Fast monostable with a transistpor array

"Bill Sloman" snipped-for-privacy@ieee.org wrote in message news:v4f2nm$2augj$ snipped-for-privacy@dont-email.me...

I put the NPN model parameters into a circuit designed by yourself. It's late now but I got as far as Time step too small.

Version 4 SHEET 1 2848 1304 WIRE -592 528 -752 528 WIRE -240 528 -592 528 WIRE -64 528 -240 528 WIRE 32 528 -64 528 WIRE -64 544 -64 528 WIRE 32 560 32 528 WIRE -592 576 -592 528 WIRE -240 592 -240 528 WIRE -752 624 -752 528 WIRE -64 656 -64 624 WIRE -592 704 -592 656 WIRE -464 704 -592 704 WIRE -384 704 -464 704 WIRE -240 704 -240 672 WIRE -240 704 -320 704 WIRE -128 704 -240 704 WIRE -464 736 -464 704 WIRE -240 736 -240 704 WIRE -752 784 -752 704 WIRE -64 784 -64 752 WIRE 32 784 32 640 WIRE 32 784 -64 784 WIRE -592 832 -592 704 WIRE 32 832 32 784 WIRE -512 880 -528 880 WIRE -464 880 -464 816 WIRE -464 880 -512 880 WIRE -448 880 -464 880 WIRE -384 880 -320 704 WIRE -320 880 -384 704 WIRE -240 880 -240 816 WIRE -240 880 -256 880 WIRE -208 880 -240 880 WIRE -32 880 -208 880 WIRE -208 912 -208 880 WIRE -512 928 -512 880 WIRE -592 1008 -592 928 WIRE -512 1008 -512 992 WIRE -512 1008 -592 1008 WIRE -208 1008 -208 976 WIRE 32 1008 32 928 WIRE 32 1008 -208 1008 WIRE -592 1040 -592 1008 WIRE 32 1040 32 1008 FLAG 32 1040 0 FLAG -592 1040 0 FLAG -752 784 0 SYMBOL RES -608 560 R0 SYMATTR InstName R1 SYMATTR Value 1Meg SYMBOL res 16 544 R0 SYMATTR InstName R10 SYMATTR Value 2Meg SYMBOL cap -384 864 R90 WINDOW 0 0 32 VBottom 2 WINDOW 3 32 32 VTop 2 SYMATTR InstName C1 SYMATTR Value 1.5p SYMBOL cap -256 864 R90 WINDOW 0 0 32 VBottom 2 WINDOW 3 32 32 VTop 2 SYMATTR InstName C2 SYMATTR Value 1.5p SYMBOL VOLTAGE -752 608 R0 SYMATTR InstName V1 SYMATTR Value 5 SYMBOL npn -32 832 R0 SYMATTR InstName Q1 SYMATTR Value NUHFARRY SYMBOL npn -528 832 M0 SYMATTR InstName Q2 SYMATTR Value NUHFARRY SYMBOL res -256 720 R0 SYMATTR InstName R3 SYMATTR Value 4.7Meg SYMBOL res -480 720 R0 SYMATTR InstName R4 SYMATTR Value 4.701Meg SYMBOL cap -528 928 R0 SYMATTR InstName C3 SYMATTR Value 0.68p SYMBOL cap -224 912 R0 SYMATTR InstName C4 SYMATTR Value 0.68p SYMBOL npn -128 656 R0 SYMATTR InstName Q10 SYMATTR Value NUHFARRY SYMBOL RES -256 576 R0 WINDOW 3 30 76 Left 2 SYMATTR Value 1Meg SYMATTR InstName R2 SYMBOL RES -48 640 R180 WINDOW 0 36 76 Left 2 WINDOW 3 36 40 Left 2 SYMATTR InstName R11 SYMATTR Value 1Meg TEXT -768 1176 Left 0 !.tran 0 150u 50u startup TEXT -776 1112 Left 0 ;This example schematic is supplied for informational/educational purposes only.\nJuly 7, 2008 by Bill Sloman (but blame the Q10 output stage on James Arthur). TEXT -480 1176 Left 0 !.model NUHFARRY NPN\n+ (IS = 1.840E - 16 XTI = 3.000E + 00 EG = 1.110E + 00 VAF = 7.200E + 01\n+ VAR = 4.500E

• 00 BF = 1.036E + 02 ISE = 1.686E - 19 NE = 1.400E + 00\n+ IKF = 5.400E - 02 XTB = 0.000E + 00 BR = 1.000E + 01 ISC = 1.605E -

Eight additional line breaks will need to be removed.

That is neither fast nor a monostable.

There's no reason to use many-GHz parts in micropower circuits. One reason is that the sim breaks.

I do wish people would comment Spice sims properly, with author, date, and what-the-hell-this-actually-is-supposed-to-do.

I wonder what happened to James. I think all the flaming here drove him away. James was fun. He always skis in short pants.

Yes ok it's an astable and certainly not intended to be fast. It was intended only to see if the model worked. Nothing more.

It doesn't work!

Most versions of the HFA3046 are obsolete. The only one left at Digikey costs $12. It would be risky to design around that part. Wasn't HFA originally a Harris part?

There are lots of dual (non-monolithic) diodes and bipolars and fets around, but the monolithic arrays are mostly gone.

The duals save space, but aren't well matched so are electrically and thermally equivalent to buying two separate parts.

I wonder if one could design really fast stuff with the HFA3046. Package parasitics look nasty. And the big package will force ugly PCB routing. You can plop a single tiny transistor wherever you want.

The HFA parts are either 5 NPNs or 5 PNPs, which is also awkward.

If it's obsolete or going obsolete then yes it would make no sense to use it. I would guess that you get to do plenty of work because a manufacturer has suddenly decided to discontinue a part.

What part would you recommend (single or dual) and does it have a working LTSpice model?

Yes that is certainly an advantage of single transistors. Is a suitable single transistor still available for use in new designs?

Am 14.06.24 um 18:03 schrieb john larkin:

I did design a 1:1000 dual slope timer stretcher with them featuring

With a space proof ceramic flat pack with even worse parasitics. And yes, Methinks it was Harris before, bought up by Renesas.

cheers, Gerhard

Just pull up some parts from the standard library, some low current NPNs like BC547 maybe. Look for low capacitances if you care about that; the HFAs are good in that respect.

Why do people avoid making clean 4-wire connections on schematics?

There are zillions of single transistors around. I like BCX70 as an npn gumdrop, although I use mosfets more often for simple things.

Yes but BC547 sounds almost as ancient as BC107 Are they still in production? What would work well at a GHz or two or more?

Some places have rules about that. I had trouble once, even though I put a very obvious and big blob where the join should have been. When doing schematics of my own I prefer to avoid a one-blob 4-wire join. Use two blobs.

It was just an example of a library part, but Digikey will sell you a million pieces.

For switching? Tuned RF?

On 14/06/2024 17:03, john larkin wrote:

Yeah, there is something very wrong with that transistor model.

Here is an emitter coupled multivib that can run fast. Should be possible somehow to gate it into being a monostable - worst case wrap a latch around it so it runs just one pulse?

Version 4 SHEET 1 2848 1304 WIRE -512 528 -752 528 WIRE -336 528 -512 528 WIRE -336 576 -336 528 WIRE -752 624 -752 528 WIRE -512 672 -512 528 WIRE -336 688 -336 656 WIRE -224 688 -336 688 WIRE -336 720 -336 688 WIRE -336 720 -448 720 WIRE -336 752 -336 720 WIRE -752 800 -752 704 WIRE -656 800 -752 800 WIRE -224 800 -272 800 WIRE -656 832 -656 800 WIRE -224 848 -224 800 WIRE -752 880 -752 800 WIRE -512 880 -512 768 WIRE -448 880 -512 880 WIRE -336 880 -336 848 WIRE -336 880 -384 880 WIRE -512 944 -512 880 WIRE -336 944 -336 880 WIRE -752 1072 -752 960 WIRE -512 1072 -512 1024 WIRE -512 1072 -752 1072 WIRE -336 1072 -336 1024 WIRE -336 1072 -512 1072 FLAG -224 848 0 FLAG -656 832 0 FLAG -224 688 out SYMBOL VOLTAGE -752 608 R0 SYMATTR InstName V1 SYMATTR Value 5 SYMBOL npn -448 672 M0 SYMATTR InstName Q1 SYMATTR Value 2SC3838K SYMBOL npn -272 752 M0 SYMATTR InstName Q2 SYMATTR Value 2SC3838K SYMBOL VOLTAGE -752 864 R0 SYMATTR InstName V2 SYMATTR Value 5 SYMBOL res -352 560 R0 SYMATTR InstName R2 SYMATTR Value 390 SYMBOL cap -384 864 R90 WINDOW 0 0 32 VBottom 2 WINDOW 3 32 32 VTop 2 SYMATTR InstName C1 SYMATTR Value 5p SYMBOL res -528 928 R0 SYMATTR InstName R3 SYMATTR Value 1k SYMBOL res -352 928 R0 SYMATTR InstName R4 SYMATTR Value 1k TEXT -816 1112 Left 0 !.tran 100u TEXT -120 976 Left 2 ;EPW SED JUN 2024 TEXT -264 1064 Left 2 ;FAST ASTABLE MULTIVIBRATOR TEXT -120 1032 Left 2 ;EMITTER COUPLED TEXT -144 848 Left 1 ;TRADITIONALLY SHOWN WITH TEXT -176 872 Left 1 ;CURRENT SOURCES FOR EMITTERS TEXT -160 896 Left 1 ;TRY CHANGING R3 R4 FOR 5mA ? TEXT -152 552 Left 1 ;ALTERNATIVE OUTPUT CAN BE TEXT -192 576 Left 1 ;TAKEN ACROSS R IN Q1 COLLECTOR

piglet

I meant a specific npn part and maybe also pnp which is not going obsolete.

Now we're getting somewhere. Thanks piglet. 2SC3838K typical ft 3.2 GHz and available. However it's not recommended for new designs if you believe the rohm site. So I wonder if it has a successor which is available.

NSVF6001SB6T1G looks interesting but no sign of a model other than symbol, footprint and 3D that I can find.

You can search Mouser or Digikey for a multi-sourced part.

Sadly, fast PNPs are mostly going EOL, faster than NPNs. And all the good things, like GaN and SiC and phemts, are n-types. Nature must dislike P-things.

Designing with disctetes is educational, but nowadays ICs are the practical winners. Like wideband MMICs for 10 cents, opamps and switching regulators for 15.

Gotta move up the abstraction stack.

I like BFS17. It's fast, but not so fast as to get cranky. It generally behaves.

There's this new trend of breaking the circuit schematic up into little boxes like this, it's nuts:

But looks like it's going obsolete. 55GN01FA is on the list of alternatives.

That's criminal.

Digikey shows two active sources. My inventory program shows MMBTH10 as an alternate.

Mouser has a million BFS17s in stock.