LMH6629: Survey of opamps based on a SiGe process for cryogenic applications

Hello, 

Not sure about the LMH6629, but the following might be useful for you

OPA837 (upgraded OPA836)

THS4551 (FDA)

OPA838 (decomp like the LMH6629)

Not really sure that the SiGe thing is what is making this better at lower temps, can you attach your reference of using some of these at 77k?

Hello,

thank you for the list! The 836/837 input voltage noise is quite high, not sure how it will beheave at low temperature... OPA838 could be worth a try... I'm also considering a hybrid with a discrete stage in front

AFAIK BJT and JFET fail at low temperatures, while SiGe HBT as well as HEMT remain usable (HBT beta in fact rises). For the LMH6629 see the following paper: https://arxiv.org/pdf/1706.04213

Hello, I started skimming through that article and it seems they might have missed one really important point, 

Figure 2 shows the amplifier GBP significantly increasing going cold. That means, for transimpedance design, as you generated these figures 10 and the GBP was increasing going cold, the feedback cap needed to be adjusted to avoid response peaking as shown in the noise plot in figure 11 - more room to improve you design methodology it would seem. We tried to simplify this design flow in this presentation. The simple summary is that the ratio of the feedback pole to the Fo described in the ppt is the Q of the closed loop response. Also, I did not see what the Cd was in the article, but adding 10pF on the inverting node is unnecessarily hurting the noise and bandwidth. 

1856.Transimpedance design flow using high speed op amps.pptx

Thank you for the hint - definitely an insightful read!

I am wondering at what point the GBP would byte you due to the parasitics of the feedback, no matter the gain...

Well yes, at higher Rf you eventually get into fempto Farad comp cap solutions, I go to Tee networks at that point, 

more specifically, having measured quite a few surface mount resistors I use 0.18pF parasitic as a good guess. If you need lower, at lower values (<20k) Precision Resistive Products makes what we called Red Buf axial resistors that are <.05pF parasitic. They may have surface mount versions now. Those were used in a lot of the early HP network analyzers. Not sure what they use now but would likely be instructive. 

Hi,

I've tested some opamps and discretes at 77 K and below:

ADA 4897: at 77 K LF noise goes up ~10x

THS4303: at 77 K LF noise goes up ~2x

OPA211: noise increases at 77K, does not function at 4K

For discretes, NXP's SiGe BFU series and the discontinued SiGe NESG3031 work at 10 mK, with increase in beta for large Ic.

For HEMTs, ATF36136 works at mK temperatures, as do Yong Jin's HEMTs 

I have a question for Michael Steffes:

Discrete SiGe transistors work down below 4 K many SiGe opamps I've tested fail. This probably has to due with internal resistors whose values diverge at low temperature. When a VFA is decompensated, you said in your planetanalog article that the degeneration resistor is reduced. This may allow them to work at lower temperatures. Do you know of SiGe op-amps with "minimal" internal resistance?

Çağlar Girit