MEP805A regulator

[fredy2]

I traced out the circuitry of the Libby regulator. I may have found a design error. They rectify the power used to run the reference regulator and the unijunction pulser that pulses a pair of SCR's that generate the field current. This rectified supply is pre-regulated with what appears to be a 29 volt Zener diode. The supply doesn't appear to have any filtering. I suspect that the raw full wave rectified is impressed upon the regulation system. This is NOT good engineering practice. I will do some more testing. My regulator blew up the MOV varistor that is across the field. This causes major damage to the board and the voltage drops to around 40 volts instead of 208. Tracing the circuitry was the only way to understand how the regulator works. Essentially a the rate of pulsing the SCR's is done by a unijunction transistor which the timing capacitor is fed from a PNP transistor that is biased on by a pulldown resistor. A rectified sample from the metering source is or'ed into the base of the pnp transistor, turning it off if the sensed voltage is higher than the reference set by an LM723 regulator chip.

 

[Suprman]

Is this why a voltage spike kills the regulator? There is a mod that was supposed to be done to all the 30 and 60 kw gens it adds a fuse before the regulator.

 

[fredy2]

I had a one page reply set to send and just then our power failed... The gist of my reply was that the fuse added protected the field from being burnt up during failure of the regulator.
I see two design defects in the regulator. One is lack of filtering on the rectified source powering the reference regulator and Unijunction pulser that drives the SCR gates which in turn powers the field. The second is they have series diodes in the SCR gates which WOULD make them subject to transients. I suspect they intended these diodes to be reverse gate protection but due to the SLOPPY PCB layout they just got them connected up wrong. The PCB layout is a disaster! I am about one step away from redoing this board I am refurbishing to correct these mistakes.
If you have information of the failures you have seen, I would like to evaluate it against these deficiencies I have found. I wonder how many of these things there are out there?
Fred in Renton WA

 

[Suprman]

The 805 and 806 plus the 400hz variants of the same wattage. Lots of them. Some type of protective component or components that could be added before the regulator would be a nice upgrade. The military is working on phasing out this series of generator.

 

[fredy2]

The problem is poorly engineered gate drive to the SCR's that rectify the field voltage in a manner controlled by the regulating signal. Until I traced out the circuit I would not have seen the problem. I first have to design the gate drive resistors when I remove the series diodes to give the right drive current and move the diodes to between the gate and cathode to prevent reverse gate drive. I suspect that is what was intended originally but their sloppy board layout resulted in improper connection. What will happen with the diodes in there is the gate will float. This allows capacitive coupling between the SCR anode and gate to trigger the SCR when transients occur... The regulator will run away as a result. Also the supply to the regulator IC depends on only a Zener diode and this voltage will collapse as the AC reverses polarity. I will design for a capacitor to help keep this voltage change within bounds.
I have other obligations for about two weeks but I intend to make a video of the retrofit to correct the problem. It is hard for me to fathom how this problem could be permitted to continue without someone finding and analyzing and coming up with a fix. These regulators are hanging on by a thread.
FGred

 

[fredy2]

I just bench tested the regulator and found that they do let the reference regulator go out of regulation during each half cycle by design. The main cause of failure I believe is letting the gates of the SCR's to float with a high impedance due to the series diodes in the gate feed. By removing the diodes and changing the 10 ohm resistors to 15 ohms to compensate for removing the 0.7 volt drop of the diodes will keep the drive level about the same and hold the impedance of the gates low as it should be. It appears that this can simply be done by removing two 10 ohm resistors and two diodes and installing two 15 ohm resistors diagonally between one of the resistor pads and one of the diode pads at two places on the circuit board. The failure that happened on my board wiped out the 2N2647 unijunction (Q4), the 39 ohm resistor (R20), and the MOV Varistor. CR16, CR17, R18 and R19 will be removed and two new resistors 15 ohms will be installed. One will be from where CR17 cathode pad was to far pad of where R17 was, and the second 15 ohm resistor will be from where CR16 cathode pad was to the far pad where R18 was.
If your regulator fried the MOV varistor, likely the 39 ohm resistor at R20 and the unijunction at Q4 is blown. The traces to the MOV varistor on my board were destroyed and were repaired by hardwiring to turret terminals installed to support the new MOV which I used a V150-10.
I have not implemented the change or tested the repaired board on the generator yet.
Fred in Renton WA

 

[Suprman]

I think the boards are covered in conformal coating. Bench work on that board isn't the easiest thing to do. I have a good solder station and have done this type of repair before. Still a good chance a trace will lift on me. Don't the diodes serve a purpose though?

 

[fredy2]

Yes, covered in conformal coating... use lots of heat and heat from the component side and suck from the circuit side. I had only minimal problem de-soldering. Used a regulated iron at 479 degrees. I had to carve out cancer where the MOV charred the board.
The only possible thing the diodes would do is to lower the drive to the gates or try to prevent triggering from small pulses. By leaving the gates float with the diodes violates good design practice and leaves the SCR's vulnerable to dV/dT triggering and the regulator to go into runaway behavior.
Fred in Renton WA

 

[Guyfang]

Is this why a voltage spike kills the regulator? There is a mod that was supposed to be done to all the 30 and 60 kw gens it adds a fuse before the regulator.

The fuse comes between the AC volt regulator and the Exciter windings. The AC regulators are the problem. They were destroying the windings.

ALL TQG gen sets are supposed to be modified, to prevent this from happening. Like about 10-12 years ago? As we all can see, it didnt happen. This is but just one reason why I keep telling everyone to take the word RESET with a grain of salt. If you own a TQG, any TQG, and you haven't looked to see if the MWO is applied, shame on you! We have talked this problem into a hole.

The Libby regulator caused all the problems, at the beginning of the 15-60 KW TQG program. So The CECOM went to another Volt regulator, made by TRC. Fixed the problem. BUT, caused more problems. Because you had to not only change the regulator, but several other parts. You CAN NOT MIX their parts!!

Libby, AC Volt Adjust, Rheostat, 10K Ohms	5905-00-556-3350
Libby, Resistor, R16, 60 Hz. Volt Regulator	5905-00-106-9354
Libby, 60 Hz. Volt Regulator	6110-01-363-0494
TRC, AC Volt Adjust Rheostat, 20K Ohms	5905-00-800-6696
TRC, Resistor, R16, 60 Hz. Volt Regulator	5901-01-460-2205
TRC, 60 Hz. Volt Regulator	5905-01-384-7063
60 Hertz Volt Reg Kit	6110-01-477-4851

The kit is for the TRC parts kit. It has everything you need.

Then later, everyone discovered that ANY regulator could screw up the windings, if it went bad in an overvoltage situation.

 

[Suprman]

Well that answers why the regulator kits state you must use all components when changing the regulator. They should have put specifically why on the note.

 

[fredy2]

I don't have access to the TRC kit... so I will just correct the design defect in the Libby I have. Removing four parts and putting in two. A total of ten cents of parts. When I traced out the Libby circuit I could not believe they would make such a stupid mistake. It would cause the regulator to run away and without the protective fuse it would fry the exciter winding. I also have considered putting a relay that would break the field current if overvoltage was detected.

 

[Guyfang]

If you just install the MWO, its good enough. No need to install a relay. The fuse is in line with the Volt reg, to F-1. In the very early stages of this problem, some CECOM guys were telling us to put TWO fuses in, one between F-1 and the other between F-2.

 

[Suprman]

I think I have the 30k trc kit.

 

[fredy2]

I am surprised they didn't design the TRC to be a drop in replacement rather than requiring changing of the feedback scaling. I think these regulators were tinkered rather than engineered. In power equipment subject to electrical noise it is usually better to opt for lower impedance rather than high. I would like to see the schematic for the TRC to see if there are latent design flaws. It is a good thing they don't put designs like this in aircraft. How did any of these pass acceptance testing and inspection?

 

[Guyfang]

I think I have the 30k trc kit.

The kit fits the 30 or 60 KW gen set.

 

[Suprman]

I know I have the 60s. The 30s and 60s are marked as such. Are they interchangeable?

 

[fredy2]

I am re-working the Libby regulator and have taken the series diodes out and replaced the gate drive resistors with 15 ohm resistors. Also have discovered another weak point. Due to the sloppy board layout, the path for the 39 ohm pull down for the Unijunction drive to the SCR's was a big circuitous route making the SCR gates subject to spike type noise. This I fixed by a short jumper between the near end of the 39 ohm resistor to the cathode of Q1.
Ebay had a TRC kit but they wanted $500... I will fix mine after cleaning the board up for ten cents of resistors and a wire jumper, How did these generators ever pass acceptance by the military?
It will be a week before I can get thus installed back in the generator. Going to do bench testing first.
Fred in Renton WA

 

[DieselAddict]

Looking forward to seeing your test results.

 

[Guyfang]

I know I have the 60s. The 30s and 60s are marked as such. Are they interchangeable?

I humbly Beg Forgiveness!!!!!!!!!!!!!!!!!!!!!! Sorry, ten thousand times forgiveness. NO THEY ARE NOT!!!!!!!!!!!

I had opened up a file, for another gen set, and did not look at the heading.

30 KW
TRC volt reg. NSN: 6110-01-384-7192
Libby volt reg. NSN: 6110-01-374-0836
TRC Kit NSN: 6110-01-442-8622

60 KW
TRC volt reg NSN: 6110-01-384-7063
Libby volt reg NSN: 6110-01-363-0494
TRC Kit NSN 6110-01-477-4851

 

[fredy2]

Bench test shows same performance as the unmodified one in regards to the trigger pulses and regulation cutoff. The modification was easy. This should make the regulator far less sensitive to transients that could put it into run-away. I will still leave the 3 amp fuse in the field lead (terminal 1) as protection for the winding. The unit originally smoked the MOV Varistor as well as the traces to it. The symptoms when it failed were the voltage went up to the regulation setting, then continued upward to about 300 at which point the MOV and traces blew up. The voltage as read on the indicator then dropped to 50 volts and stayed there. I assume that was due to residual magnetism in the field. I have traced the complete schematic of the Libby regulator and basically they made a mistake of putting series diodes in the gate circuit of the SCR's. This allows transient spikes to couple to the gates from the anodes of the SCR's. Also the long circuit path from the SCR cathodes to the 39 ohm pull down resistor can result in random triggering and run away. Maybe they put those diodes in to try to get more noise margin... That would be a Band-Aid that would still not fix the basic problem.
I would like to get a schematic of that TRC fix... Hopefully it doesn't have latent mistakes.