Ed's Blag

We 3d print a lot of ABS. I have had some bad experiences with ebay sellers who provide poor quality filament.

Do not buy printer filament from: lillyvale4u.co.uk, ebay sellers “lillyvale”, “lillyvale2″, or any other seller found to link back to a company “dotcomliving”. If you find you have purchased ABS filament from these sellers, request to cancel the order right away.

We regularly print the same job which runs for several hours. Using our preferred “EFORINK” black ABS filament we have no issues with bed adhesion, and it prints self-release from the bed at about 45°C during cooling. We coat the glass heated bed with hairspray and this provides a great balance between good adhesion at bed operating temperature and easy release when finished.

Our main issue with lillyvale4u.co.uk/lillyvale2/dotcomliving filament is a complete lack of bed adhesion. Everything seems fine for the first hour or more, but the print will detach resulting in printing spaghetti at best, a complete machine crash at worst. A secondary issue we have noticed is a lot of popping during printing, indicating water/bubbles present in the filament.

I have been operating an Anet A8 3d printer for some time and have made some major upgrades to it along the way. This blog post serves as a place to document my trials and successes with this very low cost printer.

We print a lot of parts for a customer and have been getting good results but have also been spending increasing amounts of time on maintenance and replacing worn out parts.

I printed a lot of parts to replace the worst bits of the acrylic frame, add some decent belt tensioning parts and the obligatory Z height adjustment screw. In my first week of printer run time I upgraded belt grips etc because I couldn’t get sufficient tension to stop ringing in the belts and poor print accuracy.

I bought borosilicate glass for the bed and use hairspray for bed adhesion control and easy release of the prints.

The linear rails supplied in the kit were chrome plated rods, which quickly wore out and revealed the steel under the chrome plating. The bearings ate a lot of the chrome and were ruined. I upgraded to stainless steel precision ground rods which cost nearly half the amount of the original purchase of the printer. I also replaced all the bearings at the same time.

Having the printer sitting on top of other gear I own, due to lack of table space was probably the biggest mistake I made initially. This meant that the printer wasn’t on a totally flat/level/solid surface and was noisy and difficult to square up.

I fixed this by lapping three paving slabs together until they were all making good contact with a straight edge along it’s length. These are definitely not as high quality as granite reference surfaces, but I managed to achieve flatness I was happy with at a low cost. A flat heavy surface in itself is an enormous upgrade for quiet operation, but the possibilities it provides are worth more.

I can now square the whole machine using engineers squares and other measurement/test tools that need a flat surface to sit on. I can compare heights in different locations because the surface is very flat and the equipment can be placed repeatedly with hardly any error.

I got a cheap dial test indicator from eBay, with a magnetic base and positioning arm. I ground the bottom of a heavy piece of steel flat enough to sit steadily on the surface of the slab. This serves as a base to attach the magnetic indicator stand to, and can be moved as needed on the slab.

Levelling the bed is a process that uses the indicator, and it results in the bed being level to a very close tolerance, relative to the surface the printer is sitting on. I adjust the Z height separately using the corner of the bed and paper for checking the gap between the bed and nozzle.

Because the bed is set up so flat, I know that the levelling is good before starting any Z height adjustment and can avoid the long task of checking each corner against the nozzle.

Levelling the X axis (with the two z motors disabled) is a similar process, the axis is levelled by comparing the height of the two leadscrew nuts, relative to the flat surface of the slab.

The levelling process described in most 3d printer setup guides is painfully slow compared to my method. Typically the methods described require multiple iterations of enabling steppers, homing, disabling, carefully moving the machine by hand one axis at a time to check the bed-nozzle distance at each corner of the bed. This is error prone and tedious. I now do a similar process to check but I never disable the motors and I directly issue gcode commands to move the machine to a suitable location to check the bed-nozzle gap.

I now get perfect bed adhesion and the prints are easy to release and have a consistent thickness of the brim. To avoid upsetting the relatively bendy printer frame, I remove the glass bed before pulling prints off.

Around the time I was getting really happy with the results, the X axis motor completely locked up during a print and ruined the print and messed up the squareness of the machine. I found the glass bed on the floor and completely lost two of the bulldog clips that secure it. They probably landed in a box of components and are still yet to be discovered.

I replaced the motor and carried on printing. About three prints later, there was a slight issue with X axis losing a few steps and the offset ruined the print. It probably ran into a bump in the print while making a rapid move, the small offset was enough to spoil a visible part of the print.

I decided closed loop X and Y axis control was now a must for printing this volume of parts. I ordered two of the MKS Servo42 hybrid servo-stepper motors from eBay. Investigating how I’d fit these to the existing electronics of the printer, I spotted that RAMPS boards have the stepper drivers as modules on sockets and the Servo42 units are supplied with a breakout board from this module socket. This means that with a RAMPS main board I can plug in the new Servo42 motors directly where the stepper modules would fit, and for Z and extruder drive I can use the normal stepper driver modules.

Looking around at what was available, I found an ARM based board running smoothieware, compatible with the arduino mega pinout. This means at the same time I can upgrade from the Atmega 16 bit CPU to an ARM cpu running at 100MHz.

I should be able to get higher step rates due to the faster processor, which in turn means I could use more microstepping and still reach higher speeds than before.

I recently needed to remove the black potting compound covering an electronic module to repair it. This is usually a no-go for repair jobs but this was a special 'black box' someone had made and we needed to repair or replace. With it not functioning, it was necessary to look at it to determine what to do.

Removing potting from PCBs is difficult at the best of times but we have a technique that seems to work well for most things.

Do this in a very well ventilated area or better yet outside. You need a hot air gun and scraping implements.

Initially, gently heat the potted device with the hot air gun and wait for the potting compound (usually an epoxy resin) to soften. Poking at the potting compound with something pointy will show where it's soft.

Begin by removing any soft areas that appear, usually any exposed corners or bumps will soften first. A small flat screwdriver is helpful, but be very cautious not to dig in too deep and scrape off PCB components.

When small lumps are easy to remove, take the heat gun away and try to remove small pieces until it starts to harden again. Repeat heating and scraping. Wear gloves and try not to get everything too hot.

When the potting compound is hot enough to be pliable, it will often peel away from flat areas of PCB without much trouble. Repeatedly heating and picking away the remaining areas will expose the whole device. If through hole components have been used, these will pose most difficulty as the resin will usually flow right around the parts. Heat more persistently to deal with these. The epoxy needs to really soften to be possible to remove.

I am working on an ongoing project with a friend building an arcade machine with multiple control panels and a lot of LED lighting.

The control panels have a number of acrylic edge lighting parts recessed into the panel. We found out the hard way that it’s hard to get enough light into the acrylic to look bright enough next to other more directly illuminated controls.

I thought we could make light pipes from casting acrylic, and 3d printed some mould formers with angled front faces.

We made a silicone rubber mould and then cast test light pipes, each one with an LED poked into the casting acrylic after filling the mould. A couple of the test castings had voids under the LED at one side, so I will make some more to have a complete test set of these.

The tests were initially difficult to compare the light output pattern as light was lost from the side of the slightly rough castings, but painting them on all the surfaces except the exit face improved the pattern.

I hope these will let us make brighter looking edge lit controls.

The difference between the exit face angles is quite apparent holding the light pipe touching the back of a sheet of paper. In the photos below the LED is operating at the same current, aimed close to vertically upwards. The light is more and more spread to one side as the face angle increases, as I had hoped. This should allow us to pick the design with the best light coupling to the acrylic for the edge lighting.

The tests we did before showed almost no light from a single LED, so this is a massive improvement. Hopefully with these secured in place in all the led mounting holes, it will give much brighter edge lighting than we were able to get before.

After testing this with such good results, we made more formers with the final designs of light pipes, in arrays to make it fast to cast the right number of the final light pipes we need.

One of the light pipe designs is going to be tricky to get out of the mould. The light pipe has a 4*4mm entry for the LED, while the exit face is 15*1mm meaning the approach used for the other designs may not work. We’re trying it anyway hoping we can wrestle former and eventually the lightpipes out of the mould without destroying anything. Should be interesting!

We live in a very hard water area. Our kettle and coffee machine both require regular descaling.

After our last coffee machine died a horrible death (blocked by scale particles shortly after “descaling”) we decided we had to do better than dilute vinegar as our descaler.

I took apart the coffee machine completely when it became blocked. Scale particles had been freed from the inner surfaces but had not sufficiently dissolved to leave the machine. This left me with a pile of machine parts, a pile of limescale and no coffee.

Even boiling the affected parts in dilute vinegar did not rescue it, the valve is still blocked and the coffee machine is dead. Several other issues made this bad enough to replace the machine instead of persisting.

Replacement coffee machine ordered, now I’m determined to stop this happening again.

I tried to dissolve small samples of the limescale in vinegar and citric acid. The citric acid was clearly far better.

I tested additives such as steriliser and dish soap. The steriliser has no observable effects, which is ideal as it’s there just to ensure there’s nothing growing inside the machine. The dish soap very effectively suspended particles in the solution, but also produced too much film that was difficult to rinse off the container. This is obviously not desirable inside a coffee machine, but the suspending effect of the surfactants was very desirable.

The final result of my testing is this descaler which works well in hard water areas:

1l water

10g citric acid

5ml Steriliser (1 tablet per 100ml water concentration)

It is even possible to see how much scale the descaler removed from the machine, if you catch the used descaler. Gradually adding small amounts of a solution of sodium hydroxide will neutralise the citric acid and precipitate out solid particles of limescale. This is what was removed from the machine by the descaler!

All the components of this descaler are available on ebay or from a supermarket. I used Asda sterilising tablets.

To get started with copper plating I bought a kit from classic plating as my initial research showed lots of people getting similarly poor results with DIY methods. Lots of polishing would be needed to get the bright finish needed for a PCB.

Copper sulphate alone is a poor plating electrolyte. Sulphuric acid is needed to increase the "throwing power" of the electrolyte. Sulphuric acid can be produced by the electrolysis of copper sulphate solution, but the common method of doing this involves a graphite electrode that contaminates the electrolyte and requires careful filtering to prevent poor results. Alternatively, a platinum electrode can be used in place of the graphite electrode.

Additionally, plating additives seem to be closely guarded secrets.

Anecdotal evidence in forum posts suggest glycerol and urea or thiourea as additives to help level or brighten plating. Thiourea accumulates in the thyroid gland and may be a cancer risk. Yikes.

Surfactants also appear to play a role in the additives.

Chloride ions are added too.

Because of all these variables and my need for quick and high quality results plating a small quantity of PCBs, opting for a commercial plating kit was the most sensible option.

From the outset my test results were better than those seen in YouTube videos of people plating with DIY electrolyte.

I quickly discovered I needed anode bags (j-cloth is great) to stop sludge from the anodes making too much work filtering the solution. I was not using the supplied anode plates, instead I was using smaller bundles of copper wire.

The cleaning and surface preparation solutions provided in the kit quickly made the PCB copper ready for plating. I had not seen a proper "water break test" pass until I used the alkali degreaser and acid pickle for the first time.

For PCB copper which is already quite clean, it is not necessary to use full strength cleaner or acid pickle. 5 minutes in each at 50% the recommended concentration was perfect for making test PCBs.

I manually agitated the small test tank every few minutes.

By testing I determined that I could produce a desired copper thickness in a predictable duration, and that it was level enough for my testing needs.

A small amount of sanding was needed to level off the edges of the test pieces but they quickly measured as level after the raised perimeter of the board was sanded flat.

Perfect for testing thick copper PCB layouts without having to order special test PCBs, or resorting to expensive research laboratory testing.

For an electronic device that I'm developing, I needed to determine temperature rise information that I had trouble finding any trustworthy reference for theoretical calculations to be fully trusted.

My dilemma is that we have high current traces with very thick copper on the PCB, but the connections between the traces and the components are the narrowest parts of the circuit.

Conduction and radiation of heat, heating caused by the resistance of the conductors, heating from components etc all play a part in the resulting temperature rise, as such it's a difficult thermal systen to model with confidence.

To get good quality results and be confident in design choices, I decided to carry out real temperature rise testing.

Ignoring test models completely, the test environment is also just as important, as the data requires being able to make a confident measurement of the temperature rise above ambient temperature.

I looked on eBay for a small thermal test chamber and soon found a small Electrothermal test chamber.

Photos of the ebay listing, this is how it was when I got it.

There was no temperature controller in the eBay listing, just a test chamber with some damaged connectors. I decided that the chamber was large enough for the tests I needed so I bought it and also ordered a cheap PID temperature controller, REX C100.

The temperature controller had an annoying mechanical relay as the main output control so I removed the relay and bridged the relay coil drive connections to two of the existing relay output terminals as this is perfect for driving an SSR input. I wired up an SSR to control the chamber heater.

Fitting a thermocouple to the outside surface of the inner wall of the test chamber with kapton tape and heatsink paste gave me the required temperature sensing for the controller.

The chamber has some damage to the terminals used to connect to the device(s) under test inside.

I used 7*0.2mm stranded PTFE insulated cable to replace the 12 low current test connections that run from inside to outside the chamber, as it was the only cable thin enough cable to fit along existing cable routes. Rubber grommets which protected the existing cables were brittle and cracking so I replaced them. The connectors were damaged, missing some screws and corroded so I also replaced these with 12 way phenolic barrier terminals.

I don't expect to operate this test chamber at very high temperatures or with very high test voltages so I decided that tri-rated cable would be suitable for the additional connections I need for high current tests. Tri-rated cable has an operating temperature of 90°C. I added 8 ways of 4mm² cable and another pair of phenolic barrier terminals. These run through a hole in the rear wall of the test chamber with a rubber grommet to protect them from the metal edges.

I also added a couple of k type thermocouple cables into the test chamber, and ordered some plugs and sockets to enable temperature measurements from devices under test.

There is a hole through the top of the test chamber which I assume was for a thermometer, but it is large enough for several thermocouple cables to be fed through allowing extra data logger channels to be wired up. This is less convenient than plugging in thermocouples, as it anchors the thermocouple to the device under test and the test chamber, needing it to be disassembled before the device under test can be properly removed.

Tuning the controller to operate the test chamber took a couple of attempts to get the PID autotuning to work, I had to increase the temperature set point to 50°C before it would successfully complete the autotuning, my previous attempts did not have enough difference between ambient and setpoint temperatures for it to work. I found there are too many variants of these temperature controllers which are similar but not identical, leading to confusion about manuals and menus.

I’m happy with how this turned out, it is a useful addition to the lab gear. I expect to tidy up the rear cable mess at some point, but for now, it has done a great job and I have the test data I required. Tidying can wait ;)

I will make a rear connector panel to replace the original, but for now I have added layers of masking tape inside and outside the rear panel to keep the insulation in place and allow a cable exit.

The three terminal screw block is the heater L and N, and the case ground.

The tubes are there to support a test piece away from the base of the test chamber.

In an earlier post I mentioned making sulphuric acid by electrolysis of copper sulphate solution. This results in copper deposited on an electrode which is poorly formed and has a large surface area which will quickly oxidise. To tidy up from this process, I attempted to plate the copper from one electrode to another, this time in a copper electroplating electrolyte with suitable additives for bright plating.

Because my first attempts at making sulphuric acid used carbon (graphite) rods as the anode, a lot of graphite bits became included in the copper deposits. Re-depositing the copper on a fresh electrode with an anode bag to retain the solid impurities should make a much more pure deposit.

With a couple of repetitions, swapping the electrodes when most of the copper has been plated onto the cathode, the purity will quickly approach 100%.

This seems to be a useful waste recovery step from the production of sulphuric acid from copper sulphate.

I am experimenting with copper plating and as such I need sulphuric acid. Since this is somewhat difficult to buy on it’s own in the UK, other options are worth exploring.

Copper Sulphate is readily available and a solution of copper sulphate in distilled water can be electrolysed, depositing the copper on the cathode and producing sulphuric acid.

The copper sulphate is split into copper and negative sulphate ions at the cathode. The copper is deposited. At the anode the sulphate ions lose an electron, releasing bubbles of oxygen gas and sulphuric acid.

The cathode of choice is not too important, copper is a good choice. The anode must be carbon or platinum. Carbon anodes will erode fast in this application if a high current is used, resulting in lots of bubbling. A platinum electrode is a far better choice for the anode as it is much more resistant to damage by the oxygen bubbles and the acid.

The solution should be electrolysed until the blue colour disappears and the copper anode starts to produce hydrogen bubbles, this signifies that the process is completed, now instead electrolysing the water to hydrogen and oxygen.

My initial tests with graphite electrodes were successful, giving a usable sulphuric acid solution with pH of around 1-2. It required a lot of filtering to remove the particles of graphite. I am not sure of the concentration of the results, but the pH is encouraging and shows it’s certainly a useful process.

This photo shows how badly the carbon electrode is eroded in the process.

I have now ordered a titanium mesh electrode plated with platinum for my next testing. I hope this electrode will give a better result, faster, by allowing higher current operation without the problems of the disintegrating graphite rods. When this electrode arrives I will post an update.

Update: platinum plated electrode has arrived.

I continued as before with the same solution from my first large scale test. The graphite electrode crumbled so I filtered everything carefully through cotton wool in a funnel.

At first I found the solution went quite yellow, then green. I was surprised at this and reduced the voltage from about 3v to 1.2v. Even the platinum electrode eroded at 3v. I switched to the spare, which I was glad I ordered but surprised to be using it so soon.

At 1.2v the process is much slower, depositing smaller harder copper lumps on the anode, but the blue colour is gradually lightening and all the yellow seems to have disappeared.

I needed a low cost Constant Current power supply able to operate at 0-2V at up to 10A from a supply of nominally 12v. Searching eBay I found several different power rating versions of a similar design that looked very suitable.

The eBay listing I bought is titled “DC-DC DC Buck Converter Module CC CV Power 7-32V to 0.8-28V 12A 300W Step-down” ebay item number 223048486036. Many other sellers are also selling this range of modules.

My main needs are an adjustable constant current mode, and the ability to set the maximum constant voltage the PSU will output.

Because this module is a buck regulator it is very efficient compared to a linear regulator.

Looking at the module I noticed that it is a sensible design making use of a few opamps controlling the feedback pin of the buck regulator IC.

While I haven’t done any great analysis of the power supply, I am happy with the constant current regulation and it’s meeting my needs so far. I haven’t found any sort of drop out towards zero voltage/current, the adjustment range is smooth all the way to zero in both CV and CC operation.

I needed a low cost meter displaying voltage and current. Ebay is full of small panel meters which looked suitable.

I chose the 100V 10A meter DSN-VC288 which was suitable for my needs.

I was pleased to see the meters have a calibration trimmer to adjust the displayed voltage and current value. When I set the first meter up, at low voltage and current (1V, 1A) I noticed the displayed current value was about 25% too high comparing to my DC clamp meter. Adjusting the trimmer, I ran out of travel before reaching the correct value.

Examining the PCB, I found the current shunt resistor is made of what looks like 1mm² copper wire, I realised it was possible to trim the value of this shunt by adding solder.

To correct the displayed current, I set up a simple calibration circuit. I set the calibration trimmer to the middle of the range. Using a constant current supply, the panel meter and a load resistor, I set the supply to 1A constant current. I confirmed this 1A with my DC clamp meter. I gradually added solder to the shunt, repeating testing each time until the displayed value was close to the 1A shown on my clamp meter. Leaving the power supply on and allowing it to thermally stabilise, I then adjusted the calibration trimmer to show the correct value.

Increasing the power supply settings to 2A constant current, I was pleased the displayed value was also correct.