The PPA's ccs, seems more applicable for the ad843.

 

[tangent]

It's not so much a question of precision, it's an ease-of-use issue. We haven't spent a lot of time on working out the consequences of the source resistor beyond what you see in my op-amp bias article. However, it seems evident that this will eliminate the need to match JFETs. You could use any random 2N5486es for the JFETs and then pick R8 such that you hit whatever arbitrary current draw level you want. You could leave R9 out temporarily and put a milliammeter in its place, turn the amp on, and try various R8s until you got the current draw you wanted. Then you could go through the other two channels and do the same, setting all three to the same value.

The only way you can do that in the META42 is to have a whole bunch of tested JFETs on hand and pick 2-way matched sets. In the PPA, it would have to be 3-way.

It's a whole lot more palatable to keep a variety of 1-100 ohm resistors on hand for R8 and do the test described above than buy dozens of JFETs and measure them all to hit arbitrary current values. Resistors are cheaper than JFETs and they're more generally useful besides.

 

[puppyslugg]

Hi tangent

By 'precision', I only meant the ad8610/20 and others, require a 1-3ma of bias. As such, there is a broader target to hit. In case of the ad843, there's more of a need to reach a min. of 5ma, requiring the need to be more 'precise'.


BTW: Have you tried more than 5ma of bias for the ad843?

 

[tangent]

I've not yet bothered to try and bias the AD843.