Jfet folded cascode with diamond buffer opamp

[sonnya]

Eyecandy

😀

Here are the boards. I dont think i will be able to assemble them before next week :/

You can see an DIP8 and a SO-8 on one of the pictures for comparison.

 

[sonnya]

First samples made...

😀

Now the first samples are made...

I am going to test them shortly...

 

[sonnya]

In the first test i discovered that the PMBFJ620 jfet has a vgs difference of ~440mV...

I need to find a fix for it...

 

[sonnya]

Ok, anything else is like it should be.....

Sound is very very detailed even at low level...

Despite of the offset of 440mV is does not drift more than a mV...

 

[Luke]

that's impressive.

 

[sonnya]

I did try mount them in an Quad 34 running at +/- 7.5V. Works also...

 

[sonnya]

Okay, the problem with the offset is due to a poor vgs matching in the PMBFJ620 . At id=~3mA the is a difference in the package of 430mV..

Sadly enough there is no matching jfet with the same layout... Also no dual npn bjt with the same layout.

So they are for scrap... They work, but the have the hugh DC offset...

The fix for this module is a trimmer between the rails with a series resistor to either in- or in+.

Or add an external servo...

 

[nelsonvandal]

How much is much DC? Can't you adjust it with R2/R3 (or maybe R5/R6)?

If you make a version with and one without the VAS cascode, it would be interesting to hear your oppinion on the sound of folded cascode vs cascoded folded cascode. I've tried it. It reduces distortion both in sim and in real life, but I find that it softens the sound for good and for bad.

 

[sonnya]

Hi Nelson, Thanks for showing interrest...

I would Maybee have expected less than 100mV vgs difference but it shows ~2V and 1.6V..

I am lucky that i did not buy a full reel of the pmbfj620 from NXP ...

I will try make a version with ssm2212 from analog instead of jfet's .. It has high gain of ~600 so offset current Will be small compared to bc8xx or that3xx

I Will certainly make it as a module but not that small..

They Will be mounted on a motherboard.. I Will try compare different version of the folded cascode.. And i Will let you know on the outcome

 

[sonnya]

After a long period have had sparetime to test the modules.

The attached measurement show a well behaved THD.

No load, the 2. harmonic is the only present distortion. 0.003% or as low as my equipment goes.

into 220R load 3. is present at 1KHz and 3. and 4. harmonic is present at 10KHz.

The same circuit will be no different with an bipolar input (SSM2212 from analog devices).

More to follow.

 

[sonnya]

I Will test with ssm2212 tomorrow .

The input bias current Will be less than 5uA.

- Sonny

 

[sonnya]

Just a shot of the difference on input and output on the squarewave signal.

Lightblue is the input.

 

[sonnya]

I have tested the JFET opamp with NPN pair instead of jfet's.

Some of the data is:

• Less than 2mV input offset
• less than 4uA input offset current
• Distortion 10 times lower at 10KHz into 220R load

The picture below is as with all the other measurement

• 47pF compensation
• 680mVrms out
• 10KHz Sine
• 0dB gain

Distortion is 0.0018% - Generator is: 0.0013%

 

[AndrewT]

still the post1 schematic?

 

[sonnya]

still the post1 schematic?

No, it is changed a bit.

See attachment.

instead of folded cascode, i use current mirrors now. But performance is exactly the same.

 

[sonnya]

Here is the schematic of Bipolar version of the Jfet opamp