Run 4

1. Replaced Qi gate wire with low noise copper coaxial cable ( TexCal 250-3820-0000, 30pF/ft) in the longest side coax. The length of the wire/Cable is about 5.6“.
2. Tested noise performance with DCRC v2 and SR760.
   • DCRC and SR760 gave the same result for Qi. link
   • Low noise coaxial cable has comparable performance with vacuum coax when quiet. link
   • A conventional coaxial cable is prone to microphonic noise while the vacuum coax is not. link
   • DCRC noise spectra have a frequency dependent slope while Soudan data is flat. link

matlab file

The plot above shows the Qi (conventional coax) noise PSD from the SR760, which monitored the test points (TP5) on the feedback line, and from DCRC. Both are normalized to the feedback line output. They overlap well at lower frequencies. It is not clear why the baseline of the curve from SR760 is slightly higher around a few kHz. Below 1kHz, DCRC curves have more peaks around mutiples of 60Hz. Sten suggested to ground the digital ground to chassis. There is no effect I can tell by grouing the digital ground.

Qo (vacuum coax) and Qi comparison
matlab fig. Data taken with DCRC v2.

Qo (vacuum coax) and Qi comparison at low frequencies
matlab fig. Data taken with SR760.

Low noise copper coaxial cable has comparable performance with vacuum coax when quiet. The noise PSD at low frequencies was taken with the SR760 and an oscilloscope probe to monitor the feedback line. Replacing the probe with regular wire showed that the attenuation effect from the probe was insignificant. All other results detailed here were taken with the regular wire.

Qo (vacuum coax) microphonics test
matlab fig

Qo (vacuum coax) noise psd from DCRC.

• 'Knocking': a 1/4-20 x 11/2” hex socket cap screw was used to knock on the Al plate to which the fridge dewar is bolted with a frequency of a few hertz.
• 'Big knock': a hand size plier was used to hit the Al plate during the data taking.

These curves almost overlap across the frequency range.

matlab fig

Qo noise psd taken by DCRC with trace length 128ms. Used a soft hammer knocking on the Al frame to introduce microphonics while taking data.

Qi (regular coax) microphonics test
matlab fig

The curve with 'big knock' shows various peaks and baseline change due to mircophonics compared with the quiet one. After this, we decided to quantify the interference we introduced by dropping a 1/4-20×11/2 screw on to the fridge dewar at controlled heights. We looked at the SR760 while taking the PSD using the DCRC. There was not much change upon dropping the screw from 10“. However, dropping the screw from heights of 20” and 30“ resulted in new peaks at around 6kHz and 4kHz. The curves on the plot were from average over 100 traces. The peaks are washed out a bit by this average.

matlab fig

Qi noise psd taken by DCRC with trace length 128ms. Used a soft hammer knocking on the Al frame to introduce microphonics while taking data.

matlab file

After conforming that DCRC and SR760 have consistent results. A comparison is made between Qi and Qo using DCRC with a higher frequency resolution. The three peaks around 26kHz on Qi are from the fluorescence lights. They can reduced by turnning off the lights near the fridge.

Microphonic noise is related to charge transport near the conductor/insulator interface in a cable. This noise may be reduced if the number of conduction electrons in the conductor can be reduced. Super conductors have a small number of conduction electrons (temperature dependent) when temperature is below Tc. Would a superconducting coaxial cable be able to reduce/remove microphonic noise?