Complementary JFET-JFET cascode input, BJT VAS-Drivers, Lateral Output

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Jump to Latest
N

Neutrality

Member
Joined 2004
AndrewT said:
I was working with k170 & bf244 as cascode.

I came up with Idss of cascode should be ~ double the Idss of the amp device. (actually trying to remember, I think Cascode Idss ~ 2* {k170 Id +1mA}, not idss).
If you are using a 10mA k170 and you bias it inside a 20mA cascode device then you are possibly going to get an Id ~ 60% to 80% of Idss, i.e. somewhere between 6mA and 8mA for that 10mA jFET.

Now to back track from these numbers to Jung's advice.
If you pass say 7mA through a 10mA k170 then that same current must pass through the cascode.
What Vgs of the 20mA cascode will allow ~7mA of Id to pass? (If my alternative formula applies then a 16mA cascode may be about right)
That cascode Vgs must be at least twice the Vp of the 10mA k170.
A 10mA k170 probably has a Vp <1V and more likely towards 700mVp

The only way to be sure that you meet Jung's guidance is to set up the two devices on a plug board and apply voltage and measure the Vgs of the cascode and measure the Vp of the k170.
Unfortunately no one has come in to confirm that Jung applies, when working with other than a Cascoded CCS.
Click to expand...

My example with 10 mA Idss JFETs biased to 4 mA has a Vp of -200 mV at 4 mA, from looking at the 10 mA Idss curve in the Id-Vgs chart in the spec sheet.

Measured Vgs of the cascode device in simulation is 600 mV.

So basically, with a input JFET bias of 4 mA for each 10 ma Idss 170/74, a cascode device with Idss of 8-10 should be sufficient.

That is more or less the same as 7 mA through the input JFET with cascode of 16 mA Idss.

2 x 4 mA = 8 mA , close to 7 mA in your example and 2 x 8-10 mA = 16-20 mA being close to your example of 16-20 mA Idss for the cascode device.
 
Last edited:
N

Neutrality

Member
Joined 2004
AndrewT said:
Why are you choosing a 10mA k170 to run at your 4mA bias current?

Most designers here, who all know far better than me (I just regurgitate what they tell us) suggest 60% to 80% of Idss is a good bias range and some promote much higher percentages of the order of 90% to 100% of Idss for the operating bias Id.
Click to expand...

Well, good question. I found that a bias of 4 mA worked well with 10 mA JFETs but I'll see what else I have available. Think I have 4mA models as well, and 8 mA models.
 
Last edited:
N

Neutrality

Member
Joined 2004
Neutrality said:
Well, now using 4.1 mA models for SK170 and 4 mA models for 2SJ74. Works just as well as the 10 mA models, just slightly lower distortion. Source resistor is reduced to 22 Ohm which equals 6 mA bias in each pair or 3 mA bias per JFET.

That is 75% of Idss.
Click to expand...

So biasing JFETs at 60-80% of their Idss seems to give an improvement in distortion.

It seems the 60-80% rule of thumb for biasing JFETs isnt that far from the truth.
 
N

Neutrality

Member
Joined 2004
Doing some further testing with 2SK246 and 2SK103 cascode devices, the 2SK246 model is a 5.6 mA Idss device while the 2SJ103 model is a 5 mA Idss device.

Interestingly enough, even though the Idss's are quite close to the 4 mA Idss/3 mA bias for the 170/74 input pair models, it seems to work quite well.

Now, I need to fint out how to change the 2SK246/2SJ103 models to a higher Idss and see what kind of performance I get with those.

Testing with cascode device models ranging from from 6 to 8 mA Idss should be sufficient
 
A

AndrewT

R.I.P.
Joined 2004
Neutrality said:
My example with 10 mA Idss JFETs biased to 4 mA has a Vp of -200 mV at 4 mA, from looking at the 10 mA Idss curve in the Id-Vgs chart in the spec sheet.
Click to expand...
Vp is measured at virtually zero current for Id. Maybe a 1M resistor or even a 10M resistor in the source to -ve supply.
A low (6mA) Idss k170 will have a lower Vp than a 10mA k170, since both are measured when Id<1uA.
 
N

Neutrality

Member
Joined 2004
AndrewT said:
Vp is measured at virtually zero current for Id. Maybe a 1M resistor or even a 10M resistor in the source to -ve supply.
A low (6mA) Idss k170 will have a lower Vp than a 10mA k170, since both are measured when Id<1uA.
Click to expand...

Measured with LTSpice test circuit.

+10 V or -10 V supply, gate grounded, 10M resistors in source to ground.

N-Channel :

Vp for 2Sk170 4 mA Idss model is around 500 mV.

Vp for 2sk246 5.6 mA Idss model is around 2.6 V

P-Channel :

Vp for 2Sj74 4 mA Idss model is around -468 mW

Vp for 2Sj103 5 mA Idss model is around -2.02 V

Looks like the Vp of the cascode devices are 5 times as high for the N-channel and 4 times as high for the P-channel.

Anyway, the input 170/74 4 mA Idss models biased at 3 mA, cascoded with 246/103 5.6/5 mA Idss models works very well.

THD-20 at 60 W 8 Ohm is 0.0065XX%, slewrate is a respectable 120 V/uS, phase margin is 72 degrees, gain margin is 13 dB.

However getting 4 mA Idss 170/74 might be somewhat difficult. I could go for a higher Idss but I really have to watch the Pd of the cascode devices. Current power dissipation in the 246/103 pairs is 141-144 mW already.

Would have liked it to be under 100 mW for long term reliability but 145 mW should be ok.

EDIT : Power dissipation derating curve. At an ambient temperature of 75 degrees ambient, power dissipation drops from 300 mW to 150 mW. Seems like it could end up being a problem. Which would force me to use higher rated JFETs for the cascode. Another headache.
 
Last edited:
A

AndrewT

R.I.P.
Joined 2004
Neutrality said:
Vp for 2Sk170 4 mA Idss model is around 500 mV.
Click to expand...
according to Jung, the Vgs of the cascode must apply at least 1V to the K170.
What Idss version of a K246 shows 1Vgs @ ~3 to 4mA?


Select devices around what the sim or datasheets have predicted and test on a plug board.
bl grade are >=6mA and <12mA Idss.

I agree on your concerns for temperature.
What about changing the jFET cascode to a BJT version?
Or simply adding a BJT cascode outside the jFET cascode?
 
Last edited:
N

Neutrality

Member
Joined 2004
AndrewT said:
according to Jung, the Vgs of the cascode must apply at least 1V to the K170.
What Idss version of a K246 shows 1Vgs @ ~3 to 4mA?
Click to expand...

Well, datasheet only shows a curve for 5.6 mA maximum Idss but 2SK246 goes up to 14 mA(range is 1.2 mA to 14 mA).

I would estimate that a 8-9 mA mA Idss device would show 1 Vgs at 4 mA.

However, in my amplifier circuit I get about 975 mA applied to drain of the sk170 and 710 mW to drain of the sj74. That is measured at the drain of the sk170/sj74 devices.

I'm not in a position to say anything about Jungs theories and if they even apply to a JFET-JFET cascode.

All I can see that in my sim it just not works, it works really, really well.
 
Last edited:
N

Neutrality

Member
Joined 2004
AndrewT said:
I agree on your concerns for temperature.
What about changing the jFET cascode to a BJT version?
Or simply adding a BJT cascode outside the jFET cascode?
Click to expand...

Problem with BJT cascodes is slightly more distortion. THD-20 60 W 8 Ohm, 0.0065xx% to about 0.0085xx%. Also it requires atleast 3 more components for each cascode(2 transistors for a voltage divider and a filter/decoupling cap).

The added complexity of an extra BJT cascode outside of the JFET cascode could work, but that is even more complexity and I'm almost sure that it will not work any better than just a single BJT cascode.
 
Last edited:
D

dodo

Member
Joined 2004
Isn't true the rule of thumb ( based on Borbley jfet infos)
Vgs of the uper j-fet cascoding device (sk246 ) must show at least 2V at the bias level complementary differential input jfet pairs (sk170/sj74)
in your case sk246 /sj103 must show at least 2V vgs on Id=2,5mA ?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Top Bottom