Time for an ASCII drawing, to capture the circuit.
HV Regulator +800V | JL, Apr 2015 '--|| ||--, ---+--- R ---, | | ,-| / _|_ | |/ | | --- PD -> | | | | | |\v | | R gnd | | | | | | +--------+-- Rf ----------+ | | | | Ri +------' | | gnd OUT
--
Thanks,
- Win
To be fair, that's in saturated (well.. sort of) mode, and the linear mode bandwidth is better. A little bit, anyway.
A 4N35 can do ~300ns edges, if helped;
Still, I would be amazed if the darlington bettered ~5us edges.
My experience is, 4N35s are helped the most for around 47k B-E. For lower values, the CTR is much more nonlinear (low currents are shunted and CTR ~= 0, while high currents are more or less as lumpy as normal), and switching performance slows back down. Above 100k, it's slower due to excess storage (as usual).
I want to say, 1nF is on par with, if not the junction capacitance, then either the Miller effect equivalent that the phototransistor sees, or part of the equivalent circuit of the photocurrent + diffusion model itself (commonly SPICE'd as a separate RC element). So, the MOSFET will slow down the opto, more than the opto will slow down the MOSFET, I guess is one way to put that!
Tim
-- Seven Transistor Labs Electrical Engineering Consultation Website: http://seventransistorlabs.com
That's an interesting circuit.
Thanks!
Not so much: The opto's collector load in this cascode is 1/gm of the MOSFET, which I estimate to be about 25 ohms at 10mA. That's in parallel with the capacitances. So you could argue, the MOSFET will speed up the opto.
--
Thanks,
- Win
I might need as much as 1 mA, if the customer pulses the load really hard.
-- John Larkin Highland Technology, Inc picosecond timing laser drivers and controllers jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
I prefer Sharpie Art.
There might be a hazard if the +800 rail shorted to ground. But don't do that!
-- John Larkin Highland Technology, Inc picosecond timing laser drivers and controllers jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
The 1N823 is a hybrid IC, with two dice (the zener, and a diode) packaged together. So, it oughtn't be symbolized as a Zener diode; it doesn't test like one, either. SPICE ought to recognize it as a subcircuit...
ymbol, but I can't put in 1N829 as the component value
ard-
ch you use the collector-base diode as the forward diode, and the reverse b ased emitter-base diode as the zener. Transistors aren't integrated circuit s.
What's your evidence for this implausible claim? Have you taken one apart? Or are you just guessing? There are lots of ways of getting semiconductor j unctions in series that aren't usually described as "creating integrated ci rcuits". SCR's are examples of four-layer NPNP structures - more complex th an the three layer structure required for the reference diode function - an d nobody calls them "integrated circuits", while the cheaper ones definitel y aren't any kind of hybrid.
Farnell stocks the 1N823 for $A 3.27, which would be pretty cheap for a hyb rid.
Why? A zener looks like a 0.7V drop forward diode if you reverse the curren t direction, while the 1N823 won't conduct current in the "wrong" direction until you break down the compensation diode. The 1N823 data sheet doesn't say a word about the reverse breakdown voltage of the compensation diode.
To that extent you are right - using a zener symbol for a 1N823-style refer ence diode is misleading. It's less misleading than using a simple diode sy mbol, and more informative than using an arbitrary sub-circuit block.
There are circuits where a 1N823-9 would work where a zener diode wouldn't
- in the circuit you referred to earlier, with the op-amp boot-strapped to the reference diode, a zener-based circuit has two stable states with the z ener either reversed biased or forward biassed. The undesired state is easy enough to design out, but you don't have to bother if you are using a 1N82
3-9.The ring-of two circuit that I sketched doesn't have this problem, so using the zener symbol isn't going to create any difficulties. Half-wits who nev er bother to read the data sheet might run into trouble, but no amount of c areful representation is going to stop them from screwing up.
-- Bill Sloman, Sydney
ed
? Or are you just guessing?
Guessing. The 1N823 is in an axial glass package, where one face of the d ie is the anode and the opposite face the cathode. So unless some extraordinary two-sided wafer di ffusion was employed, it has to be two dice, stacked. You can't use that package exce pt for face/substrate diodes. The Motorola data sheet says forward breakdown is '
15V to 400V' which also doesn't suggest a monolithic IC-like design.ren't usually described as >"creating integrated circuits". SCR's are examp les of four-layer NPNP structures - more complex than the three layer struc ture required for the reference diode function - and nobody calls them "int egrated circuits", while the cheaper ones definitely aren't any kind of hyb rid.
ybrid.
Not compared to optoisolators and stacked-pellet HV rectifiers, which are u nder $1.
aged
rt? Or are you just guessing?
die > is the anode and the opposite face the cathode. So unless some extr aordinary > two-sided wafer diffusion was employed, it has to be two dice, stacked. You > can't use that package except for face/substrate diodes.
It does sound more like two devices stacked, but again, nobody would call t hat a "hybrid" - a bit of gold on each of the mating faces, and you'd get d iffusion bonding while the glass package was being fused.
The term "hybrid" is usually restricted to assemblies that take a bit more assembly.
It does sound more like "any small forward diode that happens to have been overproduced", but that kind of open specification is more like a declarati on that they aren't committing themselves to any particular process than a revelation of what they they were doing when they wrote the specification.
aren't usually described as "creating integrated circuits". SCR's are exam ples of four-layer NPNP structures - more complex than the three layer stru cture required for the reference diode function - and nobody calls them "in tegrated circuits", while the cheaper ones definitely aren't any kind of hy brid.
hybrid.
under $1.
Cheap for a low-volume hybrid, then. For sufficiently high volumes you can automate quite complex assembly operations.
-- Bill Sloman, Sydney
According to page 5-4 of the 1967 Motorola Zener Diode Handbook reference diodes are assembled by stacking dice before packaging. More modern production may differ however.
piglet
Significantly less than 1 PPM is sometimes quite measurable if it is periodic ripple on the tuning voltage (or, in this case more likely the operating current) of a microwave LC VCO. Of course you measure it with a spectrum analyser on the output.
My little LED-transistor current source will have lots of noise from thermal effects, PSRR, and Johnson noise in the emitter resistor and the transistor Rbb. It's not a super-precise low-noise current source. A couple more PPB from room lighting won't matter.
1 millikelvin temperature mismatch between the transistor and the LED will change the current roughly 1 PPM.Most electronics runs in the dark anyhow.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
At 100 nA photocurrent, it's a part in 10**4 of the forward current, if you're running at 1 mA. That's good for a couple of microvolts' noise.
Cheers
Phil Hobbs
That would put the photo-noise in the output current in the 1 PPM ballpark.
My original app was to make a very fast ramp by charging a small cap, hence the RF PNP transistor. Its Rbb Johnson noise in my working bandwidth is going to be well over 10 uV, which gets integrated by the cap, so it's complex. Bottom line, what dominates is the TC and PSRR, not the noise.
Anybody who wants a precise current source should not use this one!!
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
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.