That won't help here. I changed the collector load and it made no difference.
What I did is back off on base current, reducing it to 30uA. Not nice on a pulser but still within hfe range. It fixed the problem so I guess it really must have be something in the B-E path.
Very little current gain? Oh my -- you may want to verify that the thing works over temperature and part variations before you break out the champaign. No, wait -- you already know that.
Anyway, I'd certainly want to investigate the available space between "doesn't work" and that nasty saturation effect, to make sure that it's not just clinging by one fingernail to proper functioning, with a variation in either direction leading to disaster.
--
Tim Wescott
Wescott Design Services
http://www.wescottdesign.com
Yep. If you give it a 5-10x margin that will mostly suffice. This is the stuff we'll do in the fine tuning, setting component values to their sweet spots, or close. The first set was just to arrive at a functional system, which we now have.
Sure. I just received the first pulse-echo plots from my client and they look really great. Meaning we might never lay out and build the $20 deluxe version ...
The transistor won't go into saturation anymore because it's running out of steam. The baker clamp does the same thing but for a wide range of collector currents. For a test this was good enough.
On a 700psec pulser? That's like mounting a moped tire on a Porsche :-)
I know they work but I need zero point zero additional capacitance on the collector node. For the same reason I was fighting every femtofarad in the sampler. Anyhow, I got it to work by reducing the base current.
I've seen that. I was doing pretty well exactly the same thing--trying various BJTs, CMOS driven, trying to make a fast pulse.
If you measure trr of the b-e diode of this very fast transistor, you'll be amazed.
The solutions are two: one, use the speedup cap to deliver almost all the drive, preferably roughly matched to the transistor's need. If you make the cap too large, it won't be charged enough to rip the BJT off when you need it to.
2nd, a hot schottky c-to-b really wasn't all that bad. If you let the BFR92A saturate, turn-off is much slowed. The diode's small turn-on penalty was pretty small, and the turn-off improvement was huge.
A 2nd best method was ripping the base charge out with a 74ACxx driver, with a cap chosen to apply negative bias.
Don't put a lot of d.c. into the base once it's on--that makes it harder to turn off.
I tried a *bunch* of common R.F. transistors. They all behaved similarly. They're made to run linear--not for fast switching--but they'll switch fine if you bang 'em on & yank 'em off.
That's the problem. I ended up with a few (single-digit) pF.
In my case turn-off is not important. Actually, I want that to be as slow as possible because else it kicks out a pulse of opposite polarity. Can be calulated out but then the SW folks would throw stuff at me.
Yep, t'is exactly what I did. Just like you have to be very light-footed on the accelerator pedal if ice is encountered.
They are marvelous. Can be bought everywhere for pennies yet they switch in tens or (very few) hundreds of picoseconds.
Yes, this stuff is nearly useless once things are well above a GHz. BTW, we found out that sampler diodes can be very sensitive to ESD, like the princess on the pea. Look at them the wrong way ... *PHUT* ... gone.
As in--from my logged waveforms--108nS for an MPS571 (ft=8Ghz).
Huge, as in ~50:1.
I missed that. You could want something as simple as this:
+5v | 220 | __ 74AC00 +--->
-| '. |/ | o--||---+---| PN5179
-|__.' C1 | |>. 5p6 3k3 | | === ===
This registered a 240pS fall time on a 7S14 1GHz plug-in (350pS spec'd rise time), followed by a 22nS rise. That's >1000:1 rise/fall.
I was using jelly-bean transistors for a millions-qty consumer application, but you've got faster choices now.
You might speed it up by pre-biasing the base to, say Vbe/2. Or, for an inductive load, you could bias the BJT linear.
Running the drive into the emitter and operating the BJT common-base is another option. That's FAST (if you don't need current gain).
It might not be so bad if you *want* a slow off time.
The difficulty is that the same capacitor that supplies enough charge to turn the transistor "ON", does not store enough charge to remove that the same charge later PLUS the accumulated charge from the d.c. bias.
For a fast-on, long delay, then fast-off pulse, someone might use the q(rr) of an appropriate diode in series with the base-drive to correct that. Kind of an 'asymmetrical capacitor.'
That won't work well in my case becuase the turn off is too snappy.
Good idea but its only fast on paper. Mainly because then the path to ground includes a trace to the capacitor, driver, and on to the ground pin on the driver chip. Right now I've got around 100psec ramp time which suffice for this app.
Actually, if you keep the current really, really low, as in less than
50uA, it seems to be fine. I can't get it to zero but to around 400mV which is plenty. I only need to get it slightly below the base voltage level where current in the B-E junction stops and the transistor shuts down enough. Even if it have a residual 10% conduction it would be ok.
Yes, in another version I have Schottkies and stuff in there but this one has to be bare bones "every penny gets turned round and round".
Each of the samplers has two mercury batteries that back-bias the sampling diodes. They are a real bear to replace. I'm surprised they've lasted 20 years.
7S14 is remarkable in getting 2 GHz bandwidth using an avalanche transistor - no SRDs - to make the sampling pulses.
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Precision electronic instrumentation
I sort of doubt that -2.5V V_BE would cause any damage, though. You need some minimum momentum to move dislocations or interstitials. It would be interesting to measure beta degradation vs time at such low breakdown voltages.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
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.