Date: March 25, 1990

TO:Boris Seidel

FROM:Tom Kuiper

SUBJECT:Considerations for HP Spectrum Analyzer

SYSTEM NOISE CONTRIBUTION

The figures are based on the following calculations. The noise temperature of a system is due to the noise temperature of the first stage (T₁) plus the effective noise temperature of the stages which follow the first (T_1f) divided by the gain of the first stage (G₁):

T_sys = T₁ + T_1f/G₁.

Similarly, the effective noise temperature of the ith stage can be written as

T_if = T_i + T_if/G_i

so that, for example, a four stage system, would have this equation:

T_sys = T₁ + (T₂ + (T₃ + T_3f/G₃)/G₂)/G₁,

= T₁ + T₂/G₁ + T₃/G₁G₂ + T_3f/G₃G₂G₁.

This may be summarized by saying that the contribution made by a given stage to the overall system noise is the noise temperature of that stage divided by the total gain leading up to that stage. Figure 1 shows this relationship graphically.

The HP8562A spectrum analyzer has a 44 dB noise figure. We see that if the spectrum analyzer is not to contribute more than 1K to the overall system noise, the gain leading up to it must be at least 70 dB. If we assume that the low noise amplifier (LNA, stage 1) has at least 25 dB of gain, then we need a minimum of 45 dB of gain between the LNA and the spectrum analyzer. Hence, we have decided to add two staged HP8449B broad band (2 - 26 GHz) amplifiers, each with 25 dB of gain, between the LNA and the spectrum analyzer.

What about the noise contribution of these post-amplifiers? Figure 2 should help to analyze this. If a given stage is allowed to contribute a specific amount, say 1 K, to the overall system noise, then Figure 2 shows the noise temperature (or noise figure) which a given stage is allowed to have as a function of the gain leading up to that stage.

For the case of the low noise amplifier (LNA; stage 1) being a K-band maser, which we think has a gain of 30 dB, if the succeeding stages are not allowed to contribute more than 1 K to the system noise, then the effective noise temperature of the succeeding stage may not exceed 1000K (6.5 dB NF). This clearly illustrates the importance of the gain of the first stage. Since we are considering using an HP8449B broadband amplifier, with a 10 dB noise figure, as thesecond stage, we should try to have an LNA gain of 34 dB or better. This should be readily achievable, although we will need to be a bit careful with the K-band maser tuning.

Figure 1 can also help us to decide whether the spectrum analyzer with external mixers will be usable as a receiver above 26 GHz. Assuming some gain and noise temperature for the LNAs we are likely to use above 26 GHz, we can determine what would be an allowable noise contribution from the mixer and hence an acceptable mixer noise figure.

If the noise figure is acceptable, then we will still need additional gain between the mixer and the spectrum analyzer, as discussed above. Note that in the case of mixers, the insertion loss is usually included in the noise figure specification, so that the effective gain of the mixer is unity. However, we should be careful that the information we get on the mixers follows this convention.

PHASE STABILITY

According to Bob Preston, there would be considerable interest in using this configuration for VLBI. It would open up the possibility of using the DSN 70-m antennas at such standard VLBI frequencies as 5, 10, and 43 GHz for which we are currently not equipped. HP's literature does not provide a spec on this. However, we have the verbal assurance of an HP engineer atthe factory (Boris, Manuel: do you remember his name?) that the receiver is phase stable if a 10 MHz reference is provided. This seems reasonble since the spectrum analyzer is fully synthesized. Still, it would be good to have some documentation. Even better if someone in the VLBI area (Bob Preston, Don Spitzmesser, Lyle Skjerve?) could devise a test to establish this.

BANDWIDTH

The spectrum analyzer as it comes off the shelf has a bandwidth of 30 to 35 MHz, depending on the frequency of operation. For spectroscopy, a good rule of thumb is that the bandwidth should be 0.1% of the frequency, which corresponds to 300 km/s of velocity coverage. We often use less than that. So, the spectrum analyzer is suitable without modification for these kinds of observations.

Bob Preston told me that the current Mk-III VLBI terminals have 42 MHz of bandwidth. The next generation of terminals (now called the VLBA terminals) may have up to 112 MHz of bandwidth. Hence, we need to investigate a way in which the spectrum analyzer and external mixers can have wider bandwidth. The most immediate need for VLBI is to achieve the 43 GHz capability. If we need only to purchase the LNA, and not the rest of the receiver, we're likely to achieve this capability sooner.

REMOTE OPERATION

We have verified that the spectrum analyzer is fully remotely operable via its IEEE-488 interface. I have been able to read the spectrum and display it on a computer monitor. The programming required is quite trivial. However, the time to read in the spectrum is quite slow, about 5 seconds. This may be adequate for many diagnostics but is awkward for maser tuning. Having someone in the cone to report on the display works (Manuel and I tried it) but does not meet our goal of being able to operate the receiver remotely. The spectrum analyzer has video and trace blanking outputs at the back. We need to verify that these, or some other scheme, can be used to have a remote real-time display of the spectrum analyzer.

CURRENT STATUS

At present, we know for sure that the scheme of using two staged HP8449B amplifiers and an HP8562A spectrum analyzer works satisfactorily as a diagnostic tool for any of the DSN front ends. If this is sufficient justification for their acquisition, then you should go ahead with the purchase. However, this would not be our decision to make.

We promoted the acquisition on the basis of using this configuration a general purpose receiver up to 50 GHz. Where do we stand now? We have established that the configuration and a 300 Mhz IF amp with 30 dB of gain is usable as a receiver below 26 GHz for spectroscopy, continuum observations, and SETI.

At DSS-43 , we need to come up with a receiver to cover the 11-18 GHz HEMT LNA for spectroscopy. That, in itself, justifies the purchase for DSS-43 .

For DSS-13 , we have no receivers capable of covering all of the spectrum which ispotentially available. There, too, the acquisition seems justified .

The purchase seems justifiable to me for DSS-14 as well if the configuration is phase stable . The K-band receivers for DSS-14 and DSS-43 have given us years of trouble and we can expect to expect more than the cost of the proposed purchase ($50K per station) in keeping these things working over the years. However, if the configuration were not phase stable, we'd have to keep the existing K-band receivers for VLBI. In that case, our justification for buying a spectrum analyzer and amplifiers for this station is weak. (Bob Preston: can you take charge of getting this issue resolved?)

If we have three units in the field, a net spare at JPL is also justified.

A radio astronomy justification for DSS-63 seems to me to hinge on the questions of whether a) we can cover 43 GHz and b) we can have at least a 42 MHz and preferably at 112 MHz bandwidth. Note that we have these as goals for the other stations as well. Having such capability would greatly expand the radio astronomy uses of these antennas.

Note that both the DSS-13 Project (Mark Gatti) and the SETI Project (Mike Klein) have budgets for equipment acquisition. If you fall short, because of the cost of modifying the bandwidth for example, then you may be able to negotiate with them.