LFLNA-80

[Kofen]

For low frequency use, e.g 0.1 - 10 Hz use, you would be really hard pressed to get any IMD products from the chopper clock difference, which is around 200k for the ada4528. So I would not worry about it.

[mawyatt]

The classic 2 Tone IMD shows the results around the 2 tones but often doesn't tell the overall story where the 2 tones also mix down to around DC. This was our concern, not at or around the chopper clock frequency, but the clocks mixing and appearing around DC. Since these clocks are generated on-chip they will have high levels of Phase-Noise, Edge Jitter and poor overall Frequency Stability. Add to this the possibility of having 4 clocks on-chip close in frequency with on-chip limited isolation and you begin to see our concern.

BTW long ago we experienced this effect where the harmonics of two separate higher frequency clocks within the system were mixing down to around DC and corrupting the low frequency sensitive signals of interest.

Anyway, we'll wager the chopper multi op-chip op-amps types all use the same on-chip clock source or same clock frequency. Maybe one of the analog op-amp designers on here can comment wrt to such.

Best

[mawyatt]

Should mention that we made some electrolytic capacitor measurements noted over here:

https://www.eevblog.com/forum/metrology/diy-low-frenquency-noise-meter/msg6018343/#msg6018343

These Rubycon ZLH 1500uF 16V capacitors were just laying around and we made some leakage measurements after apply a DC bias. The trend is showing the leakage is significantly reduced by DC bias aging, and after ~15hrs bias at 16V the leaking is below 11na 8.5na, and still very slowly trending downward. Edit: now below 4.5na after about 40hrs.

Of course the low frequency capacitive noise contributions remains to be seen but leakage effects are encouraging!!

Best

[iMo]

This is why you need a mechanism to not discharge the capacitor when manipulating with DUT. As I mentioned in one of my previous posts - we had a switch in our shielding box which maintained the DC at the capacitor during DUT replacement (and also reduced the time to settle after the replacement). Do not ask me how it was wired, it is long time back I mounted it in..

..The trend is showing the leakage is significantly reduced by DC bias aging..

[2N3055]

SAMWHA RD series 1000µF/50V is steady at 5nA/10V. But there is large tempco to the current.

[mawyatt]

Not only the relatively high TC but we're seeing the SD is holding ~1.1na while the Mean leakage is dropping (~40hrs below 4.5na with the Rubycon ZLH 1500uF 16V), which indicates the dielectric healing process is introducing some self-generated charge noise. Of course measurements includes any voltage fluctuations from the DC Bias source and a battery would obviously be a better source for the dielectric charge "noise" measurement.

This is not surprising as the dielectric healing in itself is a random process filling in dielectric oxide micro fissures and cracks, while thickening the overall oxide layer.

Anyway, surprising how low leakage these GP electrolytics can achieve with some "bias aging". Vaguely recall way back in 70s when we needed some larger value low leakage capacitors for the MSP project, all the electrolytics we tested had significant leakage and why we used the more expensive wet-slug tantalums.

@2N3055. What SD leakage current are you seeing with your 1000uF/50V SAMWHA RD capacitor and measurement setup?

Best

[2N3055]

Not only the relatively high TC but we're seeing the SD is holding ~1.1na while the Mean leakage is dropping (~40hrs below 4.5na with the Rubycon ZLH 1500uF 16V), which indicates the dielectric healing process is introducing some self-generated charge noise. Of course measurements includes any voltage fluctuations from the DC Bias source and a battery would obviously be a better source for the dielectric charge "noise" measurement.

This is not surprising as the dielectric healing in itself is a random process filling in dielectric oxide micro fissures and cracks, while thickening the overall oxide layer.

Anyway, surprising how low leakage these GP electrolytics can achieve with some "bias aging". Vaguely recall way back in 70s when we needed some larger value low leakage capacitors for the MSP project, all the electrolytics we tested had significant leakage and why we used the more expensive wet-slug tantalums.

@2N3055. What SD leakage current are you seeing with your 1000uF/50V SAMWHA RD capacitor and measurement setup?

Best

I was playing with it, tried direct current measurement and by using 10MΩ in volt mode as a shunt.
In current mode you are at the very low end of range and instrument noise is large.

Now it is back in volt mode and stabilizing. I will write down stdev once it settles.

[mawyatt]

I was playing with it, tried direct current measurement and by using 10MΩ in volt mode as a shunt.
In current mode you are at the very low end of range and instrument noise is large.
Now it is back in volt mode and stabilizing. I will write down stdev once it settles.

This is why we are using the KS34465A, it has very low charge injection and pa resolution. Now that we've reset the measurement it's showing ~4na bias and ~850pa SD using GPP4323 as 16V DC Bias source.

Best

[2N3055]

I was playing with it, tried direct current measurement and by using 10MΩ in volt mode as a shunt.
In current mode you are at the very low end of range and instrument noise is large.
Now it is back in volt mode and stabilizing. I will write down stdev once it settles.

This is why we are using the KS34465A, it has very low charge injection and pa resolution. Now that we've reset the measurement it's showing ~4na bias and ~850pa SD using GPP4323 as 16V DC Bias source.

Best

KS34465A has low current modes 1µA and 10µA full scale. That is useful in this case.

I have only 200µA full scale meters at hand.
On SDM4000A  with a bit of statistics and nulling out the 200µA range I was able to measure 5 nA DC repeatably with some care.
But Stdev (you asked for) is >1nV (10NPLC) so not good for this.

[mawyatt]

The lower current range is why we got the 34465A awhile back, and the DMM6500 has the lower ohms range and why we have that also. BTW we used 100NPLC for a little more averaging.

Honestly we did not expect an electrolytic of this size to have such low leakage!!

The new SDM4000A looks like a nice DMM!!

Best

[2N3055]

After 12 hours it stabilized...
It is sensitive to temperature, so me entering the room is clearly visible...

When looking into part where it is relatively flat, I see 5µV Std., conservatively... sometimes 3µV Std
That converts to 0,3-0,5 pA RMS noise.

It is close to the point of measuring voltage source noise.

That is through the 10MΩ input resistance of DMM.

[mawyatt]

Just had a nasty storm come thru (Tampa Bay seems to be the US lightning capital) and had to unplug the TE bench. Decided to use the DMM6500 which only has a 10ua lower range (KS34465A is 1ua) when reconnecting, readings are similar. Leakage is ~3na with ~650pa SD, so downward trend continues.

Maybe time to start another thread on this topic?

Best

[LooseJunkHater]

Maybe time to start another thread on this topic?

Best

If this happens, share a link here? I've been actively following-along this thread because it has been super interesting to me and I've been wanting to DIY my own LFLNA-80.

[mawyatt]

Maybe time to start another thread on this topic?

Best

If this happens, share a link here? I've been actively following-along this thread because it has been super interesting to me and I've been wanting to DIY my own LFLNA-80.

https://www.eevblog.com/forum/projects/electrolytic-leakage-measurements-and-discussion/

Best

[ZhuraYuk]

Just want to mention one more thing about my clone ZLNA, the main problem is integrator launch. It is high impedance circuit so it operates on very low current. If integrator path has high leakage the integrator will not start. Maybe in original LFLNA the FR4 material is better than in mine from jlcpcb, I do not know. You need to wash your board properly. However I have added one more improvement - remove soldermask around integrator circuit path. Also added updated gerber file to google drive https://drive.google.com/drive/u/2/folders/1QSpUXKXbVdz7Gt5mVSjBKUxbcK__LReI

Gerber_ZLNA-10_V3_PCB_ZLNA-10_V3_route2_2025-12-05_low_leakage.zip

[LooseJunkHater]

Just want to mention one more thing about my clone ZLNA, the main problem is integrator launch. It is high impedance circuit so it operates on very low current. If integrator path has high leakage the integrator will not start. Maybe in original LFLNA the FR4 material is better than in mine from jlcpcb, I do not know. You need to wash your board properly. However I have added one more improvement - remove soldermask around integrator circuit path. Also added updated gerber file to google drive https://drive.google.com/drive/u/2/folders/1QSpUXKXbVdz7Gt5mVSjBKUxbcK__LReI

Gerber_ZLNA-10_V3_PCB_ZLNA-10_V3_route2_2025-12-05_low_leakage.zip
(Attachment Link)

I'm honestly super appreciative that you continue to make minor improvements to this project! Thank you!!!

[aronake]

Not sure if it matters, but adding bypass capacitors close each OP amp may also be an idea. Probably two for each OP, + to ground and - to ground. Or maybe one + to - is enough. I added these myself with some creative soldering.

I have used the one i build quite a bit and it works very well.

[BD1QMP]

it isn't 810, it is 01B,  = 100X10=1k。
01B = 1kΩ
01C = 10kΩ
01D = 100kΩ
01E = 1MΩ
01F = 10MΩ

[/quote]
There are mistakes on schematic definitely, in schematic R210 is 1k but on PCB I clearly see that R210 is 810 - 81Ohms. You need to double check it.
[/quote]