Just went there and saw they span a range. You should match them yourself.
"8 pieces of LSK170 ULTRA LOW NOISE SINGLE N-CHANNEL JFET"
"Grade A (Idss 2.6 - 6.5)"
I will measure them tomorrow. My girl was complaining that I locked myself in the man cave 🙂 thanks for the help tonight and good night! Hopefully we will fix this tomorrow ...
Ok, they are not even close!!! In millivolts over the 10 ohm resistor it is 24,2 and 45 mv... So IDSS is 2.4 and 4.2..... I will start matching now.... This is kind of fun 🙂
Read Dennis Feucht.
He describes how to get exactly zero output offset by just adding some low value source resistors.
This is intended for selected jFETs with the same Idds and running them at less than Idss.
But can be applied to unselected jFETs that differ significantly in Idss.
He describes how to get exactly zero output offset by just adding some low value source resistors.
This is intended for selected jFETs with the same Idds and running them at less than Idss.
But can be applied to unselected jFETs that differ significantly in Idss.
If you adopt Feucht, then solder the source resistor to the jFET first.
Then bend the G + D + resistor to mimic the 3 leg 0.1" pin pitch required on the PCB.
The Source leg does not get inserted into the PCB.
Set up a small jig for the CCS & Follower jFETs and use a VR to find the required Rs.
You can use the DCB1 PCB as a jig if you fill the 12 PCB holes with SIL sockets.
Then bend the G + D + resistor to mimic the 3 leg 0.1" pin pitch required on the PCB.
The Source leg does not get inserted into the PCB.
Set up a small jig for the CCS & Follower jFETs and use a VR to find the required Rs.
You can use the DCB1 PCB as a jig if you fill the 12 PCB holes with SIL sockets.
Matching results:
Pair 1:
2.45 and 2.47
Pair 2:
2.48 and 2.5
That should be good? The ones that were in were not matched at all. So this should do it?
Pair 1:
2.45 and 2.47
Pair 2:
2.48 and 2.5
That should be good? The ones that were in were not matched at all. So this should do it?
Looks promising now:
Left has 0.6MV
Right has 1.7MV
I'll let it warm up a bit and measure again.
Left has 0.6MV
Right has 1.7MV
I'll let it warm up a bit and measure again.
It was a matching issue, obviously. Although it was good you rectified the rails difference also. It was rather high in yours. You did not match LEDS too I guess.
Those A grade are like the GR Toshiba grade. They will work, but B1 & DCB1 look for higher IDSS in order to have more reserve bias current to drive loads down to the 5K-10K input impedance amps. Its also better for THD. If your amp is high impedance the A grade will still make do. So look to substitute with matched Grade B (Idss 6 - 12) LSK170 at a future point.
Those A grade are like the GR Toshiba grade. They will work, but B1 & DCB1 look for higher IDSS in order to have more reserve bias current to drive loads down to the 5K-10K input impedance amps. Its also better for THD. If your amp is high impedance the A grade will still make do. So look to substitute with matched Grade B (Idss 6 - 12) LSK170 at a future point.
Hi Salas,
It's fun to match those fets 🙂
I did match the led's with my DMM. They are all 1.756V at 20MA. Maybe I selected wrong ones... they are clear and illuminate orange.
I think I will replace them with red ones. I read somewhere they are less noisy.
I have a fully populated B1 here with matched fets from Nelson Pass. I will take them out and see what LDSS they have.
It's fun to match those fets 🙂
I did match the led's with my DMM. They are all 1.756V at 20MA. Maybe I selected wrong ones... they are clear and illuminate orange.
I think I will replace them with red ones. I read somewhere they are less noisy.
I have a fully populated B1 here with matched fets from Nelson Pass. I will take them out and see what LDSS they have.
Leave the JFETs for 3 minutes to warm up before characterization, its closer to circuit life. Maybe bcs you matched LEDS at a DMM mA level (actually low), when it should have been at around 5mA to remind their circuit use they had shown bit different results. But anyway the leg trick ameliorated the tolerances. So you are done. Its not a problem in such a signal level, the small noise difference between LED models I mean. Its not only the color, you never know from maker to maker exactly without lab grade noise characterization. Just don't use super bright types and blue in the Vref subcircuits. Red, Green, Yellow, Orange are OK.
Ok, I put the nice ones from Pass DIY store in... Te results are:
Offset left 0.8MV
Offset right 3.0MV
Is this acceptable?
Offset left 0.8MV
Offset right 3.0MV
Is this acceptable?
Yes. But do a last tricky move if you feel lucky, mutually exchange place for the ones in the pair that does 3mV. There is a chance to do better (or worse).😀