facebook youtube

Date published: 2022-06-18

Improving the periodic error and guiding ability of a Losmandy GM8 mount

Why bothering about improving Losmandy’s GM8 mount

Fig. 1: Losmandy GM8 guide star drift while auto-guiding. Overall the mount produces round star images and guiding works. However, the guide star drifts are relatively large (compared to seeing).

As mentioned in an earlier post, I have purchased a used GM8 mount for less than 800 EUR (including motors, digital drive system, tripod and polar scope), hoping that this mount will allow smooth auto-guiding. Indeed, the GM8 does not show severe problems such as large jumps in declination (like many other cheap and light mounts such as EQ5). Thus, guiding principally works (see Figure 1). However, the Losmandy GM8 mount I got does show high-frequency peaks in RA on top of the periodic error (see Figure 2 and 3 and previous post), which negatively affect image quality, i.e. it blows up the star images when doing long exposures. For that reason, I decided to make some improvements (that’s what this post is about), hoping that this will finally allow multi-minute exposure times and perfectly round stars close(r) to the seeing-limit.

Fig. 2: RA tracking accuracy of an unmodified Losmandy GM8 mount. The gray points indicate star (zeta Peg) positions (y-axis) recorded over a period of time (x-axis). The periodic error (PE) associated with the worm gear period of ~8 minutes is seen after smoothing the data (magenta curve). Data was obtained using MGEN’s “save pos” functionality and the Lacerta tool (to convert pixel offsets into arcsecs depending on the star’s declination).
Fig. 3: FFT frequency analysis (in units of the worm period of 479s on x-axis) showing several peaks/amplitudes (y-axis) of an unmodified GM8 mount.

Step 1: Upgrade to high-precision ABEC-7 worm bearings

The very first step of improvement consists of disassembling, cleaning and re-greasing. This is necessary as with time lubricant and/or grease accumulates on the nylon discs that are located between the pressure plates and gear surfaces (see Figure 4). The slippy nylon discs then prevent the clutches to work properly, i.e. only when clutches are very strongly tightened the axes can be fixed. After cleaning (see Figure 5), the original worm bearings were exchanged with high precision ABEC-7 bearings that I purchased from the Boca Bearing Company under part number SR4-ZZC #7 PS2 (see Figure 6). Finally, the mount was re-greased and re-assembled.

The periodic error of the mount after this upgrade is shown in Figures 7 and 8. As a result, the amplitudes of the highest frequencies have decreased by a factor of ~2; compare left parts in Figures 3 and 8 where periods are <0.2T, with T being the worm period of 479s. However, the tracking curve is still dominated by high-frequency signals with peak-to-peak amplitudes of roughly +/-20arcsec.

Fig 4: After disassembling the GM8 mount, lubricant and grease was seen on the nylon discs that prevent the clutches to get tight enough.
Fig 5: Losmandy GM8 disassembled and cleaned.
Fig 6: New high precision ABEC-7 worm bearings on the RA worm
Fig. 7: RA tracking accuracy of a Losmandy GM8 mount with ABEC-7 bearings. The gray points indicate star (epsilon Hya) positions (y-axis) recorded over a period of time (x-axis). The periodic error (PE) associated with the worm gear period of ~8 minutes is seen after smoothing the data (magenta curve). Data was obtained using MGEN’s “save pos” functionality and the Lacerta tool (to convert pixel offsets into arcsecs depending on the star’s declination).
Fig.8: FFT frequency analysis of the GM8 after the upgrade to ABEC-7 worm bearings.

Step 2: Sanding the bearing blocks and preload the worm bearings with disc springs

I recognized that the worm bearings sit very firmly in their mounting cubes. Thus, I decided to sand and polish (using a very fine sanding sponge) the inner parts of the RA and DEC worm bearing mounting cubes and preload the bearings using disc springs (see Figure 9), which I purchased from Key Bellevilles Inc under part number 625 634 EL. The specifications of the disc springs are: outer diameter: 0.62in, inner diameter: 0.32in and thickness: 0.0098in. Due to the extra space needed by the disc springs, the distance between the mounting cubes of the modified worm blocks has increased (see e.g. Figure 10 for RA). Thus, the holes to mount the worm block on its mounting plate were extended using a round file (see e.g. Figure 11 for RA).

After re-assembling the mount, the periodic error measurement (Figures 12 and 13) shows that all the efforts payed off. The high-frequency peak-to-peak amplitudes could be reduced by a factor of ~2 to a level of roughly +/-10arcsec.

Fig. 9: one of the worm bearing mounting cubes after sanding/polishing and with disc spring inserted
Fig 10: RA worm block with preloaded worm bearings sitting in their mounting cubes
Fig 11: RA worm mounting plate with extended mounting holes
Fig. 12: RA tracking accuracy of a Losmandy GM8 mount with ABEC-7 bearings, sanded/polished bearing blocks and preloaded RA bearings using disc springs. The gray points indicate star (delta Vir) positions (y-axis) recorded over a period of time (x-axis). The periodic error (PE) associated with the worm gear period of ~8 minutes is seen after smoothing the data (magenta curve). Data was obtained using MGEN’s “save pos” functionality and the Lacerta tool (to convert pixel offsets into arcsecs depending on the star’s declination).
Fig. 13: FFT frequency analysis of the GM8 after the upgrade to ABEC-7 worm bearings, subsequent sanding/polishing of bearing blocks and preloading the RA bearings using disc springs.

Step 3: Lapping the worm and wheel

The final step is to lap the worm and the worm wheel together to smooth off any imperfections. To do so, I used Macolaepp K2000 diluted 1:50 with Liqui Moly LM47 grease. After applying the mix to both, the worm wheel and the worm gear, I used a cordless screwdriver to spin around the worm and worm wheel several times in both directions until the worm could be turned easily by hand at all positions of the worm wheel without shifting and no recognizable backlash. Once both axes were lapped, I carefully removed the lapping paste, re-greased and re-assembled the mount.

Finally, the tracking curve does not show any high-frequency disturbances and is characterized by the periodic error only, with an amplitude better than +/-2arcec (see Figures 14 and 15). That’s quite impressive!

Fig. 14: RA tracking accuracy of a Losmandy GM8 mount with ABEC-7 bearings, sanded/polished bearing blocks and preloaded bearings using disc springs. Additionally the worm gears and worm wheels were lapped (using Macolaepp K2000 diluted 1:50 with LM47 grease). The gray points indicate star (alpha Boo) positions (y-axis) recorded over a period of time (x-axis). The periodic error (PE) associated with the worm gear period of ~8 minutes is seen after smoothing the data (magenta curve). Data was obtained using MGEN’s “save pos” functionality and the Lacerta tool (to convert pixel offsets into arcsecs depending on the star’s declination).
Fig. 15: FFT frequency analysis of the GM8 after the upgrade to ABEC-7 worm bearings, subsequent sanding/polishing of bearing blocks, preloading the RA and DEC bearings using disc springs and lapping the worm and wheel.

Conclusion