Log sweeping a current with SPICE.

That looks very good to me, Fred.

It did have a CE sticker. Someone cheated.

Try running it. It generates one curve per variable chosen.

Cheers

Phil Hobbs

st

is not the same as sweeping a source. For that he wants. .dc dec I1 100p 10

The question is what is the curve.

I think he's doing this:

Those upper case F and R on the beta should mean DC current transfers. So h e wants to logarithmically sweep the IC. For a smooth graphic, the .DC swee p on IC allows for 100 points per decade or more. Then he does the arithmen tic on his IC/IB ratio to plot his DC BF or BR versus IC, as the case may b e, to get what he's looking for. Try doing that with STEP.

If there's some component parameter sensitivity he's looking for, dunno lik e doping density, then he can step the above DC sweeps and plot his Beta co mputation through that parameter.

So run it and see.

Cheers

Phil Hobbs

I thought you had sold that place and moved somewhere far from San Francisco. Have you moved yet?

We moved, but less than a mile, so we're still about the same distance from Mt Sutro. The EMI is if anything worse.

We use a diff amp circuit similar to the one in AoE2 fig 5.80, that uses an LT1124 dual opamp. That is a great amp, but it rectifies ambient RF so has resulting quirky DC offsets. We just hacked one channel to use an ADA4522-2, an EMI-hardened, zero-drift dual opamp. The EMI sensitivity dropped at least 100:1.

What size disk? Does someone make a usb replacement? SRS found one, but you have to let it format your usb for 720 kB or some such number.

George H.

Amazing gizmo though!

Cheers

Phil Hobbs

I rarely have the need for measuring semiconductor parametrics, but at one point I had to, so:

I used a function generator and 3 HP34401A DMMs, controlled them with LabWIndows SW and fed it into an Excel sheet

Takes a long time, and needed to add pulse loading only for higher currents, but cost me nothing, except time coding

Cheers

Klaus

I do a lot of DC measurements, for things that are not specified for RF parts. Like transfer curves, drain curves, Rds-on, C-V curves. RF part data sheets are criminal.

I usually do it all manually and draw the graph on paper as the measurements are made. You usually only need a few points to see what's happening.

Or you can let a scope do the graphing for you:

I need the transfer curve of Mini-Circuits SAV-551; I'll do that today. Of course there is no Spice model.

You'd save a lot of time and improve your analysis, if you had a proper two-channel SMU. My 2602A cost $4880 on eBay,* and is a live-saver. It works up to 10A in pulsed mode.

Using its LAN connection, you run it from your computer and jam the data into a spreadsheet. There you can not only graph your data but process it in various ways to learn more about your part. Yes you could get a 2401 for $3k, but you do need two channels to do anything serious. The name of the game is flexibility, speed and efficacy.

• I also have a 2612A that goes to +/-200 volts.

We have a bunch of Keysight SMUs. My people often use them, with Python programs, to test our gear. I can get the SAV-551 transfer curve in maybe 1/20 of the time it would take them to automate the test and play with spreadsheets and stuff. I don't need publishing-quality graphs (until I write my own book!)

Which all argue for doing this one by hand in 15 minutes. I only need a few points to get the transfer curve, and there's no processing to be done. Actually I only need a single data point.

I could use a few points to make a Spice model, but all I'm doing is making fast logic gates, so I just need a rough curve. The bigger concern is how consistent the parts will be.

You didn't look hard enough...:-)

• At this point we have a super-wide-range log sweep of base current.

Oh dear....

Sure, I added that to SuperSpice years ago. The DC sweep setup form has a dec options. It's a proper points per decade sweep of the volt or current source

There is also a hfe example that autoplote Ic/Ib

HFETester.sss

After the run, change the graph to a log scale. It will remain that way after setting.

-- Kevin Aylward

st

is not the same as sweeping a source. For that he wants. .dc dec I1 100p 10

That's some illogical, unspecified and possibly unreliable program default. You're just hacking around when you use stuff like that.

Looks like the gate voltage will be near 0V, why not return your resistor to a negative voltage, then the exact gate voltage won't matter.

Thanks Kevin, I didn't want to use LTSpice.

I am close to designing my gadget with no negative power supply. The nand gate thing is easy if a negative voltage is available. What I really need is a 0.75 volt bandgap to put in the source. Maybe a silicon diode would work. A c-load opamp could drive the source to any voltage, but that's not sporting.

I want the fastest possible nand gate, and the available tiny-logic things are slow.

It's just an interesting puzzle.

I assume you still have a password file. If not, email me and I will send you one.

Did you ever check out the additions I did to my version of the VDMOS model ?

-- Kevin Aylward