[Possum]

I started assembly of a Gilmore PSU circuit and wanted to just test out the toroid PSU and diodes to see how much DC voltage I would get out of each of the secondaries. The toroid's specs are, 2 18V secondaries at 5 Amps each. The diodes are 600V, 8 Amp Hexfreds, with one full bridge for each of the transformer's secondaries. When I plugged the transformer into the AC outlet, the transformer got very hot after a few seconds, and so did the diodes. The 2 diodes that appear at the +VDC and -VDC, from the secondaries, got a lot hotter than the other two sets (which were going to the circuit's Ground.

Any ideas? Is it possible that I wired the primaries incorrectly? (2 primaries of 117V)

 

[Apheared]

Yikes!

Yes wire them in phase. It's a good thing. Your transformer datasheet will show the way...


Talema's are:
http://www.amveco.com/images/2x117v-50-60hz_diagram.gif
or for the mini/PC mount:
http://www.amveco.com/images/mini_drawings.jpg

 

[Possum]

Ok, I was following the diagram on my transformer's spec sheet, and I can use the first Amveco diagram to show what I did.

Referring to the FIRST Amveco picture:

Primaries: (hopefully wired in parallel)

1) brown and orange connected together, then to one of the AC plug's pins.
2) red and yellow connected together, then to the other AC plug pin


Secondaries:

1) green and red wires going to the "middle" of one diode bridge
2) brown and blue going to the "middle" of the other diode bridge

Any obvious faults? This is all I had connected when I first tries to test for DC voltage at the diode outputs. Then I attached a wire from Earth-ground to the circuit ground (between the 2 diode bridges).

 

[thomas]

I don't think you should connect earth ground to circuit ground...

Circuit ground is at 18V above earth potential, so i think you're causing a short... To my knowlege, earth ground is only connected to the chassis for safety, and should be isolated from signal ground, unless a ground loop breaker circuit is used...

 

[Apheared]

yep that's all good, but remember that on one of the bridge setups, the "negative" side is really gonna be the positive... or the least negative, anyway. So the spot between them becomes the virtual ground.

Wiring both "low" sides together will probably shoot a spark and dim the lights when it connects.

 

[Possum]

Yeah, checked the wiring now about 5 times and it's correct with all of the diode's anode-cathode directions and connections.

Problem solved:

Ah well, those outter diodes that got really hot very quickly are burned out. There goes $5.64.

Luckily I bought JUST enough extras to replace the burned out ones.

 

[Apheared]

[ah note to self: post new message when coming back to a thread.]

You edited that last response and I didn't know this was done... so what was it? Why did they fry in the first place? Any clue? I mean we kinda know why they did, but any ideas as to what event/when? Initial miswiring or something?

 

[Tomo]

Hello,

Now, please look at the I-V graph for P-N junction. (I know you have it, Pussum.)

Forward Biased Diodes's Id climb sky high at ~0.7V. Note also that slope of this climb is so huge you only see it as vertical climb. (But slope is NOT infinity ... Leave this to your calc textbooks.)

On the other hand, look at Extremely Reverse Biased PN junction at after BreakDown. Id start to climb up fast at the BreakDown voltage. Note this voltage is much higher than 0.7V.

Please note that power is related to voltage and current is the following form:

P = V*I.

Assume the slope for both forward biased and breakdown are approximately the same. (Yes, they have the same sign.) The slopes are huge and the difference is the slopes are insignificant to the magnitudes of the slopes themselves. So in this exercise, we will be safe to ignore the difference.

The diodes/rectifiers heat up a lot, correct? If you have build the PSU correctly, rectifiers should not be heating up this way cosidering the fact that your PSU current draw is respectably small around 1~5A max.

It is highly likely that your rectifiers are working heavily in reverse biased and breakdown region. Note breakdown voltage is for your diodes are about 12~15V. 18*1.414 = 25.5V is going to your diodes so your diodes will definitely be in breakdown region if you have connected diodes incorrectly.

You can check this by taking advantage of large-signal characteristics of diodes/rectifiers. When the diodes are on, voltage across the diodes is ~0.7V. Use multimeter and confirm this.

Tomo

 

[Possum]

Quote:

Initial miswiring or something?

*Hides*

Yes

I placed the secondaries at the bridge output for my very first test, then corrected it afterwards, but that first test is what blew them.

BTW, power supply is now done, fitting a 6"x~2.5"x12" extruded aluminum enclosure. I'm trying to figure out why my BUF634 ground driver in my amp isn't working right with this PSU. I'll have to recheck WIRING!

 

[Tomo]

Hey,

What is the current draw on the BUF634? I know ground driver does not work when current draw is large.

Tomo

 

[Possum]

The amp that's using the BUF634 for the ground driver is my Apheared #42 setup, except now it has 1 AD823 and 4 EL2001's. Back when it only had 1 AD823 and 2 EL2001's, my meter measured somewhere around 10 mA (add about 1.6 mA for the LED). That was for 18V's to the rails. Now there's 32 volts to the rails, so the LED is up to about 3 mA.

I got so frustrated with the BUF634 that I completely remade the BUF634 ground driver circuit only, with the resistor voltage divider feeding the Vin, and the external PSU's + and GND feeding the V+ and V- rails. Now, at the output of the buffer, between V+ and pseudo ground, there's ~8V measured; between pseudo ground and V-, there's ~24V measured. I don't understand this. I had this once before when dealing with the BUF634, but don't remember what happened (I went back to the TLE2426 after that).

 

[Tomo]

Hey,

10~20mA huh? Then it should work fine. Can you check those voltage dividers? Color codes are very unreliable. Personally, I always check resistance even if the resistors are coded so. Check voltage of the input of the buffer. This voltage should be equal to the output. Lastly, pseudo-GND should never ever touch chasis GND.

This is my opinion ... So you can ignore ...
-----------------------------------
Don't take this in the wrong way, but why do you need ground driver if you already have dual power supply? I understand your PSU is the design posted by Mr Gilmore. It should give you really accurate and constant power.

I totally understand and I will use GND driver for Battery amps. However, this is home amp and you can take luxury of having enough space for lots of pre-regulator and post-regulator capacitors. (Ok ok, caps cost a lot ...)

Perhaps you are trying to come up with a good way for isolating signal GND to power supply GND. Then this is a great start in your research ... (Then again, Mr Meier's GND Breaker is pretty darn good. ...) Besides, you are using a voltage divider to set the GND, this is not hardwired. So the reliability should not be that extraordinary.

But don't get me wrong, I like GND drivers since it saves a lot of space for my portable amp. I can't smash more than 2200uF's in my porta amp. It's just that I can use many caps to stabilize the GND for home amps and I think doing this as passively as possible is good idea according to the way we talk about regulators.

Tomo

 

[Apheared]

Tomo, hold tight man... you're mixing two of his things together... the dual rail supply is for his kgilmore, the external single supply is for his 823 2001... that's the one with the buf634 as the virt gnd driver...

Now, at the output of the buffer, between V+ and pseudo ground, there's ~8V measured; between pseudo ground and V-, there's ~24V measured. I don't understand this.

Oooh I did this once. Check that the buffer itself is getting power. You'd have to be way off with those resistors to make that much difference, but if the buffer's unpowered... anyway, check that.

 

[Possum]

Thanks for the help guys!

Quote:

Can you check those voltage dividers? Color codes are very unreliable. Personally, I always check resistance even if the resistors are coded so. Check voltage of the input of the buffer.

Yeah, I used a meter to check the resistors when I put them in, so I could match two to get the closest to the midpoint voltage.

Quote:

Lastly, pseudo-GND should never ever touch chasis GND.

Not touching here. All I had coming from the power supply are the two PSU rails.

Quote:

Don't take this in the wrong way, but why do you need ground driver if you already have dual power supply?

This may seem silly, but I have a neat little rocker switch for my amp's power, and it's only SPDT, so I was hoping to control power to the amp with just this switch. I did finally end up just wiring the rails and ground (thus, no ground driver) directly to the amp's circuit, with one of the rails going through the rocker switch, but I don't know if this switch will completely shut off the amp or partially disable it. The switch will only cut off the V- rail, but the V+ and ground are still intact. I guessed the amp would be off since there would be nothing going to the ICs' V- pins.

Quote:

Perhaps you are trying to come up with a good way for isolating signal GND to power supply GND. Then this is a great start in your research ... (Then again, Mr Meier's GND Breaker is pretty darn good. ...)

I'll definitely have to look into this. Right now, I'm getting a short random noise every once in a while through the amp (didn't happen with the old power supply). So I'm guessing it might have to do with the GND from the PSU circuit. I'm hoping that isolating the GND might help.

EDIT: Just checked Jan's Ground Loop Breaker and it answered my last two questions.

 

[CaptBubba]

If you want to keep your switch, you could use it to control a couple of small relays. Keeping the V+ rail on doesn't seem to me to be a great idea, perhaps because I just like my amps to be compleatly off when I turn thiem off.