Universal Driver 2.1b

Some of this content is originally created by Steve Ward (stevehv.4hv.org) and is re-posted with his permission.

Steve Wards work with phase lead compensated drivers is most likely based off the work that Finn Hammer did on his driver modifications called the “Prediktor” that was a DRSSTC driver with phase lead.

The content of the folder stevehv.4hv.org/stevehv/lead_comp/ did not contain much written information, except the quotes below here with some details on the circuit. The pictures are from Steve, but the descriptions of them is the interpretation of Mads Barnkob.

It is recommended to read the article on the Universal Driver 1.3b to understand how the driver works and the changes from 1.3b to 2.1b will not seem to radical afterwards.

26th October 2009

Steve Ward built and tested the universal driver 2 with phase lead on his “DRSSTC Magnifier – Mark I” Created 4/15/05: http://www.stevehv.4hv.org/DRSSTCmag1.htm

The first tests was done on an unknown IGBT, and not the magnifier system above. Green is primary circuit current and light blue is inverter output voltage. Yellow is presumably gate drive signal.

For the tests that are known to be done on a CM300 IGBT brick, on the magnifier system, the following collection of oscilloscope screenshots was available. Green is primary circuit current and yellow is inverter output voltage.

26th March 2011

R26 added in series with C33 because when it was only C33 the peak voltage of the positive feedback (hysteresis)
was outside the save range for the comparator inputs. R26 and C33 provide short positive feedback pulses which helps keep the comparator from self-oscillating.

R27 added across C5: There seemed to be some instability in start up, sometimes the TL3116 would start off with
output HI, sometimes output LOW. R27 restores proper bias to the comparator input by discharging the DC offset that may be left across C5. I picked 100k for R27, but perhaps values as low as 10K may be required, i would not suggest going lower than 10k.

Steve Ward, 26th March 2011

The files for the quoted text above is the old files called UD2_1 and is listed at the bottom of this article.

2nd September 2012

You need EAGLE by Cadsoft (google it) to look at the board and schematic. You can get a free version on the web.

This project was created before i knew the importance of linking a schematic and board file. These files are not linked, consequently errors do happen!

I changed some biasing resistors for the phase lead comparator, if you have referenced the older versions you might notice this.

Im not sure the parts list is 100% up to date, i had to make some mods to the design after i ordered everything.

C33 controls a “no switch” time after each output transition on the comparator. Originally i found C33 could be just 220pF, but recently using CM600DU-24NF modules with a loooong 1uS or so delay in switching, i found i had to boost C33 up to 2.2nF.
This provided a longer period where the comparator has high immunity to noise, and without this there was severe “glitching” where the IGBT switch noise caused the comparator to switch several times instead of once per half-cycle.

R11 R12 and C9 with IC4E form a pulse-width limiter that is frequency dependent. You can disable it by shorting out C9.

Steve Ward, 2nd September 2012

The files for the quoted text above is UD2_1revb files, which are the latest and those that you should use!

1st May 2013

This is pictures of a finished universal driver 2.1b as I made it for my own large Tesla coil called Kaizer DRSSTC 3.

Freewheeling driver

The /leadcomp/ folder also contained a folder called /UD freewheeling driver/ which contained schematic, PCB layout, Atmel microcontroller code and gerber files for a freewheeling driver add-on

Old files that should not be used, only listed for historical reference

AGFA ADC 5155 X-ray Scanner Teardown (Part 1 of 4)

Part 1: Teardown of a AGFA ADC 5155 x-ray image scanner / digitizer. Also known as Computerized Radiography (CR) it fulfills the job of scanning …

Universal Driver 1.3b

This content originally created by Steve Ward (stevehv.4hv.org) is re-posted with his permission.

Having built a few DRSSTC drivers using IGBT bricks, I realized what a pain it was to build the large gate driver circuit I had previously developed. In search of something simpler, I turned back to our old friend the Gate Drive Transformer (GDT). I used to believe that it was not possible to drive large CM IGBT modules with a GDT, but I proved myself wrong. There are a few important details that must be done properly to get good results from a GDT drive. They are:

  • Low leakage inductance GDT. This requires a pretty large ferrite toroid.
  • Low impedance gate drive circuit using high quality ceramic decoupling and DC blocking capacitors.
  • Make all long wiring coaxial (or at least twisted pair), particularly on the primary side of the GDT where the impedance is 2 or 4 times more critical (depending on half-bridge or full-bridge).

This is the files for Universal Driver 1.1. Note, the original design (including PCB) have been modified.  The files listed here reflect these changes.  Please see further down this page for the newer design files.

This driver is very similar to my other ones, with a few important exceptions. The gate driver section is now very robust, using 16A P/N mosfet pairs. The input to the gate driver is set at 24VDC, so that a 1:1 GDT ratio can be used and still achieve 24V on the IGBT gate. I no longer believe its necessary to drive the gates to 30V unless you are absolutely pushing the IGBT to the extreme. Otherwise, the higher gate drive voltage just makes the possibility of gate oxide failure more likely. Even with 24V drive, TVS diodes should be placed right at the IGBT gate terminals to clamp the voltage to less than 30V.

In order to support the new gate drive setup, some additional logic was required, and was provided with a quad AND gate. I also switched over to the 74HC74 D-Flip Flop instead of the 74HC109 JK flip flop, because its 14 pins instead of 16. I also used some hacks to replace the 555 astable timer with just an RC on the input to an AND gate, in the overcurrent detection circuitry.

A picture of a completed driver and controller.

Another example of this driver in use can be seen in the DRSSTC-4.  This coil uses a full-bridge of 40N60 mini-bricks from fairchild semi.  The tank cap is a pair of special CDE caps, each rated 100nF at 8kVDC (estimated VAC rating of 2kV).  The GDT was wound with cat-5e network wire, using the “white” conductors as the primaries (4 in parallel) and the other 4 wires as secondaries.  A 5.1 ohm gate resistance is used, and is bypassed on the discharging edge with a 1N5819 diode.  The primary is tuned low to 170kHz.  This coil is a solid performer producing up to 38″ sparks with only an 11″ long secondary winding. 

IMPORTANT: The over-current detection comparator has a limited operating voltage range for its inputs. When operating from 5V, as in this circuit, the trip level set by the potentiometer should NOT exceed 3.75V for reliable operation.  Setting it above this voltage may completely disable the OCD functionality.  Lower the burden resistance (on the ODC input) if this is a problem.

A note on using fiber optic receiver OPF2412:

This one seems to come back and haunt me from time to time.  Im using the OPTEK OPF2412 fiber optic receiver.  These are noise sensitive devices, and not surprisingly, problems show up when I use them around tesla coils.  Firstly, they really should be in a metal enclosure.  Secondly (and quite surprisingly), the metal sheath that is at the end of the ST fiber cable should be bonded to the metal box.  It may seem absurd that a small piece of metal just a few millimeters from the lens of the Rx would be able to effect the thing, but surely enough it does.  The issue is either that this metal ring picks up some RF and couples it into the Rx, or its simply that the metal ring, when grounded, extends the electrostatic shielding of the Rx.  Im gonna go with number 2.  Anyway, the fix for me was to just put a small steel spring that presses against the box and the ST end when its plugged in.  It may be wise to stick all of the low voltage electronics into a metal box too.  One more odd problem discovered and solved! 

UPDATE 11/23/2008 (some of this may not make sense as i have removed the original schematics):

There were a few quirks with this design that i wanted to correct.  The RC circuit for the current limiter was fussy and hard to get working just how i wanted, also it turns out the difference between HCT and HC logic made a big difference in the circuit performance.  So i present an updated schematic that is a much more robust design.  Terry Blake gave me the inspiration for using the extra D flip flop to set the “disabled” state of the controller.  Operation is simple, when the comparator senses a trip condition, its output drops low, causing the D flip flop to clear its output, which disables the interrupter pulse via the AND gate.  The controller stays in the disabled state until a rising edge from the next interrupter pulse puts the D-FF back high.  This also has the added benefit of keeping the indicator LED ON for the maximum permissible amount of time (rather than just 1mS or something).  Below is a updated schematic that documents the surgery necessary for my PCB design. 

Also, the BOM is probably a tad outdated.  Most of the important parts are there (all the ICs and caps) but it may be lacking a resistor or 2 or something like that.  I suggest you check over the new schematic before trusting the BOM solely. 

Id also like to report that this driver is now used on my revised edition of the DRSSTC-0.5.  It works great even at 300kHz!  Truly a universal driver.

UPDATE 2/11/09:

I have re-spun the PCB for this driver.  The main reason was that I wanted to put the fiber optic receiver on the PCB, and also because i modified the over-current detection inhibit circuitry.  Below are some helpful files for this new board.  Another afterthought modification to this design deals with the current sensing circuitry.  Having the burden resistance (R5) before the bridge rectifier (D8-11) and having yet more burden resistance (R4) after the bridge, made it less than straight forward to pre-determine the volts-to-amps scaling of the circuit.  So now the burden resistance is moved to be only after the bridge rectifier, and to handle the larger currents, I suggest using 1n5819 schottky diodes instead of the 1n4148 signal diodes.  Furthermore, the somewhat large .1uF filter cap on the sensing input to the LM311 comparator was causing other scaling errors (its a significantly low impedance at 100khz) so I’ve changed it to 1nF, or I find it can be left off altogether.

Pictures of the new board:

UPDATE 6/25/09:

While analyzing something with this driver, I realized the schematics were drawn incorrectly, the inputs to the UCC27423 dual gate driver were swapped.  Updated schematics have been posted.  This is a minor error that shouldn’t impact any design significantly.

DRSSTC III update #2

Due to a busy work year and not many days at home the progress with the large DRSSTC have been very limited. I have however …