Groner's Low noise measurement amp from Linear Audio vol 3 - spare boards?

[jan.didden]

I was looking at the differences between the BF862 and the CPH3910 regarding noise performance. The '862 has a voltage noise spec (0.8nV/RtHz) and the '3910 has a noise figure in dB (2.8dB).

How does this compare? Is there a simple way to convert one into the other?
What is the reference for the noise figure?

Jan

 

[BSST]

Hi Jan,

There's no conversion to noise voltage, especially since Onsemi only provides data at 100MHz.

I can only report that in my transimpedance application, performance at low frequencies is sensitive to both current noise and especially voltage noise. Even at low low frequencies (50KHz and up), noise performance of the two devices seemed indistinguishable. I was content--- available and cheaper in production.

Steve

 

[jackinnj]

LSK170 in SOT23 is available from NAC Semi in A,B,C and D grades.

(Much higher input capacitance, 30pF, than BF862)

 

[jan.didden]

Low input cap is important in my app, and the BF862 is almost double the CPH3910.

Jan

 

[BSST]

My transimpedance amp (TIA) was the front end for a photodiode, so low capacitance was an additional benefit for low system noise. I never measured voltage noise at 1 kHz and the LSK170 spec looks outstanding.

Semiconductor physics is a weak suite for me, but I believe high forward transconductance is a necessary but not sufficient condition for low voltage noise. CHP3910's high transconductance should bode well re low Vn.

 

[syn08]

I was looking at the differences between the BF862 and the CPH3910 regarding noise performance. The '862 has a voltage noise spec (0.8nV/RtHz) and the '3910 has a noise figure in dB (2.8dB).

How does this compare? Is there a simple way to convert one into the other?
What is the reference for the noise figure?

Jan

Noise figure doesn't make any sense without specifying the source impedance. I the particular case of JFETs, where the input current noise can usually be ignored, the source impedance is in fact it's real (resistive) part.

OnSemi doesn't specify the source impedance in the CPH3910 datasheet, but given the main application for this device (RF input stages) and the frequency at which the noise figure is specified (100MHz), we can infer that the source impedance is the RF standard 50ohm.

Using 50ohm as source impedance and the 2.8dB noise figure we can immediately calculate (following the great National Semiconductor AN-104 http://www.electro.fisica.unlp.edu.ar/temas/pnolo/p1_AN-104.pdf) the equivalent noise of the CPH3910 as 45ohm, which is about 0.9nV/rtHz. This value is very close to what I measured for the CPH3910, just a smidge noisier that the BF862, but with lower Ciss. Take a look here: https://www.diyaudio.com/forums/equ...measurement-amp-ikoflexer-51.html#post5656050

Be aware there is a dual version of CPH3910 in a 5 pin SOT23 (source is common) you can always parallel JFETs without any matching, they won't hog current and the noise dependence on the actual drain current is very weak, so these devices are a space and cost saver if you plan to massively parallel them. There is also a CPH5905 which is a low noise 2SK3557 cascoded on the chip with a NPN bipolar device, also very useful for keeping the JFET Vds low and avoid extra gate current due to hot carrier injection.

 

[BSST]

@syn08

Agreed on all points!

Best,

Steve

 

[jan.didden]

Steve, attachment for your son ;-)

Jan

 

[jan.didden]

Noise figure doesn't make any sense without specifying the source impedance. I the particular case of JFETs, where the input current noise can usually be ignored, the source impedance is in fact it's real (resistive) part.

OnSemi doesn't specify the source impedance in the CPH3910 datasheet, but given the main application for this device (RF input stages) and the frequency at which the noise figure is specified (100MHz), we can infer that the source impedance is the RF standard 50ohm.

Using 50ohm as source impedance and the 2.8dB noise figure we can immediately calculate (following the great National Semiconductor AN-104 http://www.electro.fisica.unlp.edu.ar/temas/pnolo/p1_AN-104.pdf) the equivalent noise of the CPH3910 as 45ohm, which is about 0.9nV/rtHz. This value is very close to what I measured for the CPH3910, just a smidge noisier that the BF862, but with lower Ciss. Take a look here: https://www.diyaudio.com/forums/equ...measurement-amp-ikoflexer-51.html#post5656050

Be aware there is a dual version of CPH3910 in a 5 pin SOT23 (source is common) you can always parallel JFETs without any matching, they won't hog current and the noise dependence on the actual drain current is very weak, so these devices are a space and cost saver if you plan to massively parallel them. There is also a CPH5905 which is a low noise 2SK3557 cascoded on the chip with a NPN bipolar device, also very useful for keeping the JFET Vds low and avoid extra gate current due to hot carrier injection.

Thanks O, very useful. My source impedance is very low (< 1 ohm).
I'll check out that dual.

Jan

 

[syn08]

My source impedance is very low (< 1ohm).

Then you would be much better with paralleling bipolar op amps like Gerhard does; much simpler, no compensation tweaking required, PSRR is huge so no low noise power supply required. Two ADA4898-2 chips (4 op amps) will give you about 0.4nV/rtHz (which you could get with a minimum of 10 CPH JFETs) and an input current noise of 4.8pA/rtHz, meaning that up to a source impedance of 30 ohm the input current noise can be safely ignored.

 

[BSST]

@Jan

Thanks Jan! The Concorde Brits are are great supporters of Boom and have had mutual visits--- they're eager for the of return supersonic travel and continuation of Concorde legacy.

 

[jan.didden]

So, that CPH5905 is specced at 1dB NF with 1k source. Since a 1k source resistor represents about 4nV/RtHz, that CPH5905 is about 4.5nV/Rt/Hz?
Not super low.

Jan

 

[syn08]

^^^

Jan, I strongly suggest reading AN-404 on the relationship between the Noise Figure and the voltage/current noise sources. Or you missed a SQRT somewhere? You will get that 1dB NF @1K source is about 2nV/rtHz.

En^2=Eg^2*(10^(NF/10)-1)<=>Rn=Rg*(10^(NF/10)-1) in the same bandwidth (4kT and the bandwidth are reduced), where Rn si Rg are the JFET and source equivalent noise resistances. Therefore Rn=0.26*1000 or Rn=260ohm. Meaning that the JFET is SQRT(26) times noisier than a 10ohm resistor, or about 5 times, rendering 5*0.4nV/rtHz=2nV/rtHz. I admit noise calculation can be confusing...

I have no idea why 2SK3557/CPH5905 are spec'd so high, though, I used perhaps 100pcs and they are all around 1-1.2nV/rtHz, about the same as the CPH3910 (I am suspecting they are same silicon, sorted in classes).

 

[jan.didden]

I messed up somewhere. I used an online noise calculator for 1kOhm in 1Hz BW gave me 4nV, so since I specified a 1Hz BW I concluded that's a density of 4nV/RtHz. ...
And the Rt of 1 being 1. My calculus is rusty. I check that AN-104.

Jan

 

[jackinnj]

Be aware there is a dual version of CPH3910 in a 5 pin SOT23 (source is common) you can always parallel JFETs without any matching, they won't hog current and the noise dependence on the actual drain current is very weak, so these devices are a space and cost saver if you plan to massively parallel them. There is also a CPH5905 which is a low noise 2SK3557 cascoded on the chip with a NPN bipolar device, also very useful for keeping the JFET Vds low and avoid extra gate current due to hot carrier injection.

NSVJ5908 is a dual too, (2 of the NSVJ3557) but I found that the IDss of each half were quite different. It is also a bit smaller package, MCPH5.

2SK2145 --Idss of each half were about equal. Super Mini SM5 package.

 

[kn0ppers]

I finally got around to populating some PCBs that have been lying around here for a while. I have two more PCBs that have pretty much all the SMDs apart of the JFETs mounted already. So far I used 2SK3557-6, but if anyone feels like selling some of their BF862s, please let me know

 

[NicMac]

I can send you BF862s matched for the Groner LNA.
Just drop me a PM.

 

[slartibartfasst]

Hi all,

Any recommended subs for the MMBF5842 in this application? Perhaps a high Vp 2SJ106? I also have a selection of three legged pFETs...

I'd like to build this board as close to stock as possible, any advice would be greatly appreciated.

(also if anyone has a few left I would be grateful...)

Thanks!

 

[kn0ppers]

I don't know if anyone is still building this design, but just a note about the Gerber files: I was looking over them and spotted a small error on the top layer. The leftmost JFET has one trace that is barely connected, as indicated by the circle. Depending from which board house people got their PCBs, that error may have been spotted and fixed during review (in my opinion it should have, but one can never be sure).

About MMBF5462 replacements: Mouser stocks SMP5462 which seems to be a substitute made by InterFET. 2SA1312 is still stocked by Digikey. The JFETs that are suitable to replace BF862 are a bit thin in supply right now but not EOL as far as I can tell. 2SK3557/2394 seem to be the Idss ranked parts which are basically unobtainium right now, but the NSVJ3557/2394 which are not Idss ranked (just like CPH3910) are available. All in all, as of mid 2023 this LNA can still be built fairly close to original. I have attempted to fix the Gerbers and ordered another set, lets see how it goes.

 

[tombo56]

Excellent detective work. I was sweating for the several minutes required to check on spare PCBs. Huh, all is good.

PCBs were manufactured at JLCPCB. I doubt that they correct anything at received Gerber files. Seems that standard minimum tolerance is enough to fill in small gaps. Here is actual macro shot of the PCB. The same error (?) is present at the right side of input JFETs row.