Learning to foucault

[skybadger]

Got my 6" blank that's been ground to a f/5 on the bench. 

While digging it out and getting stuff ready, found that my 12" tool I acquired which was bonded by pitch to the blank was actually a mirror close to being polished out.  I separated them using a few taps from a mallet and recycled the substantial amount of pitch. That's now been put on one side for later.. 

In the meantime, I poured a sub-diameter lap for the MoM polishing of the 6" . The reason I stopped a while back was because the lap pour wasn't working for me. In this case, it poured easily into the masking-tape wrapped tool mould and once firm enough I pressed onto the cerium-wiped mirror on top of a sink protector sheet to create the lap grid. That went so well, I took the grid off and continued pressing for a good contact. The squares closed up so the pitch was still warm internally and soft. So I put the grid back and let it continue to press on the mirror. 

I then went around the lap edges with a chisel to just tap off the overlap. 

Tomorrow I'll put it on the machine for a spin for a few hours and see how it goes. 

While that happens I'll get on with making a new part for the foucault with the source and the window closer together. 

 

[skybadger]

I re-poured the pitch lap once I saw that the backing disk for it was too tall- it had a concrete pad bonded to the  plywood, done at the same time I made the grinding lap. When I add the pitch, its just too tall and tips and wobbles across the machine. 

So I junked that and made cut some new lap backing disks from scrap plywood. I chose a 120mm diameter for the 150mm mirror , so I could do spin polishing ToT on the MoM machine I have 

In the process I also re-jigged the gearing for the pulleys driving both the eccentric and the mirror platter. The platter now runs at 60rpm and the eccentric at 8.. The overarm wasnt quite centered on the platter, so that is also now fixed. 

I'm 4.5 hours into polishing  using cerium and my stock of pitch. I recycled this pitch from a 16" lap I acquired, it seems to be very hard and resistant to the fingernail test, which might explain why its taken as long as 4.5 hours to get fully polished out to a sphere. Also, I'm still learning when to press and whether to do it hot or cold. I did it hot pressed this am and then cooled and brushed with a steel brush which seems to have made a decent difference both in degree of contact and extents of good contact - all the way out to the edge. 

I'm doing 15minute wets and rotating the mirror roughly 90 degrees every whet. 

I've waited for a good polish before putting it on the tester. The polish goes out to the edge but could bear some more according to the laser test. Since this is a practice piece, I'm pushing on. 

Here's a short vid of the mirror grinding machine: 

Shows a 16" platter, 12"step and then an 8" slotted disk with mirror stops. The lap is 3/4" plywood, varnished in 10% diluted varnish for penetration and counterbored for a rotating foot to drive. 

The overlap below is 25mm on a 120mm diameter. 

The mirror under test is suspended in a sling mounted on a frame on my garage wall and retained using an aluminium catch plate at top. . 

Heres the ronchi from inside ROC:

At ROC 

outside ROC

Outcomes: 

Based on the hooks at the end of the ronchi lines inside ROC t looks like I have turned down edge. That seems unexpected for a MoM machine. If anything I was expecting TUE. 

You can also see a zone at the centre , about R/4 . Closer inspection reveals a raised zone,  a dip ands then another raised zone very close to the middle. Assuming Im reading this right, they could be dips. Reading the MoM operations guide, I expect some zoning towards the centre. I'd like to confirm whether its a raised zone or not. 

I rotated the mirror through 90 degrees and inspected it again - I didnt see anything different . Maybe I wouldnt. 

I think this  looks like a smooth surface  - bar the zones - 

Looking for confirmation and  feedback of reading. 

I also used the foucault test to view and with the knife from the left, the highlight was to the left  - is that raised or dipped ? 

Her's a close up of the central zone. 

 

[davidc135]

I agree that there is a tde and the overall central quarter looks a hill and focusing long. if it's possible to Foucault test with a knife edge instead of a Ronchi grating all the different slopes of zones are easier to identify, at least it is for me.

david

[skybadger]

I should be able to photo the foucault , I'll give it a go. 

[skybadger]

Here's a foucault of the mirror from this am, so its had all night to acclimatise.  

At RoC, knife from right , appearing from left ( so a touch inside RoC ?  ) . 

I think this shows : 

- rolled edge at 90% zone

- some shallow concentric zones from the spin polishing that isnt ripple or dog biscuit. 

- a raised central hump but with a surrounding trough. 

 

What I'd like to get is input on the relative scales. In order to get a view on whether this is worth continuing to polish ( vs return to grind) I need a view on how large each component is. 

Can I treat the turned edge as partial parabolisation and excavate the centre using CoC strokes to smooth the zones out from there to the edge ? 

Is the reason the whole mirror isn't nulled due to not quite being at at Radius of Curvature ? Still feeling my way to finding that spot. Currently using the change in direction of the shadow ingress as the indicator of moving through the centre of curvature. Is there a better way  or do I just need to practice more ? 

[Astrobits]

Firstly, vary the overlap. Fixed positioning leads to zones.

Secondly, don't rotate the mirror by ( nearly ) exact and equal amounts. Fixed angle only turns could lead to astigmatism. Rotate a little bit this time and a lot more next time then somewhere in-between the third time. Don't repeat the sequence.

Cold press often. If your lap is hard enough you can leave it on the mirror for days between polishing sessions. Just make sure it is well wetted with slurry and put it into a polythene bag to prevent it drying out.

The more random you are the better.

I once had triangular astigmatism because I thought that bubble wrap was random and I didn't need to rotate the mirror, until I looked at it afterwards and it has triangular patterning.

Nigel

Edited April 2, 2023 by Astrobits

[skybadger]

Understood re the rotation, that's where I am at right now. 

The same for the spin polish, adding some eccentric to blend the zones. 

Spent an extra hour this afternoon with a large eccentric to blend right across the face. Have some more pics to load. Seems to have reduced the hump and pushed out the tde. 

I'll have to see how much effort is required before considering a regrind to sphere

 

[davidc135]

6 hours ago, skybadger said:

Here's a foucault of the mirror from this am, so its had all night to acclimatise.  

At RoC, knife from right , appearing from left ( so a touch inside RoC ?  ) . 

I think this shows : 

- rolled edge at 90% zone

- some shallow concentric zones from the spin polishing that isnt ripple or dog biscuit. 

- a raised central hump but with a surrounding trough. 

 

What I'd like to get is input on the relative scales. In order to get a view on whether this is worth continuing to polish ( vs return to grind) I need a view on how large each component is. 

Can I treat the turned edge as partial parabolisation and excavate the centre using CoC strokes to smooth the zones out from there to the edge ? 

Is the reason the whole mirror isn't nulled due to not quite being at at Radius of Curvature ? Still feeling my way to finding that spot. Currently using the change in direction of the shadow ingress as the indicator of moving through the centre of curvature. Is there a better way  or do I just need to practice more ? 

With k.e coming in from the right you are a little outside the RoC of the main area of the mirror but obviously inside the RoC of the rolled edge and the sides of the small central plateau. I think that you have to reduce the turned edge somewhat before it could be absorbed into a paraboloid. Hopefully astrobits' advice will also smooth out the many shallow zones.

David

The central hill needs at least to be shrunk before it can be hidden behind the secondary

[dark star]

I am working on a 20 inch f/4.1 mirror.  I have the same problem of a turned down edge after spin polishing on a fixed post machine. Using a 15 inch pitch lap. I tried varying the offset of the lap but it did not seem to make much difference. I read that the offset should be around 15%. My outside ROC Ronchi test looked quite similar to yours.

I am currently working on the center zones as they are flatter than the middle and outer zones, using a 10 inch pitch lap and a centre through center stroke, not going over the edge. But I need to fix the TDE. I am now working by hand with the turntable rotating at around 6 rpm.

I will be interested in what stroke you use to get rid of the TDE.

David

[Astrobits]

If you are trying to flatten a nigh zone then the centre of the lap should, on average, be over the peak of the high zone. Use quite small laps for narrow zones and use circular motion of the lap rather than straight motion as the mirror turns as this will help blend with the lower areas either side of the high zone. Don't forget to check progress as often as possible. When getting near the end I would check the figure after just one turn of the mirror with a little overlap of the start and end points and restart at a different position for another wet.

Nigel

[skybadger]

This is the first photo of the last session - taken after 15 mins of polishing using a 120mm diameter lap set to swing across the centre from edge to edge so that the 80% zone of the lap sits on the starting zone of the TDE. As far as I can see I have smoothed out the central peak and trough to some degree.  Platter was 60rpm and eccentric 8rpm. 

I used a 1.5kg weight on top and movement was smooth. I had also cold pressed it for 10 mins with a 5Kg weight before starting while I did other stuff. 

 

This picture is after the second session, continuing the same settings but adding more weight - now at 2+1.5=3.5Kg. I think the smoothness is suffering due to the weight - I need  to reduce it. Other parameters remain the same. 

Outcome:

Hill is slightly smaller and trough better blended. This seems a slow process which probably indicates hill is quite large. No apparent change to starting zone of TDE . Surface now rougher than before. 

Plan: 

Continue and observe

This is after two more 15 minute wets later with the same settings as before.

I didnt leave the blank much time to cool down but did give it a good wash in cold water. 

Outcome:

I may have deepened the trough while trying to remove the central hump. I can't see the hump reducing and the surface remains rougher than when starting the session. 

Plan:

Finished this session . Re-measure after back from travels and cold press. 

Remove the central hump using a 50% (80mm) lap operating across just the centre hump ( i.e. 10 mm eccentric, 0mm offset from centre) and then smooth out using the 80% lap with wider eccentric.

Check for hump reduction then take curve out to edge using tangential strokes on the eccentric, increasing offset, starting at 10mm  going to 35mm with a 80 mm lap.

If the hump doesn't show reduction, re-grind to spherical 'cos polishing ain't enough. 

 

 

Edited April 6, 2023 by skybadger
spelling

[davidc135]

I don't see the point in regrinding as, after the mirror is repolished, you must eventually face the same questions of technique. Do you need to persist with the machine? Although I can see the attraction of mastering it.

The figure may look a mess but it's probably not miles away if the right laps, strokes, decisions etc are made.

David

[skybadger]

For me going to regrinding is a practical act after learning what happens with the initial lap too small. 

If the hump can't be polished down because it's too large, grinding is the only option.

The next step after grinding would be to polish with the 80% lap and see how things differ there. 

I do want to master the mom.. 

[skybadger]

I spent another 30 min session trying to reduce the central hump yesterday. 

I say its a hump because the ronchi looks like this , outside of ROC. The kink to the outside in the central zone , outside of ROC, indicates the hump ( by reference to this: https://nicholoptical.files.wordpress.com/2016/11/the-ronchi-test.pdf ). The ronchiimage  isnt very clean because the one I have to hand is 250LPI and I need to find my other, more coarse one (~100lpi) to remove all the edge diffraction effects. 

Two wets of 15mins each with zero offset and 20mm length strokes across the centre

Knife edge is from right. 

I'm having a hard time recognising when a zone is nulled. Its either bright. dark or gray/nulled as I understand it. 

If I was using a coude mask - the 50% zone would be grayed in the picture below ?

So I think the central hump is reducing and blending - ill give it a final hour of polishing to try to confirm and reduce out and then blend with the larger lap before making a final decision on going back to grind. 

I also spent yesterday re-making the foucault measurement device. Its still tripod mounted since I haven't got a bench for this. 

The radial movement is now on a micrometer driven slide giving +- 7.5mm at 0.01mm resolution, the tangential direction on a photographic rack mount. 

If its still too sensitive to movements I'll mount it on a XY table I have for the mill. The led I need to move closer to on-axis and so eventually I'll move that in front of the slit. 

 

I want to move to making measurements to graph the shape out objectively. I have FigureXP and sixtests. I'm also looking at FoucaultUnmasked. Any recommendations ?

Cheers

Mike

 

[Astrobits]

It strikes me that there is something basically wrong with your process. All the images appear to have a similar shape indicating a recurring pattern. This could be caused by the pitch lap, the strokes in use, the mirror support or some combination of those. Looking back at your image of the mirror on the MOM there does not appear to be any soft base ( i.e. carpet ) under the mirror.

Firstly, I would change the support for the mirror. Put some thick, soft  material under the mirror. Make sure that the material does not have any sort of symmetry ( carpet is a popular choice, don't use bubble wrap). I would then switch to hand polishing the high zones using a very small, 2" dia, lap to reduce those somewhat. Be aware that small polishers like this will change the shape very quickly so use sparingly. when the zones have reduced re-press the large lap and continue with variable strokes.

Don't forget: regularity = zones, randomness = perfection.

Nigel

 

[skybadger]

Thanks Nigel. 

There's a grey polycell mat under the mirror which is the material used for underlay for wooden floors, so moderately dense and very even. 

I've made a smaller 80mm lap to focus on the small zones and using direct strokes over the centre, moving out  increasing the offset and eccentric to bring the central zone down and spread it to the outer zones.

Then once I get positive change in the desired direction  I'll follow it up using the larger lap to smooth it out. 

For the MoM, the direction is to bring the platter speed down from 60 to 6 and bring the eccentric up front to 15 which moves from polishing to figuring configuration.

The place I'm at now is trying to get good quantitative measurements after each wet using the Foucault to confirm the qualitative interpretation of the surface shape. The micrometer stage is starting to give good readings and I'm using an Everest pin mask to identify the null zones reliably. 

In short, taking your input and adjusting accordingly. 

[skybadger]

I think I would summarise progress so far  as introducing and now trying to remove the central hump. The hump was introduced by a poor lap which had poor contact, was too tall and was too small in diameter, introducing the central hump and ditch.

That was addressed by creating a larger lap (120mm)   with much better contact, less tall so good drag on the mirror but still unsuited for removing the hump. 

So I made a new lap of 80mm, softened the pitch a tad and moved to focusing on reducing the hump which seems to be progressing slowly but positively.  

I'll cut out and prep a 2" lap and give that a go as well. 

As part of that process I've started capturing measurements and gaining practice in the Foucault method in order to make them repeatable and representative so I can track the figuring progress. My problem there seems to be capturing a reliable zone 0 measurement and making sure the mask is symmetric for repeatability. 

So I think I'm making progress slowly, even if its not showing well in the mirror yet, mostly because I'm snatching half hour sessions in the garage between other activities. 

cheers

Mike

[Astrobits]

Perhaps you might like to think about making a Dall Null tester. You will need a plano-convex lens about 25mm dia with a focal length around 150-200mm and a bit of engineering to make up the tester.

One of the drawbacks of the Foucault test on faster mirrors is that the light areas are very light and the dark areas are very dark making it difficult to see subtle zones. With a null tester the knife edge shadows appear as that of a sphere, i.e. the parabolic mirror will grey out much more gradually than with Foucault. Any high or low zones will stand out much more.  I have used my Dall tester for my mirrors up to 500mm F/4.

Personally, I would concentrate initially on the outer areas of the mirror as the central area will be partially covered by the shadow of the secondary and it contributes little to the image anyway.

Nigel

 

[skybadger]

20 hours ago, Astrobits said:

Perhaps you might like to think about making a Dall Null tester. You will need a plano-convex lens about 25mm dia with a focal length around 150-200mm and a bit of engineering to make up the tester.

One of the drawbacks of the Foucault test on faster mirrors is that the light areas are very light and the dark areas are very dark making it difficult to see subtle zones. With a null tester the knife edge shadows appear as that of a sphere, i.e. the parabolic mirror will grey out much more gradually than with Foucault. Any high or low zones will stand out much more.  I have used my Dall tester for my mirrors up to 500mm F/4.

Personally, I would concentrate initially on the outer areas of the mirror as the central area will be partially covered by the shadow of the secondary and it contributes little to the image anyway.

Nigel

 

Isn't one of the issues with the Dall null technique the necessity to very accurately place the lens at a certain distance from the foucault ? 

I'm beginning to regret not picking up the testing frame I was offered when I went to collect some glass and grits from a retiring ATM. I just didn't have the room for it then.  Nowadays I have room but its full of other stuff instead. I wonder if I could build one upside down in the roof of my garage ? Ie suspended from the ceiling. Feels diversionary.

 

 

[skybadger]

Further progress 

According to my grinding diary I took these after the last two wets of 5 mins which  are digging out the hill from the centre and attempting to smooth out the trench before moving on to take the curve out to the edge. 

The pin mask lies across 7 zones and shows the trench being on the inner edge of zone 2 

I used the small lap across the centre to reduce the hill and added some offset to blend that outwards. 

I also took some actual measurements but not all make sense - I clearly wasn't reading the micrometer correctly. more practice required in the dark garage. 

Here's the table data from Figure XP

First Q: the difference in range between measured zone focus distance and expected is real. Why is it so different ? I haven't polished away all those mm since I started near spherical. I set the zone settings to moving slit. I will double check with a second micrometer to validate. 

Regarding the Dall null test- I read up on the test on optics.net but I think I have enough contrast to see the shadows as I currently am. I am aiming to being the line of shadow bisection to the centre and measure the null when its on a zone marker. for example. the null in the right hand picture is between zones 1 and 2. I am getting more familiar now with what I am looking for!

Second Q: Considering Zambuto describes the most effective polishing 'sweet spot' for the lap as being where the lap has greatest effect at 80% diameter of the lap and then uses offsets to target  the zone requiring work,  I should be offsetting the lap to make the sweet spot coincide with the zone I want to reduce  - which means for an 80 mm lap , using a -30mm offset for the central hill and a +15mm offset for the zone 3 reading. Sound about right ? 

I did some more wets last night . the first 4 for small lap fixup and the second for blending across the mirror. 

Cold pressed for 5 mins with 5Kg	Time	Offset	Eccentric	(Overhang)	Notes
80mm lap	5mins	0	25	0	forgot to rotate blank
40 mins in total	5mins	15	20	0	forgot to rotate blank
	5mins	30	10		forgot to rotate blank
	5mins	45	5		rotated blank
Moved to 120mm lap - warmed for a warm press before use. channels not closing up yet. Probably not warmed enough yet.	5mins	5	20	~10mm at extremes	rotated blank
	5mins	15	15	~10mm at extremes	rotated blank
	5mins	25	15	~10mm at all times	rotated blank
	5mins	30	10	~10mm at all times	rotated blank

I hope to measure the outcome of this, this afternoon and I'll show those measurements here.. 

 

[dark star]

What pins did you use and how did you work out the position of the pins? I would like to try an Everest pin mask on my 20 inch. I have always had problems seeing the shadows with a Coude mask.

 

David

[skybadger]

The pin locations are from both the stellaphane web calculator and the calculator in figureXp. 

I marked a sheet of polystyrene plastic using a compass  and used a scalpel to cut the fingers out by allowing some materialaround each mark. Means a new mask per mirror but styrene sheet or just plain card is cheap. 

Edited April 19, 2023 by skybadger
Expansion

[Astrobits]

I hope that I am reading your post correctly. I think that Zambuto is referring to the "sweet spot" of a lap as being in the area either side of the fore/aft stroke direction encompassed by the central diagonal and either side within 80% of the diameter.

The maximum effect of a lap occurs along the fore/aft diagonal, reducing effectiveness towards the sides. If you are offsetting the 80mm lap by 30mm then that puts the maximum effect at the edge of the central hill. This would do nothing to reduce the central hill and will probably make it worse by targeting the low zone around that central hill. You need to have the centre of the lap oscillating about the centre of the zone you wish to reduce. This is one area where simple machinery fails to give us the best results as it does not vary the position of the lap enough to prevent the lap effectively digging a hole for itself.

As I said previously, I would concentrate on the outer zone first. Using the smaller lap by hand in a circular stroke, no more than about 1/2-1" in dia, over the high zone as the MOM rotates the mirror, making a flower petal pattern in the polishing medium. At present you are still trying to get to an optical sphere before going for parabolising. In parabolising you will be attacking the centre anyway and hopefully doing less and less work as you get towards the edge of the mirror.

It is a fallacy that the surface after fine grinding is a sphere to optical standards. I once made a grinding tool that used rather large pieces of glass and it so happened that there were more gaps in a particular zone than elsewhere. Despite the mirror appearing to be a smooth surface, a sphere it was not. After polishing there was a large high zone at the location of the gaps in the grinding tool. Lesson learnt.

Nigel

[skybadger]

Always good input. thanks Nigel. 

That was why I was asking. I plan on making a 50mm lap tonight to do that job. 

The latest set of figures after polishing in the above method are below. The surface error plot tells me I have a conic of about -3! Which might explain the difference in sizes of figures betwween ideal and measured. 

Overall Im happy that I am getting a better understanding of what is going on, getting better at manipulating laps and the test tools and starting to get some repeatability in my foucault measurements.

 

 

The Foucault sequence is shown below moving from zone 0 to zone 7 outwards. : 

zones 1, 3, 5, 6, (8)

 

And finally a ronchi just outside RoC

 

 

Edited April 20, 2023 by skybadger
expansion

[fwm891]

I was never really happy using the centre as my zero position and tended to use the 70% zone as my zero with + and - figures either side. It was always easy to position a pair of visible shadows and take that reading as zero or with a digital scal easier still…