[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Terry's DRSSTC - All Ready!!
- To: tesla@xxxxxxxxxx
- Subject: Terry's DRSSTC - All Ready!!
- From: "Tesla list" <tesla@xxxxxxxxxx>
- Date: Wed, 02 Mar 2005 17:22:38 -0700
- Delivered-to: email@example.com
- Delivered-to: firstname.lastname@example.org
- Old-return-path: <email@example.com>
- Resent-date: Wed, 2 Mar 2005 19:32:31 -0700 (MST)
- Resent-from: tesla@xxxxxxxxxx
- Resent-message-id: <RJaoQD.A.5vE.9cnJCB@poodle>
- Resent-sender: tesla-request@xxxxxxxxxx
Original poster: Terry Fritz <teslalist@xxxxxxxxxxxxxxxxxxxxxxx>
I finally think I have the DRSSTC all ready to go!!!
First some old issues...
At full power to the buss, I did note it now has a significant "sound" as
the bridge switches. Magneto-constriction I think. I did not get my ears
next to it since another explosion would be "bad" with an ear near
it. But something is far noisier now. Might just be a loose heat sink or
two parts clattering too. No big deal for now...
Hhahaha!!! - Funny!! It was the over range beeper on the Tek 5205
Differential probe. It must have been getting a little noise in the
circuit and every time the thing fired it chirped just like a ferrite
transformer clicking!! I guess Tek designed it to sound like that :o))
Note that the voltage driving square wave is super clean!! Almost "too"
clean... It started out clean with the last bridge too, but then got
noisy... Hmmmmmmm ?;-| I wonder if there was some damage to the old
bridge I did not detect... Gee whiz, I checked and double checked
everything... Think more on that... But clean is good for now ;-)
It "might" be getting a little worse now as I have been playing with
it. Even though the two 10nF caps should be taking most of the high di/dt,
there may still be some cap damage going on. I took Gerry's advice and
ordered up the 2.5uF metal foil caps poly buss caps and I will replace them
once the new caps arrive. That should fix that once and for all. Poly
buss caps across the bridge should certainly be the full metal foil
types!!! Not a big deal for now and it should be fine till the new caps
The current limit circuit still trips too soon. Probably noise on the
protection inputs due to these super fast switching times!!!!
This is the big gorilla I have been working on for the last month now!! I
finally fixed it :-)))))))))))))
So here is "how I fixed the over current protection noise" or "what I have
been doing the last month" ;-)
I did add a grounded shield to the CTs during the down time and "special"
caps to the protection circuit:
The shield works great and should be grounded the Chassis ground. This
stops most of the capacitive noise pickup back to the electronics.
Just a conductive tube between the noisy output wires off the H-bridge and
the CTs. This stops most of the capacitive noise pickup back to the
electronics. Solved a lot of noise problems right there!!
The "Magic Caps"... I found that a 1000 ohm pull up resistor could be
pulled down just from noise! The HC and complimentary driven stuff is rock
solid but be very careful of trusting "pullups"...
100 to 1000pF caps fixed that!! But in all cases with such bypass caps,
put them right from the IC ground pin to the pin you are protecting!! Very
close and directly grounded at the IC ground... The ground "buss" on the
card is pretty noisy when playing with 100mV signals. Also, 0.1uF bypass
caps are probably the bare minimum for most things. I would go to 1uF
This is a big one!!!
Drive the CT output at high voltage!!! None of that sub one volt
crap!! To find this, I hooked in a "Test CT" that was just like the real
that I could fiddle with. I could then pull all the differential probe
tricks and found this was how the signal looked (yellow) compared to the
Pearson probe (blue):
Isn't that just "special" :-P The scope was even averaging to reduce the
So Instead of using a 1 ohm burden resistor I used 10 ohms. 500A gives 50V
The signal cables just pick up too much noise even with only 1 ohm of
source impedance! The signal is just is a mess around the H-bridge. So if
you want accurate signals, don't try for 1V =100A. Step up to 10V=100A so
you can see the signal in all the noise.
Maintain the current loop!!
Let me explain that... Before I had a resistor directly across the CT and
I looked at the voltage across it over a long cable:
Problem is, all the voltage wiring and impedances pick up noise like made...
So I change it so the current loop went clear back to the ICs and only got
converted to voltage at the very last right next to the ICs. The "current
loop" is extremely immune to noise unlike voltage lines.
Now the signal and the Pearson compare like this :-)))))))))
Just about perfect!!!
So it still did not work... I had begun to worry a week or so back... As
I fiddled and mess with things, the level that the current trip activated
never changed much... So I tore through every node in the protection
circuit with differential probes (I am real good at that now...). It came
down to the LM339 was false tripping at 120 amps.
Here is the false trip output (red) with all the noise. The noise actually
gets better at higher current. You can see it trying hard and harder to
trip the LM555 timer on the peaks until it finally does...
Here is the LM339 comparator input:
The trip is at 80uS... but obviously there is 0.8V to go. It tripped at
120A but was set for 200A...
If I shorted the input it was fine but if a touched the input with a probe
it was "sensitive".... Whaa!! All the signals were perfect!! How could
it false trip??? If your really good... Here is the schematic:
The circuit is in the top left...
I am using the center tapped CT to get complimentary outputs monitored but
the LM339. The LM339 is ground referenced so I chop the negative part of
the signal with the 1N4148 diodes.... And there was the error!!!!!!!!!
The 1N4148's forward voltage is about 0.6 volts (-0.55 in my case). The
LM339's lower input voltage for ground operation is -0.3
volts... And.. They are NOT kidding!!! Since that was the only thing
that could be causing it, I changed the diodes to Schottky 1N5819 with Vf =
The clouds cleared, the birds sang, and the darn things stared working
perfectly!!!!!!!!!! Too much negative input to the LM339 messed it up and
made it false trip. The Schottkys fixed that right up!!!
So I set the current limit to 175A and gave it a try into a bare primary:
It's perfect!!!! So these five little fixes were all that was wrong
:o))) I will have to get a new protection card with the new fixes built in
it but, the kludged up old on is fine for now. The card is open and
unprotected but only Fix #2 seemed to have been caused by that. Even
analog stuff "can" be forced to work in there without a shield box. But
enclosing the stuff would not hurt. While running at 100 amps,
I did manage to short out the box's +5V!! It dropped the fans and
everything and came right back when the probe came away. Amazingly, that
did not blow the H-Bridge. The thing managed to live though the loss and
reaccusition of the +5V line under full power ;-)))
The final box looks like this:
The crossover delay is just under 400nS:
You can see the ringing is starting to come back so maybe the new caps are
going bad slowly. The caps are only 105v/uS rated. The new ones are
628V/uS and will not fail in the same area...
The ring down has just a sight delay as the anti-parallel diodes pick up
The transition dV/dT is here:
about 400V in 20nS (the voltage probe is across 1/2 the bridge) so
So after a long month fiddling with problems, It seems to be all together
now ;-)) It needs those new buss caps and the protection card redone, but
those are not critical to running it. I can also now clean up all the
schematics and parts lists...
The lab is a mess so I need to clean it up to run a coil. I think I can
take it to the big lab this weekend too!! So maybe it will finally put out
more that 6 inch sparks ;-)))