Fusor Forums

Sounds like polarity was reversed, according to my limited experience.

Definitely sounds like installed in reverse. Trolyics don't like that.

Richard Hull

You guys were right. That was reckless. Fortunately I ordered several new caps. I installed a new cap and dialed a low current demand and slowly ramped up a voltage demand of 1 volt or ten percent output. The pin out for voltage out followed along nicely. My next order of business is to build a proper control box, but for now I consider my power supply in working order.

I'm glad you got it up and running.

Richard Hull

I finally gave up trying to make my DXM work and returned it. Although I was not happy that the vendor had listed these as working on eBay, Jon at NTC has been very cooperative. He even sent me another unit (marked "scrap") for spare parts to try to get one working. On my first unit, I was able to communicate via the web interface and could control the voltage up to 1400 volts. For any higher setting it failed. The spare parts unit had a blown fuse, which suggested problems on the PC board, so I tried the high voltage potted section with the PC from my original unit. I got the same results. I'm inclined to believe that both units have a failure in the potted HV section. Cliff had indicated that the ceramic capacitors in the potted section tended to fail when the units were operated in excessive ambient temperature. Although I have no way to know if the PC in my original unit was fully functional, I was definitely able to communicate with it. So if anyone is interested in a board that does communicate, the serial number on the communicating unit is 100785082-A00323. My guess is that Jon would be very amenable to an offer, since it is likely destined to be labeled "scrap" also.

I suspect the failure mode of my non working unit went something like this.

1. High ripple current causes C149 to overheat, rupture, vent electrolyte, finally shorting internally becoming a resistor
2. Diode CR34 feeds C149 and with the fault current overheats and also becomes a resistor
3. The internal 24VDC supply that runs the DSP and other low voltage circuitry starts to drop
4. R218 a 0.05ohm 5W current sensing resistor overheats and cracks in half
5. The 24VDC supply collapses and K3 relay contacts open inserting inrush current limiter RV2 in series with the rectified 120VAC mains. This temporarily reduces the fault current
6. The fault current finally exceeds RV2 max continuous current. It heats up and cracks in half partially melting adjacent relay K3
7. The input full wave rectifier CR35 survives the fault current and the fuse finally opens
8. FET Q10 is destroyed (three leads get shorted together) while parallel component Q6 survives. An interesting failure transient. (FYI at lower output energy the DXM is fully operational on just Q6)

The two units I have are not identical. My older unit that arrived working was manufactured 23rd week 2013 and has a factory mod, see pic. My newer not working unit was manufactured 5th week 2014 and does not have the mod. The factory mod (and current design) puts a small restive load on C160 which reduces the ripple voltage/current on C149 (the issue is two separate 400VDC rails share one switching regulator). The two extra diodes in the picture are a belt suspenders plus rope kind of thing. Without the factory mod the ripple was about 100V every 0.75 seconds. BTW you need a 120V line isolation transformer if you want to connect an oscilloscope and probe the power factor correction circuitry looking for problems.

Not having the parts for the factory mod I put a 1N4007 from C160 to C149. My mod seems to work fine and reduces the ripple to 0.5V at 120Hz when the HV output is off and disappears with any HV output energy. To be honest I prefer my mod but I will continue to keep an eye on it because “Only the paranoid survive”.

I replaced C149 with a higher ripple current capacitor (on both my supplies) that has over twice the original capacitor rating, Digikey 450BXW100MEFR18X35, it is physically taller and a tight fit for the available space. Electrical tape above the capacitor on the lid inside reduces any shorting possibility if the capacitor top should ever bulge. I also replaced 4 questionably functioning parts K3 and CR35 and Q6 plus its driver IC in addition to the obviously failed parts. The total repair cost is about $40.

The two internal fuses will be replaced with one slow blow 5A and one fast blow 6 or 7A (lower values provide better component protection).

Lastly the service guy with not great eyesight (me) caused the problem in pic 3 requiring a scope to locate. The fix was simple just shove the wires into the connector…Doh Have fun
-Peter

Andrew,

Are you willing to make and sell a C11 replacement connector to single conductor? I contacted claymount and they want $190 for p/n 12096 (recommended by spellman)

That sounds way too expensive, can you send me the cad file of the cable they were quoting?

I don't have time right now to manufacture more connectors, though here is some ordering information

Ask for a quote on:
PN 14896, a 2m cable version should be $69.21
PN 10008 ring nut w/ r16 cutout should be $10 ea

I have ordered parts from claymount, it's generally straightforward to order with a credit card, they will want you to submit a purchase order to them, I have attached a template I used, It probably helps to have a fedex account number to give them to simplify shipping.

Andrew,

They quoted $160 each and a 40 day lead time. (Also "Please note that the 14896 does already include 1 of the 10008 ring nuts in the hardware kit that accompanies the cable assembly. ") I've asked them to match the $70 price.

Garrett,

that does sound weird, there was some variation between quotes on certain parts (one quote to me was $10ea for the ring nut, the other was $4) so they don't seem to have fixed prices, but such a large jump in price sounds unusual.

Andrew

I am waiting for a quote from Claymount for both the part numbers you mentioned.

The web page product inquiry email address for North America is:
Philippines.Claymount.CS@vareximaging.com

Do you have a different contact address?

-Peter

My sale's contact was:
Karen Albright
email: Karen.Albright <at> varian.com

If you were being quoted through a different division, particularly an international one, they might quote a different price.
The 40 days lead time on the cable sounds correct, they quoted the same lead time to me.

Thanks. Just sent an email with fingers crossed.
-Peter

Andrew,

The price you were quoted was given as a "one-time, educational institution courtesy".

What contacts are you using in your replacement connector?

I bought two of these units from the same eBay seller [ntc_tech]. The main difference was the fact they were listed as [For parts or not working] and it was true; both broken as expected..

I have both up and running and bench tested to 5k volts. I have been unable to get a response on the HV cable so I can't test them at full throttle.

Thanks to Cliff S. and his generous contribution in time and sharing partial schematics the repairs went like this.

Unit Serial Number 411: (showing 13.8 hrs)
Replaced Q11 and Q8 [IXFH26N50] Symptoms: showing shorts on all 3 pins.
Replaced Q10 [65F6110] Symptoms: showing shorts on all 3 pins.
Replaced C149 (working and tested OK, but slight bulldog in top, you know how sensitive CAPs are to heat and they will fail eventually)

Unit Serial Number 341:
Replaced Q6 and Q10 [IXFH26N50] Symptoms: showing shorts on all 3 pins.
Replaced C160 and C161 (preventative maintenance, they tested fine)

Time:
Learning schematics and finding physical PCB part locations: 1 hour
Tracking down faulty components with Fluke meter: 1hr on first unit, 30minutes on second due to learning the layout.
Desoldering and verifying parts were defective: 30minutes (20 of which was switching to leaded solder so i could actually get them desoldered)
Ordering components: 30min all Mouser
Replacing faulty components: 30min.
=========
Total: 3 hrs for 2 units, or 1.5hrs / unit.

Cost ~$57 for two units. I actually used one MOSFET from unit 341 in 411 that I switched out both so they would match, but then found the same component bad in the other unit that I missed at initial debug. Thus I had one extra part, and needed it.

Shopping List:
C160 and C161
Mouser #: 5985-450V150
Mfr. #: 381LX151M450K022
Desc.: Aluminum Electrolytic Capacitors - Snap In Aluminum Electrolytic Capacitors - Snap In 450V 150uF 30X30
2 $3.80 $7.60

C149
Mouser #: 667-EEU-EE2W560
Mfr. #: EEU-EE2W560
Desc.: Aluminum Electrolytic Capacitors - Leaded Aluminum Electrolytic Capacitors - Leaded 56uF 450volts AEC-Q200
2 $3.26 $6.52

QQ8 and Q11
Mouser #: 747-IXFH26N50P
Mfr. #: IXFH26N50P
Desc.: MOSFET MOSFET HiPERFET Id26 BVdass500
2 $5.91 $11.82

OEM Fuses - Ceramic + extras
Mouser #: 576-0314015.HXP
Mfr. #: 0314015.HXP
Desc.: Cartridge Fuses Cartridge Fuses 250V 15A Fast Acting
10 $0.744 $7.44

Q6 and Q10
Mouser #: 726-IPW65R110CFDA
Mfr. #: IPW65R110CFDA
Desc.: MOSFET MOSFET N-Ch 650V 31.2A TO247-3
2 $8.22 $16.44

Trent,

Nice! What do you plan on doing with 2 supplies?

I have 3x operational HV supplies
Bertan 30 kV @ 33mA (older analog Bertan 106 3040)
2x Spellman 70kV @ 8.56mA (DXM70Ns)
This will give me flexibility to have 70kV @ 17mA. I also want to do some testing of direct conversion even if just simulated but thats beyond the typical fusor applications. I also wanted to make sure I ended up with at least one working supply, so buying two increases my odds of success.

I really like the fact I can completely automate these supplies via ethernet, RS232, or USB. I realize a simple DAC will automate even the older analog supplies, but repeatability and remote-control are two features at the top of my wish list and these supplies are perfect for both.

A word of caution on supply ratings; a power supply that has a 70 kV rating and has a max current of 8 or so milliamps does not generally mean it can supply both at the same time. You might want to check what the supply can do before counting on this type of performance. Critical is the power it can supply within the 25 - 45 kV range.

Also, using two supplies simultaneously is very difficult due to issues of impedance matching and ground loops. Be careful trying that with a fusor since resistance can jump all other the place and balancing load is, as a result, tricky.

Your 30 kV, 30 ma supply does appear to be a good unit for fusion (again, assuming it can handle close to that current in the 25+ kV range.)

I am curious to know what you mean by "direct conversion"?

Finally, your repair of those units is impressive and I am sure you will get good results once you start with your project.

Putting two supplies in parallel can be tricky. However if they are both the same make and model and each has its own ballast resistor (to help balance the load evenly) it can be done. You might also consider putting a diode in series with the output of each, but they would need to be 80KV PIV diodes and you would have a roughly 150V drop across each diode.

Dennis:
Yes; 70kx @ 8ma spec != empirical. As Ronald Reagan used to say; Trust But Verify. They are what they are, and I am not trying to save the world, so if they under-perform, I will have to accept it as fact or buy new supplies.

As for direct conversion, I am referring to directly converting energy from a fusion device into electric potential. Not suited for D-D, but for p+-B11 (and others). It is really the holly grail since it has high theoretical conversion efficiency 60-88% (depending on financial tolerance) and removes the water-steam-condensation cycle which is inherently inefficient, large, and while not overly complex, not as "solid state" as direct conversion. Here is a decent 2 page read. http://www.askmar.com/Fusion_files/Dire ... actors.pdf

Bob:
Thanks for the 80KV PIV diode idea; unidirectional semiconductor isolation sounds like a good idea and for 150v penalty seems smart. And yes I have ordered matching ballast resistors as my initial order was one of each, not its two of the target units. I am using Surplus Sales of Nevada, great selection of ballast resistors and very reasonable pricing. I will test 40kohms - 100kohms ranging from 80-200watts, big guys. Right now for testing the 60-65kohm range at 120watts seems fine for my 30kv setup.

Purchasing a connector is not looking so good.

I was ready to buy the $160 part but then on Friday they had cold feet. I think it was because I was the end user and paid all taxes.

So don’t hold your breath and get ready to make your own HV cable termination.


BTW I measured the filament waveform (no load) thinking it might be useful as a floating power supply for a mass flow controller or something.

The waveform looks like the output of a pulse transformer. You get a positive +20V pulse then some time later a -20V pulse. A faster pulse rate is a higher filament current.

-Peter

I was also thinking about the filament power supply. I would not use it for mass flow devices because they generally need 24vdc and those supplies are cheap and readily available.

However the filament supply is unique in the fact it is on the -70kV side of the supply, thus Spellman has performed the necessary steps to have a low voltage, high current supply that can operate at the -70kV potential. Thats non-trivial for most. So I was thinking of using the filament supply as intended; to drive a filament. Seems like a easy supply for the hot filament of an ion source. Assuming the ion source was implemented near or inside the cathode. I don't have any brilliant ideas on any such implementation at the moment, but have added it to my mental toolkit.

As for the cable; I just ordered one directly from Claymount (now Varex) for $170 +S&H+Tax. Yes, almost as much as the supply. I too got cold feet on ordering two, and grabbed some decent HV connectors and a nice cable from HVStuff to see if I have the skills to contain 70kV (likely not).

Andrew,

I attempted a HV cable and it arcs at about 28kV to the power supply case. Any chance I could buy a single conductor 2.5 ft cable assembly from you?

I don't have any complete cables for sale, but I do have a cable plug. You would have to pot in you own HV cable with an RTV silicone and buy the required pins. Send me an email at seltzman <at> wisc.edu

I might be willing to do a mass production run of cable plugs if there is interest.

I fabricated a physical controller for the Spellman supply, thought it would be a one day job, but took two. The main snags were;

1. The pin 11 - 12 that is labeled "X-Ray Enable" is the interlock (no big news there)
2. Pins 15 and 12 must also be connected but only after 11 and 12. So if you wire them to one switch as I did the power supply won't enable HV out.
The rest was straight forward.
Other notes;
3. If you want to get +10V from the Spellman to simplify your controller, you have to turn the tiny trim pots on the front of the supply all the way to the right, then you will have +10 V on pins 5 and 7

For real use I don't need this controller but for development and testing nothing beats a knob and switch. For automation reasons I will be using the rs232 and/or TCP for remote operation. During "Oh C4@p" moments I prefer not to have to fumble with a mouse and keyboard. I love slap-off switches and paddles as you can see. I had one 1.5kV supply go "fireworks" on me this week and I paddle-shutoffed the impending fire.

You can also see the controller I made for my 30kV Bertan Series 106 (smaller brown controller with one LCD) has been a good workhorse.

*A special thanks to Andrew and Dan Knapp for publishing every detail; it saved me a few hours I am sure.