1. A bit of history:

The 4m dish pedestal is originally designed as one component in mobile 2.8-3.3 GHz radar radioset SCR-584-B around 1945-1946.

Some of original components of this particular unit have been made in 1960, which gives an impression that the unit itself is made around that time.

The hardware underneath our 4.0m dish is:

• Az-/El- mechanics with 260V/2A+0.2A DC motors
• mechanics has also syncro/selsyn based axle orientation feedback
• Tube based servo amplifier with motor-generator type post-amplifiers that actually feed the motors.

The pedestal has been in use at Metsähovi Radio Laboratory from 1966 until about 1976-1980, from where it was given to Viestikallio group in 1998.

It has not accumulated very much operational hours, probably less than about 1000 hours total, and practically none of them in radar scanning like modes, that is, mechanics have essentially idled.

2. Plans (for replacing things)

2.1. Pedestal

• Iron is ok.
• External cover gaskets need replacing. (Silicone?)
• Paintjob needs to be reworked.
• Motors were changed (summer 2001), new motors are:
  • 24V DC motors with 1:10.17 reduction gear
  • The motors have built-in 512 pulses/round incremental (TTL level output!) rotation feedback sensors (no index signal). These must be changed into RS-422 differential for transfer to the controller.
• Axles need absolute orientation sensing; indices + increments may also do, but will require possibly 360 degree turning before finding said marker index.
  Update: There is readily usable mechanical connection point with 6.00 mm axle diameter at both Az- and El-axles which connections do turn 16 times the respective main axle.
  This gives 4 bits "for free" - but only when using multiturn absolutes!
  • As a surplus we happened to get two (a bit weathered) Stegmann AG-661-09-4096x4096 multiturn absolute encoders.
  • As a surplus we happened to get yet one Stegmann AG-626-WSR multiturn absolute encoder.
  • Electrical interface for Stegmann sensors (two); a bi-directional RS-422 differential signals set.
• For the Azimuth-axle a ± 2 turns (alltogether 4 turns ?) limit switches with mechanical construction! (Using inductive(?) 4-20mA current-message limit switches.)
• Inductive proximity limit switches for the elevation axle
• Interface circuitry for reading 4-20mA current-message devices
  • Feeds the system power (24V DC) with over-current limit (30 mA max)
  • Detects lack of operational voltage
  • Detects short-circuit
  • Detects open-circuit
  • Gives signal: limit/not (around 10 mA ?)
  • Gives signal: 0 .. 20 mA -> 0 .. 4.5V
  • Gives signal: FAULT ( + possible additional indications - high/low-fault )
• ==> 2.3.4. Controller
• ==> 2.3.5. Interface hardware for DC motors

2.2. Tracking details

For some analysis about tracking resolution requirements, see: 13.7m tracking needs page. The pointing at 10 GHz will need 12.73 bits of resolution, tracking calls for a bit more, say 3.322 bits more -> 16.05 bits.

The 3dB beam width is 0.532° at 10.5 GHz, and 0.233° at 24.0 GHz. (1/677 and 1/1547 of full circle, respectively.)

Depending on accuracy, possibly anything from 15 bits up will do.

Gearing ratios:

• At Az-axle the gearing from motor axle to main az-shaft is: 1:562.5
• At El-axle the gearing from motor axle to main el-shaft is: 1:1030

Talking with Michael, OH2AUE, about what the dish is good for, he said we should try to be able to operate pointing (very least) for 24 GHz.

At old motors the 3dB beam-width means:

• Az: 1/1.2036 rotations for 10.5 GHz, and 1/2.7502 rotations for 24.0 GHz
• El: 1/0.6573 rotations for 10.5 GHz, and 1/1.5019 rotations for 24.0 GHz

At old motors the 1/10 of 3dB beam-width (pointing) means:

• Az: 1/12.036 rotations for 10.5 GHz, and 1/27.502 rotations for 24.0 GHz
• El: 1/6.573 rotations for 10.5 GHz, and 1/15.019 rotations for 24.0 GHz

At old motors the 1/100 of 3dB beam-width (tracking) means:

• Az: 1/120.36 rotations for 10.5 GHz, and 1/275.02 rotations for 24.0 GHz
• El: 1/65.73 rotations for 10.5 GHz, and 1/150.19 rotations for 24.0 GHz

With new motors integrated speed encoders the 1/100 of 3dB beamwidth (tracking) means:

• Az: 43.25 pulses at 10.5 GHz, and 18.93 pulses at 24.0 GHz
• El: 79.19 pulses at 10.5 GHz, and 34.66 pulses at 24.0 GHz

The old DC motors can be driven to about 1/10-1/20 of the motor axle turn.

Both the AZ and the EL axles have selsyn axles spinning at the rate of 16:1 faster than the axle:

• There is excellent coupler interface with 6.00mm diameter axle for modern measuring devices.
• The gearing is backlash free (per documentation) at these axles

As it turns out, Heidenhain makes absolute encoders which have also incremental (analog!) signal of quadratudic kind:

• ROQ 426 (2¹³ bits/r abs, with 2¹¹ increment phases): EUR 680,- + VAT (FIM 4980 incl VAT)

Comparing this with e.g. Stegmann:

• DG60L,XSR5000,F: 5000 steps/r increment: FIM 2000,- incl VAT
• AG626 WSR: multiturn SSI, 13 bits/r: FIM 3843,- incl VAT
• Connector + 5m cable: FIM 354 incl VAT

2.3. Controller/Driver

Control computer drives H-briges with speed feedback for controlled speed ramp-up/-down.

The bridges need to have current limiting circuitry, possibly with controller driven current envelope.

For example NatSemic LM628 can do this quite well with Heidenhain ROQ426/2k.

... but!

It really needs at least as much of incremental speed data as pointing resolution is! Preferrably more!

In fact, the Az-motor axle needs to have about 195 separately identified orientations for smoothish tracking.

Huh . . .

2.3.1. Picture 1: Overview

The Controllers ("Ctrl") below may be LM628, or they may be something else.
Changing the motors and having new power drivers in place of "H-bridges" can mean interface alternates there:

• A ± 10V voltage signal
• Possible sign + PWM

Note: This picture precedes the decission to get new DC motors of 24V
Note: This also precedes the decission to use MPC555 controller

2.3.2. Picture 2: Control loops

• Innermost: Gate-drives with hardwired overcurrent protection
• Current control: Controls (limits) motor moment with measuing/PWMing its current
• Speed control: Commands movement around
• General command: Reads absolute encoder (and does fixups), and compares its report to that of the commanded direction setting.

The gate-drivers with their over-current protection are hardware stuff, all other boxes are more or less software.

2.3.3. Picture 3: Az axle/motor/pot view from under

The pedestal mechanics itself from underneath on top of the cylinderical base "tube" (d ~ 1.0m, h ~ 0.7m)

The bright white at the right side is crawl-door to the outside.

The image is retouched (to clarify it a bit), click on it to see the original.

2.3.4. Controller (no picture)

Intention (as of 27-jun-2001) is to use a MPC555 based "ITIpower555" controller made by (www.)ITI(.fi).

The controller has following features:

• Ethernet (10base-T)
• Two CAN-2.0B interfaces
• LON interface
• Connection to text/graphics LCD display modules
• 16 analog inputs with 10-bit A/D converter
• 16 analog/digital (dualmode) inputs with 10-bit A/D converter
• 8 dedicated PWM outputs (MIOS)
• 16 general purpose digital IO pins (MIOS)
• 32 TPU IO pins
• CPU internal Flash: 256 kB + 192 kB
• CPU internal SRAM: 16 kB + 10 kB ( + 6kB TPU SRAM)
• External Flash: 512 kB (8-bit wide)
• External SRAM 256 kB (32-bit wide)

To make good use of things (in the pedestal mechanics), we can:

• Do synchro resolving (120° 3-phase selsyns!); at most about 12 bits/r
  • For both axles use both the coarce- and fine-selsyns! ( = 4 selsyns! )
  • Fine-selsyn gives effectively 4 bits more -> 16 bits/r of AZ/EL axles.
  • The accuracy of this thing is uncertain!
  • Using the A/D converters in the controller needs:
    • Selsyn exitation by TTL pulse, or some sine generator (400 Hz?)
      (The old system PPI scope uses sharp pulses for exitation)
    • Analog signal conditioning circuitry from selsyn output (±? V) to A/D input
      Possibly doing (positive/negative) peak detection/hold so that the transient-like exitations effect can be studied leisurely (and with better accuracy!)
    • 3 A/D input pins for each used selsyn! ( -> 4x3 = 12 A/D inputs )
    • 1 D output for each selsyn exitation (or single one for all four ?)
• We have 10 bits/r + 64 revs absolutes (10+6 bits, parallel!), which can be used to yield (for AZ) the sector knowledge, and also about which rotation the thing is in.
• We have one Stegmann AG-661-09-4k/4k multiturn absolute encoder (12+12 bits) serial communication models. It has enough resolution for this project at 16x "fine-selsyn" axle. More are looked around for..
• We have two AD2S83IP syncro-digital converters; 16 bits/r. These can be used for some resolving backups..
• We have also some older syncro-digital converter hybrides:
  • Analog Devices: "SDC1704 option 611" -- educated guess: 6=temp-range ?, 1=RefFreq 400 Hz, 1=11.8V Ref/26V Signal; 14 bits, 2-phase
  • Some unknown maker from 1979!, 16 bits out, 3-phase syncro

Plan: (before Stegmann AG661 fell to our hands..)

• Use 10+6 bit parallel interfaced absolute at the "AZ-potentiometer-axle" (see picture 3) to determine coarse Az-axle orientation; for the rest:
• 1st order aproximation: Using surplus syncro-digital resolvers:
  • Resolve El-coarse-selsyn with 14 bit AD's SDC1740 getting the rough idea of the axle orientation
  • Resolve Az- and El- fine-selsyns with 16-bit syncro/digital converters.
    (These might get 16 bits out of fine-selsyns.. Add there the 4 "free" bits from the fine-axis gearing --> in theory it gets 20 bits for a turn of main axles.)
• Backup plan: Use MPC555's 10-bit A/D converters along with frontend analog stuff to preprocess syncro signals:
  • Resolve Az-Fine-Selsyn with exitation pulse + 3 A/D inputs.
    (Gets at most to 16 bits/r (or main axle), probably only about 14-15 bits/r, we want 16-18 final bits)
  • Resolve El coarse- and fine-selsyns with exitation pulse + 3 A/D inputs each. (6 A/D for these)
    (Gets at most to 16 bits/r of main axles, probably only about 14-15 bits/r, we want 16-18 final bits)
• Future option: Replace (augment) both axle fine-selsyns with 13 (or more) guaranteed bits/r single-turn absolutes, or even multiturn absolutes, and obliviate the need for any other resolving.
  (Money question, of course... 1300 EUR (incl VAT) for a pair of such encoders is a non-trivial sum..)

2.3.5. Picture 5: Interface hardware for the DC motors

Interface hardware in between ITIpower555 and the motors:

• One PWM output per motor
• TTL output commands for each motor: (switch gate-driver controls, 2 per motor, 4 total)
  • Run/Break
    (At break both low side transistors conduct, high-sides are off.)
  • Left/Right
  (A clumsy attempt to denote logical one with bold, and logical zero with italic -- presuming also that powerup/reset condition will be logical zero.)
• One Analog input to A/D for motor current measurement (0..+5V) per motor (2 total)
  The shunt current (voltage) (see picture 5 below) is amplified by factor X by an opamp using 0/+5V power.
  (Back-EMF created during breaking is ignored, and won't affect gatedrives.)
• Switch gate-drives include hard wired over-current limiting circuitry
  (Software fault won't be allowed to burn the system)
  Over-current detection must force the PWM signal to zero until the next raising edge of the PWM cycle. (D-latch with RESET(s) ?)
  D-latch: D-in = "1", CLOCK = PWM, /RST = /OverCurrentDetect. The ProtectedPWM = (Q-out) AND (PWM).

The Back-EMF from motor can not go to the power-supply, thus the system needs something like a large crowbar Zener with trigger level of circa 25-26V. Both motors need their own shunts.

Missing: Directional limit sensor inputs for ELEVATION upper/lower edges.
(We have lots of 4-20 mA current message / 24V inductive proximity sensors to use.)

2.3.6. Interface hardware for resolvers

Interface hardware in between ITIpower555 and chosen resolving method:

• Existing 16 and 14 bit syncro/digital converters:
  • Resolver exitation sin source (50 Hz ? 60 Hz ? 400 Hz ? 4000 Hz ?)
  • TTL(?) IO adaptation - 16 bits wide I/O ? (not just I ?)
  • add controls/strobes/muxes

• Using MPC555's embedded A/D converters:
  • Resolver exitation sin source (50 Hz ? 60 Hz ? 400 Hz ? 4000 Hz ?)
  • Resolver receiving load resistors (circa 50 Ohms)
  • Amplifiers/scalers (from ± something to 0..+5V range); three per resolver

• Using e.g. Analog Devices AD2S83IP for syncro/digital conversion:
  • Resolver exitation sin source (50 Hz ? 60 Hz ? 400 Hz ? 4000 Hz ?)
  • Signal receiving preprocessing to convert from 120°(60°) 3-phase resolver to 90° 2-phase converter
  • TTL(?) IO adaptation - 16 bits wide I/O ? (not just I ?)
  • add controls/strobes/muxes

• Digital absolute encoder at fine-selsyn axles:
  • CAN-bus ?
  • SSI interface (RS-422/RS-485)

• Digital incremental encoders at fine-selsyn axles
  • Quadratude + index (3 signals) in RS-422 differential, or 11µApp differential, or 1.0Vpp differential (each differential needs 2 wires) (MPC555 TPU), ja:
  • The analog methods are received both by differential receivers with hysteresis (RS-422) feeding them to Quadratude tracking (MPC555 TPU), and:
  • Difference amplifier scaling the result into 0..+5V range for A/D inputs (1 per signal)

• Digital incremental encoders at motor axles
  • Quadratude (2 signals) in RS-422 differential (each differential needs 2 wires)

The SELSYN exitation/output signal ratio has been measured (with 4.7 kilo Ohm loads) to be quite accurately 2:1.

The SELSYN exitation was also observed to work with pure sin signals of 50-5000 Hz.

2.3.7. Interface hardware for current-loop devices

Burr-brown RCV420, Burr-brown AB-018, Burr-brown AB-031, Burr-brown AB-041

Simple approach is to use current-limited current sources at 24V level (max 50 mA ?), and have a low-side shunt resistance (100 Ohms) which voltage is measured.

Simultaneously detecting: over/under current, line break, and the actual "message", too.

2.3.8. Bigger picture of the interfaces.

2.4. Pedestal base

Underneath the pedestal mechanics, there is cylinderical base, which dimensions are in the picture below.

As a part of a plan in which we make a rain-proof hut under the mirror (see 2.5. below) options are:

• Creating a closable door into the "crawl-door" in the picture. Likely similar to what is now at the edge of the Az-mechanics. It will also likely need safety sensors to prevent system running while the door is not in place - e.g. when it might stick out in a way which comes into mechanical contact with the reflector.
• Also the big 1.2m hole under the base cylinder needs closing (but adding there cable pass-thru channel.)
• Alternatively closing that "crawl-door", and water+heat proofing whole cylinder inside to the top making cable pass-throughs etc. at the top of it.

2.5. Base structure

At the top of the Monument is this kind of 3x3 meter structure, on top of which is the 4 meter dish pedestal, base, and the mirror.

A side view shows that to put any decent door into this structure we must move the diagonal trusses at least at one edge, along with possible adding some vertical beams.

At the top of the picture there is galvanized walking-grating, below it a trellis desribed latter, and most of the picture consists of four triangular base structure.

Material for this triangular base structure is 50x50 L-bar.

From top appears a trellis made of UNP100 beams, on top of which there is a 1600x1600x5 steel plate on top of which the pedestal base sits.

In the picture there is a "key" shape trying to show where beams have welded bolt fixtures.

This whole trellis, and that plate are flexing structures, so that the whole structure bends considerably, and needs diagonal support stiffener trusses, of which one is shown with break-line.

Those stiffeners are, however, inside the "upper hut" demanding space in it. They also cause the making of waterproofing to be a bit more challenging under this trellis.

Designtask:

To stiffen this trellis without slanted support trusses inside the "room space".

Layout for walking grating on top of the trellis is shown in the picture below. Top (or bottom) of those bits exist, others need to be purchased.

For stiffness it is important how those bits are cut!

Optional for two 750x1500 bits is to make 750x750 bits, which are always placeable into right orientation. (We shall by a 3000x1500 bit of grating, cut it length-wise into two, and then other of those is cut to 750x750 squares.)

The structure has been place diagonally on top of the Monument, because otherwise there would not have been sufficient base support beams -- perhaps there has been some other reason too, because this structure could be oriented also other way.

"Upper hut" could be rotated 45 degrees, but then it needs to be moved also to other place so that it has support beams below it!

For reorienting the upper hut, there might be needed also expansion of the upper base area, to north or to south ?

TODO:

• Design of the water proofing
• Design of the heat insulation
• design of the floor
• Windows
• Door/doors
• Walk space
• Stairs!

2.6. Motor adapter models

In the pictures:

• The motor is pink
• Round adapter made from Machine-shop aluminium is orange diagonal lines
• Stainless steel flange is of green diagonal lines
• Adapter axle with oldham-coupling element is black stair-line
• Brearing is of blue triangles
• Dimensions are in millimeters

Cross-section pictures along X and Y axis.

The latter has motor axle keys, and oil filling plug hole.

Slanted view from above:

Slanted view from below.

A new set of pictures with rain-cover for Elevation motor:

3. Work log:

3.1. 16/Jun/2001: Opening the boxes..

We wanted know some technical things from deep inside, and thus opened several cover plates to explore a bit, and pull out couple SELSYN devices.

3.1.1. Work in the lower AZ-SELSYN compartment begun with vacuuming dead insects:

Left (in front of the pedestal itself): Kai Forssen, right: Matti Aarnio (OH2MQK).

We are about 11 meters above ground at a fairly small ledge...

3.1.2. A surprise in the upper compartment...

Yes, a birds nest among the nest of the wires...

This has been built there in between summers of 1998 and 2001! (The box was last opened in 1998 when the system was installed at its present location.)

Do note the "fine-potentiometer" (to be taken out, pictures below) on the top of the "fine-selsyn" -- left edge of the compartment, second unit from front.

The green/gray-brown stuff at lower left corner is out of focus moss at brown rock some 12 meters below the cameraman...

3.1.3. After the covers have been closed:

From rear of the dish:

No more birds here! A standard plastic reusable soft-drink bottle of 1.5L blocks the passage that they used.

3.1.4. SELSYN devices, and potentiometers that we took out...

The mounting hole diameter for these devices is about 92 mm (3+3/8 inch ?)

3.1.5. On top of two so called "fine selsyns" there are potentiometers

With this coupler we will place there modern digital absolute encoders; multiturn models, of course. (16 turns at these axles matches one turn of the Az or El axle..)

The axle diameter at this coupler is 6.00 mm.

3.2. 3/Aug/2001: Taking old Az-motor out..

The new motors have arrived, and we took old Az-motor out for getting adaptation mechanics done for it. (This is on video tape, no still photos.)

The take-out seremony involved:

• Opening two of four of the conical cover plates in the Az-assembly of the pedestal.
• Roping the dish down so it won't (much) move freely around
• Taking the Az-motor out, and covering the gearbox below it with cardboard plate so that it won't get much garbage (like dead insects) flying around in there.
• Carrying the 20-30 kg motor out and down (that was heavy lifting in difficult workplace, indeed!)
• Testing with a mechanical mockup to see, if the new motor fits into the intended installation space, and seeing that it will indeed fit.

The next day the wind strengthened, and we learned of the biggest mistake that we made -- when we took the motor out, the Az became loose to spin around in the wind.

The limitter ropes were in danger of snapping, and we had to raise the dish to its maximum elevation for it to present less opportunity for the wind to torque it around.

3.3. 11/Aug/2001: Various measurements..

Among other activities a moment was spent at measuring and checking the dimensions of the top platform of the monument.

This resulted in chapter 2.5. above.

3.4. 25/Aug/2001: Return of the Az-motor..

The Az-motor was returned from workshop, and got installed into the pedestal while waiting for the new adaptation mechanics.

3.5. 29/Sep/2001: Fitting test of the new motors..

Fitted the new motors into Az-motor install place (no photos taken of that session). Below are photos of these motors.

In rushed returning of the original Az-motor something went wrong, and it appears not to work at the moment.

The last picture has the schematic of things inside the connection box beside the motor gearbox. The integrated incremental encoder has TTL interface, and to move its signals a bit further away will most likely necessitate a small TTL->RS422 driver circuitry to be added into this box. The isolation/protection class of the motor is: IP-55.

3.6. 10/Nov/2001: Cabling the new motors..

Chairman Leskinen is working at the old controller in downstrairs -- disabling all unnecessary components, e.g. all except the analog directional displays.

We are driving the elevation up with a 12V power supply, and it is definitely raising slowly...

El: 42°	El: 50°	El: 59°

Matti Aarnio <matti.aarnio@zmailer.org>; OH2MQK