From: Roger Cook To: Grant Saviers Hi Grant, I started work on my EDM project, mainly the mechanical side and researching the electronic side. My project stopped when I had to move to a new location for employment and I have not been able to restart the project yet. I was interested in Ananian's power supply because I thought it was more elegant and had much greater capacity than the one in Robert Langlois' book. I am a physician with a hobby interest in electronics but rather limited knowledge. My plan was to use Ananian's circuit as he designed it but with the addition of a two KVA transformer on the input side for isolation from the mains. Maybe the output is dangerous enough in any case that the isolation would not make any difference. In case you are not familiar with Langlois' book, it is:"Build an EDM: Electrical Discharge Machining - Removing Metal By Spark Erosion" Village Press, Inc, 1997. Mr. Langlois wrote several follow-up articles in "Home Shop Machinist" http://www.homeshopmachinist.net/index.cfm I was not able to get the back issues from Village press but I was able to get copies of the articles through a University Library. I was able to contact Mr. Langlois once at this address: RLanglois
fanshawec.on.ca
I have heard from several other people who are interested
in, or actually working on, EDM machines. They may be
worth contacting:
Steve Scampini [EMAIL PROTECTED]
Malcolm Cox [EMAIL PROTECTED]
David Pape [EMAIL PROTECTED]
E. Maurice [EMAIL PROTECTED]
Jaroslav Chadim
[EMAIL PROTECTED] (limited English)
Ric Wade [EMAIL PROTECTED]
I am eager to get back to my EDM project myself but for
now I would be very interested to hear of your progress
with EDM. As an electrical engineer, your thoughts on
Ananian's power supply and your experience if you do
build it would be very interesting to me and maybe to the
other contacts I listed above.
I will add your name and email address to my list of
people who have emailed me about EDM and pass it on to
any other EDM contacts I might have in the future (unless
you do not want me to do this).
Good luck,
Roger Cook
Date: Mon, 22 Aug 2005 10:49:48 -0700 From: Grant Saviers To: Roger Cook, edm
cscott.net
Subject: Re: Ananian's EDM power supply
Roger,
Good to hear from you and you may pass my contact info and comments
along to others that have contacted you. I've also copied Scott
Ananian, so he can weigh in on my comments.
I've been researching home shop EDM for a while and bought a copy of
the Langlois book (usually advertised in Home Shop Machinist
magazine) which has all the articles in one place. I've also
purchased the wire and sinker edm plans from the University group in
Europe that developed them. They are long on mechanical design and
short on the electronics. There is also a recent article in HSM or
Machinist's Workshop re a simpler than Langlois design, but I don't
think much of it.
All of these are short on the theory and Ananian does delve into this
nicely. I'll try to get copies of his references and have found a
few others on the web. Langlois sketched out a MOSFET based
switching design at the end of his articles but nothing more has been
published. I was sketching out the concepts of a 60 volt 50 amp
MOSFET based design of my own but will take a step back now that I
have seen Ananian's thinking. Probably the best course at this time
is to look at the research papers and think some more before buying
more parts.
A few comments about the design and building of Ananian's design:
It will be quite a challenge for the first builder, particularly if a
PCB isn't used. My first boss gained some fame on the internet with
"Vonada's maxims", the most popular being "Every picofarad has a
little nanohenry all its own". High power pulse circuits are
terribly layout sensitive, so much thought and much trial and error
usually go into the layout and debug.
You're correct to be concerned about safety with his direct line
coupled voltage doubler. It's ok IF neutral and line are correct,
but I would NEVER trust regular house/shop wiring to have kept them
straight. Best to use an isolation transformer, a backwards 240::120
step down would yield same bulk DC supply voltage (330v) and be safe
and not too expensive for 15 kva. Find one on ebay or try Grainger.
His line doubler also will do nasty things to the AC 115v primary
power waveform and contemporary regulatory codes don't allow this
kind of uncorrected capacitor input power supply in production
products.
I think some more work is needed on the power capacitor and inductor
components. The ripple current is too high for the DigiKey specified
47uf C7. I also doubt the doubler capacitor C3 will stand the RMS
current if this design is used, it also needs to be much higher
voltage rated and protected from line voltage spikes. Special ultra
low ESR (equivalent series resistance) capacitors are made for
switched mode power supplies which are analogous to the Ananian
design. Appropriate ones cost about $25 so if it is in stock this is
not such a big deal. I think the inductors will also need to be more
sophisticated than "a few turns on a bolt" given the high frequencies
and currents involved. At the specified pulse duration and
frequency, eddy current losses in a bolt would be very high, making
it very hot in a continuous EDM application. A ferrite toroid is the
standard approach and not a big deal from a design or cost
perspective. See www.amidoncorp.com for some basics on ferrites, and
they are happy to sell small quantities. I've used their parts for
RF applications (I'm an Extra Class ham radio license holder).
I think provisions should be made to vary the pulse width, say 50 to
200 us and pulse interval, say 200 us to 2 ms. Other articles point
out that different electrodes and different materials being eroded
need power adjustments and also depending on how well a discharge
site is being flushed with solvent it will require different pulse
intervals. Variable voltage may also be nice to adjust the pulse
energy, it would be straightforward to have 330 or 165 volts as the
bulk DC supply if the backwards step down transformer were used.
Ananian mentions adjustable voltage but it isn't in the schematic.
Ananian makes an interesting comment about "tickler" designs and
added the boost circuit to his design. I'm speculating about a
design that uses the output inductor also as a pulse transformer but
haven't worked out the magnetic calculations yet to see if it
feasible. If it is possible, then there is no reason to have a 330
volt supply to maintain a 20 volt nominal arc voltage. Note that the
120 hz 60 volt RC Langlois style EDM designs do work, obviously the
electrode is very close to the work to start the arc. If a 330 volt
or higher few microsecond long pulse can be superimposed on the
leading edge of a 60 volt 100 us power pulse, the circuit gets
simpler, safer, and cheaper. (Power MOSFETs are $3 vs $18+ for the
high voltage IGBT's) But, that is a whole new design approach.
In other words, there is a lot yet to do!
regards,
Grant
Date: Mon, 22 Aug 2005 16:01:02 -0400 (EDT) From: C. Scott Ananian To: Grant Saviers Cc: Roger Cook Subject: Re: Ananian's EDM power supply Your comments sound (in general) right on. To my regret I still haven't had the time to sit down and finish this design out, so there are undoubtedly corrections & improvements left to be made. Do you mind if I post your email to the web page? Here are some brief comments: > I think some more work is needed on the power capacitor and inductor > components. The ripple current is too high for the DigiKey specified > 47uf C7. I also doubt the doubler capacitor C3 will stand the RMS > current if this design is used, it also needs to be much higher voltage > rated and protected from line voltage spikes. Special ultra low ESR > (equivalent series resistance) capacitors are made for switched mode > power supplies which are analogous to the Ananian design. Appropriate > ones cost about $25 so if it is in stock this is not such a big deal. I had a very hard time finding appropriate components from the catalogs I had available; I'm sure I was just looking in the wrong places. > I > think the inductors will also need to be more sophisticated than "a few > turns on a bolt" given the high frequencies and currents involved. Again, it was very hard (for me) to find appropriate commercial inductors. The frequencies aren't *that* high, and the duty cycle can be kept as low as required if necessary. I used an inductance meter to verify that my "few turns" had the right base inductance, but frequency response was a concern I wasn't able to fully investigate with the tools I had available. I was aware this might be an issue. (Note that using a ferrous core introduces its own problems. In particular, saturation at high currents can lead to a sudden drop in inductance, which would cause the current to shoot up & quickly burn out the protected transistor. I was using a non-ferrous (aluminum) core (read, bolt) in an attempt to mitigate this possibility. Magnetics design is a black art; I've forgotten most of the thick book I read in an attempt to teach myself the basics.) > I think provisions should be made to vary the pulse width, say 50 to 200 > us and pulse interval, say 200 us to 2 ms. Other articles point out that > different electrodes and different materials being eroded need power > adjustments and also depending on how well a discharge site is being > flushed with solvent it will require different pulse intervals. > Variable voltage may also be nice to adjust the pulse energy, it would be > straightforward to have 330 or 165 volts as the bulk DC supply if the > backwards step down transformer were used. Ananian mentions adjustable > voltage but it isn't in the schematic. It is, it's just clever. =) The comparators (U3) sense the voltage across the capacitor and compare it to the PWM D/A voltage coming out of U4-1 (VSET_PWM) to set the discharge level. There's a separate DISCHARGE_EN signal which might let you clip short the discharge pulse if you're clever about the sequencing (watching the SPARK signal and de-asserting DISCHARGE_EN), but you obviously can't make the pulse longer that way. There's also separate dedicated current limiting for Q1 and Q2 (U3-4 protects Q1, and U3-1 protects Q2). The BOOST circuit is the 'risky' part of the venture, which I added for adventure's sake (but see below). The idea is that the controlling microprocessor toggles BOOST_EN active for some small (experimentally determined) fraction of a second to leverage the 100uH inductor's ability to generate high voltage -- and then you cross your fingers and hope Z3 protects Q4/the environment (and Z1 protects PS2501/Q3) if the 'tickle' doesn't actually induce breakdown across the workpiece, since that current has to go *somewhere*. The risks are: a) Z3 might not be fast enough, and b) Q4 may well burn itself out if (due to software error or otherwise) BOOST is left asserted for too long or at an inappropriate time. (The inductor will limit the crowbar current through Q4 iff BOOST is asserted for "short enough" *and* there is not already current flowing through the coil.) While I've got the details paged in, note that there are two 100 uH inductors in the circuit. They are both for transistor protection: we need to limit the max dI/dT (current rise) in the circuit so that we have time for our feedback-based limiter circuits to work (in particular, we can't turn off the transistors faster than their minimum switching delay). This added inductance could actually inhibit sparking across the workpiece, because even though we have high V the inductor will initially act to prevent the flow of current. Several of the references (IIRC) mentioned inductance-related issues with their machines. This was the secondary motivation for the BOOST circuit -- to get current flowing in the inductor so that its sluggishness doesn't prevent spark current from flowing. With BOOST working, any inductance-related pcb-layout issues should be similarly mitigated. I never laid out the microcontroller control circuit, because frankly just about any current microcontroller will do fine. Only 4 TTL-level outputs and 2 inputs are required. Some of the signals are fast/short enough that a dedicated controller is probably required; I don't expect you'd be able to (say) just hook this up to the parallel port of a Windows machine and make do. Do be careful with the layout to maintain isolation for the digital control, though. > $3 vs $18+ for the high voltage IGBT's) But, that is a whole new design > approach. And a very interesting one! --scott Dictionary blowfish CLOWER TASS overthrow Sigint PBHISTORY UNIFRUIT ( http://cscott.net/ )