INSTALLATION
Plug the wall transformer into a 115 vac outlet.
Plug the barrel connector into the back of the 504.
While viewing the front panel LED's turn on the power switch (red handle, up is on). The blue power light should come on followed by a one-half second flash from each of the four green LED's in order from channels 1 to 8. The yellow Tx LED should flash briefly.
Go to Hyper Terminal setup.
With Hyperterminal running:
1. Enter "cofv" for configure format volts output.
2. Enter "r" for rate mode (default is channel 1 rate 1 per sec)
3. The default range mode is 5 volts bipolar so you should see:
4.995
4.996
.
.
.
4.995
The 4.994 volts is due the range divider resistors with the input open circuited. Connecting Pin 1 to Pin 2 (leftmost two pins) on the input plug will cause the readings to go to near zero: 0.002, 0.003, ...., 0.002.
Typing a '?' and pressing the Enter key will cause a summary of the A/D settings to be printed.
HYPER TERMINAL
(Included with MS Windows. Updates are on Cnet along with many similar programs such as Tera Term)
From windows
1). Click on programs
2). Click on Accessories
3). Select hyper terminal if available. If it does not appear on the list of accessory programs, then go to: How to install Hyper terminal.
HOW TO SET UP WINDOWS HYPER TERMINAL FOR A COM PORT
1) In hyper terminal create a "new connection" this can be done by clicking the Hypertrm.exe icon or via the menu within HyperTerminal "File:NewConnection".
2) Type in a name for the connection (for example, cyq504).
3) Under "connect using" select "Direct to Com 1" if your available com port is com 1 otherwise select the appropriate com port. Modems are relentlessly selfish about interrupts. So if a modem is present then avoid the modem port and it's cousin. That is, if the modem is on com 1 then avoid both com 1 and 3. Com 1 and 3 share an interrupt while com 2 and 4 share a different interrupt; in this case avoid 2 and 4.
4) Set "Bits per second" to 9600.
5) Set "Data Bits" to 8.
6) Set "Stop Bits" to 1.
7) Set "Flow control" to None.
8) Click OK - the hyper terminal screen should appear.
9). Select properties from the menu bar.
>8) Click on "settings", use default: autodetect and back scroll 500 lines.
9) For "ASCII setup" use
a). Sending - check: "echo typed characters locally" box. This lets you see the commands you type in. You may also issue the "ck;" command. This will cause the 504 to echo received characters (keystrokes).
b). Receiving - check: "wrap long lines".
10) Under "File" menu select "Save As" and save the configuration file to a convenient folder.
You can create a shortcut to the configuration file and put it on the desktop, this makes life a bit simpler. The configuration files have a naming convention of *.ht where star is the name you chose when setting up the new connection. For example, "cyq504.ht".
Connect a serial cable from the DB-9 on the back of the 504 to the serial COM port that you intend to use.
Apply power: the word "CyQ504" will be sent by the 504. That is why the yellow Tx LED blinks on power up.
You now have a basic DATA ACQUISITION SYSTEM
The transfer button on the hyper terminal menu bar will allow you to save received data to a file. You can put the CyQ5XX A/D in continuous acquire mode and the data will go to a file name that you select (default is programs/accessories/capture.txt). You can then import this file into a spreadsheet or whatever. You can also direct the output to a line printer for hard copy. The funny character (y umlaut) at the beginning of each line represents the invisible 0xFF start character from the CyQ5XX.
Notes:
1. To change the baud rate DURING a session you must (after going through properties to change the rate) click the 'call' button then 'disconnect' and then 'connect'.
2. The ANSI terminal option will display all ASCII characters including symbols for the unprintable ones.
HOW TO INSTALL HYPER TERMINAL
1). Click on Control panel
2). Click on My Computer
3). Click on Install/Remove Programs
4). Click on Windows settings
5). Click on Communications
6). Select Hyper terminal and click apply or OK at the bottom. You will be prompted for your Windows CDROM.
7). Done and out'ta there.
WinWedge to Excel Quick Startup
Goal: Two clicks to data into Excel
1. Install WinWedge
2. Place our CyQ floppy disk in a drive
3. Plug in wall transformer and turn the AD power on.
EXCEL Operation Summary
1. Start Excel
2. Start WinWedge, and click
File, Open "floppy:CyQ.SW3" where floppy is either A or B.
3. Click Activate
Click Test
You may need to Click on the Excel window to shift it into focus for keystrokes.
How to get to the two click goal:
Clicking on any file with an SW3 extension (that is anyname.SW3) will start Winwedge with that configuration file). Therefore, you can drag an SW3 file to the screen for a shortcut. Then it is just click Excel, click the SW3 shortcut and you are up and running (don't forget to turn the power on).
So create a directory, copy CYQ.SW3 into the directory, create a shortcut to CYQ.SW3 and drag it onto the main window.
Done! Click Excel, then click the shortcut to CYQ.SW3.
Windows Newbie Details:
1. Click on start - then programs - then Explorer.
2. Click on C: - then Files (left side of explorer menu bar) - click on New, and then Folder. Rename the folder from NEW FOLDER to CYQ (click on, right button, choose rename, and then change the name).
3. Copy CYQ.SW3 from the floppy to this directory using either drag or copy and paste.
4. Click on CYQ.SW3, right button click, choose create shortcut.
5. Click and Drag the shortcut to the main window.
Example: Grass recorder to 504 Set Up
1. Turn Power switch off (red handle down) and Plug in wall transformer.
2. Plug channel 1 (big green block top leftmost 2 holes; left hole is ground, right is signal) mini phone plug into J6 (right hand side lower corner) on the Grass Amplifier.
3. Connect the serial interface cable to the COM 1 port on the back of the computer, and to the back of the 504 (DB9 connector in the middle of case).
If COM1 is not available then connect to com 2, 3, or 4 and one small change will be needed later in WinWedge- Port- setup ).
4. Start Excel and minimize it.
5. Start Winwedge.
6. Click file - click open - click on the file CYQ.SW3.
7. Click on Port, and then analyze.
8. Turn on the power switch on the 504 - the yellow Tx light should blink and CyQ504 should appear in the input string box on the screen.
(If no success then you have the wrong COM port - click on port and then setup and then COM2 or whatever and try again. To save this port change click on file and then save.)
9. Click activate and then test mode.
10. Maximize Excel. You may need to click on the title bar or cell 1,1 to shift the focus to Excel.
11. You should now be feeding data into spreadsheet column 1.
Note that touching the "s" key (stop) will stop the AD data, and touching the "g" key (go) will resume the data acquisition. Turning the OFFSET knob on the Grass should vary the number about +-500.0. Unless told otherwise we normalize the range to +-1000.0.
12. Changing things: The first things you may want to change are channels or sampling rate - see FAQ section. The most difficult part of setting up Wedge has been done for you - define input data format. You may wish to add or modify fields, however, leave the Define, Input Data Record Structure, Start and End menu box (with the start character, variable length, cr-lf items) alone; just click continue, and go on to the Input Record Definition Editor. In the Input Record Definition Editor there is a Math Expression Box. We multipy the data field by {field value/2048} to normalize the +-2048 output from the AD to 1.0. The default scale factor is 1000.0 for the second multiplier - you may change this to a more convenient value.
CyQ 504 COMMANDS
There are three kinds of commands. Configure commands are used to change parameters such as interval between conversions or number of points to average. Acquire commands cause or enable conversions to occur. Utility commands include memory, baud rate, and other housekeeping chores .
Typing ?; or ?<ENTER> will print out the current status.
All commands MUST end with either a semi-colon ';' or the ENTER key. The command is buffered, but not processed by the A/D until a ';' or '\r' is received. The backspace key may be used for corrections on screen.
Where possible the command format allows a value to be assigned with an '=' following the command code or for the feature to be turned on or off with a logical t, f character instead of the '=' sign. This allows setup values to be defined and preserved for future use. It allows the feature to be reactivated with a minimum of hassle. For example: "cfm=3;" will set median points to 3, but it will not affect the usage of the median filter. If the median filter was initially off, then it may be turned on at a later time by "cfmt;" (or the abbreviated "cfm;') To stop using this filter issue the command "cfmf;" Again, note that "cfm=5" neither turns the filter on or off.
In the following N is an integer number; L is a logical true or false, and X is a hexadecimal number. A logical, L, may be T, t ,1 or ';' for true, or F, f, 0 for false.;
COMMAND SUMMARY
aNNNN; acquire where N is a channel number(s).
s; stop.
g; go
Configure
Mode
cmr: acquire mode rate
cmt; timed
cmp; polled
cmr=N; Sampling rate in rate mode
cmt=N; Time between samples in timed mode
Span
csb; Span bipolar +/- 5 volts
csu; unipolar (0 to 10 volts)
Format
cofi; Integer output format -2048 to 2048 or 0 to 4096
cofv; Voltage format
cofx; Hexadecimal format
cofb; Binary format
Filters
cfbL; Burst mode averaging on-off
cfb=N; Number of conversions to average in burst mode
cfr=N; Burst mode sampling rate
cfsL; Sample averaging on-off
cfs=N; Number of samples to average
cfmL; Median filter on-off
cfm=N; Median size (1 to 11)
Utility:
e; error light off.
ck; keystroke (Rx char) echo.
?; Transmit command summary and current settings.
mss; Memory save setup.
mls; Memory load saved setup.
msd; Memory save current setup as default.
mpd; Memory Purge default.
$@R Restart.
ACQUIRE COMMANDS
. Acquire: "aNNN;" where NNN... are 1 or more channel numbers. This assigns the channel numbers to be converted in either rate, timed or poll mode. If no channels numbers are present, "a;", then the previously defined channel numbers will be used. In polled mode this will cause a poll to occur. In rate and timed modes this command is used to define the channels. Modes are assigned in configure using "cmr;" or "cmt;" or "cmp;" commands for rate, timed or poll.
2. Stop: "s;" Stops conversions from occurring. If in rate or timed modes and poll cannot occur, conversions are stopped. CAUTION: Stop MUST be exited with a GO ("g;") command for the poll (or rate or timed) acquires to work. You will forget this, and frantically pound the keyboard, wondering why nothing is happening.
3. Go: "g;" Allows poll and continuous commands to resume. If stop occurred while a continuous conversion was in effect then the continuous mode is resumed. The DEFAULT mode on power up is go.
Configure commands
Current configuration: "?;" causes modes, rate, and the number of points to be averaged to be printed followed by the interval between conversions for averaging. This is helpful during the development phase to make sure that the command you thought you entered was the command you entered. It also serves as a Help file for the command mnemonics.
Basic acquire mode commands
"cmr;" Configure acquire mode as rate. Sampling rate conversion. Channels following the acquire command ("aNN...;" will be converted repeatedly at the default rate(1 per sec) or previously set value for rate (see setup command "cmr=N;").
"cmt;" configure acquire more as timed. Channels following the acquire command ("aNN...;" will be converted repeatedly at the default interval (1 sec) or previously set value (see setup command "cmt=N;").
"cmp;" configure acquire mode as polled. An aNN..; will cause channels NN... to be polled.
Rate in rate mode: "cmr=N;" where N ranges from 1 to 4000 samples per second. DEFAULT=1 sample per second. This command sets overall conversions per second.
Time between conversions in timed mode: "cmt=N;" where N ranges from 1 to 60,000. DEFAULT=1000 ms. This may be an averaged conversion.
Filters
Median
cfmL;" enable ("cfb;" or "cfbt" or "cfb1;") or disable (cfbf;"or "cfb0;") median filter.
"cfm=N;" Set median filter size to N where N is 2 to 12.
Burst mode averaging.
Enable, disable burst mode averaging "cfbL;" where L is logical.
"cfb;" enables burst mode for N samples set by "cfb=N;".
"cfbf;" disables burst mode averaging.
Number of conversions to average in burst mode averaging: "cfb=N;" where N ranges from 1 to 255. Conversion results are summed until this number is reached then divided, and the result submitted to the sample averaging filter. DEFAULT= 10.
Burst mode sampling rate: "cfr=N;". The rate in sample per second that the active AD channels are sequentially sampled and summed. A sampling rate of 600 with a count of 10 will attenuate power line noise by >50 db.
Sample averaging
"cfsL;" Enable or disable sample averaging.
"cfs=N;" Set number of samples to average to N. N samples will be acquired and summed at the rate set for rate mode or at intervals set by timed mode. The sum will then be divided by N and the average transmitted on the N'th acquire. For example, if cfs=10 and cmr=5 conversions per second, then an averaged sample will be transmitted every 2 seconds.
SPAN
Bipolar span: "csb;" The input voltage range is -5 to + 5 voltsThe corresponding output range is -2048 to +2048 in integer format mode. In binary mode the output data is offset binary with zero volts in corresponding to integer 2048. Toggles unipolar off.
Unipolar range: "csu;" The input voltage range is 0 to 10 volts. The corresponding output range is 0 to 4096 in ASCII integer format mode. This toggles bipolar off.
DATA OUTPUT FORMAT
All output begins with the special character all bits set, hexadecimal 0xFF, binary 1111 1111 (all 8 bits set to 1).
'v' volts output: "<0xFF>#.###....." Channel values are printed in volts with 0.001 volt (1 mv) resolution. Values range from -4.999 to 4.999 in bipolar, and 0.000 to 9.999 in unipolar mode (precision +/- 2.5 mv. typical accuracy +/- 10 mv . Channels are comma separated followed by a CR LF at he end, where CR is carriage return and LF is linefeed. In other words: start separator followed by comma delimited variable length numbers (due to the sign), and ending with a carriage return. Makes nuts and bolts sense, especially if you are using the AD to actually measure a voltage instead of, say, a pressure related to voltage only as an intermediate scale factor.
'i' causes the output format to be integer characters: <0xFF>number, number,....CR LF . In other words: start separator followed by comma delimited variable length numbers, and ending with a carriage return line feed. This is the DEFAULT. The range is -2048 to 2048 in bipolar and 0 to 4096 in unipolar. (varies slightly with A/D offset error) Easy to see what is going on with the A/D.
'x' causes the output format to be in hexadecimal notation: <0xFF>number, number,....CR LF. In other words: start separator followed by comma delimited variable length numbers, and ending with a carriage return. Included for the comfort of programmers.
'b' binary output: "<0xFF>####....." each channel value is represented by two ascii characters; many unprintable. The digits are concatenated without delimiters. There is no CR at the end. This is a fixed length format with the length determined by the number of active channels. This allows much higher conversion speeds. Each character transmitted requires 10 bits. In integer mode one channel requires at least 5 digits plus a start character and final CR,LF; that's 80 bits. In binary mode only 2 digits plus a start character are used or 30 bits. Almost three times as fast. However, it is completely unreadable nonsense on a screen. For a computer a fixed length format with a start byte is quite easy to decode to an integer using either basic or c languages (wait for the start byte, shift left 8 next byte, add next byte, repeat for each channel).
"cofc;" causes channel numbers to be included in the transmission
ch:value,ch;value....
"cofiL; " will cause the first number to be an index number. This number will increment by 1 for each transmission. 000 001 002 003,....255 and roll over to 000. This feature is helpful in reassuring you that there is no missing data due to the communication link.
Turning it off reduces the bandwidth requirement which is in the direction of goodness, that is, if it works correctly with the index then it is even more likely to work without it.
Utility commands
1. Baud rate:"cqX;" where X ranges from 0 to A. Sets the transmit and receive baud rates for the A/D serial output. The DEFAULT rate at power up is 9600. There is always 1 start bit, 8 data bits, and 1 stop bit; there is no parity. The computer baud rate MUST be changed to match after this command is executed (in hyper terminal you must click properties, and then configure. Baud rate will not change until you click disconnect and then connect on the toolbar).
The values are: 1=2400, 2=4800, 3=9600, 4=14,400, 5=19,200, 6=28,800, 7=38,400, 8=57,600, 9=115,200, A=230,400.
2. Error reset: "e;" turns off the error LED.
3. Memory:
a. Save setup: "mss;"
Memory write setup .... stores the current setup including modes, channels, and configuration. "mss1;", "mss2;" save additional setups.
b. Read and load setup: "mls;", "mls1;", "mls2;" Memory load setup. Loads a saved setup if one is present.
c. "msd;" Memory save default. Stores the current setup (baud, mode, configuration)
as the power up default. Replaces (over writes) a previously stored default. You
may wish to first test using the write and read saved setup commands,
However, should you accidently save a setup that renders the AD inaccessible (high data rate, etc.) or a year has passed and you forgot the baud rate ... there is an out. The device always powers up at 9600 baud; it checks for and, if present, loads the stored default after the led's flash (about 2 sec). During the time the led's are flashing pressing the escape (ESC) key will cause the module to skip the load stored default. It will come up with the original safe defaults shown in this manual.
d. "mpd;" Purge stored default:
Restores the power up defaults of the module to those shown in this manual.
e. "mld;" loads default settings.
4. Reset the module: $@R will cause a power on restart.
5. Echo received characters: "ck;" Keystroke echo. Incoming characters are
transmitted back to the serial source sending the characters.
Why both rate and interval modes?
The two modes allow for the precise control of data collection from over 3000 samples per second to less than 1 sample every 60,000 ms (or sec). At low data rates the resolution of rate mode is poor. For example, you can only choose between 1 or 2 samples per second, corresponding intervals of either 1000 or 2000 ms. In interval mode you can choose 1001, 1002....1999 ms, you have all 1000 choices.
On the other hand, at high data rates in interval mode, you can only have either 1 or 2 ms corresponding to rates of either 1000 or 500 samples per second. Kind of frustrating when you need high data rates and the highest is between 500 and 1000 samples per second. In rate mode you can choose 500, 501,....999 samples per second.
So between the two modes you have the capability to precisely optimize the A/D to your situation.
Averaging modes?
The different methods allow you to choose the most effective combination for the noise content in your environment. They relieve the main computer of performing these common tasks thereby freeing it up for other tasks. The filters reduce the need for high serial bandwidth. Timing accuracy is improved because the 504 microcontroller is a dedicated single task device as opposed to the multitasking time sharing OS in larger computers.
The sample averaging filter is the classic add them up and divide filter for gaussian (normal distribution) noise reduction. The median filter is for popcorn noise, and signals near the limits artifact. Burst mode averaging is over sampling and is helpful in reducing aliasing problems, specifically in removing power line noise.
Median Filter
I have always been amazed at how badly one outlier can trash the calculation of an average. If you pitch it, then the data is tight. Often it was one point in a series of repeat measurements that were averaged at data collection time. If you take the median of the repeats then the single measurement falls out; the mean and the median now coincide. This is as it should be; because the outlier was due to a defect in the measurement. It is perfectly valid to reject the outlier since the median is a more accurate estimate of the central tendency in narrow distributions. I digress, back to the question of how did the crazy number get there in the first place.
Shot (or popcorn) noise is due to very short spikes polluting the signal. In audio systems this is heard as a click or pop. The spike amplitude is generally sufficient to max out the electronics. In processing a signal to remove noise by averaging this is the most feared type of noise.
Let's say we have a perfect signal conditioner and a perfect A to D Converter. The input signal is one volt to the AD. The range of the AD is zero to five. We are acquiring five points and averaging them so: 1+1+1+1+1 =5 divided by five equals one volt; perfect. A defective bulb in a fluorescent light fixture two floor down begins to flicker. Now we have 1+5+1+1+1, 1+1+1+1+1, 5+1+1+1+1 or averages of = 1.8, 1.0,1.8 . . . 1.0 volts. Increasing the averaging to say 10 (and assuming this doesn't pick up even more pops) only reduces the average to 1.4. We are definitely in deep doo, doo, and are faced with two choices, either find the source and remove it (politically difficult if some big dude is using the equipment) or filter out the noise somehow.
Enter the median filter. The median of a bunch of numbers is the number that has as many numbers below it as above it . For example, the median of 1,2,3,4,5 is 3; the median of 5,1,2,3,4 is 3; the median of 1,1,5,1,1 is 1 and so forth. We simply sort the numbers just as we would sort cards, and pick the one in the middle. If several readings are the same (you can get shot for this in Nevada), then we simply list them as they are encountered in the sort, for example, 5,2,1,2,3 becomes 1,2,2,3,5 when sorted; the median is 2, the one in the middle. (This is called a 5-point median, five is an odd number so it is an odd median; even medians, for example four, are defined as the average of the two numbers in the middle.)
From a statistical perspective what we are trying to find is the "central tendency" of a bunch of numbers (a distribution). If the distribution is normal (as would be expected with Gaussian noise) then the distribution will have a tail. To get rid of the noise, you want the value at the peak of the distribution. Clearly this can be done by averaging all the numbers. However, in the real world, you only get to average a sample of all the numbers, say 10 points to estimate the mean. Ever so often you will get a value from the tail of the distribution in your 10 points; this trashes the average for that 10. There is no way to know when or how often this will occur when rolling the dice. This is why a sample median is a better (more reliable) estimate of the central tendency, when the distribution is narrow (low gaussian noise), than a sample average.
Fig. 1 Popcorn noise spikes appearing as AD converter values at the high range limits. The blue bars show the effect on the normal distribution of under range and over range noise.
Yes, but if there are an equal number of high pops and low pops then the noise cancels out when the signal averaged.
The pops will cause fixed low or high values that average to zero in bipolar mode or half-range in unipolar mode. This tends to pull the signal average to the center of the AD range. If the signal average is near midrange, then the effect is minimized. The further from center the signal is, the more the average is pulled toward midrange.
For examples consider these numbers in bipolar mode, range five volts: -5, -5, -5, 1, 1, 1, 1, 5, 5, 5; this is 10 readings about a true signal of one volt with six spikes. There are equal numbers of high (five volt) and low (-5 volt) noise pops. It seems like they should cancel, but the series adds up to four on 10 readings giving an average of 0.4 volts. This is a lousy estimate of the one volt actual value. Note that a five volt true value would average to two volts in this case. Of course a zero volt true would be dead on.
Median versus averaging at lower and upper AD limits
An analog to digital converter has definite minimum and maximum outputs. For a 12 bit AD these are 0 and 4096. In unipolar mode, with a range of 0 to 10 volts, any signal at or below zero will be a 0, and any signal at or above 10 volts will give 4096. So what happens to our estimates if we input a signal with a normal distribution about a DC level of 1 volt, and gradually lower the DC level toward 0 volts? To find out let us average the output from the AD and compare that with the median. As we approach 0 the tail of the distribution falls below 0 giving us more and more 0's in the AD output, and therefore in our average. When the peak is at zero we still have a positive value from the averaging filter due to the upper half of the distribution.
Fig. 2 AD lower range limit effect on Mean and Median of a low level normally distributed signal.
So the bad news is that to get a zero output from averaging we have to have the whole distribution below zero. The good news is that the median filter tracks the peak and goes to zero as the peak crosses zero; there are as many zero AD readings below the peak as there are positive readings above, and it is the middle reading in the sort that matters. A similar situation occurs at the upper end of the range or in bipolar at the lower end.
Things to watch out for
A median filter should be used with an averaging filter when the peak of the distribution is wide and flat. The median is a mediocre estimate of the central tendency when the distribution is flat topped (look up kurtosis in a statistics text).
Burst Filtering (over sampling)
Aliasing occurs when a spurious signal corrupts the data signal due to the periodic sampling rate of the AD. A noise signal is aliased into the data if it is periodic, and a multiple of the sampling rate. This is the strobe light effect where the motion of a fan blade may be made to appear to be stopped, slowed or reversed. It also makes stagecoach wheels appear to turn backwards in old movies when the frame rate of the camera synchronizes with the rotation of the spokes in the wheel. With an AD converter it often looks like a slow sinusoidal drift in the sampled data.
Fig. 4 The slowed case. A high frequency alias appearing as a low frequency signal.
One way to prevent this is to place a low pass filter ahead of the AD. This is usually a simple first order resistor capacitor network. We deliberately restrict the bandpass on our signal conditioners to reduce the aliasing problems for attached AD's. The design rule is to limit the bandwidth to only that needed to amplify the signal to be measured. Using a wider bandwidth amplifier than is needed increases noise contamination and the likelihood of aliasing due to unknown interference.
Another way is to increase the sampling rate to the point where the interfering signal frequency would have to be beyond the bandpass of the AD input signal conditioners. The Burst Averaging option on the 504 provides this over sampling capability. For example, the 103 signal conditioner has a 3 db bandpass or 300 Hz. A burst sampling rate of 600 Hz would push the lowest frequency potential aliasing signal to 600 Hz; this spurious signal would be severely attenuated by the 103 signal conditioner and never reach the AD input.
The requirement for periodicity in the interfering signal reduces the likelihood of aliasing problems. The requirement that it be a multiple of the sampling rate also makes the problem less likely; furthermore the frequency response of signal conditioners ahead of the AD generally decreases at higher frequencies.
The Governor
(Remember when rental trucks had a governor? No matter how hard you pushed the gas pedal - 50 mph was tops)
The maximum attainable conversion rate is affected by a confusing array of settings. Specifically, the data output rate cannot exceed the baud rate. This is dependent on the baud rate setting which can assume 10 values. The time to transmit data is also affected by format. One channel in binary mode requires 3 bytes: the 0xFF start character and 2 for the number. A byte requires 10 bits to be transmitted. So you got to divide the baud rate by 10 or multiply the bytes by 10 to get the bytes per second limit. However, there are several formats with volts being the longest: 0xFF start, followed by sign, followed by a digit, decimal point and four more digits, and finally by a carriage return. More than one channel will cause a comma to be inserted, and if you are printing the channel number, then this also needs to be counted. Internal code time (?) and conversion time (60 us) also need to be factored in.
Okay, so it becomes a guessing game. You try it - if it fails, then you drop the rate. You end up bouncing between trial, error, calculator and scratchpad (and the local pub).
The CyQ 504 automatically adjusts the rate downward to the maximum that will allow the data to be reliably transmitted. If you enter a rate that is too high (interval that is too short) then you will see a series of error messages on the screen - the red error light will flicker. Then the error light will stay off and smooth data output will resume. If you enter the ? command you can see what the maximum safe rate (or shortest interval) is for your current baud rate and format settings.
In interval mode the top available rate is 1000 samples per sec at 1 ms interval. The governor may need to increase the interval to accommodate channels or formats. At low data rates the governor functions mainly when averaging is being done. The number of samples to average and the averaging interval are fixed. The governor automatically increases the interval between data outputs to accommodate the averaging. We assume that the averaging is being done to reduce noise just before data transmission, therefore we leave it alone.
Note that we let you run off of the road a few times before we set a speed limit.
If an error occurs then a line feed will be followed by one or more 's followed by an error message. The error light will come on, and can only be turned off with an 'e;' command.
cmd Command buffer overflow. More than 16 characters were received before a command could be processed. Either a ';' was dropped, or there was noise, or a key is stuck on the keyboard.
Rx Receive buffer overflow. Too many commands were sent to fast. Usually happens when commands are being sent by a computer program. Place some wait(1 or more ms) statements in the code.
Tx Transmit buffer overflow. Everything was not transmitted before the next data was ready to be loaded for transmission. The buffer will go into a wait loop for about 1 ms to give the uart buffer time to clear; if this fails then this error will be set. The delay may trigger the governor producing a speeding error. Baud rate needs to be increased or the sampling rate decreased.
Speed Acquire rate is too high or baud rate is too low ... the rate will automatically be decreased.
Other errors will result in the following:
The command will be printed up to the offending character, an underline '_' will be printed, this will be followed by a letter where:
'?' means the last character was invalid
'L' a logical was expected: t,f,0,1.
'=' an equal sign was expected. (In some instances both an L and = error will occur.)
'N' a number was expected, or the number is out of range, e.g., a bad channel number.
'X' a hexadecimal number is out of range or the character is not a hexadecimal number.
For example:
type cz;
***cz_?
is the error message because 'z' is not a valid configure command character.