Taylor-Hobson Talyvel 4 - Page 2
Close
Login to Your Account
Page 2 of 2 FirstFirst 12
Results 21 to 32 of 32
  1. #21
    Join Date
    Sep 2011
    Country
    GERMANY
    Posts
    3,436
    Post Thanks / Like
    Likes (Given)
    2215
    Likes (Received)
    1328

    Default

    One of the cables from read head to display unit was missing a connector. For the next person who has to fix or make a Talyvel 4 cable, here's the details.


  2. Likes CharlyDE liked this post
  3. #22
    Join Date
    Jul 2007
    Country
    UNITED KINGDOM
    Posts
    5,265
    Post Thanks / Like
    Likes (Given)
    3178
    Likes (Received)
    2676

    Default

    I didn't realise that NiCd batteries were still available except for military applications!

  4. #23
    Join Date
    Sep 2011
    Country
    GERMANY
    Posts
    3,436
    Post Thanks / Like
    Likes (Given)
    2215
    Likes (Received)
    1328

    Default

    Quote Originally Posted by Mark Rand View Post
    I didn't realise that NiCd batteries were still available.
    NiMh and Li batteries are so much better that NiCd batteries are only used for "legacy" applications. If I needed to replace these I would modify the charger circuit and install a pair of 3300mAh 8.4V NiMh packs intended for model airplanes. But the current NiCd batteries still work well, so no need for that yet. Possibly no need ever since I intend to use this only in my shop where I have electrical power.

  5. Likes Mark Rand liked this post
  6. #24
    Join Date
    Sep 2011
    Country
    GERMANY
    Posts
    3,436
    Post Thanks / Like
    Likes (Given)
    2215
    Likes (Received)
    1328

    Default

    Here's what the display unit looks like (turned off, sorry).



    I spent some time this weekend going through the calibration procedure. The gain calibration requires shifting the angle by 300 seconds of arc, which is 1.454mm in one meter. I wanted to be able to do this in many steps of around 20 arcseconds so that I could also test the linearity. At first I did this using a small sine plate, but it was very time consuming to keep shifting the gage blocks and very fussy and error-prone, so eventually I bit the bullet and made this long sine bar out of a piece of 30 x 40mm extruded aluminum I had kicking around (4mm wall thickness):









    I glued a hardened ground 10mm pin into a machined slot on the left side and fixed a micrometer screw with an 8mm ball on the right side. The ball is glued into a turned recess in the non-rotating micrometer screw shank.



    The length between the contact points is 1 meter. So I can smoothly shift the angle from -300 to +300 arcsecs just by dialing the micrometer between -1.45 and +1.45mm of level. Each graduation of the micrometer (0.01mm) is about 2 seconds of arc.

    The error of the bar over the +-300 arcsec range is around 0.3%, dominated by the accuracy of the micrometer screw and my ability to read the markings. The other major source of error is the determination of the distance between the dowel pin and the contact point of the ball, but the error here is less than 1 mm, so below 0.1%.

    When I first got the meter the gain for head A was about 65% of the correct value and the gain for head B was about 89% of the correct value. With this setup I was easily able to set it within 1% of the correct value. I was also able to test the linearity and repeatabilty of the level, which are remarkable. Taylor Hobson claims that the errors are below 2% of reading + 0.2 arcsec, but I am finding the linearity is within about 1% over +-300 arcsecs. It also repeats rock solid within 0.1-0.2 arcsecs, something which I found very challenging with an autocollimator.

    A nice test is to line up read heads A and B on the sine bar, put the display in B-A mode, and then tilt the bar, checking that the display does not vary as the sine bar is moved. With the gains set correctly the display of B-A fluctuates in the +-0.2 arcsec range as A and B vary over the range from -30 arcsec to +30 arcsec.



    This verifies that one can use the differential mode to remove table tilt when doing surface plate mapping.

    Fantastic device!
    Last edited by ballen; 08-05-2019 at 02:58 PM.

  7. Likes andrewmawson, CharlyDE, Mark Rand liked this post
  8. #25
    Join Date
    Sep 2011
    Country
    GERMANY
    Posts
    3,436
    Post Thanks / Like
    Likes (Given)
    2215
    Likes (Received)
    1328

    Default

    I'm going to continue extending this thread with info that might be useful to others with Talyvel4 meters. This post concerns the computer interface. The manual provides good detail of what is needed: a DB-25 (male) connector to DB-9 (female) serial port connector cable, set up as follows:

    Code:
    DB25 (Taylvel) end              DB9 (Computer) end
    
    BRAID --------------------------------- BRAID 
    
    RxD 2 --------------------------------- 3 TxD
    TxD 3 --------------------------------- 2 RxD
    DTR 6 --------------------------------- 8 CTS
    GND 7 --------------------------------- 5 GND
    
        18--|                           |-- 1 DCD
        25--|                           |-- 6 DSR
                                        |-- 7 RTS
    What the manual completely fails to mention (and I figured out after spending 20 minutes with a scope tracing the circuit and scratching my head) is the link shown on the bottom left above. On the DB-25 side pins 18 and 25 must be jumpered!! This is done to reduce power use: the computer interface circuits get their supply current via this jumper. This is not documented anywhere except on the schematics! So if you unplug the DB-25 side (or if the jumper is missing) then the computer interface circuitry remains unpowered and non-functional.

    Another important detail that the manual gets wrong concerns the "remote" contacts. After sending the unit an ASCI "X" character (ASCII decimal 88) the unit does NOT begin steadily transmitting values as the manual claims. Instead, after sending the "X" the unit stands by until you close the "remote" switch contacts. Then it transmits a single measurement value. Again, the only way to figure this out is to look at the latching circuit in the schematic.

    In gradient mode the format of the transmitted values is ("S" means "slope"):
    Code:
    S_A-0.029  (channel A)
    S_B+1.237  (channel B)
    S_D+2.234  (channel B-A)
    In arcseconds mode it is:
    Code:
    __A-234.5  (channel A)
    __B+023.7  (channel B)
    __D-123.4  (channel B-A)
    The underscores above "_" means an ASCII blank space character, and "D" is used to denote "difference".

    I think there is no way to change from A to B to D (B-A) mode under computer control but need to look at the schematics with this in mind.

  9. #26
    Join Date
    Sep 2011
    Country
    GERMANY
    Posts
    3,436
    Post Thanks / Like
    Likes (Given)
    2215
    Likes (Received)
    1328

    Default

    I want to make my surface plate more accurate by lapping away the high areas. The first step in this was to get a good map of the topography. I have finally gotten the Taylor-Hobson software working and used it to make a grid-map of the plate (as opposed to the Union Jack maps I had made earlier).

    My plate (800x1200mm) is sitting on a proper three-point support stand, but that stand itself in sitting on an ordinary workshop floor which tilts as I move around it. So I always do differential B-A measurements of the plate, with one level (A) fixed to the plate and the other (B) moving around the surface. This subtracts away the effects of any changing tilt of the base.

    To make a grid map requires that the moving level retains a constant offset while doing all of the the north-south lines and then while doing all of the east-west lines. To make this faster and more accurate, I added a structure under the plate (see photos) to hold the A head so I could move the B head anywhere on the plate. This is attached below the plate plate via three 25mm pads epoxied to the underside of the plate (Stabilit Express, Methylmethacrylat, good for metal and granite). The location just above the frame of the stand puts it out of sight and out of the way, but it's still easy to access. It has a further benefit: the support plate is designed so that if the epoxy lets go, it will only drop the metal plate about 0.5mm onto the metal frame. If you look closely at the picture below you can see that I have put a thin layer of green foam above the 0.5mm gap, so that the head won't be damaged by impact shock if that happens.





    Here the head is set up for the north/south lines. I rotate it 90 degrees into the other position for the east/west lines.

    The plate is ground tool steel, 10 x 130 x 130mm, and the posts are 12mm stainless round with a 6mm through-hole for an M6 screw. The lengths of the three posts are tuned to about 0.01mm (0.00004") so that the steel plate is fairly parallel to the table top directly above it.

    The reward for this effort was a total closure error of 0.55um (about 20 micro-inches).

    Here's the result of a grid of 8 x 12 measurements on 90mm squares, which produce 9 x 13 heights (in microns).



    It's exactly what I had expected from the Union Jack measurements, but I am reassured to see that it's consistent with that. Basically the surface is bowl-shaped, and about 10 microns low in the middle.


    I 10.42 8.65 7.44 6.53 6.57 6.32 6.21 7.54 7.62 8.09 8.40 9.40 10.42
    H 8.05 6.65 5.70 5.17 4.82 4.48 4.40 5.18 5.95 6.14 6.39 7.22 8.13
    G 7.18 5.66 4.99 4.36 3.93 3.28 2.96 3.87 4.64 4.95 5.49 6.17 6.60
    F 6.18 5.10 4.57 4.22 2.27 1.78 1.62 2.91 3.76 4.73 4.93 5.43 5.88
    E 5.85 4.67 3.94 3.17 1.12 0.00 0.96 2.69 3.66 4.74 4.97 5.13 5.72
    D 6.15 3.94 2.75 2.14 1.64 1.47 1.78 2.76 4.29 5.21 5.75 5.93 6.25
    C 7.23 5.29 4.25 3.68 2.05 1.85 1.81 2.67 3.44 4.09 5.64 6.44 7.09
    B 8.40 6.52 5.08 4.12 2.77 2.89 1.96 2.56 3.43 3.45 5.62 6.62 7.88
    A 10.42 8.60 5.94 4.41 2.66 2.83 1.19 2.10 3.14 3.31 4.58 5.98 8.06


    The next step is to start lapping away at the high areas on the periphery of the plate -- sort of like step-scraping but with a lap. If anyone here has done something similar, I'd be grateful for advice about lap composition, and SiC grit to use. I initially want to remove about 3-5 microns from around the periphery of the plate.
    Last edited by ballen; 11-09-2020 at 02:45 PM. Reason: improve formatting

  10. Likes eKretz, CharlyDE liked this post
  11. #27
    Join Date
    Jul 2007
    Country
    UNITED KINGDOM
    Posts
    5,265
    Post Thanks / Like
    Likes (Given)
    3178
    Likes (Received)
    2676

    Default

    I can't help with SiC abrasive, but I was lapping my 600x900mm surface plate last week, using a 300x450 cast iron plate as the lap. With that lap size, 40-60m diamond grit cuts very fast, 30-40m diamond grit cuts almost as fast. 15-30m cuts more slowly and 6-12m and finer cut very slowly.

    When I say fast, I mean that a freshly charged lap can remove 2g of granite in a minute with little effort. Slow is down to about 0.1g per minute. I have been using a 0.1g resolution electronic scale to estimate the amount of granite dust I am creating, because it is a way to estimate how much height I am removing, knowing that granite is about 2.75g/ml. e.g, to remove about 10 of height from half the area of your plate, you would aim at 80x120x0.001/2= 4.8ml=13.2g of granite.

    Aim at pushing the lap well over the edge of the plate (30-50%) in order to remove more material from the outside than the centre of the plate. You can even diagonally 'plane down' the corners of the plate if you need to remove material from the very outside of the corners, but the lap will cut much faster then because the granite dust falls to the floor and doesn't prevent the lap from cutting.

    When you get to finer abrasives, the lap needs to be flat, otherwise the abrasive does not come into contact with the granite. The other problem is that the lap will tend to float on a layer of air very easily if you get down to 20m and smaller abrasives. I'm thinking of grinding some grooves across the lap i'm using with a parting disk on the surface grinder to prevent this. Obviously, if you are using an oil based SiC abrasive, that problem might happen sooner.

    I found that I could improve the'bearing' of the lap by using Timesaver compound a few times instead to diamond. This cuts the lap as well as the granite. With your SiC abrasive, you might get this effect anyway if the abrasive isn't embedded in the lap.

    The reason I'm lapping my 600x900mm plate is so that it is flat enough to scrape a 450x600 cast iron plate to be a lap on the 900x1200 surface table . I was habing great difficulty scraping it flat using a smaller granite plate as a master because it was very hard to avoid scraping a twist into it.

    Using your Moody program, the granite plate started at 12m low in the middle and quite bumpy.
    It then went to 22m low in the middle. That's when I discovered that you have to really push the lap past the edge of the plate to remove material from the outside edges.
    It then went to 6m high on two opposite corners and flatter over the rest.
    It's now at 3.5, but low on the other two corners (I over corrected!)

    It actually takes longer to make the measurements than to do the lapping.

    I have been forced to do some painting and decorating work in the house by my wife, but I hope to get back to lapping during the evenings.

    PS:_ Thank you for the program, it saved me from having to re-learn my 25 year old C skills.

  12. #28
    Join Date
    Sep 2011
    Country
    GERMANY
    Posts
    3,436
    Post Thanks / Like
    Likes (Given)
    2215
    Likes (Received)
    1328

    Default

    Hi Mark,

    I was thinking about making a plaster-and-tile lap like those used for telescope mirrors. I can cast them on my smaller (300 x 450) surface plate, which is good to 2um.

    I was planning at least initially to use a much smaller lap, says 150x150, to do the equivalent of step scraping, to knock down the high areas. Then to move to a larger lap as I finish. Is that a mistake? I have a 300 x 300 cast iron plate which I can use at the end.

    As you say, I need to remove about 4 microns x 80 cm x 120 cm = 4ml = 13g of granite. In order to have control of the process I want to start with the SiC equivalent of your 15-30 micron diamond, so I am removing at most a fraction of a gram per minute. From looking at this table, it seems that 400 grit SiC is about 17 microns, so that should be a good starting point.

    Yes, the measuring process will be more time-consuming than the lapping for sure!

    I am very happy to hear that you are using the Moody program that I wrote. Are you using a level or an autocollimator to get the data points?

    Cheers,
    Bruce

    Edit: I just got a good deal on three cast iron rounds, 250mm diameter x 40mm thick, weight about 14kg each. I am going to flatten these on the lathe, and use a bandsaw or tablesaw to cut swarf slots. Then I'll lap the three of them flat against each other using Whitworth's 3-plate method. Those should give me a good set of tools for leveling the areas around the periphery of the surface plate. If needed I can use a 300 x 300 square surface plate to smooth over any transitions.
    Last edited by ballen; 11-10-2020 at 08:46 AM.

  13. #29
    Join Date
    Jul 2007
    Country
    UNITED KINGDOM
    Posts
    5,265
    Post Thanks / Like
    Likes (Given)
    3178
    Likes (Received)
    2676

    Default

    That all sounds good. I think the cast iron rounds may be less work than the 'fabricated' lap. (He said the after one whole week of experience )

    I'm using an autocollimator. I did do one run using a 200mm 5 second/div spirit level. It is hard to interpolate to 1 second (compared with 0.1 sec on the autocollimator) and is very 'noisy'. Only good for rough measurements.

    A set of Talyvel levels would be more flexible/universal than the autocollimator, I think and similar accuracy.

  14. #30
    Join Date
    Sep 2011
    Country
    GERMANY
    Posts
    3,436
    Post Thanks / Like
    Likes (Given)
    2215
    Likes (Received)
    1328

    Default

    0.2 arcsec is 1 micron in a meter. So if you want to calibrate a reasonable-sized plate at the micron level, you need 0.1 arcsec accuracy and resolution.

    I initially mapped out my surface plate using an autocollimator, but it was very fussy and time-consuming. With the Talyvel 4 hooked to a computer, I can do it more accurately and much more quickly.

    I'll come back here again when I have the three circular laps prepared. Then we'll see if my plan to do the lapping equivalent to step scraping can succeed.

  15. #31
    Join Date
    Sep 2011
    Country
    GERMANY
    Posts
    3,436
    Post Thanks / Like
    Likes (Given)
    2215
    Likes (Received)
    1328

    Default

    The Talyvel 4 was designed to run off either wall power or off a pair of internal rechargable NiCd batteries. These batteries are obsolete, and the internal charging circuit is primitive, so I've improved the charging circuit and replaced the batteries with modern Li-Ion 18650 batteries that have much longer life and none of the undesirable (memory, self-discharge) issues of NiCds. Here are the details for anyone who wants to carry out a similar modification.

    The stock "charging circuit" is a 39 ohm 2W resistor in series with the positive or negative 12V rail (actually about 11.3V after the series diode). This is a lousy way to charge NiCd batteries: it cooks them with too much current when they are undervoltage and then continuously, even when they are fully charged. Remove these two resistors:



    and replace each one with one of these two-pin constant current modules (125 mA):



    If you want to use these with the original NiCds, that's possible. In this case, add another diode (or two) in series with the one shown, so that the charging cuts off when the NiCd reaches about 8.4 to 8.5V.

    Alternatively, if you want to replace the batteries with Li-Ion ones, then drop the resistor to 6.2 or 6.8 ohms to provide a larger current (around 200mA) and replace each of the two stock NiCd batteries with a pair of 18650 Li-Ion batteries connected to a "2S BMS balance" board, which provides overcharge and undervoltage protection.

    Here's what it looks like when installed. You can see the two modules I made in the middle. These are hand-wired on a bit of prototype board then potted in epoxy to make them robust. The red and black stripes to identify the positive and negative sides. The LED lights when charging, use a low-voltage one (red is a good choice, I didn't have any on hand so used green which is higher voltage and not such a great choice).


  16. #32
    Join Date
    Feb 2010
    Location
    ch
    Posts
    2,913
    Post Thanks / Like
    Likes (Given)
    243
    Likes (Received)
    396

    Default

    so when you use the level and have mapped out a straight line, how do you decide what the slope of the "reference" line should be?


Tags for this Thread

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •