The following information represents configuration and operations status of NCAR's Vaisala PTB22 barometers during field project HVAMS-Sept-Oct 2003 and subsequent testing performed to determine why the barometers have difficulty in RS485 bussed operations, and what solutions may improve the performance

Links within this doc.:

Typical Setup Commands

Vaisala Email / Notes / Hints

Note: Vaisala never provided schematics of complete answers to all questions, although did provide some helpful info.

PAM HVAMS03 and RS485 Operation Barometer Checks……

Item

Box-1

Box-2

Box-3

Box-4

Box-5

Box-6

Box-7

Box-8

Box-9

Setup Notes:

TRH sbus01 tyCo/3

502

701

206

702

RMY sbus12 tyCo/3

319

355

315

318

38770

38768

245

313

39769

AD sbus86 tyCo/3

PTB

NCAR0003

NCAR0002

NCAR0001

S0610003 NO-SBUS

S0610001 NO-SBUS

U4110004

U4110001

U4110003 NO-SBUS

S0610002 NO-SBUS

sbus

TearDown Notes:

TRH sbus01 tyCo/3

701

502

703

206

702

RMY sbus12 tyCo/3

319

355

315

318

38770

38768

245

313

39769

AD sbus86 tyCo/3

PTB

S0610002

NCAR0002

NCAR0001

NCAR0003

S0610001

S0610003

U4110001

U4110003

U4110004

polled-sio

fmt160-sbus

fmt163-polledSio

fmt161-polledSio

fmt162-sbus

fmt161-polledSio

fmt163-polledSio

fmt162-sbus

Problem 1,2 observed in Field

Known Problem 1 in Field

See repair note below. Edge connector prob.

BAROMETER cfgs at teardown

Software version

PTB220 / 1.04

PTB220 / 1.01

PTB220 / 1.04

PTB220 / 3.05

Serial number

S0610002

NCAR0002

NCAR0001

NCAR0003

S0610001

S0610003

U4110001

U4110003

U4110004

Configuration

Linear adjustments

OFF

Multipoint adjustments

Calibration date

1/30/1997

12/13/1995

????-??-??

3/8/1995

1/30/1997

10/11/1999

Baud Parity Data Stop Dpx

9600 N 8 1 F

9600 E 7 1 F

9600 N 8 1 F

Echo

OFF

Sending mode

POLL / OPEN

POLL / CLOSE

POLL / OPEN

STOP

POLL / OPEN

POLL

Measurement mode

NORMAL

Pulse mode = OFF SLOW LO

Address

Output interval

1 s

5 s

10 s

1 s

5 s

0 s

1 s

0 s

Output format

"B1 " 4.2 P1 " " 3.1 T1 #r #n

"B1002 " 4.2 P1 " " T1 #r #n

"B1001 " 4.2 P1 " " 3.1 T1 #r #n

Error output format

SCOM format

Pressure unit

hPa

mbar

hPa

Averaging time

0.5 s

1.5 s

1.0 s

0.5 s

Problem-1

Work if set internal dip switch S1-8=ON="Use Factory Defaults", Won't work if set OFF=user settings, Factory=9600E71

Problem-2

Communications on either SBUS or RS232 die when voltage (on station) drops below ~12.8

Dec,2003: RS485 and bussed Operations Testing

BAROMETER cfgs during tests:

PTB

NCAR0001

NCAR0002

NCAR0003

S0610001

S0610002

S0610003

U4110001

U4110003

U4110004

labeled

Software version

PTB220 / 1.01

PTB220 / 1.04

PTB220 / 3.05

Serial number

N0000001

N0000002

N0000003

S0610001

S0610002

S0610003

U4110001

U4110003

U4110004

Configuration

Linear adjustments

OFF

Multipoint adjustments

Calibration date

????-??-??

12/13/1995

3/8/1995

1/30/1997

10/11/1999

Baud Parity Data Stop Dpx

9600 N 8 1 H

Echo

OFF

Sending mode

POLL / CLOSE

POLL / OPEN

STOP

POLL / OPEN

POLL

Measurement mode

NORMAL

Pulse mode = OFF SLOW LOW 0.0

Address

Output interval

5 s

10 s

1 s

5 s

0 s

1 s

0 s

Output format

"B1 " 4.2 P1 " " 3.1 T1 " " ERR #r

Error output format

SCOM format

Pressure unit

mbar

hPa

mbar

hPa

Averaging time

1.0 s

1.5 s

1.0 s

0.5 s

1.0 s

0.5 s

NOTE-1a

NOTE-4a

Initial Test1-2:

EVE electronics on sbus with all 9 barometers setup to 'half-duplex' mode per manual. Initially tried with sbus attached and then removed it entirely to just poll barometers.

Fmt PTB220sbustest: polling barometers every 10 sec. ~12.0 vdc

Results:

Every configuration ended up with all barometers (except maybe 1 locking up). Power cycle did recover them every time. In some cases they locked up sooner than other cases. It appeared to be better with removal of other sbus items, but not by much.

Power = 12.0vdc no problems noticed per field

SBUS signal looked crummy whether using/installing a 120ohm/33nF terminator at the end of the chain, or not having it. Probably better with it but didn't have effect on lockups. Tried removing termination on EVE driver side with same result. Signals went roughly 0-4vdc, mirror image btwn 485+/-

Note-1a

baro NCAR0003 was having trouble with 485. discovered that the edge connector of the 485 module (HMP230RSB) was not soldered and actually pulled out of the board…..fixed this and it appears fine now.

Conclusions:

Comms contention and multi-cast on the bus caused barometers to lock up requiring power cycle. Definitely observed contention on sbus caused by EVE implying improper task/io segregation: ie code inadequate.

PTB220's unable to recover from lockup condition…..need watchdog.

Test 3

Procomm script (PTB220 send9) written to poll all barometers every 15 seconds through EVE using 'talk' when system stopped. The reason for this was to get EVE's software out of the way and see

Results:

Barometers for the most part kept running for about 8 hours with some drop-outs. At that time it appears as though EVE's io locked up and crashed. After a reset of EVE, a talk session revealed that all barometers were working.

Conclusions:

See EVE code inadequacy per above, only this time the only process running was EveTerm / talk.

Test 4

Modified B&B 485 driver to run all barometers straight off PC. Modified procomm script to reduce the output messages and columnize it for ease of handling. However boosted script message i/o to fastest rate possible.

Changed PTB config in all barometers to: 9600 seri N 8 1 F in order to see if that would (despite vaisala instructions) overcome contention lockup.

Baud Parity Data Stop Dpx

9600 N 8 1 H

Output interval

1 s

Averaging time

1.0 s

Running at 12.1vdc

Results:

Note-4a

Lockup of baro U4110001 during setup of test. Definitely caused by low voltage. Confirmed power cycling didn't recover. Tried on EVE box (ie by itself) and discovered that it would work on 'RS232' down to 10.9vdc, but not on 485 even though the Vcc was 5v for the LTC491. Tried several times raising and lowering voltage noticing it stopped at about 12.35. Decided to put it back on full baro-bus and run at 12.1 for giggles, and naturally it came up....ie flakey somewhere.

All barometers locked up after roughly an hour………perhaps due to fastest possible polling by script…..

Test 5

Same as 4 except put polling back to 1msg/second and changed PTB config back to 9600 N 8 1 H

Results:

Barometers ran quite a bit longer but then did have groups go out and eventually return, only to repeat. And final situation appeared that only baro B7, U4110001 continued report by itself.

Test 6

Same as 4. Removed baros NCAR0003 and U4110001 from bus (note U4110001 had been only one running at end of last test).

Results:

Initially saw similar individual 1 message drop outs. Eventually by morning all but 2 sensors were down.

Test 7

After receiving email from Vaisala, I removed all the 485 termination on the HMP230's (probably should have been done long ago…) and changed polling to Crsend xCR

Results:

Initially quite good….signal on 485 bus was much improved and the ensemble reported very well, however by morning once again all but 3 had locked up.

Conclusions:

Important to remove 485 termination for our topology. Very helpful to send leading CR to help clear out barometers command buffers just in case.

Test 8

Per email from Vaisala, all barometers setup at 4800bps, and retained the polling Crsend xCR. PC polling, short 485 bus, no termination on end.

Results:

All barometers stayed up over the weekend!

Conclusions:

The barometers perform much, much better at 4800bps for bussed operation.

Notes on RS485 Waveform

B&B converter with 120ohm/33nF termination on end of bus

PAM 485 with 120ohm/33nF termination on end of bus

Without PTB220 internal 485 Termination: improved signal

Setup command

seri 9600 N 8 1 H

half duplex recommended for 485 bussed operations

seri 9600 N 8 1 F

typically used in EVE

open x

open dialog to barometer

Reqd after an open (or do a power cycle) to allow polling to work properly

echo on/off

allows you to see stuff

intv

sets reporting interval

dnos

self diagnostic message (only higher versions)

stop if in run mode

smode poll

sets it to reply to polling commands

form "Bx " 4.2P1 " " 3.1T1 " " ERR #r#n

scom

sets up to use a specific 'poll' command besides 'send x'

reports all status info. (don't do this on a bus….all reply)