YAG power supply woes / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

YAG power supply woes

Move Thread

LAN_403

...

Mon Oct 29 2007, 02:46AM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

After getting my big yag laser working with a few MOTs, I decided it was time to get my big power supply working.

Now, one would assume that would have been easy, since I snagged the power supply along with the laser. Unfortunately, the maker of the power supply (Converter Power Inc), has been bought a few times, and no one relies to my e-mails.

So I decided to try and make it work by reverse engineering...
What I have found:
The 'remote' db-25 connector goes to a signal conditioning board that has diode arrays on the lines to clamp the voltages. One of the pins also goes through an op-amp (presumably the voltage set).
From there it goes to the control board, which has a lot of 4000 series logic chips, and a number of opamps. Also fed into the board is a feedback connector, which sends voltage proportional to the voltage and current being output from the supply.
The board then has a 16 pin connector which goes to both of the power boards (which are 100% identical). In this connector, there is power being fed back into the control board (+/-15v), a pin for the soft turnon (there is a 60r resistor with a relay across it to limit the current to the caps), a pair of pins for the fans, a pair of pins for the thermal cutoff switch, and a pair of pins that feed the gate drivers.
Each of these pins go through 3 paralleled inverters in a cd4049, and then into a irf512 (with a zener across the gate and a freewheel diode on the drain), and then off to a GDT with 2 secondaries.
The output of this feeds a H-bridge of irgp56 IGBTs with a mr2406 freewheel diode. There is also an active pulldown on each of the fets (I have schematic for it on paper, if anyone wants it I can digitise it). The fullbridge has 2800ufd worth of electrolytics, and 20ufd of film caps (2x ec5mp16) for filtering on the DC side, and a on the AC side there is a ~10uH series inductor and a .5ufd dc bocking cap.
This output is then fed into the primaries (paralleled) of 2 ferrite transformers which have about 50 turns (rough guess) on the primaries, and about 500 on the secondaries.
The output of these is then seriesed, and fed into a rectifier board that has strings of 4 to-220 diodes (that have house part numbers) through 2 ohms of resistance, and then off to the output. This board also has a current sense resistor and voltage divider to measure the output voltage/current.
Keep in mind that there are 2 complete power boads, so a total of 8 transistors, 4 transformers, and 16 diodes making up the rectifiers.

All of the connections between the 20ufd worth of filter caps, the 10uH series inductor, and dc block cap are super thick (the inductor has a .18" diameter litz wire, the caps are attached to the board using about 10 pieces of #16 wire), presumably due to skin effect related problems. There should 'only' be about 20a runing through them.

Now, the supply came with a cable for the 'remote' connector, which only had 10 pins pupulated--so I narrowed my search to just those pins. A bit of tracing on the controller circuit board showed that most of the pins were connected to an open collector output, presumably the overtemp/etc outputs. 3 other pins were connected into the logic parts of the circuits, but all of the nets had an output on them, so I would assume they are also some type of output. In the end, there was only the 1 line that went through the opamp (I know that was an input), and 2 pins that didn't really seem to do anything useful on the board. But I connected LEDs to the suspected output, and tried all of the possible combinations on the inputs, but alas I couldn't get the supply to do anything.

So... I am thinking about either rebuilding the controller board (using some more modern control chips), or scrapping the existing design, and just building a monster mazzilli driver using all of the parts. I would probably end up using just 2 of the cores (they are monsters!) and running 2 completely separate system in parallel (so that I can use both of the 20a/120v services in the lab). Also, I would dump the old fets and switch over to something like a 40n60 and use a ucc27321 to drive it (I am running so slow the delays shouldn't really matter too much), and then use the huge litz wire that currently makes up the series inductor for my primary, and the old DC blocking cap as the resonant cap.

Now, one would assume that I could just drive the board with a fixed frequency (presumably somewhere around 10khz, since that is what the IGBTs are rated at) and then either use deadtime or just turning it full on/off to stop charging... But, judging by the number of opamps on the board and the current sense line, the original board was changing the frequency, or at least the duty cycle, as it charged. And I have no idea how to reproduce that, or what I am trying to reproduce for that matter.

Also, right now it is set up to run off of 220v, and I am using about 50lbs worth of MOTs as a big isolation/stepup transformer (which also lets me suck power off of both of the 20a/120v circuit I have in the lab

), but if I rebuilt it as a big mazzilli I could design it to run off 120v, and eliminate that iron. Also, the overall charging system would be smaller (even not counting the MOTs) using the mazzilli driver.

Of course, if I were to rig my own charger, I would have to deal with all of the little things--ballasting, turning it off, interlocks, etc...

Comments?

Steve Conner

Mon Oct 29 2007, 08:21AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

With all due respect I think it would be a lot easier to just get it working than try to gut and rebuild it. There'll be some pin that you forgot to pull up/down/connect to some other pin etc. In particular, there are always two terminals that need to be joined for the interlock.

You can easily convert it to 120V by using two electrolytics in series on the DC bus and rewiring the input rectifier as a doubler. Since it has two power boards you can separate the rectifiers and DC buses, and connect one to each of your 20A outlets. Then you can say "No Mr. Bond, I expect you to die."

...

Mon Oct 29 2007, 10:38PM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

Thats the thing, I have spent about 10 hours meticulously tracing pins trying to figure out what everything does, and still haven't managed to convince it to do anything. I was considering making a full schematic of the controller board, but there are just to many 'hidden' traces that are under the ICs or connectors, etc, that it would take forever to get a workable schematic. And then I would still have to figure out what exactly all of those chips are supposed to be doing.

At this point, I am starting to think that the cable that they sent with it got moxed up with something else, and there area actually a ton more wires that I need to use (since all 25 of the pins on the db25 'remote' connector do in fact go somewhere)--so I would have to trace out the rest of them and see if any of them look like an interlock.

It just seems like if I had spent those few days working on my own design I would have an at least somewhat functional power supply...

Yurtle

Sat Nov 03 2007, 09:44AM

Registered Member #1099 Joined: Sat Nov 03 2007, 09:38AM
Location:
Posts: 1

Is this what you have? If so, I've got two manuals with complete pinouts, as well as a schematic for a controller. I used an old VTVM to make my controller.

Adam

...

Sat Nov 03 2007, 04:19PM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

Nope, this one is about 4x the size of that one

But it doesn't realy matter now, since I have started rebuilding this one with the new topology.

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob