http://ramirovm.eu/en-devices

A very old motor

Posted on 2010/12/20/Monday by Ramiro

This engine came with my lathe, It was used to move a grinder.
This is a unmodified- 100 year old engine. so it uses chatterton as insulator.
When I first sow it was really dirty. Then I cleaned it with oil and with petrol. Then I took thise photos:

Then I didn’t know what had I in may hands then I started cleaning until I could imagine it.

I used an old bucket, oil and petrol Na(OH) and soap.
Then I lastly could see that the device was made by Siemens, had a few readable inscriptions on its plate. Then I saw that its windings were not varnished but insulated with chatterton. then I avoided everything that could damage this product. It consist on a mixture of pitch ( a sort of vegetal rosin ) rosin (Kolophonia rosin) and a few more products.
The most difficult part was cleaning the case. Lastly, when i could see the cast iron I covered it with turpentine oil and then painted it with an anti rust enamel paint. Lastly I covered it with normal green enamel paint

The main problem during the mending was that i could not remove the end bearing from the shaft. I have to ask for help.
I thought I needed a bearing extractor but at the end i could not do it. I asked a friend of a friend for help, just in case he had a bigger extractor, but, it still was impossible. Then we cut the bearing and measured the shaft. It had at its end a little tapering. Then it was obvious why I couldn’t remove the bearing. We made a cylinder from it in his the lathe. Then I revised the whole device. I revised a very threaded hole with a tip then I bought new bolts made of AISI 316. There are two left. I could not change them because the last shop which sold them, does not any more. So it is almost impossible to buy AISI316 or brass bolts in the city of Madrid unless you buy a whole box (with 100 bolts). This shortage is not desirable not normal in a main city nor in a 4x 10^6 inhabitants supposed metropolis and cultural focus but is true. Unluckly true.
Then I decided to wait until I could make the bolts or buy them from a shop (maybe an online one or similar).

The connection panel, also made of chatterton was broken. I then machined again (with a die) every threaded part (bolts and nuts) I added new nuts and glued the broken panel it with a mixture of rosin and shellac, then every bolt was machined again (with a die) and nuts were revised in the same way.
New terminals were added to replace the bare wire, so now wires are soldered instead of just rolled.
I got this result:

Posted in Uncategorized | Leave a comment

conmutation 2

Posted on 2010/11/17/Wednesday by Ramiro

Here we are going to build a part of what we saw in the previous post, a conmutation device and some of its filters .

Taken into acount that we will not exceed 500V, we will use mosfets.
For lower intensities and higer voltages it would had been better to use IGBT for example 1500V.

We could choose a MOSFET suited for a tram o a tube train (in madrid 600V in the older parts, I don’t know the new ones) but it wuld be difficult to find and its behaviour would be exactly the same. but for a much expensive price . Almost everery thing else is almost the same.
The choosen one is IRF 540. because it is easy to find and cheap.

This MOSFETs has been mounted in a copper plate. This cooper plate is intended to act as a a holder as well as a heat radiator in case the device can get hot. Its machinig process was the following:
First of all a line in the middle of the plate and mosfet data were engraved (by hand, with a burin). The line is intended to serve as a reference in further processes.

cipper plates being drilled

holes were drilled but there were a little mistake. I took measuremets from the oposite side, then the terminal is too close to the mosfet. this is not a big problem althoug will require some care in further maching (special terminals …).

Plates were sanded and nearly mirror polished:

polished plates

And some thermal fluid was placed betwin mosfets and plates:

MOSFET about being placed in the plate

And then everything was hand adjusted.

Filters were bulid in two parts. the first one in a PCB that will be placed near the conmutation device and the second one will be point to point build in a piece of wood.it is supposed to be reconected in deppending on the used motor.

The point to ponit part has been build in a prismatic piece of wood. its measures were 80 -in the fiber direction- and 20×20 in the basis.

Temporary some devices were temporary palced with silicon.

Later devices were placed using brass bolts for resistors and brass screws for lampholders

The other filter we spoke about will be discussed in a further post. this filter is usud only to avoid catastrophic faliure (mosfet destruction, personal injuries…). Optimal filters will be looked for later. this is intended only to protect people and the device itself.

Posted in motor control | Tagged 3-phase devices, asynchonous machines, introduction, motor control, pwm | 1 Comment

The board

Posted on 2010/11/16/Tuesday by Ramiro

As we could see in the general introduction, the circuit is mounted over a shell lach varnished laminated wood. The soportes will be made in wood. it is not n a perfect isolator, it can not cope very agressive environments but con be easily worked.

(Since it will be placed (at least for a long time ) in a house, will allow a culturally acettable finishing).

It is olso esaier to find, and cheaper than other feasible options i could choose for the point to pint assembly

Once varnished the board looked like::

Tdhe devices, (mosfets diodes…) werw placed on 3 x7 wood pieces. those pices were foxed throw 6mm dowels.
The stands for vitrifiead ceramics resistor and lamp holders (other resitor) were made of 20mm square section wood. They were fixed to the board throw brass screws. tohse pieces were drilled and trteaded with am M4 coarse screw thread.

on those holes M4 brass threaded rod was placed:

Board

In order to reach the shown finishing , the boarrd and the wooden piedes were sanded up to 800.

Then varnished with a shellac in ethanol solution.and then sanded again up to finding a pfat surfice (with no oil or water just in case it wew not absorved by the wood) then i followed soaking the sandpaper and wood in oil. in that way i could go on up to 1600. at the end it was washed with ethanol to avoid any problem with the sand paper.

a bit more advanced stage of the some board

The board again

It was decided to use mains AC because it is esay to find, but is likely (even desiderable) for it to work with 3-phase supply. so the best option was using a power conector whose change for a 3 phase one would be easy. Then the obvious option is the CEE form (IEC60309) and the rectifier has been built letting holes for the extra diodes.

The plug has been fitted into the wood by an ad hoc pice of wood. When I did it I thougt on more decortive ones , but at the end i choose the simplest one,

It was done in a piece of willow trunk but any other wood could have been used
At the beginnign I sterted with:

Raw woodblock

Wodblock

This became:

finished wood block

The power connector was fixed using two brass screws whicho crosse tehe connector up to the wood. When possible I will add a computer generated cut of the piece.

Then conductive bars were put there by the use (again) of brass screws.

Every bar is mirror or half mirror polished, as can be seen :

busbar were fixed with brass screws and has been mirros polished:

The coper srew was made with a die because it is almos impossible to find. in Madrid. At this frequency polishing bus bars is useless. acutally it has been done because I have no way to cover them with tin and preotect them from decay. .

I obtained:

Posted in motor control | Tagged 3-phase devices, asynchonous machines, introduction, motor control, pwm | Leave a comment

Other works

Posted on 2010/11/16/Tuesday by Ramiro

It is easy to realise that the average home is not prepared to undertake this project. so as you may imagine neither my home is nor my family wants ito be, so here i will tray to speak about those other labuors needed (or very convenient, or complementary) to complete this project or similar things (tools, places, procedures…)

There will be few tags here abaut those topics. One of them is for places (exactly how to make places usable, outdoor elctrical equipment , anti-rust protection…), other for tools and their use, other for non standard procedures and one for the lathe and similar things.

This history about the lathe begins when a waterpolo mate (¿or fellow?) offered me the late his grandfathen bought (as a 2nd hand item) when he was young.

then he founded a car-repair workshop. then his father died and the son put the lathe on sale, but he wanted the lathe not to be scraped or let in open air to rust, then he sold it me and to a fried an I who share it now.

This category is devoted mainly to show how a 100 year-old lathe that has not been used for at least 20 year can be broubgt back to a usable situation.

Some of the the proccedures will not be exactly the best ones but many times will be the only availble option given the situation (not much money). But for that, it could be used as a referece on how to do one’s best in machine reparation without causing any permanent damage to it with the constrits oa a short budget.

Posted in Uncategorized | Leave a comment

conmutation 1

Posted on 2010/11/16/Tuesday by Ramiro

Think about an electrical machine. Each one of its circuits has an impedance given by:
$L= R+ s L+ \frac{1}{sC}$ >

In almost any existing machine (and in every useful one) capacitances are negligible, so there’s no use in keeping the 1/sc. This is true because the Laplace transform of 0 is 0, not because you just find that by simple subtitution. The ohmic resistance is very often negligible, so we will neglect it now. Let’s imagine that we put an inductance through a function like this one:

$1-\theta (t)$

Then, obviously:

$E=\frac{\partial \Phi}{\partial T}=\frac{\partial \Phi}{\partial I} \frac{\partial I}{\partial T} =� L \frac{\partial I}{\partial T} \Rightarrow V= \delta(t)$

is found in the inductance, so the switching device shall be able to cope with:

$V_0 +\delta (t)$

Then, the signal is not less than any finite number, so it is necessary to add a device able to filter those peaks. If the engine is small, it may be enough with a diode in short circuit with the windings:

If the device is a BJT, it will filter some of this current (leak current), but this is not true for a MOSFET. If the engine is not that little (or you don’t like to take any risk) you may want to add a RC filter, or even a mixture of the above options. If the budget is a concern, choosing an incandescent ligthbulb as a resistor for the filter is a very good choice.

In our example, the circuit is as follows:

We will consider R2 as 0. It could be used for the most safe performance, but unless the motor is a very big one, it may noy be necessary (this will be explained somewhe else in this web page)

We have a RLC circuit, so:
$\frac{1}{z(s)} = \frac {1}{R}+� \frac {1}{Ls} +sC$
$z(s) = \frac {1}{\frac {1}{R}+� \frac {1}{Ls} +sC}$
$z(s)= \frac {s}{\frac {s}{R}+� \frac {1}{L} +s^2C}$
If we divide over C:
${z(s)} = \frac {Cs}{\frac {s}{CR}+� \frac {1}{CL} +s^2}$

We can rewrite it as:

$z(s) = \frac {Cs}{Bs+� {\omega_0}^2 +s^2}$
with $B=\frac {1}{CR}''$ y $\omega^2=\frac{1}{CL}$

Where we can substitute (Remember the relation betwen Fourier and Laplace transformations)

$z(j \omega) = \frac {Cs}{j\omega s+� {\omega_0}^2 - \omega^2}$

And a good complete set to analise the phenomena found could be:

[latex} \Phi_i= e^{- K B_w t} e^{j i \omega_o t} / i \in N [/latex]

Where j is the complex unit, chosen mainly to avoid mistakes with the i on the index of eigenfunctions. This is a separable problem (in fact it is already separated) as a product of two set of functions: one time dependent and one frequency dependent. We will only normalize the frecuency dependent part for obvious reasons.

[latex] V= A_i F(t) \frac{1}{\sqrt{2 \Pi}} e^{j i \omega_o t} [latex]

Now we know how the filter works, but we dont know the signals it will find. Applyng the Faraday Lenz Law:

{latex} E=- \frac{ \partial \phi }{ \partial t} [\latex]

We will discuss this as well as a set of eigenfunctions in a future post ¿Conmutation 3?

Posted in motor control | Tagged 3-phase devices, asynchonous machines, introduction, motor control, pwm | Leave a comment

a PWM example

Posted on 2010/10/22/Friday by Ramiro

Let’s build an example
This will be a circit able to control electrical amchines throw PWM. This exmple is clearly oversized considering every use it may have.

Most of what will be shown here it’s done in order to show somo of the procedures typically found in power electronics. For example we will use bus bars for less than 20A. just to show how to use them instead of using just cables. Is such a way if the conmutation devices were changed, a 100KW device could be controlled instead of the 5KW one we will be able to control.

¿Can’t it be done with a printed ciruit board PCB?

No. An if I would suggest you not to tray. Up to 5-8 its possible (if you don’t mind 1-2-wide tracks) but for higer intensity can be very difficult (altogh sipecilly thick-coopered PCB are feasible, are not a common intem in electronic components shop ). From those values on it is advisable to use point to point assembly as it was done in our grandparents’ time.

I choose buliding the example over a laminted wood board. Obviously it is not a good chice for agressive envirnoments such as hight-moist ones… This circuit is not intended to be put in such a place. In fact it probably end up being used at home. Then, it is not a problem.

If the circuit is intended to be used in a different environment we could use plastic materials, ceramics or other suitable material. Anywaw the changes are only trivial ones.

Instead of using special isolators, the circuit will be built with pieces of varnished wood. If we used higer voltages, (that creaded higer potencial differences over the isoltors), it would have been necessary to use glass or ceramical ones… (o even plastic if there were impact risk but is not as good as the ones above)
If given, the really unlikely case of very height voltage the device could be placed in oil or even in hydrogen hexafluorride (with IGBT instead of MOSFET) . But it’s not the case. ( Even in railroad systems, usual voltages are from 600V to 30KV)
Every wooden part shuld be impermeabilised. we are going to varnish them with shellac. Shellac is a known isolator (it was used up to very recntly in transformaers and engines) and it’s also a known varnish.

I will try to made the information shown here the most correct and complete possible, but I can make mistakes. Anyway it is a power device so it is dangerous. So dangerous that in case of major fail or missuse can even cause death to the people who uses it o others directly as well as trhow fire, explosion of nearby sustances, cutting tools…

It is also not negligible le the risk that is udertaken while it is build, specially while it is tested. and many other risks i can’t even think of.

Then if its constructione is attempted, it is under the only responsibility of the builder.
Anyway this text’s autor takes no responsibility over the use it can be made of it.

I would recommend it’s construction only to people with some elctricity or electronics-related experience or with scientifical or technical education.

Posted in Uncategorized | Tagged 3-phase devices, asynchonous machines, introduction, motor control, pwm | Leave a comment

PWM

Posted on 2010/10/11/Monday by Ramiro

Lets consider a signal. its mean over a period of time T shuld be:
$V= \frac{1}{T} \int_0^t f(t) dt$

So if a given value of V is needed you only have to keep this intrgal unchanged (it is not necessary for it to be a voltage, it can also be an intensity or whatever).

The first way to achieve thtat cuold be keepeng the value of f(t) as close as possible to de value of the mean. This also keeps all the moments of the signal close to 0 for example the variance.This parameter very important as far as noise is concernd. This way is almost perffect but for is very inefficient. alwys less than the 25%, so it should not be used for hight power devices unless there is no choice.

Everythng that keeps the integral unchanged coud be used. Since it is a linear system on f(t), it remains unchanged if we change the complete set to a different one. it is also linear in dt (we could even choose the absolute value and consider it a scalar product).
But as far as this topic is concernd we could take its definition instead of the integral.
$V= \frac{1}{T} \lim_{h-0} \sum f(t) dt$

Then, a good way to do it could be splitting every intervel in two different intervals, so as to
$V= \frac{1}{T} \lim_{h-0} \sum f(t) dt =V= \frac{1}{T} \lim_{h-0} \sum (V dt_1)+ 0 (dt-dt_1)$
Then ist just switching on and off. Then if we choose a signal bigger than the maximum of F(t) and take the product with (actually convolution) a fucntion as the following
$f= \sum_i a_i \theta(t_{ai})(1-\theta(t_{bi})$
which switch on in a and off in b.
Hera comes the problem of choosing the time T in order to tihnk of the mean.
If there ire periods, its obvious that should be the shortest (of them).
If ther is no period, the time shuld be related the most relevant for the system, thus the problem is the the same as the one we find when passing from the micorscopical to macroscopical scales, as happens in electormognatism or thermodynamics. Then a time longer tha the one needed for switching (each individual switch should be negligible to the time the mean is computed) and much faster than every relevant phenomenon in the controlled sysytem.

So it can be different in different systems for example a oven (or e furnance, for what matters ist the same), this used to be done with 5- 15 secondes on and the mean, then culd be computed over 10 minutes or so. This is possible because the system has a huge heat capacity.

The oven is a very special case as a computer power supply also is. this reqeuires a really fast answer and has other constraints, mainly a voltage limit that shuld not be surpassed and a lot of noises coming from higer fourier components that should be filtered.

Those contraints (like for axample an additional current limitation) and filters for such a things shuld be taken into account as we will do in a further post.

Conmutation
The switching times could be choosen in different ways, form pre-computed values to all sorts of feedback systems (including the simple and obsolete bimetallic thermostat)
Some of those thigs should be treated in a further post but not every possible system. some of them will be treated if and only if i were asked for.

The mean is a function

We could think it in meny ways. but it implys that for every time, the mean shouldd follow a function. Then if those derivatives are very differnt (one to each other.) then is a case of the Schwartz thoerm for oscillating integrals.
So:
$'' V(t)=\frac{1}{T} \int_0 ^ T f(t) dt$ .
This needs the norm of the derivative of the menan [latex} V(t) [/latex] should always be lesser than f(t) or if there are discontinuitues in V(t) there should be less than in f(t)
This is not very clear and should be prooved in casewise . anyway it is necessary to proove it in a epsilon-delta that for the averge case might resemble the Schwatrz theorm or just be a simple manipulation of the definition (Stieltjes, Lebesgue or even Riemann definitions).

A simple way for the typical case can be treated calculating the means locally (it it has less discontinuities than V(t) and then comparing each other and with the times (frequencies) that will be present on the system. then could be considered as continuous. (if we consider for the slower signal a interval (ball) much bigger than the period of the fastst one its quit likely for us to proove it)

¿how to get it?

Let’s consider the follwing integral:
$F=\frac{1}{T} \int_{0} ^{T} A \theta(t-t_o) dt =� \frac{A t_0 }{T}$
So for changing the mean it’s only necessary to change $t_0$

It can be achheved in a dynamical way, for example comparing the desired signal with a sawtooth. then the conmutation elment (MOSFET) will close the circuit (conduct) if and only if the sawtooth is smoller than the signal it is compared with. Later, an example will be shown.

how can we filter it?

What we have is a signal that con be written as a Fourier series. The N first components should be the same as the ones for the desired function (this lastone should be used as a control signal to the actual one). Then the differences should be smaller than the radius or ball ke spoke abaut. Then in the superior order compnents we will find the effects of conmutation. this idea is the same we spoke abaut befor but told by its effects (with fourier series).
Then we can filer it with a lowwpass filter (able to absorb the power needed…). that will cut off voltage peaks and higer fourier compnents.

PHSVM
(phase space vector modulation)
It is exactly the same idea but taking into account n independent integrals over the same ball.

Posted in Uncategorized, motor control | Tagged 3-phase devices, asynchonous machines, introduction, motor control, pwm | Leave a comment

A very old motor

conmutation 2

The board

Other works

conmutation 1

a PWM example

PWM

Archives

Meta