Total Power Digitizer Dynamic Range

Question from Clint Janes and Bill Brundage:

There is some question about the dynamic range required in switching from a 300K noise source to observing levels; does the 3-bit digitizer have the range? The basic design assumption has been constant system gain for an entire cycle of observe and calibration. If the calibration implementation includes periodically adding 300K to 400K of noise to Trcvr, what is the impact on the baseband (2-4 GHz) 3-bit digitizer (DG) and astronomical data? If the ratio (Tcal + Trcvr) / Tsysobserve is as large as 8 (might be 10?), then the DG may effectively be a 1-bit or 0-bit digitizer at Tsysobserve and also a 1-bit or 0-bit digitizer at (Tcal + Trcvr). Not good.


Here is a summary of the following email exchange, where I have attempted to insert some real numbers to put the issue on more solid ground.

In the end, Larry and Bryan came to the conclusion that:

  • The worst-case for the dynamic range of the TP digitizers was for observations at the coldest of sky conditions (35 GHz).
  • The dynamic range of interest for this discussion is that which is seen by the TP digitizers during a given receiver setup. If the band, LO frequency, or anything else in the signal processing changes, we can readjust the attenuator start over.
  • If we readjust only the attenuator, that too is a setup change. Generally, we shouldn't do this unless something else in the setup is significantly changed, but we might do so if
    • we are willing to re-calibrate the complex gain astronomically after the attenuator change, so that none of the previous calibration data applies to subsequent observations, or
    • we are doing total power only, so phase doesn't matter, and we think we have sufficiently accurate knowledge of the difference in effective attenuation between the old and new settings.

Using the formalism presented in ALMA Memo 276 (Butler and Wootten), the system temperature below the atmosphere is given by:

Tsys = Trx + Tsky + Tant + Tcmbr
     = Trx + etal*Tatm*[1-exp(-tau0*A)] + (1-etal)*Tspill + Tcmb*exp(-tau0*A)

Now assume the following:

etal = 0.95
Tatm = Tspill = Tamb < 290 K
A = 1
tau0 = 0.016 (ALMA Memo 276)

Then, the minimum value for Tsys is given by...

Tsys < 15K + 4K + 15K + 3K = 37K

For the insertion of a 400K calibration load, the maximum Tsys becomes...

Tsys,max = Trx + Tload
         = 15K + 400K
         = 415K

Therefore, the worst-case ratio of max to min Tsys becomes...

Tsys,max/Tsys,min = 415K/37K = 11

This indicates that a dynamic range of 20:1 at the TP digitizers is sufficient.

-- JeffMangum - 20 Jul 2004

The following is a resurrection of a series of emails relating to the dynamic range for the TP digitizers...

Question from Clint...


From: Clint Janes <cjanes@aoc.nrao.edu>
To: bbutler@zia.aoc.nrao.edu, awooten <awooten@aoc.nrao.edu>
CC: "ldaddario >> Larry D'Addario" <ldaddario@zia.aoc.nrao.edu>,
   Dick Sramek
 <dsramek@zia.aoc.nrao.edu>,
   "Cunningham, Charles"
 <Charles.Cunningham@nrc-cnrc.gc.ca>,
   Jeff Mangum <jmangum@tuc.nrao.edu>, Bill Brundage <wbrundag@aoc.nrao.edu>
Subject: [Fwd: Re: [Fwd: TP Digitizer]]
Date: Tue, 17 Jun 2003 15:36:45 -0600
Message-ID: <3EEF89ED.3040304@aoc.nrao.edu>
X-MailScanner-SpamCheck: not spam, SpamAssassin (score=-1.2, required 7,
   DEAR_SOMEBODY, FWD_MSG, NOSPAM_INC, SPAM_PHRASE_00_01, USER_AGENT,
   USER_AGENT_MOZILLA_UA, X_ACCEPT_LANG)
X-MailScanner-SpamCheck: not spam, SpamAssassin (score=-1.2, required 7,
   DEAR_SOMEBODY, FWD_MSG, NOSPAM_INC, SPAM_PHRASE_00_01, USER_AGENT,
   USER_AGENT_MOZILLA_UA, X_ACCEPT_LANG)

Dear Al and Bryan:

      Bill Brundage plans for a dynamic range of 13 dB for the total 
power (TP) detector/]digitizers.  It is assumed that the receiver gain 
will be adjusted for the signal to fall within this range.  I am writing 
to see if this meets requirements.
   Bryan Butler explains that the basic requirement is to go from lowest 
possible system temperature to a hot calibration load of 300K.
13 dB provides for a ratio of 20:1, or a system temperature range of 15K 
to 300K.
      Bill will provide 1% linearity for TP using EPROMs to characterize 
the performance of the sensing diodes over the operating power range.
   Presumably, FE will attenuate solar (Tsys) so that the signal to the TP 
diodes will fall within the equivalent Tsys of 15K to 300K.
   Please let me know if this plan does not meet scientific requirements, 
thanks.

                          Best regards,

                          Clint

Bryan adds a twist...


From: Bryan Butler <bbutler@aoc.nrao.edu>
To: Clint Janes <cjanes@zia.aoc.NRAO.EDU>
Cc: bbutler@zia.aoc.NRAO.EDU, awooten <awooten@zia.aoc.NRAO.EDU>,
   "ldaddario >> Larry D'Addario" <ldaddario@zia.aoc.NRAO.EDU>,
   Dick Sramek <dsramek@zia.aoc.NRAO.EDU>,
   "Cunningham, Charles" <Charles.Cunningham@nrc-cnrc.gc.ca>,
   Jeff Mangum <jmangum@tuc.nrao.edu>,
   Bill Brundage <wbrundag@zia.aoc.NRAO.EDU>
Subject: Re: [Fwd: Re: [Fwd: TP Digitizer]]
Date: Tue, 17 Jun 2003 16:05:18 -0600
Message-ID: <20030617220518.GC10020@planetas.aoc>
X-MailScanner-SpamCheck: not spam, SpamAssassin (score=-2.6, required 7,
   DEAR_SOMEBODY, EMAIL_ATTRIBUTION, FWD_MSG, IN_REP_TO, REFERENCES,
   SPAM_PHRASE_00_01)


one point is that already we are thinking that one of the loads
will have to be a "hot" (with "hot" TBD, but > 370 K) load.
so the 20:1 range will get you 20 to 400 K instead of 15 to 300 K.


   -bryan


On 2003.06.17 15:36 Clint Janes wrote:
> Dear Al and Bryan:
> 
>       Bill Brundage plans for a dynamic range of 13 dB for the total 
> power (TP) detector/]digitizers.  It is assumed that the receiver gain 
> will be adjusted for the signal to fall within this range.  I am writing 
> to see if this meets requirements.
>    Bryan Butler explains that the basic requirement is to go from lowest 
> possible system temperature to a hot calibration load of 300K.
> 13 dB provides for a ratio of 20:1, or a system temperature range of 15K 
> to 300K.
>       Bill will provide 1% linearity for TP using EPROMs to characterize 
> the performance of the sensing diodes over the operating power range.
>    Presumably, FE will attenuate solar (Tsys) so that the signal to the TP 
> diodes will fall within the equivalent Tsys of 15K to 300K.
>    Please let me know if this plan does not meet scientific requirements, 
> thanks.
> 
>                           Best regards,
> 
>                           Clint
> 
> 

Larry's input...


From: "Larry D'Addario" <ldaddario@nrao.edu>
To: Bryan Butler <bbutler@aoc.nrao.edu>
CC: Clint Janes <cjanes@zia.aoc.nrao.edu>, bbutler@zia.aoc.nrao.edu,
   awooten <awooten@zia.aoc.nrao.edu>,
   "ldaddario >> Larry D'Addario" <ldaddario@zia.aoc.nrao.edu>,
   Dick Sramek <dsramek@zia.aoc.nrao.edu>,
   "Cunningham, Charles" <Charles.Cunningham@nrc-cnrc.gc.ca>,
   Jeff Mangum <jmangum@tuc.nrao.edu>,
   Bill Brundage <wbrundag@zia.aoc.nrao.edu>
Subject: Re: [Fwd: Re: [Fwd: TP Digitizer]]
Date: Tue, 17 Jun 2003 18:04:59 -0700
Message-ID: <3EEFBABB.A49CBB00@nrao.edu>

Bryan,

I don't think you mean a hot load that covers the entire beam, so you
need to consider the diultion factor.  The antenna temperature will then
be far less than 370K.  If you really do mean a full-beam hot load, then
I question whether a reasonable design is possible.  Furthermore, the
fact that it forces a larger dynamic range on the entire receiving
system should be a *strong* negative factor to include in a pro/con
list.

--Larry 

...and Bryan sets Larry straight...


From: Bryan Butler <bbutler@aoc.nrao.edu>
To: "Larry D'Addario" <ldaddario@cv3.cv.nrao.edu>
Cc: Clint Janes <cjanes@zia.aoc.nrao.edu>, bbutler@zia.aoc.nrao.edu,
   awootten <awooten@zia.aoc.nrao.edu>,
   "ldaddario >> Larry D'Addario" <ldaddario@zia.aoc.nrao.edu>,
   Dick Sramek <dsramek@zia.aoc.nrao.edu>,
   "Cunningham, Charles" <Charles.Cunningham@nrc-cnrc.gc.ca>,
   Jeff Mangum <jmangum@tuc.nrao.edu>,
   Bill Brundage <wbrundag@zia.aoc.nrao.edu>
Subject: Re: [Fwd: Re: [Fwd: TP Digitizer]]
Date: Wed, 18 Jun 2003 09:52:54 -0600
Message-ID: <20030618155254.GO10020@planetas.aoc>
X-MailScanner-SpamCheck: not spam, SpamAssassin (score=-2.7, required 7,
   EMAIL_ATTRIBUTION, FWD_MSG, IN_REP_TO, REFERENCES,
   SPAM_PHRASE_00_01)



no, it covers the entire beam - except for band 1, which is not 
included in that scheme.  this has *always* been the plan with
the calibration loads - that they cover the entire beam.  only
the dual-load in the subreflector didn't - and there you have
the problem of the variable coupling factor.

in fact it's worse than i said yesterday, since in the 675 and
850 GHz windows the atmosphere + receivers give Tsys of several
thousand K, making the load temperature a non-issue.


   -bryan


On 2003.06.17 19:04 Larry D'Addario wrote:
> Bryan,
> 
> I don't think you mean a hot load that covers the entire beam, so you
> need to consider the diultion factor.  The antenna temperature will then
> be far less than 370K.  If you really do mean a full-beam hot load, then
> I question whether a reasonable design is possible.  Furthermore, the
> fact that it forces a larger dynamic range on the entire receiving
> system should be a *strong* negative factor to include in a pro/con
> list.
> 
> --Larry 

...and Al points out that the sky is probably our hottest "load"...


From: Al Wootten <awootten@nrao.edu>
To: Bryan Butler <bbutler@aoc.nrao.edu>
Cc: "Larry D'Addario" <ldaddario@cv3.cv.nrao.edu>,
   Clint Janes <cjanes@zia.aoc.nrao.edu>,
   "ldaddario >> Larry D'Addario" <ldaddario@zia.aoc.nrao.edu>,
   Dick Sramek <dsramek@zia.aoc.nrao.edu>,
   "Cunningham, Charles" <Charles.Cunningham@nrc-cnrc.gc.ca>,
   Jeff Mangum <jmangum@tuc.nrao.edu>,
   Bill Brundage <wbrundag@zia.aoc.nrao.edu>
Subject: Re: [Fwd: Re: [Fwd: TP Digitizer]]
Date: Wed, 18 Jun 2003 14:07:21 -0400
Message-ID: <16112.43609.984808.263841@polaris.cv.nrao.edu>
X-MailScanner-SpamCheck: not spam, SpamAssassin (score=-101.1, required 7,
   FWD_MSG, IN_REP_TO, REFERENCES, SPAM_PHRASE_00_01,
   USER_IN_WHITELIST)

Certainly the sky gives us our hottest load.  The 'ALMA Sensitivity
Calculator' is an easy way to estimate the value of Tsys for various
source parameters under good atmospheric conditions:
http://www.eso.org/projects/alma/science/bin/sensitivity.html?cmd=s&cfreq=655&velres=1&angres=1&inttime=60

For this, Tsys at 650 GHz can be expected to be 1300 K (assuming
receivers of spec performance).

-Al

Bill asks the fundamental question...


From: Bill Brundage <wbrundag@aoc.nrao.edu>
To: Bryan Butler <bbutler@aoc.nrao.edu>,
   Clint Janes
 <cjanes@zia.aoc.NRAO.EDU>,
   Al Wooten <awooten@zia.aoc.NRAO.EDU>,
   "Larry D'Addario"
 <ldaddario@zia.aoc.NRAO.EDU>,
   Dick Sramek <dsramek@zia.aoc.NRAO.EDU>,
   Charles
 <Charles.Cunningham@nrc-cnrc.gc.ca>,
   Jeff Mangum <jmangum@tuc.nrao.edu>
Subject: Re: [Fwd: Re: [Fwd: TP Digitizer]]
Date: Wed, 18 Jun 2003 13:56:21 -0600
Message-ID: <3EF0C3E5.2010402@aoc.nrao.edu>
X-MailScanner-SpamCheck: not spam, SpamAssassin (score=-3.3, required 7,
   DEAR_SOMEBODY, EMAIL_ATTRIBUTION, FWD_MSG, NOSPAM_INC, REFERENCES,
   SPAM_PHRASE_00_01, USER_AGENT, USER_AGENT_MOZILLA_UA, X_ACCEPT_LANG)
X-MailScanner-SpamCheck: not spam, SpamAssassin (score=-3.3, required 7,
   DEAR_SOMEBODY, EMAIL_ATTRIBUTION, FWD_MSG, NOSPAM_INC, REFERENCES,
   SPAM_PHRASE_00_01, USER_AGENT, USER_AGENT_MOZILLA_UA, X_ACCEPT_LANG)

All,
   Let's not overlook the fundamental question for the total power 
detector data:
   What is the required dynamic range of total power data during 
the "worst case" observing cycle of source, calibration source 
and calibration load, for any ALMA band.
   Here I define "required dynamic range" as (Tsysmax / Tsysmin) 
for the worst case cycle, assuming constant system gain during 
the cycle.
...Bill

Bryan Butler wrote:
> one point is that already we are thinking that one of the loads
> will have to be a "hot" (with "hot" TBD, but > 370 K) load.
> so the 20:1 range will get you 20 to 400 K instead of 15 to 300 K.
> 
> 
>    -bryan
> 
> 
> On 2003.06.17 15:36 Clint Janes wrote:
> 
>>Dear Al and Bryan:
>>
>>      Bill Brundage plans for a dynamic range of 13 dB for the total 
>>power (TP) detector/]digitizers.  It is assumed that the receiver gain 
>>will be adjusted for the signal to fall within this range.  I am writing 
>>to see if this meets requirements.
>>   Bryan Butler explains that the basic requirement is to go from lowest 
>>possible system temperature to a hot calibration load of 300K.
>>13 dB provides for a ratio of 20:1, or a system temperature range of 15K 
>>to 300K.
>>      Bill will provide 1% linearity for TP using EPROMs to characterize 
>>the performance of the sensing diodes over the operating power range.
>>   Presumably, FE will attenuate solar (Tsys) so that the signal to the TP 
>>diodes will fall within the equivalent Tsys of 15K to 300K.
>>   Please let me know if this plan does not meet scientific requirements, 
>>thanks.
>>
>>                          Best regards,
>>
>>                          Clint
>>
>>
> 
> 
> 

Bryan asks if the 13dB scale is a sliding scale...


From: Bryan Butler <bbutler@aoc.nrao.edu>
To: Bill Brundage <wbrundag@zia.aoc.NRAO.EDU>
Cc: Bryan Butler <bbutler@zia.aoc.NRAO.EDU>,
   Clint Janes <cjanes@zia.aoc.NRAO.EDU>, Al Wooten <awooten@zia.aoc.NRAO.EDU>,
   "Larry D'Addario" <ldaddario@zia.aoc.NRAO.EDU>,
   Dick Sramek <dsramek@zia.aoc.NRAO.EDU>,
   Charles <Charles.Cunningham@nrc-cnrc.gc.ca>,
   Jeff Mangum <jmangum@tuc.nrao.edu>
Subject: Re: [Fwd: Re: [Fwd: TP Digitizer]]
Date: Wed, 18 Jun 2003 13:59:04 -0600
Message-ID: <20030618195904.GS10020@planetas.aoc>
X-MailScanner-SpamCheck: not spam, SpamAssassin (score=-2.9, required 7,
   DEAR_SOMEBODY, EMAIL_ATTRIBUTION, FWD_MSG, IN_REP_TO, REFERENCES,
   SPAM_PHRASE_01_02)


when i was talking to clint about this, it was not clear whether
thi 13dB meant what you are describing it as - i.e., that this 
is a sliding scale, not an absolute one.


   -bryan


On 2003.06.18 13:56 Bill Brundage wrote:
> All,
>    Let's not overlook the fundamental question for the total power 
> detector data:
>    What is the required dynamic range of total power data during 
> the "worst case" observing cycle of source, calibration source 
> and calibration load, for any ALMA band.
>    Here I define "required dynamic range" as (Tsysmax / Tsysmin) 
> for the worst case cycle, assuming constant system gain during 
> the cycle.
> ...Bill
> 
> Bryan Butler wrote:
> > one point is that already we are thinking that one of the loads
> > will have to be a "hot" (with "hot" TBD, but > 370 K) load.
> > so the 20:1 range will get you 20 to 400 K instead of 15 to 300 K.
> > 
> > 
> >    -bryan
> > 
> > 
> > On 2003.06.17 15:36 Clint Janes wrote:
> > 
> >>Dear Al and Bryan:
> >>
> >>      Bill Brundage plans for a dynamic range of 13 dB for the total 
> >>power (TP) detector/]digitizers.  It is assumed that the receiver gain 
> >>will be adjusted for the signal to fall within this range.  I am writing 
> >>to see if this meets requirements.
> >>   Bryan Butler explains that the basic requirement is to go from lowest 
> >>possible system temperature to a hot calibration load of 300K.
> >>13 dB provides for a ratio of 20:1, or a system temperature range of 15K 
> >>to 300K.
> >>      Bill will provide 1% linearity for TP using EPROMs to characterize 
> >>the performance of the sensing diodes over the operating power range.
> >>   Presumably, FE will attenuate solar (Tsys) so that the signal to the TP 
> >>diodes will fall within the equivalent Tsys of 15K to 300K.
> >>   Please let me know if this plan does not meet scientific requirements, 
> >>thanks.
> >>
> >>                          Best regards,
> >>
> >>                          Clint
> >>
> >>
> > 
> > 
> > 
> 
> 
> 

...and Bill answers "yes" to Bryan's question...


From: Bill Brundage <wbrundag@aoc.nrao.edu>
To: Bryan Butler <bbutler@aoc.nrao.edu>
CC: Clint Janes <cjanes@zia.aoc.nrao.edu>,
   Al Wooten
 <awooten@zia.aoc.nrao.edu>,
   "Larry D'Addario" <ldaddario@zia.aoc.nrao.edu>,
   Dick Sramek <dsramek@zia.aoc.nrao.edu>,
   Charles
 <Charles.Cunningham@nrc-cnrc.gc.ca>,
   Jeff Mangum <jmangum@tuc.nrao.edu>
Subject: Re: [Fwd: Re: [Fwd: TP Digitizer]]
Date: Wed, 18 Jun 2003 14:36:59 -0600
Message-ID: <3EF0CD6B.2000104@aoc.nrao.edu>
X-MailScanner-SpamCheck: not spam, SpamAssassin (score=-3.6, required 7,
   DEAR_SOMEBODY, EMAIL_ATTRIBUTION, FWD_MSG, NOSPAM_INC, REFERENCES,
   SPAM_PHRASE_01_02, USER_AGENT, USER_AGENT_MOZILLA_UA, X_ACCEPT_LANG)
X-MailScanner-SpamCheck: not spam, SpamAssassin (score=-3.6, required 7,
   DEAR_SOMEBODY, EMAIL_ATTRIBUTION, FWD_MSG, NOSPAM_INC, REFERENCES,
   SPAM_PHRASE_01_02, USER_AGENT, USER_AGENT_MOZILLA_UA, X_ACCEPT_LANG)

Bryan,
   Correct, a sliding scale from cycle to cycle, where gain may 
change only between cycles.  Software anticipates Tsysmax/Tsysmin 
for the next observing cycle and sets gain (thus total power 
operating level) accordingly.  This has been the observing 
process assumed for desiging the downconverter.  If wrong, I must 
know the true process very soon.
   If radiometric tests for commissioning a new antenna require a 
precision dynamic range greater than 20:1, top-line commercial 
power meters/sensors may be required.
  ...Bill

Bryan Butler wrote:
> when i was talking to clint about this, it was not clear whether
> thi 13dB meant what you are describing it as - i.e., that this 
> is a sliding scale, not an absolute one.
> 
> 
>    -bryan
> 
> 
> On 2003.06.18 13:56 Bill Brundage wrote:
> 
>>All,
>>   Let's not overlook the fundamental question for the total power 
>>detector data:
>>   What is the required dynamic range of total power data during 
>>the "worst case" observing cycle of source, calibration source 
>>and calibration load, for any ALMA band.
>>   Here I define "required dynamic range" as (Tsysmax / Tsysmin) 
>>for the worst case cycle, assuming constant system gain during 
>>the cycle.
>>...Bill
>>
>>Bryan Butler wrote:
>>
>>>one point is that already we are thinking that one of the loads
>>>will have to be a "hot" (with "hot" TBD, but > 370 K) load.
>>>so the 20:1 range will get you 20 to 400 K instead of 15 to 300 K.
>>>
>>>
>>>   -bryan
>>>
>>>
>>>On 2003.06.17 15:36 Clint Janes wrote:
>>>
>>>
>>>>Dear Al and Bryan:
>>>>
>>>>     Bill Brundage plans for a dynamic range of 13 dB for the total 
>>>>power (TP) detector/]digitizers.  It is assumed that the receiver gain 
>>>>will be adjusted for the signal to fall within this range.  I am writing 
>>>>to see if this meets requirements.
>>>>   Bryan Butler explains that the basic requirement is to go from lowest 
>>>>possible system temperature to a hot calibration load of 300K.
>>>>13 dB provides for a ratio of 20:1, or a system temperature range of 15K 
>>>>to 300K.
>>>>     Bill will provide 1% linearity for TP using EPROMs to characterize 
>>>>the performance of the sensing diodes over the operating power range.
>>>>   Presumably, FE will attenuate solar (Tsys) so that the signal to the TP 
>>>>diodes will fall within the equivalent Tsys of 15K to 300K.
>>>>   Please let me know if this plan does not meet scientific requirements, 
>>>>thanks.
>>>>
>>>>                         Best regards,
>>>>
>>>>                         Clint
>>>>
>>>>
>>>
>>>
>>>
>>
>>
> 
> 

Larry's comments to the whole discussion...


From: "Larry D'Addario" <ldaddario@nrao.edu>
To: Bill Brundage <wbrundag@aoc.nrao.edu>
Cc: Bryan Butler <bbutler@zia.aoc.nrao.edu>,
   Clint Janes <cjanes@zia.aoc.nrao.edu>,
   Al Wooten
 <awooten@zia.aoc.nrao.edu>,
   "Larry D'Addario" <ldaddario@zia.aoc.nrao.edu>,
   Dick Sramek <dsramek@zia.aoc.nrao.edu>,
   Charles
 <Charles.Cunningham@nrc-cnrc.gc.ca>,
   Jeff Mangum <jmangum@tuc.nrao.edu>
Subject: Re: [Fwd: Re: [Fwd: TP Digitizer]]
Date: Thu, 26 Jun 2003 16:59:37 -0700
Message-ID: <16123.35049.813678.55521@brador.tuc.nrao.edu>
X-MailScanner-SpamCheck: not spam, SpamAssassin (score=-102.7, required 7,
   EMAIL_ATTRIBUTION, FWD_MSG, IN_REP_TO, REFERENCES, SPAM_PHRASE_00_01,
   USER_IN_WHITELIST)

Folks,

There was a thread of email on this subject a couple of weeks ago, and
I'm just getting around to responding after being away on a couple of
trips.  It seems to me that the situation is very confused.  I hope I
don't confuse it any further.

Bryan Butler writes:
 > one point is that already we are thinking that one of the loads
 > will have to be a "hot" (with "hot" TBD, but > 370 K) load.
 > so the 20:1 range will get you 20 to 400 K instead of 15 to 300 K.

It's the system temperature that matters, not the antenna (load)
temperature.

Al Wootten writes:
 > Certainly the sky gives us our hottest load.  The 'ALMA Sensitivity
 > Calculator' is an easy way to estimate the value of Tsys for
 > various source parameters under good atmospheric conditions:
 > http://www.eso.org/projects/alma/science/bin/sensitivity.html?cmd=s&cfreq=655&velres=1&angres=1&inttime=60
 > 
 > For this, Tsys at 650 GHz can be expected to be 1300 K (assuming
 > receivers of spec performance).

I haven't used this "sensitivity calculator" but I'll bet that it
computes Tsys*, the "outside the atmosphere" value.  While astronomers
like this parameter, it's entirely unphysical and not of interest in
the present discussion.  What we want is the total power at some
physical place in the signal processing chain, and we usually find it
convenient to refer to the receiver input, where we have $$ T_sys =
T_ant + T_rcvr $$.  When pointed at the sky, we always have T_ant <
290K, since the average atmosphere temp along the line of sight is
never more than 290K.  And the front end specs require T_rcvr < 655K
at the edges of Band 10 and much less everywhere else.  Thus the
absolute worst is Tsys=945K with an opaque sky.  But we never observe
with an opaque sky, so the worst practical value is around Tsys=800K.

If we have a hot load at 370K then certainly this is hotter than
the sky will ever be, unless we're pointing at the Sun.

Bill Brundage writes:
 >    Correct, a sliding scale from cycle to cycle, where gain may 
 > change only between cycles.  Software anticipates Tsysmax/Tsysmin 
 > for the next observing cycle and sets gain (thus total power 
 > operating level) accordingly.  This has been the observing 
 > process assumed for desiging the downconverter.  If wrong, I must 
 > know the true process very soon.

Well, it's a bit wrong.  It depends what you mean by a "cycle."  And
it depends on whether we're doing total power radiometry with the
analog detectors or doing interferometry with the correlator.

First, we are not at all concerned with the total range of system
temperature but only with the range that can occur while the receiver
*setup* remains unchanged.  If the band, LO frequency, or anything
else in the signal processing changes, we can readjust the attenuator
and we start over.  If we readjust *only* the attenuator, that too is
a setup change.  Generally, we shouldn't do this unless something else
in the setup is significantly changed, but we might do so if (a) we
are willing to re-calibrate the complex gain astronomically after the
attenuator change, so that none of the previous calibration data
applies to subsequent observations; or (b) we are doing total power
only, so phase doesn't matter, and we think we have sufficiently
accurate knowledge of the difference in effective attenuation between
the old and new settings.

Begin by assuming that, for a fixed setup, the receiver temperature is
very stable.  Then the only thing that changes is the antenna
temperature.  If we will sometimes be looking at hot loads (ambient
temperature or above), then the largest range will occur when the sky
is coldest.  Presumably this is for the lowest frequency bands.  Those
bands have the lowest T_rcvr too, so they produce the largest ratio of
maximum to minimum T_sys.  

For Band 1, we should have T_rcvr < 15K and we might have T_ant as low
as 30K.  This gives T_sys max/min of (15+300)/(15+30)=7.0 if the hot
load is at 300K, or (15+370)/(15+30)=8.55 if the hot load is at 370K.
For all other bands, the ratio is smaller.  I have not considered what
happens when pointing at the Sun; presumably an RF attenuator will
then be inserted.  If looking at active regions on the Sun, the
restrictions on re-adjusting the IF attenuation might have to be
relaxed.

Bill Brundage writes:
 >    If radiometric tests for commissioning a new antenna require a 
 > precision dynamic range greater than 20:1, top-line commercial 
 > power meters/sensors may be required.

It seems to me that, except for solar observations, the dynamic range
is far less than 20:1.


Next, consider the linearity requirements (degree of allowed
saturation or compression) over the dynamic range.  This is a bit more
complicated, so I'll cover it in a later email.

--Larry

Bryan's summary analysis


From: Bryan Butler <bbutler@aoc.nrao.edu>
To: "Larry D'Addario" <ldaddario@zia.aoc.NRAO.EDU>
Cc: Bill Brundage <wbrundag@zia.aoc.NRAO.EDU>,
   Bryan Butler <bbutler@zia.aoc.NRAO.EDU>,
   Clint Janes <cjanes@zia.aoc.NRAO.EDU>, Al Wooten <awooten@zia.aoc.NRAO.EDU>,
   "Larry D'Addario" <ldaddario@zia.aoc.NRAO.EDU>,
   Dick Sramek <dsramek@zia.aoc.NRAO.EDU>,
   Charles <Charles.Cunningham@nrc-cnrc.gc.ca>,
   Jeff Mangum <jmangum@tuc.nrao.edu>
Subject: Re: [Fwd: Re: [Fwd: TP Digitizer]]
Date: Mon, 7 Jul 2003 08:48:31 -0600
Message-ID: <20030707144831.GM32651@planetas.aoc.nrao.edu>
X-MailScanner-SpamCheck: not spam, SpamAssassin (score=-2.7, required 7,
   EMAIL_ATTRIBUTION, FWD_MSG, IN_REP_TO, REFERENCES,
   SPAM_PHRASE_00_01)



i think larry is right here, just added a few comments.


   -bryan


On 2003.06.26 17:59 Larry D'Addario wrote:
> Folks,
> 
> There was a thread of email on this subject a couple of weeks ago, and
> I'm just getting around to responding after being away on a couple of
> trips.  It seems to me that the situation is very confused.  I hope I
> don't confuse it any further.
> 
> Bryan Butler writes:
>  > one point is that already we are thinking that one of the loads
>  > will have to be a "hot" (with "hot" TBD, but > 370 K) load.
>  > so the 20:1 range will get you 20 to 400 K instead of 15 to 300 K.
> 
> It's the system temperature that matters, not the antenna (load)
> temperature.


yes, as i pointed out right after my first email...


> 
> Al Wootten writes:
>  > Certainly the sky gives us our hottest load.  The 'ALMA Sensitivity
>  > Calculator' is an easy way to estimate the value of Tsys for
>  > various source parameters under good atmospheric conditions:
>  > 
> http://www.eso.org/projects/alma/science/bin/sensitivity.html?cmd=s&cfreq=655&velres=1&angres=1&inttime=60
>  >
>  > For this, Tsys at 650 GHz can be expected to be 1300 K (assuming
>  > receivers of spec performance).
> 
> I haven't used this "sensitivity calculator" but I'll bet that it
> computes Tsys*, the "outside the atmosphere" value.


i don't know the details either, but i'm pretty sure you're right.


> While astronomers
> like this parameter, it's entirely unphysical


not quite fair, since it is *quite* physical when you are discussing
the emitting source itself.  but you are right that it is not the
proper quantity in this context.


> and not of interest in
> the present discussion.  What we want is the total power at some
> physical place in the signal processing chain, and we usually find it
> convenient to refer to the receiver input, where we have $$ T_sys =
> T_ant + T_rcvr $$.  When pointed at the sky, we always have T_ant <
> 290K, since the average atmosphere temp along the line of sight is
> never more than 290K.  And the front end specs require T_rcvr < 655K
> at the edges of Band 10 and much less everywhere else.  Thus the
> absolute worst is Tsys=945K with an opaque sky.  But we never observe
> with an opaque sky, so the worst practical value is around Tsys=800K.
> 
> If we have a hot load at 370K then certainly this is hotter than
> the sky will ever be, unless we're pointing at the Sun.


there are other sources this bright (planets come to mind [venus and
mercury anyway] - i suspect there are masers as well).


> Bill Brundage writes:
>  >    Correct, a sliding scale from cycle to cycle, where gain may
>  > change only between cycles.  Software anticipates Tsysmax/Tsysmin
>  > for the next observing cycle and sets gain (thus total power
>  > operating level) accordingly.  This has been the observing
>  > process assumed for desiging the downconverter.  If wrong, I must
>  > know the true process very soon.
> 
> Well, it's a bit wrong.  It depends what you mean by a "cycle."  And
> it depends on whether we're doing total power radiometry with the
> analog detectors or doing interferometry with the correlator.
> 
> First, we are not at all concerned with the total range of system
> temperature but only with the range that can occur while the receiver
> *setup* remains unchanged.  If the band, LO frequency, or anything
> else in the signal processing changes, we can readjust the attenuator
> and we start over.  If we readjust *only* the attenuator, that too is
> a setup change.  Generally, we shouldn't do this unless something else
> in the setup is significantly changed, but we might do so if (a) we
> are willing to re-calibrate the complex gain astronomically after the
> attenuator change, so that none of the previous calibration data
> applies to subsequent observations; or (b) we are doing total power
> only, so phase doesn't matter, and we think we have sufficiently
> accurate knowledge of the difference in effective attenuation between
> the old and new settings.
> 
> Begin by assuming that, for a fixed setup, the receiver temperature is
> very stable.  Then the only thing that changes is the antenna
> temperature.  If we will sometimes be looking at hot loads (ambient
> temperature or above), then the largest range will occur when the sky
> is coldest.  Presumably this is for the lowest frequency bands.  Those
> bands have the lowest T_rcvr too, so they produce the largest ratio of
> maximum to minimum T_sys.
> 
> For Band 1, we should have T_rcvr < 15K and we might have T_ant as low
> as 30K.  This gives T_sys max/min of (15+300)/(15+30)=7.0 if the hot
> load is at 300K, or (15+370)/(15+30)=8.55 if the hot load is at 370K.
> For all other bands, the ratio is smaller.  I have not considered what
> happens when pointing at the Sun; presumably an RF attenuator will
> then be inserted.  If looking at active regions on the Sun, the
> restrictions on re-adjusting the IF attenuation might have to be
> relaxed.


FYI, the current spec on the solar attenuator in the FE specs document
(version 2.5) is:

FEND-62920-ZZZ
A solar filter shall be provided for solar observations. Remote
controlled automatic insertion into the beam shall be provided. It
shall provide at least 20 dB attenuation of 10 micron radiation and
an RF signal attenuation of 13 dB for all ALMA frequency bands.


> 
> Bill Brundage writes:
>  >    If radiometric tests for commissioning a new antenna require a
>  > precision dynamic range greater than 20:1, top-line commercial
>  > power meters/sensors may be required.
> 
> It seems to me that, except for solar observations, the dynamic range
> is far less than 20:1.
> 
> 
> Next, consider the linearity requirements (degree of allowed
> saturation or compression) over the dynamic range.  This is a bit more
> complicated, so I'll cover it in a later email.
> 
> --Larry
> 

-- JeffMangum - 21 Jul 2004
Topic revision: r2 - 2004-08-12, JeffMangum
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback