Has anyone tried the 2SK932 JFET in a buffer, e.g., Borbely, B1, Mesmerize, or the like? The datasheet spec's compare well with the BF862. http://www.onsemi.com/pub_link/Collateral/EN2841-D.PDF But the '932 is superior in the fact that it is available in 3 IDSS grades, which makes matching more efficient than the BF862. I'm trying it in a Borbely without any luck.
Thanks,
George
Thanks,
George
Thanks for this, needs to be checked for audio frequency noise since it's for the same AM tuner application. Stocked everywhere at 20 cents in hundreds looks very good.
EDIT - Only 15V breakdown is that your problem.
Last edited:
Don't worry I'm ordering some on speculation and can test them in the same adapters I used for the BF862's. Too bad there are no P channel radios.
Based on the datasheet value of NF=2dB @Rgen=500ohm and @Id=10mA, a quick napkin scribble calculation shows this device has about twice the noise of the BF862, to about 1.6nV/rtHz.
You'd be much better with the 2SK3557 from the same manufacturer, NF=1dB @Rgen=1k and @Id=1mA, that's about 1nV/rtHz. Same SOT123 case, also comes in 2 Idss grades.
Big question for both is the LF noise and the noise corner frequency. Some HF devices have very low noise at >10KHz, but the low frequency noise (essentially GR noise) makes them useless for audio. BF862 is a lucky hit, good for audio as much as for AM radios.
The gm at 10mA is around 40mS the graph shows 2dB NF out where there can be no excess noise except from parasitic resistances which would compromise RF performance. That is as is often the case I don't necessarily trust the DS.
The data you want is spot noise at Rg = 0 not rf noise figure. At 1k Rg the difference between a 0 nV FET and a 1 nV FET is 4nV vs 4.12nV i.e. unnecessary obscuring of resolution. Easy enough to measure the noise for audio applications, then the problem becomes the yield to what you need, these might be another lucky find and maybe not.
Waly was a little optimistic my Quan-Tech noise slide rule (I'll bet I'm the only person here with one
) gives over 200 Ohms parasitic Rg to get a 2dB noise figure at high frequencies, I'm not ready to believe that without verifying.
Last edited:
Looks like I am not reading the graph correctly.The data you want is spot noise at Rg = 0 not rf noise figure. At 1k Rg the difference between a 0 nV FET and a 1 nV FET is 4nV vs 4.12nV i.e. unnecessary obscuring of resolution. Easy enough to measure the noise for audio applications, then the problem becomes the yield to what you need, these might be another lucky find and maybe not.
Waly was a little optimistic my Quan-Tech noise slide rule (I'll bet I'm the only person here with one
Reading the 932 datasheet page 3 graph NF - f, where Vds=5V and Id=10mA,
I see at 100Hz & Rg=1k that Noise Figure is just above 2dB
If I go down to ~70Hz the Noise Figure becomes 3dB. I thought that meant that the FET produces the same noise as a 1k resistor when Frequency is 70Hz. Is that wrong?
How does one read this graph?
The next graph shows that @ 1mA and Rg=1k that @ 200Hz the NF is 3dB
No, that's fine taken literally the BF862 is bad at 70Hz too. The noise figure of 2dB at 100kHz bothers me. At this point there is no more 1/f noise. A set of real measurements for an audio application are needed.
Waly was a little optimistic my Quan-Tech noise slide rule (I'll bet I'm the only person here with one
To add to the confusion, the 2SK3557 has, according to the data sheet, only 30mS gm at the same 10mA, but only 1dB noise @1k @10mA.
Your buddy syn08 measured the 2SK3557 and it is only a smidge (~5%) noisier than the BF862, has slightly (~10-15%) higher LF GR excess noise (on top of 1/f) and about the same noise corner frequency (~100Hz).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.