I'm looking at playing with some SiGe parts for a close-to-DC application (< 1MHz). These things have scary high fT like 30-80 GHz.
Any tips for avoiding oscillation that I might not even be able to see? Ground plane, run at low current?
I'm looking at playing with some SiGe parts for a close-to-DC application (< 1MHz). These things have scary high fT like 30-80 GHz.
Any tips for avoiding oscillation that I might not even be able to see? Ground plane, run at low current?
Do some of the indirect tests for oscillation that amateur radio operators have learned to do...
You may be able to put Q-killing resistors in the relevant leads -- the idea is to put low-value resistors up close, to absorb energy in the return path.
It will, of course, knock your desired properties a bit, but if you've got the headroom you may as well buy some stability margin with it.
If I didn't do that, I'd want to analyze the layout with an eye toward transmission line behavior, if not flat design a little amplifier cell with all the right loading to be unconditionally stable.
To some extent you'll be gaining stability by using 'ordinary' FR-4, but that just means that some day a board house's 'improvement' will knock your circuit for a loop.
-- Tim Wescott Control system and signal processing consulting www.wescottdesign.com
As Tim wrote, resistors. Usually one 0402 or smaller resistor around
50ohms (whatever you have there) right at the base and in series with it suffices. A 0402 bypass cap from base to emitter also helps but you could get away without one.I don't know what you want to do with those. Not having the emitter on ground can get very spooky. The collector can be separated by a snippet of copper foil as a little "wall".
Most important is to breadboard on copperclad. I prefer living-bug style with the devices glued to plastic spacer pieces. This snippets of FR4, ice cram stick, wood shim pieces.
-- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
Argh. Can't do that, 50R would ruin everything.
Might have to go with that. There are very low values available.
Thanks, Joerg, Tim.
Ok, how about a very small SMT ferrite bead? Not sure if a real bead with a wire through it will work. It might if you drill a hole right in front of the base and sort of sink the bead 1/3rd into the board, lengthwise. Make sure the ferrite doesn't touch the ground plane anywhere.
You only need a few hundred pF and it doesn't have to be NP0.
We are here to serve :-)
Thanks for the hint about the X7S datasheet in the other thread, made my life easier. Drops the size of the caps from 2220 to 1812, meaning we can get more on the board.
-- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
What do you intend to do with these things?
I breadboarded some of the Infineon 45 GHz SiGe things, soldered down to copperclad. I was exploring how fast they would turn on as fairly low current switches. They seemed stable but were surprisingly slow.
PHEMTS are faster in real life. Their feedback capacitance is absurdly low so they tend to be stable. I have breadboarded phemt circuits on copperclad and got good, fast, clean switching.
EL07 driving an NE3508:
ftp://jjlarkin.lmi.net/BB_fast.JPG
This gave me a pretty good 5-volt p-p square wave at 1 GHz.
John
I'd be looking at beads, personally--some of those nice SiGe NPNs have
1-Hz voltage noise down at the 200 pV level, which is the Johnson noise of a 0.5 ohm resistor. The beads' noise doesn't really kick in till they start to look resistive.Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal ElectroOptical Innovations 55 Orchard Rd Briarcliff Manor NY 10510 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net
I'd be looking at beads, personally--some of those nice SiGe NPNs have
1-Hz voltage noise down at the 200 pV level, which is the Johnson noise of a 0.5 ohm resistor. The beads' noise doesn't really kick in till they start to look resistive.Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal ElectroOptical Innovations 55 Orchard Rd Briarcliff Manor NY 10510 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net
The ground plane may help.
Lossy things right at the legs may also help.
Low bias currents will help.
Lowish RF impedances will help.
Oh. OK. I'd sort of come to expect RF parts to be leaky and noisy at low frequencies, but that's apparently not the case, for bipolars at least. A BFT25 c-b junction makes an incredible diode.
I'd expect an 0603 to be OK. Like maybe 100 ohms at 100 MHz. Get it working and touch a few nearby points with a pencil. If you don't see any bias shifts, it's not oscillating.
How about this:
John
Amplify some relatively low frequency sinusoidal signals. Those parts can have some interesting noise characteristics.
The ferrite bead sounds like a good idea... hmm might have to work with the &*&$&*$ microscopic 0201 parts to get much Z at a few GHz.
Square wave @ 1Ghz. Ha.
Best regards, Spehro Pefhany
-- "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
Joerg wrote: : As Tim wrote, resistors. Usually one 0402 or smaller resistor around : 50ohms (whatever you have there) right at the base and in series with it
56ohm 0402 piggybag'd with Murata LQW15 22nH at base and collector works nicely.: I don't know what you want to do with those. Not having the emitter on : ground can get very spooky. The collector can be separated by a snippet : of copper foil as a little "wall".
Although it feels spooky, it is possible to make a long-tailed pair out of those in a stable manner, without spoiling the noise. I didn't need the fence, although I had to use one in the older BF862 FET-based amplifier.
Regards Mikko
The BF862 is a moped against some of the new BJTs such as a BFP620, it has an ft about 100 times higher :-)
-- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
But it has a 1 kHz 1/f corner, starting at a flatband value of 0.8 nV/sqrt(Hz), whereas the SiGe parts' corners are 100 kHz or higher.
Some of Mikko's noise data on those SiGe things are amazing--200 pV/sqrt(Hz)--but they kind of fall apart in the audio range.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal ElectroOptical Innovations 55 Orchard Rd Briarcliff Manor NY 10510 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net
I meant WRT the tendency to oscillate. They sure aren't so great to regulate and current-steer laser diodes. For that I'd use devices such as the BCX70K.
Yup, not that great for audio. Most devices aren't. I want my tubes back :-)
-- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
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.