[amsat-bb] Re: General Telemetry Question (PITFALL!)
Perfect Timing! I'm finishing up details of my Master's Thesis and using Excel Curve fitting functions for some plotted simulation data concerning link margin vs elevation. This Pitfall sub-thread was perfectly timed to help me avoid showing useless data without enough significant figures. This listserv rocks, Thanks!! -Zach, KJ4QLP On 12/11/2013 2:03 PM, Jim White wrote: I have been using the free program CurveExpert for years for plotting and calculating calibration data. That is, for deriving the cal equation from test data. http://www.curveexpert.net/ It lets you choose linear or quadratic and if quadratic you can pick how many terms you want. It lets you very quickly experiment with the form of the formula so you can see how to get the best fit. It also will suggest the highest exponential value for the formula. I generally build my table of test data, ADC counts vs measured values (V, I, temp, etc.), in Excel. Then I copy and paste that table into Curve Expert, click one button to make it fit the curve and another to show me the coefficients of the formula. I copy and past those back into Excel next to the table. Then I add another column to the table with values calculated using the formula and coefficients. That's the double check to be sure the formula is correct. Of course you can also see the standard deviation and other 'goodness' values in the CurveExpert data. But I find I can tell by just looking at the fit plot if I got the data wrong or my measurements weren't precise enough. I find CurveExpert much easier to use and much more flexible than Excels' curve fitting functions. Jim On 12/11/2013 11:43 AM, Robert Bruninga wrote: To follow up on Bob's comment. If you send the raw analog sensor data... Change calibration values if found to be wrong after launch... We did on PCSAT! Caution to Satellite Builders: Be careful when using an EXCEL TREND LINE equation for doing Engineering Unit conversion back to original units. It was a big lesson for us back on PCSAT in 2001. The problem is, generally, EXCEL displays trend line equations in a nice GENERAL human readable form. For example, for our thermistors, the 3rd order trend line equation to convert from telemetry count back to degrees C was displayed by EXCEL as something like this: 2E-7 X^3 - 2E-4 X^2 + 1.804E-1 X + 2.379E2. One would think one is getting a very precise to 4-significant digit equation. WRONG. Notice the Cubed and Squared terms (which can be very big at warmer temperatures) are only represented to a single decimal digit(+/- 10%)!!! (2 and 2)... When this trend line is used (as displayed), to give back our temperatures from the incoming COUNT, the temperatures were way off! The key is to make sure the trend line equation is displayed in SCIENTIFIC format before you write it down and then try to use it. Then the first two terms above are properly displayed by EXCEL as 1.544E-7 X^3 and -2.069E-4 X^2. (instead of 2E-7 and 2E-4). We catch this error in a lot of student's work... Bob, WB4APR ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb
[amsat-bb] Re: General Telemetry Question (PITFALL!)
To follow up on Bob's comment. If you send the raw analog sensor data... Change calibration values if found to be wrong after launch... We did on PCSAT! Caution to Satellite Builders: Be careful when using an EXCEL TREND LINE equation for doing Engineering Unit conversion back to original units. It was a big lesson for us back on PCSAT in 2001. The problem is, generally, EXCEL displays trend line equations in a nice GENERAL human readable form. For example, for our thermistors, the 3rd order trend line equation to convert from telemetry count back to degrees C was displayed by EXCEL as something like this: 2E-7 X^3 - 2E-4 X^2 + 1.804E-1 X + 2.379E2. One would think one is getting a very precise to 4-significant digit equation. WRONG. Notice the Cubed and Squared terms (which can be very big at warmer temperatures) are only represented to a single decimal digit(+/- 10%)!!! (2 and 2)... When this trend line is used (as displayed), to give back our temperatures from the incoming COUNT, the temperatures were way off! The key is to make sure the trend line equation is displayed in SCIENTIFIC format before you write it down and then try to use it. Then the first two terms above are properly displayed by EXCEL as 1.544E-7 X^3 and -2.069E-4 X^2. (instead of 2E-7 and 2E-4). We catch this error in a lot of student's work... Bob, WB4APR ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb
[amsat-bb] Re: General Telemetry Question (PITFALL!)
I have been using the free program CurveExpert for years for plotting and calculating calibration data. That is, for deriving the cal equation from test data. http://www.curveexpert.net/ It lets you choose linear or quadratic and if quadratic you can pick how many terms you want. It lets you very quickly experiment with the form of the formula so you can see how to get the best fit. It also will suggest the highest exponential value for the formula. I generally build my table of test data, ADC counts vs measured values (V, I, temp, etc.), in Excel. Then I copy and paste that table into Curve Expert, click one button to make it fit the curve and another to show me the coefficients of the formula. I copy and past those back into Excel next to the table. Then I add another column to the table with values calculated using the formula and coefficients. That's the double check to be sure the formula is correct. Of course you can also see the standard deviation and other 'goodness' values in the CurveExpert data. But I find I can tell by just looking at the fit plot if I got the data wrong or my measurements weren't precise enough. I find CurveExpert much easier to use and much more flexible than Excels' curve fitting functions. Jim On 12/11/2013 11:43 AM, Robert Bruninga wrote: To follow up on Bob's comment. If you send the raw analog sensor data... Change calibration values if found to be wrong after launch... We did on PCSAT! Caution to Satellite Builders: Be careful when using an EXCEL TREND LINE equation for doing Engineering Unit conversion back to original units. It was a big lesson for us back on PCSAT in 2001. The problem is, generally, EXCEL displays trend line equations in a nice GENERAL human readable form. For example, for our thermistors, the 3rd order trend line equation to convert from telemetry count back to degrees C was displayed by EXCEL as something like this: 2E-7 X^3 - 2E-4 X^2 + 1.804E-1 X + 2.379E2. One would think one is getting a very precise to 4-significant digit equation. WRONG. Notice the Cubed and Squared terms (which can be very big at warmer temperatures) are only represented to a single decimal digit(+/- 10%)!!! (2 and 2)... When this trend line is used (as displayed), to give back our temperatures from the incoming COUNT, the temperatures were way off! The key is to make sure the trend line equation is displayed in SCIENTIFIC format before you write it down and then try to use it. Then the first two terms above are properly displayed by EXCEL as 1.544E-7 X^3 and -2.069E-4 X^2. (instead of 2E-7 and 2E-4). We catch this error in a lot of student's work... Bob, WB4APR ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb
[amsat-bb] Re: General Telemetry Question
David, It is much more efficient in terms of the information transmitted, hence power and bandwidth, to use the raw binary/hex for transmission. It also saves the programming and memory in the satellite CPU. The combination frees up resources which can be otherwise used. It works well given the almost universal availability of personal computers. I recall, vaguely, there have been a few birds with some quick look, real people data, but I may be in error. 73s, Alan WA4SCA -Original Message- From: amsat-bb-boun...@amsat.org [mailto:amsat-bb-boun...@amsat.org] On Behalf Of Dave Marthouse Sent: Monday, December 09, 2013 7:35 AM To: amsat-bb@amsat.org Subject: [amsat-bb] General Telemetry Question I noticed over the years that satellite beacon downlinks transmit their telemetry in a form that must be translated by a telemetry app to their engineering values. Since the information is transmitted from the satellites why not provide the engineering values in the downlink without the extra step having to be done on the ground? What is the logic of doing this? Dave Marthouse N2AAM dmartho...@gmail.com ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb
[amsat-bb] Re: General Telemetry Question
Answer: Engineering efficiency.. There is far more computing power on the ground than the satellite. Also, KISS principle. Also, calibration can be done without modifying flight code. And finally, it is far more compact to send binary or hex than human readable decimal. Bob, WB4aPR -Original Message- why not provide the engineering values in the downlink without the extra step having to be done on the ground? What is the logic of doing this? ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb
[amsat-bb] Re: General Telemetry Question
Hi, In the case of FUNcube the on board sensors give their readings as n bit values when they are interrogated. This data is agregated into a data frame for transmission using forward error correction to improve the s/n ratio. The problem with on board conversion is that you would have to store the scaling/offset/logarithmic values for all channels on board, in rom. These are usually only characterised during thermal cycling / illumination testing etc and would have to be uploaded to the satellite rom. Not necessarily a simple task. It is easier to do it on the ground where we can tweak the calculation factors. 73 Dave, g4dpz On 9 Dec 2013 13:52, Dave Marthouse dmartho...@gmail.com wrote: I noticed over the years that satellite beacon downlinks transmit their telemetry in a form that must be translated by a telemetry app to their engineering values. Since the information is transmitted from the satellites why not provide the engineering values in the downlink without the extra step having to be done on the ground? What is the logic of doing this? Dave Marthouse N2AAM dmartho...@gmail.com ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb
[amsat-bb] Re: General Telemetry Question
On 12/9/13, Alan wa4...@gmail.com wrote: David, It is much more efficient in terms of the information transmitted, hence power and bandwidth, to use the raw binary/hex for transmission. It also saves the programming and memory in the satellite CPU. The combination frees up resources which can be otherwise used. It works well given the almost universal availability of personal computers. I recall, vaguely, there have been a few birds with some quick look, real people data, but I may be in error. ARRISat transmitted some of its operating data on FM using a voice synthesizer. 73s Bernhard VA6BMJ @ DO33FL snip ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb
[amsat-bb] Re: General Telemetry Question
Hi All, Just to confuse ..FUNcube-1 transmits some telemetry in RAW and some in human readable format. The latter comes from the GOMspace EPS which is powering our baby! cheers Graham G3VZV -Original Message- From: Alan Sent: Monday, December 09, 2013 2:44 PM To: 'Dave Marthouse' ; amsat-bb@amsat.org Cc: CC Subject: [amsat-bb] Re: General Telemetry Question David, It is much more efficient in terms of the information transmitted, hence power and bandwidth, to use the raw binary/hex for transmission. It also saves the programming and memory in the satellite CPU. The combination frees up resources which can be otherwise used. It works well given the almost universal availability of personal computers. I recall, vaguely, there have been a few birds with some quick look, real people data, but I may be in error. 73s, Alan WA4SCA -Original Message- From: amsat-bb-boun...@amsat.org [mailto:amsat-bb-boun...@amsat.org] On Behalf Of Dave Marthouse Sent: Monday, December 09, 2013 7:35 AM To: amsat-bb@amsat.org Subject: [amsat-bb] General Telemetry Question I noticed over the years that satellite beacon downlinks transmit their telemetry in a form that must be translated by a telemetry app to their engineering values. Since the information is transmitted from the satellites why not provide the engineering values in the downlink without the extra step having to be done on the ground? What is the logic of doing this? Dave Marthouse N2AAM dmartho...@gmail.com ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb
[amsat-bb] Re: General Telemetry Question
To follow up on Bob's comment. If you send the raw analog sensor data to the ground you can - Fix mixed up channels if you got them wrong before launch. This happened with the 1990 AMSAT Microsats and I've seen it since then in other birds. - Change calibration values if found to be wrong after launch. I've seen this most often with ACS systems where the sign of a magnetometer or torq rod is backwards. - Change cal equations if an analog sensor or its system partially fails. Most recently I've seen this with a science mission cubesat that has been in orbit about 6 months and suffered a partial failure. Adjusting the equations on the ground allowed for a continued science mission. - Save downlink characters, hence time. You can get more data down in a shorter packet. Example: To send human readable ASCII for a telemetry value like A=3676 takes 7 bytes. If you just send the number and a space it's 5 bytes. The same value in binary is two bytes (long int in C language). - Get all values in a single AX.25 frame with a single and common time stamp. In binary you can get about 225 values in a frame (with a time stamp, ID, etc.). In ASCII you can only fit about 50ish. A typical cubesat has more than 50 TLM values (although some have less). A typical microsat may have as many as 200. - When downloading science or sensor data the amount you can get to the ground is often the limiting design factor. With current technology you can usually store as much as you want in the sat. But to get it to the ground you need to be as efficient as possible. Binary is most often used, but that's not efficient enough for some missions and further compacting the data is needed - using one or more of several other techniques. Jim On 12/9/2013 7:44 AM, Robert Bruninga wrote: Answer: Engineering efficiency.. There is far more computing power on the ground than the satellite. Also, KISS principle. Also, calibration can be done without modifying flight code. And finally, it is far more compact to send binary or hex than human readable decimal. Bob, WB4aPR -Original Message- why not provide the engineering values in the downlink without the extra step having to be done on the ground? What is the logic of doing this? ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb ___ Sent via AMSAT-BB@amsat.org. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb
