C-multipliers and simple circuit to measure npn beta.

Hi all, We make this low noise power supply. (~ +/- 12V up to ~100mA) A voltage regulator followed by a cap-multiplier. The first bunch I made had an output noise of ~>1.5 nV/rtHz. So I wrote a spec of 2nV/rtHz. Then a new batch of transistors (2N4401) and the noise was worse... ~3.5 nV/rtHz. I had a bag of the older npn's and just put those in. Now I've run low and ordered a bunch of them (200) from several different suppliers. These are in now and some of them test just fine. (transistors from the same batch seem to be very similar) And I was wondering if the difference in noise is related to a difference in Beta. (Can I do a DC measurement and ID the good ones?) So I'm looking for a simple circuit to measure transistor beta. This looks like a good start,

formatting link

The noise is highest at the highest currents, so I'd like the current source to ~100 mA. I guess that's simple enough.

Other ideas? Thoughts? Thanks.

George H.

Reply to
George Herold
Loading thread data ...

Something like this?

formatting link

GH

Reply to
George Herold

Low beta will make higher Ib, and Ib has shot noise. Ib noise makes voltage noise in whatever impedance the base sees. I guess Ib never makes it to the collector, either.

Internal base resistance Rbb directly makes Johnson voltage noise, and Rbb could vary a lot in random generic 2N4401s. See Win's chart on p

501 of AoE3. My favorite gumdrop, BCX70, is the worst in his list with Rb of 760 ohms. That makes about 3.6 nV/rtHz, but doesn't matter if I'm just turning LEDs or relays on and off. 2N4401 is in Win's table, at 40 ohms Rbb. But your transistors might have come from anywhere. ZTX718 looks good.
--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

Way overkill.

That makes no current in the transistor.

formatting link

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

You also have to worry about the extrinsic base resistance, Rbb'. I use a lot of 2SD2704K's for cap multipliers of about that power level, but they only come in SOT23. AOE3 lists a bunch of low noise BJTs on P.

501. You might look at the ZXT851, which is a fave of theirs.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

This should be simpler:

formatting link

Set the current to 100mA and hang a meter into the base connection, set to current measurement. For example, one of the $5 Harborfreight multimeters.

Running these at 100mA in linear mode is "pedal to the metal".

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

It is easiest just to buy a cheap DVM which can measure betas. But there is a quick and simple way, two resistors, a battery, ohms law about currents and a DVM. Choose resistors so that you get suitable voltages. Version 4 SHEET 1 880 680 WIRE 272 16 144 16 WIRE 368 16 272 16 WIRE 144 48 144 16 WIRE 272 48 272 16 WIRE 368 96 368 16 WIRE 272 192 272 128 WIRE 144 240 144 128 WIRE 208 240 144 240 WIRE 272 304 272 288 WIRE 368 304 368 176 WIRE 368 304 272 304 WIRE 272 336 272 304 FLAG 272 336 0 SYMBOL res 128 32 R0 SYMATTR InstName R1 SYMATTR Value 1000000 SYMBOL res 256 32 R0 SYMATTR InstName R2 SYMATTR Value 1000 SYMBOL npn 208 192 R0 SYMATTR InstName Q1 SYMBOL voltage 368 80 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V1 SYMATTR Value 9V

Reply to
LM

Wideband transistors have low base resistance. In normal use use you stick a 27R resistor close to the base to stop them oscillating, but the right inductor might do the job.

--
Bill Sloman, Sydney
Reply to
Bill Sloman

OK, I forgot to say that the noise gets higher with temperature. (current) I test 'em with 100 ohm Al body power R's 25W. The power R's sit in the same shielded box with the front end opamps and gain R's. The ~2 watts of power heats everyone up... guess what, the 'amplifier noise' goes up too. :^)

Driving home I was thinking I should just change the noise spec. in the manual/ website. And send some newsletter/ circuit, showing how to measure the power supply noise... then that can go into your calc. of the total noise. It mostly 'could' only matter when driving the incandescent light bulb... and then at the highest currents vibration noise in the bulb filament overwhelms the power supply noise. (at some frequencies :^)

Noise is fun...

George H.

Reply to
George Herold

Oh dear, The Vref is to gnd. I could float things on the input end. I'll have to try it now.. well tomorrow.

I don't want so much voltage drop across the transistor.

100 mA is a lot for the to-92 pac.

George H.

Reply to
George Herold

Yeah, So this to-92 pac transistor is also the 'fuse' if kids pull too much. (there's a ~300 mA current limit up stream.) (You know, I need to think more about protecting the b-e junction.) We send extra's and they come in phoenix terminal blocks (on the pcb)

George H.

Reply to
George Herold

In the C-mult. there's not much voltage drop across 'em, I see some small heating effect, but they are not hot to the touch, (Even with my lips. :^)

George H.

Reply to
George Herold

I didn't look at your LTspice circuit. But I want to look at different collector/emitter currents.. 1-100 mA say.

Oh I want to say I love all the different ideas! (I'll look tomorrow at work) Thanks.

George H.

Reply to
George Herold

George, forget beta and the other parameters. It's all about r_bb'. That's hard to measure, except through noise measurements, which is why we did for you. We were not impressed with the 2n4401 series for low noise.

We found that specific types, from well-known manufacturers, maintained similar performance, over decades of production. So pick a part, and don't change anything. I suggest one of the Zetez / Diode, Inc. parts from our table. Then you can give a low noise spec and keep it.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

The emitter resistor sets the current. I didn't put a value on that because I didn't know the desired operating point.

The +5 can be anything you like.

The 33R just keeps the transistor from oscillating.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

use +1V instead of +5, that'd be more representative of the operating area of a cap multiplier.

--
  When I tried casting out nines I made a hash of it.
Reply to
Jasen Betts

The thing that hacks me off is the photodiode equivalent of Rbb', i.e. the sheet resistance of the epi layer. Your average 0.1-inch square photodiode has about 60 ohms' series resistance, and some are much worse. (I'm looking at you, Vishay.)

That noise gets differentiated by the diode capacitance just like the TIA's input noise voltage, and is often the limiting factor in low-light measurements.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Hi Win, OK thanks... I'll have to buy some others and test for myself. I will say that the other npn we had in stock was the 2n3904 (r_bb = 110 ohms), and it was worse (noise-wise) compared to the 2n4401. So that's encouraging.

Hey, did you ever measure the TIP31/32? I use that in another low noise supply, and it's nice... Well that supply has a quieter voltage reference feeding the C-mult so that helps.

George H.

Reply to
George Herold

Huh, Well I've never run into that. I hardly know what r_bb is. (I'm going to re-read sec 8.3 in AoE3...)

George H.

Reply to
George Herold

If you want a quiet and power, wire the transistor as a Sziklai, with a quiet type as the input part.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

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.