1
\$\begingroup\$

I would like to create a circuit which creates nanosecond pulses for a trigger signal with a frequency from 20 to 80 MHz, at 3.3V. I have created the following cricuit as can be seen in the attatched image. Since this is a new design, would it work to get output singnals at the 50 ohm resistor of nanosecond width with each trigger signal (TTL 3.3V)?

  • Should I change the decoupling capacitors or do they seem to be correctly selected?
  • For the LT1021-5 should I change it to a lt1460-5?

I would appreciate any input, thanks a lot! created schematic

\$\endgroup\$
10
  • 2
    \$\begingroup\$ I'll only add this thought. Given \$50\:\Omega\$, and considering \$100\:\text{MHz}\$ (\$10\:\text{ns}\$ total cycle time), and considering that you should want your rise and fall times to be about \$\frac1{10}\$th of that in, say, perhaps \$2\tau\$ to \$3\tau\$.... how many \$\text{pF}\$ do you calculate you can drive? (Forget the circuit for now.) \$\endgroup\$
    – jonk
    Nov 22, 2020 at 1:00
  • \$\begingroup\$ hi @jonk, thanks for your reply. Hope I'm getting this right. If τ = R*C => C = 2τ/50Ω = 2(1/10 * 10ns)/50Ω = 40pF? \$\endgroup\$
    – ulix
    Nov 22, 2020 at 2:41
  • \$\begingroup\$ I get less. The rising edge should be ((10e-9 s)/(50 ohm)) * 0.1 (falling edge, likewise, so that you are left with 80% for the tops and bottoms) and then either divided by 2 or 3, depending on how many taus you want. I'd say 7-10 pF. \$\endgroup\$
    – jonk
    Nov 22, 2020 at 2:46
  • \$\begingroup\$ Q1 will charge your output load quickly. What will R5 do? \$\endgroup\$
    – Neil_UK
    Nov 22, 2020 at 5:48
  • \$\begingroup\$ @jonk yes you are right! For now I would change it to 10pF. \$\endgroup\$
    – ulix
    Nov 22, 2020 at 13:12

1 Answer 1

1
\$\begingroup\$

Taking a risk on misunderstanding your load, I'd suggest the following

schematic

simulate this circuit – Schematic created using CircuitLab

If your output is a laser diode, then your drive to it should be a current source. You can use slower, higher current transistors in the output than your suggested BRF92, which is too weedy for your output, and too lively to be used as an emitter follower. Connected like this, you will get most of the advertised ft out of the transistors. Choose R1 with the supply voltage to set your load current.

The use of PECL (that is, ECL driven from +5 V instead of -5 V) in the pulse-forming stage has a number of advantages. It's about the right speed for your use. The LTC6752 is a bit slow for your suggested output pulse lengths. ECL is fairly tame when driven with delays like the RC shown. If you don't like the idea of a slow edge into the next gate, then as your delay is only a few nS, you could replace the RC with two or three feet of transmission line (with appropriate termination) to get the delay, RG178 is nice and thin, or a synthetic delay line from a few Ls and Cs. PECL has the right output voltage to directly drive the output current switch. Each positive going input edge will produce a positive going pulse.

U1 is some complementary ECL output gate, perhaps a TTL to PECL converter/level shifter.

\$\endgroup\$
11
  • \$\begingroup\$ it is a great suggestion. Since I failed selecting a comperator, can you suggest TTL to PECL converter as well as a PECL? Perferable from Analog/LT since I wanted to do the simulation in LTSpice. For Q1/Q2 I would have choosen a MMBT5551? Does this seem reasonable? \$\endgroup\$
    – ulix
    Nov 23, 2020 at 22:41
  • \$\begingroup\$ mmbt5551 looks awful, they're dog slow, and their gain collapses beyond a few 10s of mA. Why a 5551, was it the first one you found in a list? \$\endgroup\$
    – Neil_UK
    Nov 24, 2020 at 5:43
  • \$\begingroup\$ fast pnps seem to be harder to find than when I was using them decades ago. I looked at Digikey, filtered on 'available in 1s', as well as DC current, voltage and speed, and 30A02CH and 2SAR502U3T106 are about the least worst. They may be fast enough for your application. See if you can find models for them, and simulate what you get, with reasonable estimates for parasitic capacitance of the diode and connections. \$\endgroup\$
    – Neil_UK
    Nov 24, 2020 at 6:11
  • \$\begingroup\$ I only MMBT5551 in mind since I think I have a package of them. I haven't found for the 30A02CH and 2SAR502U3T106 any Ltspice models. But first I wanted to create the digital part, and I'm not sure which parts to choose. For the level shifter I would have choosen a MAX9371 and for the AND gate a MC10EL05, does this make sense (wouldn't be able to simulate it with ltspice since there are no models for these)? \$\endgroup\$
    – ulix
    Nov 24, 2020 at 17:01
  • \$\begingroup\$ @ulix Yes, both of those parts look fine. Are you able to operate your laser diode from a fixed positive rail with a current pulled out of the -ve terminal. If so, an NPN output long-tailed pair gives you many many more transistors to choose from in the appropriate 1 or 2 GHz ft range. \$\endgroup\$
    – Neil_UK
    Nov 24, 2020 at 17:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.