David Monaghan wrote...
A nice effort so far, but your site needs work: there are missing
web-page drawings (empty boxes show), broken links, no description,
e.g., question: no constant I mode?
There are 14 decades of constant I control, with the max current of each
decade (FSD on I meter) set with VR3. (See Current Sense circuit PP50a)
VR3 can be disabled by removing JP3, and constant current/current limit can
be
controlled from zero to FSD of each decade with a 5k pot on the front
panel.
Your drawings need work: unlabeled connectors, unlabeled switches,
missing dots (or too small), etc. 5V ref drawing missing entirely.
Yes. You have to zoom it to see the connection dots. The 5V
reference on my prototype ended up being the 1ppmDegC Maxim 6350
And your design needs work (and/or the drawing is in error?),
e.g., U1b and U2b fight each other during current limit events
(LMC662 opamp's have identical source/sink specs)?
The LMC662 sinks 21mA and sources 22mA. Say the load in dynamic, and
goes lower resistance, and the consequent increase in load current
exceeds the Constant Current/Limit point. U2b output goes sufficiently
low to limit the DUT current (by dropping gate voltage), U1b output(gate
drive) goes high(but
now has no effect on the gate), and the Red LED (current control) goes on.
It seems to work quite well.
U2a drives
the FET's source, and U1b the gate (so the FET's gate-charge
current goes through U1b and U2a)?
Yes. Is there a problem with this? It may be easier to think of U2a
functioning to position the 12v rails in relation to the FET source so that
the fixed voltage (in relation to the 12V rails) at the output of U1a
is precisely at the same level as V+out on the DUT.
So, saying that again.... As V+out varies, then so do the 12V rails.
The importance of this will be understood when the instrument is set
to zero source voltage, and used as a femtoamp meter ...for DUTs that
generate their own current. Note also that this is a zero impedence
femtoamp meter when used this way. Have you seen the projects
that this Power Supply facilitated?
http://users.bigpond.com/dmon2001/PS/project03.htm
http://users.bigpond.com/dmon2001/PS/project05.htm
500V is obtained across 64k
Yes some confusion there.
This particular circuit is limited
to 64 volts output. It is probably a more practical range. Another
three front panel microswitches, and you get to 500volts.
Only 1mV signal across critical spots at 1pA?
???????
What's
the voltage across the 1k trim pots (the divider resistor values
seem wrong)?
The 1K trim pots are to assure that the voltage at +input of U1b (ie top
of the resistor column) is the same as V+out...is the same as the
Guard Voltage on the output of U1a. What I will do on the site
is to make a simplified circuit of the principle of operation of the
PS. It seems so simple to start with...then you have to take into
account
input offsets and all sorts of protection. I had to blow a few chips
along the way to learn what precautions to take. The PS will now
take inductive kickbacks from running and switching on and off
DC motors.
Hmm, I don't like the look of the 0.1uF across the
BUZ350 MOSFET, all set to force any HV supply noise onto fragile
DUTs. And why is the 100-ohm damping resistor a 3W part?
Ah....I wondered that too. But does not the series RC serve to
dampen any inherent tendency for the BUZ350 to oscillate?
I thought it was common practice.
And why is the 100-ohm damping resistor a 3W part?
Because once while prototyping, I was amazed to have a 1/2 watt
100 ohm resistor burn out. I understand your point. It is probably only
necessary until one gets the circuit right.
It has been my experience that the challenge of this type of circuit is to
assure its stability against oscillation.
What
happens to the floating 500V supply's AC main leakage currents,
when you're making pA measurements?
I have only ever floated the DUT along with the 500V(to take that example)
rails. Thus the DUT is tied to the Earthed negative rail and the same
earth and neg rail is attached to a faraday cage around the DUT. For small
current measurements,
the result will be messy without a faraday cage enclosing the experiment.
12V supply leakage does not matter, because the 12V rail is locked onto
1mA pulling it negative, and U2a output to FET source pulling it positive.
The 1mA constant current column is made up in part of any current leakage
through the FET during small current measurements.
Ditto for the 12V supply's
leakage currents? Floating I and V meters are awkward, hard to
power, impractical to get any analog outputs, etc.
They work well in practise here, although protection needs to be built
into
the Current Meter which is actually calibrated as a 0-1 volt meter.
Say what,
500V max, and *no* safety interlock? Is there a separate safety
voltage limit to protect parts from inadvertent knob adjustments?
A parallel protective device would be handy. I remember once, my boss
imported an expensive X-Ray tube from Italy. He tested the filament out
nonchalantly with
one of my supplies. I was sweating. A badly soldered joint, or a
intermittant pot, and the tube would be history.
But for a supply like this, one could adapt it for a purpose, rather than
build it to do everything.
How about a soft start for the voltage source? What's the green
LED for?
The green LED indicates VOLTAGE CONTROL. The interesting thing is that
if both the Red and Green LEDS dont light, it indicates that their is
insufficient power to regulate either.
That's OK, keep trying - I'll stick with my trusty Keithley 487
for now. When they say 500V they mean 500V, and more impressive,
when they say 10 femptoamps, they mean 10 femptoamps. :>)
http://www.keithley.com/servlet/Data?id=380
And when they say $3000 they mean $3000
Thanks Winfield for taking the trouble of checking the circuit out.
I agree with you...there is a long way to go.
David Monaghan
