The LT6656 is a 3-terminal device; will it work in a 2-terminal mode (ie: short input pin to output pin)? Adding an op-amp brings in too much complexity to allow a 2-terminal composite regulator.
Did i or did i not say "shunt regulator"?
Push comes to shove,i could characterize one of my Codatron(TM) HT-400V and subtract its voltage at the various currents. Requires logging the values (easy,my test program does that) of the "reference" HT-400V then later after logging low voltage parts (HT-50V thru HT-250V) placed in series with the HT-400V, write a program to look-up the reference values and subtract them out.
e pass element.. like this? (or a FET as others suggested?)
Ahh OK. Well as you found that is basically unavoidable. All diodes have shot noise. The avalanche process increases the the shot noise. In theory the lower voltage zeners (which operate via tunneling and not avalanche br eakdown) should have less noise.
The (low current) knee region is where it gets really bad (or good... I lik e to look at the excess noise as a feature and not a bug :^) where the dio de flops back and forth between being on and then off.
OK here's a crazy idea. One way to reduce the random on/off behavior is to shine some light (maybe an IR led?) on the zener. (you need a glass encap sulated package for this trick to work.) The light makes e-h pairs which ke ep the avalanche going and reduce the on/off noise.
(Of course it's no longer a two terminal device...)
Except, If you believe the Spice model, you're on the soft part of the knee. ...Jim Thompson
--
| James E.Thompson | mens |
| Analog Innovations | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| San Tan Valley, AZ 85142 Skype: skypeanalog | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
Grin, yeah I was thinking about that. The light trick really works to reduce the corner noise. I've got this circuit with a sine wave buried in the above zener noise. The way I check it is to stick an incandesecnt light over the zeners to turn them "off". ~40-60dB noise reduction (A guess, I'll measure a bett er number next time I test one.)
Hmm, I don't know much about radioactivity. There are a lot of photons per second in an IR LED. Switching times are ~1us or so... noise power should go as the size of the pulse (in Coulombs), so for -40dB I need an event ev ery ~10nS. (I think that's right.)
Okay, well, you should have titled the thread Low Current HV Shunt Regulato r. Low incremental resistance means high feedback loop gain regulating the output voltage. For typical reference voltages in the 1V range and an outpu t of 400V, this makes for an approximate 50dB feedback attenuation, so depe nding on what you call "low" incremental resistance, your error amplifier g ain may need to be in the 120dB range. Then your ultralow current coupled w ith the HV requirement pretty much eliminates MOSFETs as candidates, they h ave too much off state leakage. There are self-contained ICs meeting some o f your requirements like the LTC1541 (throw away the comparator) which can be coupled with fairly high gain BJT like the PMBTA45 500V from NPX (60 cen ts in onesies from DigiKey). Any more info than this is speculative because it's anyone's guess what the regulator is regulating.
--
"it's the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com
Use a regular TL431A. Pulse the needed 100uA current into the 431 by MOSFET or whatever for some micro second. When stable sample the voltage onto a c apacitor and turn the pulse off. Wait a looooong time and do it again. The average current is whatever the sample circuit dictates
That 4500V fet linked above has safe operating area and characteristic curves for ID vs. VDS at various gate voltage etc., not that I suggest using that one in particular. The ones that are not intended for linear operation don't have the safe operating area graphs, and can fail when operated continuously at high voltage whilst conducting current, due to the negative tempco of Vth causing current hogging or something.
I'm designing something similar to your shunt regulator at the moment, with a NMOS FET as the power device. I put a 3uF cap from drain to source to make it less susceptible to awkward loads. I also put a 100nF capacitor from drain to gate to provide fast negative feedback as well as providing Miller (or Blumlein?) compensation for the outer loop. I also drive the gate with a current from a transconductance amplifier, such as the first half of a LT1228, or for a bit more transconductance, the C terminal of an OPA861. I take voltage feedback from the drain via a high value resistor (with a small cap in parallel), to the B or + input of the transconductance amplifier. I program the output voltage with a current sink pulling current out of the low voltage end of the feedback resistor. (This is nicer than a resistive divider for feedback because it doesn't reduce the loop gain, since errors in the output voltage get back to the input of the transconductance amplifier volt-for-volt, until my protection diodes conduct.) Due to my accuracy requirements and the lousy input bias current of both of those transconductance amplifiers, I probably need to insert a fast, low input current buffer between the high value feedback resistor and the input of the transconductance amplifier. I think a unity gain buffer made from an AD8065 would do there, or you might get away with a discrete JFET follower if the output voltage tolerance allows. I also have to provide a stiff voltage reference for the E terminal of the OPA861, as I am running it from single supply rails, and the E input needs to be biased away from the negative rail. I haven't finished designing it yet but simulations are promising so far, though I don't trust TI's OPA861 spice model much further than I can throw it.
In case you haven't guessed, I am using a transconductance amplifier rather than an ordinary op-amp because the MOSFET I wish to use has many nanofarads of gate capacitance which will cause a fairly low-frequency pole with the open loop output resistance of any practical error amplifier, and I find it easier to make that be the dominant pole, which means that op-amps (with their internal 90 degree phase lag) are not much good.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.